summaryrefslogtreecommitdiff
path: root/gnu/usr.bin/perl/regexec.c
blob: c22954acdccbf8b314049a6d3fd0ec436a4b37aa (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
5910
5911
5912
5913
5914
5915
5916
5917
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
5939
5940
5941
5942
5943
5944
5945
5946
5947
5948
5949
5950
5951
5952
5953
5954
5955
5956
5957
5958
5959
5960
5961
5962
5963
5964
5965
5966
5967
5968
5969
5970
5971
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
6001
6002
6003
6004
6005
6006
6007
6008
6009
6010
6011
6012
6013
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
6037
6038
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
6063
6064
6065
6066
6067
6068
6069
6070
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
6086
6087
6088
6089
6090
6091
6092
6093
6094
6095
6096
6097
6098
6099
6100
6101
6102
6103
6104
6105
6106
6107
6108
6109
6110
6111
6112
6113
6114
6115
6116
6117
6118
6119
6120
6121
6122
6123
6124
6125
6126
6127
6128
6129
6130
6131
6132
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147
6148
6149
6150
6151
6152
6153
6154
6155
6156
6157
6158
6159
6160
6161
6162
6163
6164
6165
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
6181
6182
6183
6184
6185
6186
6187
6188
6189
6190
6191
6192
6193
6194
6195
6196
6197
6198
6199
6200
6201
6202
6203
6204
6205
6206
6207
6208
6209
6210
6211
6212
6213
6214
6215
6216
6217
6218
6219
6220
6221
6222
6223
6224
6225
6226
6227
6228
6229
6230
6231
6232
6233
6234
6235
6236
6237
6238
6239
6240
6241
6242
6243
6244
6245
6246
6247
6248
6249
6250
6251
6252
6253
6254
6255
6256
6257
6258
6259
6260
6261
6262
6263
6264
6265
6266
6267
6268
6269
6270
6271
6272
6273
6274
6275
6276
6277
6278
6279
6280
6281
6282
6283
6284
6285
6286
6287
6288
6289
6290
6291
6292
6293
6294
6295
6296
6297
6298
6299
6300
6301
6302
6303
6304
6305
6306
6307
6308
6309
6310
6311
6312
6313
6314
6315
6316
6317
6318
6319
6320
6321
6322
6323
6324
6325
6326
6327
6328
6329
6330
6331
6332
6333
6334
6335
6336
6337
6338
6339
6340
6341
6342
6343
6344
6345
6346
6347
6348
6349
6350
6351
6352
6353
6354
6355
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
6371
6372
6373
6374
6375
6376
6377
6378
6379
6380
6381
6382
6383
6384
6385
6386
6387
6388
6389
6390
6391
6392
6393
6394
6395
6396
6397
6398
6399
6400
6401
6402
6403
6404
6405
6406
6407
6408
6409
6410
6411
6412
6413
6414
6415
6416
6417
6418
6419
6420
6421
6422
6423
6424
6425
6426
6427
6428
6429
6430
6431
6432
6433
6434
6435
6436
6437
6438
6439
6440
6441
6442
6443
6444
6445
6446
6447
6448
6449
6450
6451
6452
6453
6454
6455
6456
6457
6458
6459
6460
6461
6462
6463
6464
6465
6466
6467
6468
6469
6470
6471
6472
6473
6474
6475
6476
6477
6478
6479
6480
6481
6482
6483
6484
6485
6486
6487
6488
6489
6490
6491
6492
6493
6494
6495
6496
6497
6498
6499
6500
6501
6502
6503
6504
6505
6506
6507
6508
6509
6510
6511
6512
6513
6514
6515
6516
6517
6518
6519
6520
6521
6522
6523
6524
6525
6526
6527
6528
6529
6530
6531
6532
6533
6534
6535
6536
6537
6538
6539
6540
6541
6542
6543
6544
6545
6546
6547
6548
6549
6550
6551
6552
6553
6554
6555
6556
6557
6558
6559
6560
6561
6562
6563
6564
6565
6566
6567
6568
6569
6570
6571
6572
6573
6574
6575
6576
6577
6578
6579
6580
6581
6582
6583
6584
6585
6586
6587
6588
6589
6590
6591
6592
6593
6594
6595
6596
6597
6598
6599
6600
6601
6602
6603
6604
6605
6606
6607
6608
6609
6610
6611
6612
6613
6614
6615
6616
6617
6618
6619
6620
6621
6622
6623
6624
6625
6626
6627
6628
6629
6630
6631
6632
6633
6634
6635
6636
6637
6638
6639
6640
6641
6642
6643
6644
6645
6646
6647
6648
6649
6650
6651
6652
6653
6654
6655
6656
6657
6658
6659
6660
6661
6662
6663
6664
6665
6666
6667
6668
6669
6670
6671
6672
6673
6674
6675
6676
6677
6678
6679
6680
6681
6682
6683
6684
6685
6686
6687
6688
6689
6690
6691
6692
6693
6694
6695
6696
6697
6698
6699
6700
6701
6702
6703
6704
6705
6706
6707
6708
6709
6710
6711
6712
6713
6714
6715
6716
6717
6718
6719
6720
6721
6722
6723
6724
6725
6726
6727
6728
6729
6730
6731
6732
6733
6734
6735
6736
6737
6738
6739
6740
6741
6742
6743
6744
6745
6746
6747
6748
6749
6750
6751
6752
6753
6754
6755
6756
6757
6758
6759
6760
6761
6762
6763
6764
6765
6766
6767
6768
6769
6770
6771
6772
6773
6774
6775
6776
6777
6778
6779
6780
6781
6782
6783
6784
6785
6786
6787
6788
6789
6790
6791
6792
6793
6794
6795
6796
6797
6798
6799
6800
6801
6802
6803
6804
6805
6806
6807
6808
6809
6810
6811
6812
6813
6814
6815
6816
6817
6818
6819
6820
6821
6822
6823
6824
6825
6826
6827
6828
6829
6830
6831
6832
6833
6834
6835
6836
6837
6838
6839
6840
6841
6842
6843
6844
6845
6846
6847
6848
6849
6850
6851
6852
6853
6854
6855
6856
6857
6858
6859
6860
6861
6862
6863
6864
6865
6866
6867
6868
6869
6870
6871
6872
6873
6874
6875
6876
6877
6878
6879
6880
6881
6882
6883
6884
6885
6886
6887
6888
6889
6890
6891
6892
6893
6894
6895
6896
6897
6898
6899
6900
6901
6902
6903
6904
6905
6906
6907
6908
6909
6910
6911
6912
6913
6914
6915
6916
6917
6918
6919
6920
6921
6922
6923
6924
6925
6926
6927
6928
6929
6930
6931
6932
6933
6934
6935
6936
6937
6938
6939
6940
6941
6942
6943
6944
6945
6946
6947
6948
6949
6950
6951
6952
6953
6954
6955
6956
6957
6958
6959
6960
6961
6962
6963
6964
6965
6966
6967
6968
6969
6970
6971
6972
6973
6974
6975
6976
6977
6978
6979
6980
6981
6982
6983
6984
6985
6986
6987
6988
6989
6990
6991
6992
6993
6994
6995
6996
6997
6998
6999
7000
7001
7002
7003
7004
7005
7006
7007
7008
7009
7010
7011
7012
7013
7014
7015
7016
7017
7018
7019
7020
7021
7022
7023
7024
7025
7026
7027
7028
7029
7030
7031
7032
7033
7034
7035
7036
7037
7038
7039
7040
7041
7042
7043
7044
7045
7046
7047
7048
7049
7050
7051
7052
7053
7054
7055
7056
7057
7058
7059
7060
7061
7062
7063
7064
7065
7066
7067
7068
7069
7070
7071
7072
7073
7074
7075
7076
7077
7078
7079
7080
7081
7082
7083
7084
7085
7086
7087
7088
7089
7090
7091
7092
7093
7094
7095
7096
7097
7098
7099
7100
7101
7102
7103
7104
7105
7106
7107
7108
7109
7110
7111
7112
7113
7114
7115
7116
7117
7118
7119
7120
7121
7122
7123
7124
7125
7126
7127
7128
7129
7130
7131
7132
7133
/*    regexec.c
 */

/*
 * 	One Ring to rule them all, One Ring to find them
 &
 *     [p.v of _The Lord of the Rings_, opening poem]
 *     [p.50 of _The Lord of the Rings_, I/iii: "The Shadow of the Past"]
 *     [p.254 of _The Lord of the Rings_, II/ii: "The Council of Elrond"]
 */

/* This file contains functions for executing a regular expression.  See
 * also regcomp.c which funnily enough, contains functions for compiling
 * a regular expression.
 *
 * This file is also copied at build time to ext/re/re_exec.c, where
 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
 * This causes the main functions to be compiled under new names and with
 * debugging support added, which makes "use re 'debug'" work.
 */

/* NOTE: this is derived from Henry Spencer's regexp code, and should not
 * confused with the original package (see point 3 below).  Thanks, Henry!
 */

/* Additional note: this code is very heavily munged from Henry's version
 * in places.  In some spots I've traded clarity for efficiency, so don't
 * blame Henry for some of the lack of readability.
 */

/* The names of the functions have been changed from regcomp and
 * regexec to  pregcomp and pregexec in order to avoid conflicts
 * with the POSIX routines of the same names.
*/

#ifdef PERL_EXT_RE_BUILD
#include "re_top.h"
#endif

/*
 * pregcomp and pregexec -- regsub and regerror are not used in perl
 *
 *	Copyright (c) 1986 by University of Toronto.
 *	Written by Henry Spencer.  Not derived from licensed software.
 *
 *	Permission is granted to anyone to use this software for any
 *	purpose on any computer system, and to redistribute it freely,
 *	subject to the following restrictions:
 *
 *	1. The author is not responsible for the consequences of use of
 *		this software, no matter how awful, even if they arise
 *		from defects in it.
 *
 *	2. The origin of this software must not be misrepresented, either
 *		by explicit claim or by omission.
 *
 *	3. Altered versions must be plainly marked as such, and must not
 *		be misrepresented as being the original software.
 *
 ****    Alterations to Henry's code are...
 ****
 ****    Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
 ****    2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
 ****    by Larry Wall and others
 ****
 ****    You may distribute under the terms of either the GNU General Public
 ****    License or the Artistic License, as specified in the README file.
 *
 * Beware that some of this code is subtly aware of the way operator
 * precedence is structured in regular expressions.  Serious changes in
 * regular-expression syntax might require a total rethink.
 */
#include "EXTERN.h"
#define PERL_IN_REGEXEC_C
#include "perl.h"

#ifdef PERL_IN_XSUB_RE
#  include "re_comp.h"
#else
#  include "regcomp.h"
#endif

#define RF_tainted	1	/* tainted information used? e.g. locale */
#define RF_warned	2		/* warned about big count? */

#define RF_utf8		8		/* Pattern contains multibyte chars? */

#define UTF_PATTERN ((PL_reg_flags & RF_utf8) != 0)

#define RS_init		1		/* eval environment created */
#define RS_set		2		/* replsv value is set */

#ifndef STATIC
#define	STATIC	static
#endif

/* Valid for non-utf8 strings, non-ANYOFV nodes only: avoids the reginclass
 * call if there are no complications: i.e., if everything matchable is
 * straight forward in the bitmap */
#define REGINCLASS(prog,p,c)  (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0)   \
					      : ANYOF_BITMAP_TEST(p,*(c)))

/*
 * Forwards.
 */

#define CHR_SVLEN(sv) (utf8_target ? sv_len_utf8(sv) : SvCUR(sv))
#define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b)

#define HOPc(pos,off) \
	(char *)(PL_reg_match_utf8 \
	    ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \
	    : (U8*)(pos + off))
#define HOPBACKc(pos, off) \
	(char*)(PL_reg_match_utf8\
	    ? reghopmaybe3((U8*)pos, -off, (U8*)PL_bostr) \
	    : (pos - off >= PL_bostr)		\
		? (U8*)pos - off		\
		: NULL)

#define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off))
#define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))

/* these are unrolled below in the CCC_TRY_XXX defined */
#ifdef EBCDIC
    /* Often 'str' is a hard-coded utf8 string instead of utfebcdic. so just
     * skip the check on EBCDIC platforms */
#   define LOAD_UTF8_CHARCLASS(class,str) LOAD_UTF8_CHARCLASS_NO_CHECK(class)
#else
#   define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \
    if (!CAT2(PL_utf8_,class)) { \
	bool ok; \
	ENTER; save_re_context(); \
	ok=CAT2(is_utf8_,class)((const U8*)str); \
	assert(ok); assert(CAT2(PL_utf8_,class)); LEAVE; } } STMT_END
#endif

/* Doesn't do an assert to verify that is correct */
#define LOAD_UTF8_CHARCLASS_NO_CHECK(class) STMT_START { \
    if (!CAT2(PL_utf8_,class)) { \
	bool throw_away PERL_UNUSED_DECL; \
	ENTER; save_re_context(); \
	throw_away = CAT2(is_utf8_,class)((const U8*)" "); \
	LEAVE; } } STMT_END

#define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a")
#define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0")
#define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ")

#define LOAD_UTF8_CHARCLASS_GCB()  /* Grapheme cluster boundaries */        \
	LOAD_UTF8_CHARCLASS(X_begin, " ");                                  \
	LOAD_UTF8_CHARCLASS(X_non_hangul, "A");                             \
	/* These are utf8 constants, and not utf-ebcdic constants, so the   \
	    * assert should likely and hopefully fail on an EBCDIC machine */ \
	LOAD_UTF8_CHARCLASS(X_extend, "\xcc\x80"); /* U+0300 */             \
									    \
	/* No asserts are done for these, in case called on an early        \
	    * Unicode version in which they map to nothing */               \
	LOAD_UTF8_CHARCLASS_NO_CHECK(X_prepend);/* U+0E40 "\xe0\xb9\x80" */ \
	LOAD_UTF8_CHARCLASS_NO_CHECK(X_L);	    /* U+1100 "\xe1\x84\x80" */ \
	LOAD_UTF8_CHARCLASS_NO_CHECK(X_LV);     /* U+AC00 "\xea\xb0\x80" */ \
	LOAD_UTF8_CHARCLASS_NO_CHECK(X_LVT);    /* U+AC01 "\xea\xb0\x81" */ \
	LOAD_UTF8_CHARCLASS_NO_CHECK(X_LV_LVT_V);/* U+AC01 "\xea\xb0\x81" */\
	LOAD_UTF8_CHARCLASS_NO_CHECK(X_T);      /* U+11A8 "\xe1\x86\xa8" */ \
	LOAD_UTF8_CHARCLASS_NO_CHECK(X_V)       /* U+1160 "\xe1\x85\xa0" */  

#define PLACEHOLDER	/* Something for the preprocessor to grab onto */

/* The actual code for CCC_TRY, which uses several variables from the routine
 * it's callable from.  It is designed to be the bulk of a case statement.
 * FUNC is the macro or function to call on non-utf8 targets that indicate if
 *      nextchr matches the class.
 * UTF8_TEST is the whole test string to use for utf8 targets
 * LOAD is what to use to test, and if not present to load in the swash for the
 *	class
 * POS_OR_NEG is either empty or ! to complement the results of FUNC or
 *	UTF8_TEST test.
 * The logic is: Fail if we're at the end-of-string; otherwise if the target is
 * utf8 and a variant, load the swash if necessary and test using the utf8
 * test.  Advance to the next character if test is ok, otherwise fail; If not
 * utf8 or an invariant under utf8, use the non-utf8 test, and fail if it
 * fails, or advance to the next character */

#define _CCC_TRY_CODE(POS_OR_NEG, FUNC, UTF8_TEST, CLASS, STR)                \
    if (locinput >= PL_regeol) {                                              \
	sayNO;                                                                \
    }                                                                         \
    if (utf8_target && UTF8_IS_CONTINUED(nextchr)) {                          \
	LOAD_UTF8_CHARCLASS(CLASS, STR);                                      \
	if (POS_OR_NEG (UTF8_TEST)) {                                         \
	    sayNO;                                                            \
	}                                                                     \
	locinput += PL_utf8skip[nextchr];                                     \
	nextchr = UCHARAT(locinput);                                          \
	break;                                                                \
    }                                                                         \
    if (POS_OR_NEG (FUNC(nextchr))) {                                         \
	sayNO;                                                                \
    }                                                                         \
    nextchr = UCHARAT(++locinput);                                            \
    break;

/* Handle the non-locale cases for a character class and its complement.  It
 * calls _CCC_TRY_CODE with a ! to complement the test for the character class.
 * This is because that code fails when the test succeeds, so we want to have
 * the test fail so that the code succeeds.  The swash is stored in a
 * predictable PL_ place */
#define _CCC_TRY_NONLOCALE(NAME,  NNAME,  FUNC,                               \
	                   CLASS, STR)                                        \
    case NAME:                                                                \
	_CCC_TRY_CODE( !, FUNC,                                               \
		          cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS),             \
			                    (U8*)locinput, TRUE)),            \
		          CLASS, STR)                                         \
    case NNAME:                                                               \
	_CCC_TRY_CODE(  PLACEHOLDER , FUNC,                                   \
		          cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS),             \
			                    (U8*)locinput, TRUE)),            \
		          CLASS, STR)                                         \

/* Generate the case statements for both locale and non-locale character
 * classes in regmatch for classes that don't have special unicode semantics.
 * Locales don't use an immediate swash, but an intermediary special locale
 * function that is called on the pointer to the current place in the input
 * string.  That function will resolve to needing the same swash.  One might
 * think that because we don't know what the locale will match, we shouldn't
 * check with the swash loading function that it loaded properly; ie, that we
 * should use LOAD_UTF8_CHARCLASS_NO_CHECK for those, but what is passed to the
 * regular LOAD_UTF8_CHARCLASS is in non-locale terms, and so locale is
 * irrelevant here */
#define CCC_TRY(NAME,  NNAME,  FUNC,                                          \
	        NAMEL, NNAMEL, LCFUNC, LCFUNC_utf8,                           \
	        NAMEA, NNAMEA, FUNCA,                                         \
		CLASS, STR)                                                   \
    case NAMEL:                                                               \
	PL_reg_flags |= RF_tainted;                                           \
	_CCC_TRY_CODE( !, LCFUNC, LCFUNC_utf8((U8*)locinput), CLASS, STR)     \
    case NNAMEL:                                                              \
	PL_reg_flags |= RF_tainted;                                           \
	_CCC_TRY_CODE( PLACEHOLDER, LCFUNC, LCFUNC_utf8((U8*)locinput),       \
		       CLASS, STR)                                            \
    case NAMEA:                                                               \
	if (locinput >= PL_regeol || ! FUNCA(nextchr)) {                      \
	    sayNO;                                                            \
	}                                                                     \
	/* Matched a utf8-invariant, so don't have to worry about utf8 */     \
	nextchr = UCHARAT(++locinput);                                        \
	break;                                                                \
    case NNAMEA:                                                              \
	if (locinput >= PL_regeol || FUNCA(nextchr)) {                        \
	    sayNO;                                                            \
	}                                                                     \
	if (utf8_target) {                                                    \
	    locinput += PL_utf8skip[nextchr];                                 \
	    nextchr = UCHARAT(locinput);                                      \
	}                                                                     \
	else {                                                                \
	    nextchr = UCHARAT(++locinput);                                    \
	}                                                                     \
	break;                                                                \
    /* Generate the non-locale cases */                                       \
    _CCC_TRY_NONLOCALE(NAME, NNAME, FUNC, CLASS, STR)

/* This is like CCC_TRY, but has an extra set of parameters for generating case
 * statements to handle separate Unicode semantics nodes */
#define CCC_TRY_U(NAME,  NNAME,  FUNC,                                         \
		  NAMEL, NNAMEL, LCFUNC, LCFUNC_utf8,                          \
	          NAMEU, NNAMEU, FUNCU,                                        \
	          NAMEA, NNAMEA, FUNCA,                                        \
	          CLASS, STR)                                                  \
    CCC_TRY(NAME, NNAME, FUNC,                                                 \
	    NAMEL, NNAMEL, LCFUNC, LCFUNC_utf8,                                \
	    NAMEA, NNAMEA, FUNCA,                                              \
	    CLASS, STR)                                                        \
    _CCC_TRY_NONLOCALE(NAMEU, NNAMEU, FUNCU, CLASS, STR)

/* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */

/* for use after a quantifier and before an EXACT-like node -- japhy */
/* it would be nice to rework regcomp.sym to generate this stuff. sigh
 *
 * NOTE that *nothing* that affects backtracking should be in here, specifically
 * VERBS must NOT be included. JUMPABLE is used to determine  if we can ignore a
 * node that is in between two EXACT like nodes when ascertaining what the required
 * "follow" character is. This should probably be moved to regex compile time
 * although it may be done at run time beause of the REF possibility - more
 * investigation required. -- demerphq
*/
#define JUMPABLE(rn) (      \
    OP(rn) == OPEN ||       \
    (OP(rn) == CLOSE && (!cur_eval || cur_eval->u.eval.close_paren != ARG(rn))) || \
    OP(rn) == EVAL ||   \
    OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
    OP(rn) == PLUS || OP(rn) == MINMOD || \
    OP(rn) == KEEPS || \
    (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \
)
#define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT)

#define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF )

#if 0 
/* Currently these are only used when PL_regkind[OP(rn)] == EXACT so
   we don't need this definition. */
#define IS_TEXT(rn)   ( OP(rn)==EXACT   || OP(rn)==REF   || OP(rn)==NREF   )
#define IS_TEXTF(rn)  ( OP(rn)==EXACTFU || OP(rn)==EXACTFU_SS || OP(rn)==EXACTFU_TRICKYFOLD || OP(rn)==EXACTFA || OP(rn)==EXACTF || OP(rn)==REFF  || OP(rn)==NREFF )
#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL )

#else
/* ... so we use this as its faster. */
#define IS_TEXT(rn)   ( OP(rn)==EXACT   )
#define IS_TEXTFU(rn)  ( OP(rn)==EXACTFU || OP(rn)==EXACTFU_SS || OP(rn)==EXACTFU_TRICKYFOLD || OP(rn) == EXACTFA)
#define IS_TEXTF(rn)  ( OP(rn)==EXACTF  )
#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL )

#endif

/*
  Search for mandatory following text node; for lookahead, the text must
  follow but for lookbehind (rn->flags != 0) we skip to the next step.
*/
#define FIND_NEXT_IMPT(rn) STMT_START { \
    while (JUMPABLE(rn)) { \
	const OPCODE type = OP(rn); \
	if (type == SUSPEND || PL_regkind[type] == CURLY) \
	    rn = NEXTOPER(NEXTOPER(rn)); \
	else if (type == PLUS) \
	    rn = NEXTOPER(rn); \
	else if (type == IFMATCH) \
	    rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
	else rn += NEXT_OFF(rn); \
    } \
} STMT_END 


static void restore_pos(pTHX_ void *arg);

#define REGCP_PAREN_ELEMS 4
#define REGCP_OTHER_ELEMS 5
#define REGCP_FRAME_ELEMS 1
/* REGCP_FRAME_ELEMS are not part of the REGCP_OTHER_ELEMS and
 * are needed for the regexp context stack bookkeeping. */

STATIC CHECKPOINT
S_regcppush(pTHX_ I32 parenfloor)
{
    dVAR;
    const int retval = PL_savestack_ix;
    const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
    const UV total_elems = paren_elems_to_push + REGCP_OTHER_ELEMS;
    const UV elems_shifted = total_elems << SAVE_TIGHT_SHIFT;
    int p;
    GET_RE_DEBUG_FLAGS_DECL;

    if (paren_elems_to_push < 0)
	Perl_croak(aTHX_ "panic: paren_elems_to_push, %i < 0",
		   paren_elems_to_push);

    if ((elems_shifted >> SAVE_TIGHT_SHIFT) != total_elems)
	Perl_croak(aTHX_ "panic: paren_elems_to_push offset %"UVuf
		   " out of range (%lu-%ld)",
		   total_elems, (unsigned long)PL_regsize, (long)parenfloor);

    SSGROW(total_elems + REGCP_FRAME_ELEMS);
    
    for (p = PL_regsize; p > parenfloor; p--) {
/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
	SSPUSHINT(PL_regoffs[p].end);
	SSPUSHINT(PL_regoffs[p].start);
	SSPUSHPTR(PL_reg_start_tmp[p]);
	SSPUSHINT(p);
	DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
	  "     saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n",
		      (UV)p, (IV)PL_regoffs[p].start,
		      (IV)(PL_reg_start_tmp[p] - PL_bostr),
		      (IV)PL_regoffs[p].end
	));
    }
/* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
    SSPUSHPTR(PL_regoffs);
    SSPUSHINT(PL_regsize);
    SSPUSHINT(*PL_reglastparen);
    SSPUSHINT(*PL_reglastcloseparen);
    SSPUSHPTR(PL_reginput);
    SSPUSHUV(SAVEt_REGCONTEXT | elems_shifted); /* Magic cookie. */

    return retval;
}

/* These are needed since we do not localize EVAL nodes: */
#define REGCP_SET(cp)                                           \
    DEBUG_STATE_r(                                              \
            PerlIO_printf(Perl_debug_log,		        \
	        "  Setting an EVAL scope, savestack=%"IVdf"\n",	\
	        (IV)PL_savestack_ix));                          \
    cp = PL_savestack_ix

#define REGCP_UNWIND(cp)                                        \
    DEBUG_STATE_r(                                              \
        if (cp != PL_savestack_ix) 		                \
    	    PerlIO_printf(Perl_debug_log,		        \
		"  Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
	        (IV)(cp), (IV)PL_savestack_ix));                \
    regcpblow(cp)

STATIC char *
S_regcppop(pTHX_ const regexp *rex)
{
    dVAR;
    UV i;
    char *input;
    GET_RE_DEBUG_FLAGS_DECL;

    PERL_ARGS_ASSERT_REGCPPOP;

    /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
    i = SSPOPUV;
    assert((i & SAVE_MASK) == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
    i >>= SAVE_TIGHT_SHIFT; /* Parentheses elements to pop. */
    input = (char *) SSPOPPTR;
    *PL_reglastcloseparen = SSPOPINT;
    *PL_reglastparen = SSPOPINT;
    PL_regsize = SSPOPINT;
    PL_regoffs=(regexp_paren_pair *) SSPOPPTR;

    i -= REGCP_OTHER_ELEMS;
    /* Now restore the parentheses context. */
    for ( ; i > 0; i -= REGCP_PAREN_ELEMS) {
	I32 tmps;
	U32 paren = (U32)SSPOPINT;
	PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
	PL_regoffs[paren].start = SSPOPINT;
	tmps = SSPOPINT;
	if (paren <= *PL_reglastparen)
	    PL_regoffs[paren].end = tmps;
	DEBUG_BUFFERS_r(
	    PerlIO_printf(Perl_debug_log,
			  "     restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
			  (UV)paren, (IV)PL_regoffs[paren].start,
			  (IV)(PL_reg_start_tmp[paren] - PL_bostr),
			  (IV)PL_regoffs[paren].end,
			  (paren > *PL_reglastparen ? "(no)" : ""));
	);
    }
    DEBUG_BUFFERS_r(
	if (*PL_reglastparen + 1 <= rex->nparens) {
	    PerlIO_printf(Perl_debug_log,
			  "     restoring \\%"IVdf"..\\%"IVdf" to undef\n",
			  (IV)(*PL_reglastparen + 1), (IV)rex->nparens);
	}
    );
#if 1
    /* It would seem that the similar code in regtry()
     * already takes care of this, and in fact it is in
     * a better location to since this code can #if 0-ed out
     * but the code in regtry() is needed or otherwise tests
     * requiring null fields (pat.t#187 and split.t#{13,14}
     * (as of patchlevel 7877)  will fail.  Then again,
     * this code seems to be necessary or otherwise
     * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
     * --jhi updated by dapm */
    for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) {
	if (i > PL_regsize)
	    PL_regoffs[i].start = -1;
	PL_regoffs[i].end = -1;
    }
#endif
    return input;
}

#define regcpblow(cp) LEAVE_SCOPE(cp)	/* Ignores regcppush()ed data. */

/*
 * pregexec and friends
 */

#ifndef PERL_IN_XSUB_RE
/*
 - pregexec - match a regexp against a string
 */
I32
Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, register char *strend,
	 char *strbeg, I32 minend, SV *screamer, U32 nosave)
/* strend: pointer to null at end of string */
/* strbeg: real beginning of string */
/* minend: end of match must be >=minend after stringarg. */
/* nosave: For optimizations. */
{
    PERL_ARGS_ASSERT_PREGEXEC;

    return
	regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
		      nosave ? 0 : REXEC_COPY_STR);
}
#endif

/*
 * Need to implement the following flags for reg_anch:
 *
 * USE_INTUIT_NOML		- Useful to call re_intuit_start() first
 * USE_INTUIT_ML
 * INTUIT_AUTORITATIVE_NOML	- Can trust a positive answer
 * INTUIT_AUTORITATIVE_ML
 * INTUIT_ONCE_NOML		- Intuit can match in one location only.
 * INTUIT_ONCE_ML
 *
 * Another flag for this function: SECOND_TIME (so that float substrs
 * with giant delta may be not rechecked).
 */

/* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */

/* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend.
   Otherwise, only SvCUR(sv) is used to get strbeg. */

/* XXXX We assume that strpos is strbeg unless sv. */

/* XXXX Some places assume that there is a fixed substring.
	An update may be needed if optimizer marks as "INTUITable"
	RExen without fixed substrings.  Similarly, it is assumed that
	lengths of all the strings are no more than minlen, thus they
	cannot come from lookahead.
	(Or minlen should take into account lookahead.) 
  NOTE: Some of this comment is not correct. minlen does now take account
  of lookahead/behind. Further research is required. -- demerphq

*/

/* A failure to find a constant substring means that there is no need to make
   an expensive call to REx engine, thus we celebrate a failure.  Similarly,
   finding a substring too deep into the string means that less calls to
   regtry() should be needed.

   REx compiler's optimizer found 4 possible hints:
	a) Anchored substring;
	b) Fixed substring;
	c) Whether we are anchored (beginning-of-line or \G);
	d) First node (of those at offset 0) which may distinguish positions;
   We use a)b)d) and multiline-part of c), and try to find a position in the
   string which does not contradict any of them.
 */

/* Most of decisions we do here should have been done at compile time.
   The nodes of the REx which we used for the search should have been
   deleted from the finite automaton. */

char *
Perl_re_intuit_start(pTHX_ REGEXP * const rx, SV *sv, char *strpos,
		     char *strend, const U32 flags, re_scream_pos_data *data)
{
    dVAR;
    struct regexp *const prog = (struct regexp *)SvANY(rx);
    register I32 start_shift = 0;
    /* Should be nonnegative! */
    register I32 end_shift   = 0;
    register char *s;
    register SV *check;
    char *strbeg;
    char *t;
    const bool utf8_target = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
    I32 ml_anch;
    register char *other_last = NULL;	/* other substr checked before this */
    char *check_at = NULL;		/* check substr found at this pos */
    const I32 multiline = prog->extflags & RXf_PMf_MULTILINE;
    RXi_GET_DECL(prog,progi);
#ifdef DEBUGGING
    const char * const i_strpos = strpos;
#endif
    GET_RE_DEBUG_FLAGS_DECL;

    PERL_ARGS_ASSERT_RE_INTUIT_START;

    RX_MATCH_UTF8_set(rx,utf8_target);

    if (RX_UTF8(rx)) {
	PL_reg_flags |= RF_utf8;
    }
    DEBUG_EXECUTE_r( 
        debug_start_match(rx, utf8_target, strpos, strend,
            sv ? "Guessing start of match in sv for"
               : "Guessing start of match in string for");
	      );

    /* CHR_DIST() would be more correct here but it makes things slow. */
    if (prog->minlen > strend - strpos) {
	DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
			      "String too short... [re_intuit_start]\n"));
	goto fail;
    }
                
    strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
    PL_regeol = strend;
    if (utf8_target) {
	if (!prog->check_utf8 && prog->check_substr)
	    to_utf8_substr(prog);
	check = prog->check_utf8;
    } else {
	if (!prog->check_substr && prog->check_utf8)
	    to_byte_substr(prog);
	check = prog->check_substr;
    }
    if (check == &PL_sv_undef) {
	DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
		"Non-utf8 string cannot match utf8 check string\n"));
	goto fail;
    }
    if (prog->extflags & RXf_ANCH) {	/* Match at beg-of-str or after \n */
	ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE)
		     || ( (prog->extflags & RXf_ANCH_BOL)
			  && !multiline ) );	/* Check after \n? */

	if (!ml_anch) {
	  if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */
		&& !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */
	       /* SvCUR is not set on references: SvRV and SvPVX_const overlap */
	       && sv && !SvROK(sv)
	       && (strpos != strbeg)) {
	      DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
	      goto fail;
	  }
	  if (prog->check_offset_min == prog->check_offset_max &&
	      !(prog->extflags & RXf_CANY_SEEN)) {
	    /* Substring at constant offset from beg-of-str... */
	    I32 slen;

	    s = HOP3c(strpos, prog->check_offset_min, strend);
	    
	    if (SvTAIL(check)) {
		slen = SvCUR(check);	/* >= 1 */

		if ( strend - s > slen || strend - s < slen - 1
		     || (strend - s == slen && strend[-1] != '\n')) {
		    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
		    goto fail_finish;
		}
		/* Now should match s[0..slen-2] */
		slen--;
		if (slen && (*SvPVX_const(check) != *s
			     || (slen > 1
				 && memNE(SvPVX_const(check), s, slen)))) {
		  report_neq:
		    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
		    goto fail_finish;
		}
	    }
	    else if (*SvPVX_const(check) != *s
		     || ((slen = SvCUR(check)) > 1
			 && memNE(SvPVX_const(check), s, slen)))
		goto report_neq;
	    check_at = s;
	    goto success_at_start;
	  }
	}
	/* Match is anchored, but substr is not anchored wrt beg-of-str. */
	s = strpos;
	start_shift = prog->check_offset_min; /* okay to underestimate on CC */
	end_shift = prog->check_end_shift;
	
	if (!ml_anch) {
	    const I32 end = prog->check_offset_max + CHR_SVLEN(check)
					 - (SvTAIL(check) != 0);
	    const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;

	    if (end_shift < eshift)
		end_shift = eshift;
	}
    }
    else {				/* Can match at random position */
	ml_anch = 0;
	s = strpos;
	start_shift = prog->check_offset_min;  /* okay to underestimate on CC */
	end_shift = prog->check_end_shift;
	
	/* end shift should be non negative here */
    }

#ifdef QDEBUGGING	/* 7/99: reports of failure (with the older version) */
    if (end_shift < 0)
	Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ",
		   (IV)end_shift, RX_PRECOMP(prog));
#endif

  restart:
    /* Find a possible match in the region s..strend by looking for
       the "check" substring in the region corrected by start/end_shift. */
    
    {
        I32 srch_start_shift = start_shift;
        I32 srch_end_shift = end_shift;
        if (srch_start_shift < 0 && strbeg - s > srch_start_shift) {
	    srch_end_shift -= ((strbeg - s) - srch_start_shift); 
	    srch_start_shift = strbeg - s;
	}
    DEBUG_OPTIMISE_MORE_r({
        PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n",
            (IV)prog->check_offset_min,
            (IV)srch_start_shift,
            (IV)srch_end_shift, 
            (IV)prog->check_end_shift);
    });       
        
    if ((flags & REXEC_SCREAM) && SvSCREAM(sv)) {
	I32 p = -1;			/* Internal iterator of scream. */
	I32 * const pp = data ? data->scream_pos : &p;
	const MAGIC *mg;
	bool found = FALSE;

	assert(SvMAGICAL(sv));
	mg = mg_find(sv, PERL_MAGIC_study);
	assert(mg);

	if (mg->mg_private == 1) {
	    found = ((U8 *)mg->mg_ptr)[BmRARE(check)] != (U8)~0;
	} else if (mg->mg_private == 2) {
	    found = ((U16 *)mg->mg_ptr)[BmRARE(check)] != (U16)~0;
	} else {
	    assert (mg->mg_private == 4);
	    found = ((U32 *)mg->mg_ptr)[BmRARE(check)] != (U32)~0;
	}

	if (found
	    || ( BmRARE(check) == '\n'
		 && (BmPREVIOUS(check) == SvCUR(check) - 1)
		 && SvTAIL(check) ))
	    s = screaminstr(sv, check,
			    srch_start_shift + (s - strbeg), srch_end_shift, pp, 0);
	else
	    goto fail_finish;
	/* we may be pointing at the wrong string */
	if (s && RXp_MATCH_COPIED(prog))
	    s = strbeg + (s - SvPVX_const(sv));
	if (data)
	    *data->scream_olds = s;
    }
    else {
        U8* start_point;
        U8* end_point;
        if (prog->extflags & RXf_CANY_SEEN) {
            start_point= (U8*)(s + srch_start_shift);
            end_point= (U8*)(strend - srch_end_shift);
        } else {
	    start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend);
            end_point= HOP3(strend, -srch_end_shift, strbeg);
	}
	DEBUG_OPTIMISE_MORE_r({
            PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n", 
                (int)(end_point - start_point),
                (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point), 
                start_point);
        });

	s = fbm_instr( start_point, end_point,
		      check, multiline ? FBMrf_MULTILINE : 0);
    }
    }
    /* Update the count-of-usability, remove useless subpatterns,
	unshift s.  */

    DEBUG_EXECUTE_r({
        RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
            SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
        PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s",
			  (s ? "Found" : "Did not find"),
	    (check == (utf8_target ? prog->anchored_utf8 : prog->anchored_substr)
	        ? "anchored" : "floating"),
	    quoted,
	    RE_SV_TAIL(check),
	    (s ? " at offset " : "...\n") ); 
    });

    if (!s)
	goto fail_finish;
    /* Finish the diagnostic message */
    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );

    /* XXX dmq: first branch is for positive lookbehind...
       Our check string is offset from the beginning of the pattern.
       So we need to do any stclass tests offset forward from that 
       point. I think. :-(
     */
    
        
    
    check_at=s;
     

    /* Got a candidate.  Check MBOL anchoring, and the *other* substr.
       Start with the other substr.
       XXXX no SCREAM optimization yet - and a very coarse implementation
       XXXX /ttx+/ results in anchored="ttx", floating="x".  floating will
		*always* match.  Probably should be marked during compile...
       Probably it is right to do no SCREAM here...
     */

    if (utf8_target ? (prog->float_utf8 && prog->anchored_utf8)
                : (prog->float_substr && prog->anchored_substr)) 
    {
	/* Take into account the "other" substring. */
	/* XXXX May be hopelessly wrong for UTF... */
	if (!other_last)
	    other_last = strpos;
	if (check == (utf8_target ? prog->float_utf8 : prog->float_substr)) {
	  do_other_anchored:
	    {
		char * const last = HOP3c(s, -start_shift, strbeg);
		char *last1, *last2;
		char * const saved_s = s;
		SV* must;

		t = s - prog->check_offset_max;
		if (s - strpos > prog->check_offset_max  /* signed-corrected t > strpos */
		    && (!utf8_target
			|| ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos))
			    && t > strpos)))
		    NOOP;
		else
		    t = strpos;
		t = HOP3c(t, prog->anchored_offset, strend);
		if (t < other_last)	/* These positions already checked */
		    t = other_last;
		last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
		if (last < last1)
		    last1 = last;
                /* XXXX It is not documented what units *_offsets are in.  
                   We assume bytes, but this is clearly wrong. 
                   Meaning this code needs to be carefully reviewed for errors.
                   dmq.
                  */
 
		/* On end-of-str: see comment below. */
		must = utf8_target ? prog->anchored_utf8 : prog->anchored_substr;
		if (must == &PL_sv_undef) {
		    s = (char*)NULL;
		    DEBUG_r(must = prog->anchored_utf8);	/* for debug */
		}
		else
		    s = fbm_instr(
			(unsigned char*)t,
			HOP3(HOP3(last1, prog->anchored_offset, strend)
				+ SvCUR(must), -(SvTAIL(must)!=0), strbeg),
			must,
			multiline ? FBMrf_MULTILINE : 0
		    );
                DEBUG_EXECUTE_r({
                    RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
                        SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
                    PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s",
			(s ? "Found" : "Contradicts"),
                        quoted, RE_SV_TAIL(must));
                });		    
		
			    
		if (!s) {
		    if (last1 >= last2) {
			DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
						", giving up...\n"));
			goto fail_finish;
		    }
		    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
			", trying floating at offset %ld...\n",
			(long)(HOP3c(saved_s, 1, strend) - i_strpos)));
		    other_last = HOP3c(last1, prog->anchored_offset+1, strend);
		    s = HOP3c(last, 1, strend);
		    goto restart;
		}
		else {
		    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
			  (long)(s - i_strpos)));
		    t = HOP3c(s, -prog->anchored_offset, strbeg);
		    other_last = HOP3c(s, 1, strend);
		    s = saved_s;
		    if (t == strpos)
			goto try_at_start;
		    goto try_at_offset;
		}
	    }
	}
	else {		/* Take into account the floating substring. */
	    char *last, *last1;
	    char * const saved_s = s;
	    SV* must;

	    t = HOP3c(s, -start_shift, strbeg);
	    last1 = last =
		HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
	    if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
		last = HOP3c(t, prog->float_max_offset, strend);
	    s = HOP3c(t, prog->float_min_offset, strend);
	    if (s < other_last)
		s = other_last;
 /* XXXX It is not documented what units *_offsets are in.  Assume bytes.  */
	    must = utf8_target ? prog->float_utf8 : prog->float_substr;
	    /* fbm_instr() takes into account exact value of end-of-str
	       if the check is SvTAIL(ed).  Since false positives are OK,
	       and end-of-str is not later than strend we are OK. */
	    if (must == &PL_sv_undef) {
		s = (char*)NULL;
		DEBUG_r(must = prog->float_utf8);	/* for debug message */
	    }
	    else
		s = fbm_instr((unsigned char*)s,
			      (unsigned char*)last + SvCUR(must)
				  - (SvTAIL(must)!=0),
			      must, multiline ? FBMrf_MULTILINE : 0);
	    DEBUG_EXECUTE_r({
	        RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
	            SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
	        PerlIO_printf(Perl_debug_log, "%s floating substr %s%s",
		    (s ? "Found" : "Contradicts"),
		    quoted, RE_SV_TAIL(must));
            });
	    if (!s) {
		if (last1 == last) {
		    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
					    ", giving up...\n"));
		    goto fail_finish;
		}
		DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
		    ", trying anchored starting at offset %ld...\n",
		    (long)(saved_s + 1 - i_strpos)));
		other_last = last;
		s = HOP3c(t, 1, strend);
		goto restart;
	    }
	    else {
		DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
		      (long)(s - i_strpos)));
		other_last = s; /* Fix this later. --Hugo */
		s = saved_s;
		if (t == strpos)
		    goto try_at_start;
		goto try_at_offset;
	    }
	}
    }

    
    t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos);
        
    DEBUG_OPTIMISE_MORE_r(
        PerlIO_printf(Perl_debug_log, 
            "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n",
            (IV)prog->check_offset_min,
            (IV)prog->check_offset_max,
            (IV)(s-strpos),
            (IV)(t-strpos),
            (IV)(t-s),
            (IV)(strend-strpos)
        )
    );

    if (s - strpos > prog->check_offset_max  /* signed-corrected t > strpos */
        && (!utf8_target
	    || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos)))
		 && t > strpos))) 
    {
	/* Fixed substring is found far enough so that the match
	   cannot start at strpos. */
      try_at_offset:
	if (ml_anch && t[-1] != '\n') {
	    /* Eventually fbm_*() should handle this, but often
	       anchored_offset is not 0, so this check will not be wasted. */
	    /* XXXX In the code below we prefer to look for "^" even in
	       presence of anchored substrings.  And we search even
	       beyond the found float position.  These pessimizations
	       are historical artefacts only.  */
	  find_anchor:
	    while (t < strend - prog->minlen) {
		if (*t == '\n') {
		    if (t < check_at - prog->check_offset_min) {
			if (utf8_target ? prog->anchored_utf8 : prog->anchored_substr) {
			    /* Since we moved from the found position,
			       we definitely contradict the found anchored
			       substr.  Due to the above check we do not
			       contradict "check" substr.
			       Thus we can arrive here only if check substr
			       is float.  Redo checking for "other"=="fixed".
			     */
			    strpos = t + 1;			
			    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
				PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
			    goto do_other_anchored;
			}
			/* We don't contradict the found floating substring. */
			/* XXXX Why not check for STCLASS? */
			s = t + 1;
			DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
			    PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
			goto set_useful;
		    }
		    /* Position contradicts check-string */
		    /* XXXX probably better to look for check-string
		       than for "\n", so one should lower the limit for t? */
		    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
			PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
		    other_last = strpos = s = t + 1;
		    goto restart;
		}
		t++;
	    }
	    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
			PL_colors[0], PL_colors[1]));
	    goto fail_finish;
	}
	else {
	    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
			PL_colors[0], PL_colors[1]));
	}
	s = t;
      set_useful:
	++BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr);	/* hooray/5 */
    }
    else {
	/* The found string does not prohibit matching at strpos,
	   - no optimization of calling REx engine can be performed,
	   unless it was an MBOL and we are not after MBOL,
	   or a future STCLASS check will fail this. */
      try_at_start:
	/* Even in this situation we may use MBOL flag if strpos is offset
	   wrt the start of the string. */
	if (ml_anch && sv && !SvROK(sv)	/* See prev comment on SvROK */
	    && (strpos != strbeg) && strpos[-1] != '\n'
	    /* May be due to an implicit anchor of m{.*foo}  */
	    && !(prog->intflags & PREGf_IMPLICIT))
	{
	    t = strpos;
	    goto find_anchor;
	}
	DEBUG_EXECUTE_r( if (ml_anch)
	    PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
			  (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
	);
      success_at_start:
	if (!(prog->intflags & PREGf_NAUGHTY)	/* XXXX If strpos moved? */
	    && (utf8_target ? (
		prog->check_utf8		/* Could be deleted already */
		&& --BmUSEFUL(prog->check_utf8) < 0
		&& (prog->check_utf8 == prog->float_utf8)
	    ) : (
		prog->check_substr		/* Could be deleted already */
		&& --BmUSEFUL(prog->check_substr) < 0
		&& (prog->check_substr == prog->float_substr)
	    )))
	{
	    /* If flags & SOMETHING - do not do it many times on the same match */
	    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
	    /* XXX Does the destruction order has to change with utf8_target? */
	    SvREFCNT_dec(utf8_target ? prog->check_utf8 : prog->check_substr);
	    SvREFCNT_dec(utf8_target ? prog->check_substr : prog->check_utf8);
	    prog->check_substr = prog->check_utf8 = NULL;	/* disable */
	    prog->float_substr = prog->float_utf8 = NULL;	/* clear */
	    check = NULL;			/* abort */
	    s = strpos;
	    /* XXXX If the check string was an implicit check MBOL, then we need to unset the relevant flag
		    see http://bugs.activestate.com/show_bug.cgi?id=87173 */
	    if (prog->intflags & PREGf_IMPLICIT)
		prog->extflags &= ~RXf_ANCH_MBOL;
	    /* XXXX This is a remnant of the old implementation.  It
	            looks wasteful, since now INTUIT can use many
	            other heuristics. */
	    prog->extflags &= ~RXf_USE_INTUIT;
	    /* XXXX What other flags might need to be cleared in this branch? */
	}
	else
	    s = strpos;
    }

    /* Last resort... */
    /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
    /* trie stclasses are too expensive to use here, we are better off to
       leave it to regmatch itself */
    if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
	/* minlen == 0 is possible if regstclass is \b or \B,
	   and the fixed substr is ''$.
	   Since minlen is already taken into account, s+1 is before strend;
	   accidentally, minlen >= 1 guaranties no false positives at s + 1
	   even for \b or \B.  But (minlen? 1 : 0) below assumes that
	   regstclass does not come from lookahead...  */
	/* If regstclass takes bytelength more than 1: If charlength==1, OK.
	   This leaves EXACTF-ish only, which are dealt with in find_byclass().  */
        const U8* const str = (U8*)STRING(progi->regstclass);
        const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
		    ? CHR_DIST(str+STR_LEN(progi->regstclass), str)
		    : 1);
	char * endpos;
	if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
            endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend);
        else if (prog->float_substr || prog->float_utf8)
	    endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend);
        else 
            endpos= strend;
		    
        DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n",
				      (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg)));
	
	t = s;
        s = find_byclass(prog, progi->regstclass, s, endpos, NULL);
	if (!s) {
#ifdef DEBUGGING
	    const char *what = NULL;
#endif
	    if (endpos == strend) {
		DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
				"Could not match STCLASS...\n") );
		goto fail;
	    }
	    DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
				   "This position contradicts STCLASS...\n") );
	    if ((prog->extflags & RXf_ANCH) && !ml_anch)
		goto fail;
	    /* Contradict one of substrings */
	    if (prog->anchored_substr || prog->anchored_utf8) {
		if ((utf8_target ? prog->anchored_utf8 : prog->anchored_substr) == check) {
		    DEBUG_EXECUTE_r( what = "anchored" );
		  hop_and_restart:
		    s = HOP3c(t, 1, strend);
		    if (s + start_shift + end_shift > strend) {
			/* XXXX Should be taken into account earlier? */
			DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
					       "Could not match STCLASS...\n") );
			goto fail;
		    }
		    if (!check)
			goto giveup;
		    DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
				"Looking for %s substr starting at offset %ld...\n",
				 what, (long)(s + start_shift - i_strpos)) );
		    goto restart;
		}
		/* Have both, check_string is floating */
		if (t + start_shift >= check_at) /* Contradicts floating=check */
		    goto retry_floating_check;
		/* Recheck anchored substring, but not floating... */
		s = check_at;
		if (!check)
		    goto giveup;
		DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
			  "Looking for anchored substr starting at offset %ld...\n",
			  (long)(other_last - i_strpos)) );
		goto do_other_anchored;
	    }
	    /* Another way we could have checked stclass at the
               current position only: */
	    if (ml_anch) {
		s = t = t + 1;
		if (!check)
		    goto giveup;
		DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
			  "Looking for /%s^%s/m starting at offset %ld...\n",
			  PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
		goto try_at_offset;
	    }
	    if (!(utf8_target ? prog->float_utf8 : prog->float_substr))	/* Could have been deleted */
		goto fail;
	    /* Check is floating substring. */
	  retry_floating_check:
	    t = check_at - start_shift;
	    DEBUG_EXECUTE_r( what = "floating" );
	    goto hop_and_restart;
	}
	if (t != s) {
            DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
			"By STCLASS: moving %ld --> %ld\n",
                                  (long)(t - i_strpos), (long)(s - i_strpos))
                   );
        }
        else {
            DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
                                  "Does not contradict STCLASS...\n"); 
                   );
        }
    }
  giveup:
    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
			  PL_colors[4], (check ? "Guessed" : "Giving up"),
			  PL_colors[5], (long)(s - i_strpos)) );
    return s;

  fail_finish:				/* Substring not found */
    if (prog->check_substr || prog->check_utf8)		/* could be removed already */
	BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
  fail:
    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
			  PL_colors[4], PL_colors[5]));
    return NULL;
}

#define DECL_TRIE_TYPE(scan) \
    const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold } \
                    trie_type = ((scan->flags == EXACT) \
                              ? (utf8_target ? trie_utf8 : trie_plain) \
                              : (utf8_target ? trie_utf8_fold : trie_latin_utf8_fold))

#define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len,          \
uvc, charid, foldlen, foldbuf, uniflags) STMT_START {                               \
    STRLEN skiplen;                                                                 \
    switch (trie_type) {                                                            \
    case trie_utf8_fold:                                                            \
        if ( foldlen>0 ) {                                                          \
            uvc = utf8n_to_uvuni( (const U8*) uscan, UTF8_MAXLEN, &len, uniflags ); \
            foldlen -= len;                                                         \
            uscan += len;                                                           \
            len=0;                                                                  \
        } else {                                                                    \
            uvc = to_utf8_fold( (const U8*) uc, foldbuf, &foldlen );                \
            len = UTF8SKIP(uc);                                                     \
            skiplen = UNISKIP( uvc );                                               \
            foldlen -= skiplen;                                                     \
            uscan = foldbuf + skiplen;                                              \
        }                                                                           \
        break;                                                                      \
    case trie_latin_utf8_fold:                                                      \
        if ( foldlen>0 ) {                                                          \
            uvc = utf8n_to_uvuni( (const U8*) uscan, UTF8_MAXLEN, &len, uniflags ); \
            foldlen -= len;                                                         \
            uscan += len;                                                           \
            len=0;                                                                  \
        } else {                                                                    \
            len = 1;                                                                \
            uvc = _to_fold_latin1( (U8) *uc, foldbuf, &foldlen, 1);                 \
            skiplen = UNISKIP( uvc );                                               \
            foldlen -= skiplen;                                                     \
            uscan = foldbuf + skiplen;                                              \
        }                                                                           \
        break;                                                                      \
    case trie_utf8:                                                                 \
        uvc = utf8n_to_uvuni( (const U8*) uc, UTF8_MAXLEN, &len, uniflags );        \
        break;                                                                      \
    case trie_plain:                                                                \
        uvc = (UV)*uc;                                                              \
        len = 1;                                                                    \
    }                                                                               \
    if (uvc < 256) {                                                                \
        charid = trie->charmap[ uvc ];                                              \
    }                                                                               \
    else {                                                                          \
        charid = 0;                                                                 \
        if (widecharmap) {                                                          \
            SV** const svpp = hv_fetch(widecharmap,                                 \
                        (char*)&uvc, sizeof(UV), 0);                                \
            if (svpp)                                                               \
                charid = (U16)SvIV(*svpp);                                          \
        }                                                                           \
    }                                                                               \
} STMT_END

#define REXEC_FBC_EXACTISH_SCAN(CoNd)                     \
STMT_START {                                              \
    while (s <= e) {                                      \
	if ( (CoNd)                                       \
	     && (ln == 1 || folder(s, pat_string, ln))    \
	     && (!reginfo || regtry(reginfo, &s)) )       \
	    goto got_it;                                  \
	s++;                                              \
    }                                                     \
} STMT_END

#define REXEC_FBC_UTF8_SCAN(CoDe)                     \
STMT_START {                                          \
    while (s + (uskip = UTF8SKIP(s)) <= strend) {     \
	CoDe                                          \
	s += uskip;                                   \
    }                                                 \
} STMT_END

#define REXEC_FBC_SCAN(CoDe)                          \
STMT_START {                                          \
    while (s < strend) {                              \
	CoDe                                          \
	s++;                                          \
    }                                                 \
} STMT_END

#define REXEC_FBC_UTF8_CLASS_SCAN(CoNd)               \
REXEC_FBC_UTF8_SCAN(                                  \
    if (CoNd) {                                       \
	if (tmp && (!reginfo || regtry(reginfo, &s)))  \
	    goto got_it;                              \
	else                                          \
	    tmp = doevery;                            \
    }                                                 \
    else                                              \
	tmp = 1;                                      \
)

#define REXEC_FBC_CLASS_SCAN(CoNd)                    \
REXEC_FBC_SCAN(                                       \
    if (CoNd) {                                       \
	if (tmp && (!reginfo || regtry(reginfo, &s)))  \
	    goto got_it;                              \
	else                                          \
	    tmp = doevery;                            \
    }                                                 \
    else                                              \
	tmp = 1;                                      \
)

#define REXEC_FBC_TRYIT               \
if ((!reginfo || regtry(reginfo, &s))) \
    goto got_it

#define REXEC_FBC_CSCAN(CoNdUtF8,CoNd)                         \
    if (utf8_target) {                                             \
	REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8);                   \
    }                                                          \
    else {                                                     \
	REXEC_FBC_CLASS_SCAN(CoNd);                            \
    }
    
#define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd)      \
    if (utf8_target) {                                             \
	UtFpReLoAd;                                            \
	REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8);                   \
    }                                                          \
    else {                                                     \
	REXEC_FBC_CLASS_SCAN(CoNd);                            \
    }

#define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd)                   \
    PL_reg_flags |= RF_tainted;                                \
    if (utf8_target) {                                             \
	REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8);                   \
    }                                                          \
    else {                                                     \
	REXEC_FBC_CLASS_SCAN(CoNd);                            \
    }

#define DUMP_EXEC_POS(li,s,doutf8) \
    dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)


#define UTF8_NOLOAD(TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
	tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';                         \
	tmp = TEST_NON_UTF8(tmp);                                              \
	REXEC_FBC_UTF8_SCAN(                                                   \
	    if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
		tmp = !tmp;                                                    \
		IF_SUCCESS;                                                    \
	    }                                                                  \
	    else {                                                             \
		IF_FAIL;                                                       \
	    }                                                                  \
	);                                                                     \

#define UTF8_LOAD(TeSt1_UtF8, TeSt2_UtF8, IF_SUCCESS, IF_FAIL) \
	if (s == PL_bostr) {                                                   \
	    tmp = '\n';                                                        \
	}                                                                      \
	else {                                                                 \
	    U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);                 \
	    tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);       \
	}                                                                      \
	tmp = TeSt1_UtF8;                                                      \
	LOAD_UTF8_CHARCLASS_ALNUM();                                                                \
	REXEC_FBC_UTF8_SCAN(                                                   \
	    if (tmp == ! (TeSt2_UtF8)) { \
		tmp = !tmp;                                                    \
		IF_SUCCESS;                                                    \
	    }                                                                  \
	    else {                                                             \
		IF_FAIL;                                                       \
	    }                                                                  \
	);                                                                     \

/* The only difference between the BOUND and NBOUND cases is that
 * REXEC_FBC_TRYIT is called when matched in BOUND, and when non-matched in
 * NBOUND.  This is accomplished by passing it in either the if or else clause,
 * with the other one being empty */
#define FBC_BOUND(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
    FBC_BOUND_COMMON(UTF8_LOAD(TEST1_UTF8, TEST2_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)

#define FBC_BOUND_NOLOAD(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
    FBC_BOUND_COMMON(UTF8_NOLOAD(TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)

#define FBC_NBOUND(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
    FBC_BOUND_COMMON(UTF8_LOAD(TEST1_UTF8, TEST2_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)

#define FBC_NBOUND_NOLOAD(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
    FBC_BOUND_COMMON(UTF8_NOLOAD(TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)


/* Common to the BOUND and NBOUND cases.  Unfortunately the UTF8 tests need to
 * be passed in completely with the variable name being tested, which isn't
 * such a clean interface, but this is easier to read than it was before.  We
 * are looking for the boundary (or non-boundary between a word and non-word
 * character.  The utf8 and non-utf8 cases have the same logic, but the details
 * must be different.  Find the "wordness" of the character just prior to this
 * one, and compare it with the wordness of this one.  If they differ, we have
 * a boundary.  At the beginning of the string, pretend that the previous
 * character was a new-line */
#define FBC_BOUND_COMMON(UTF8_CODE, TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
    if (utf8_target) {                                                         \
		UTF8_CODE \
    }                                                                          \
    else {  /* Not utf8 */                                                     \
	tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';                         \
	tmp = TEST_NON_UTF8(tmp);                                              \
	REXEC_FBC_SCAN(                                                        \
	    if (tmp == ! TEST_NON_UTF8((U8) *s)) {                             \
		tmp = !tmp;                                                    \
		IF_SUCCESS;                                                    \
	    }                                                                  \
	    else {                                                             \
		IF_FAIL;                                                       \
	    }                                                                  \
	);                                                                     \
    }                                                                          \
    if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s)))           \
	goto got_it;

/* We know what class REx starts with.  Try to find this position... */
/* if reginfo is NULL, its a dryrun */
/* annoyingly all the vars in this routine have different names from their counterparts
   in regmatch. /grrr */

STATIC char *
S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, 
    const char *strend, regmatch_info *reginfo)
{
	dVAR;
	const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
	char *pat_string;   /* The pattern's exactish string */
	char *pat_end;	    /* ptr to end char of pat_string */
	re_fold_t folder;	/* Function for computing non-utf8 folds */
	const U8 *fold_array;   /* array for folding ords < 256 */
	STRLEN ln;
	STRLEN lnc;
	register STRLEN uskip;
	U8 c1;
	U8 c2;
	char *e;
	register I32 tmp = 1;	/* Scratch variable? */
	register const bool utf8_target = PL_reg_match_utf8;
	UV utf8_fold_flags = 0;
        RXi_GET_DECL(prog,progi);

	PERL_ARGS_ASSERT_FIND_BYCLASS;
        
	/* We know what class it must start with. */
	switch (OP(c)) {
	case ANYOFV:
	case ANYOF:
	    if (utf8_target || OP(c) == ANYOFV) {
		STRLEN inclasslen = strend - s;
		REXEC_FBC_UTF8_CLASS_SCAN(
                          reginclass(prog, c, (U8*)s, &inclasslen, utf8_target));
	    }
	    else {
		REXEC_FBC_CLASS_SCAN(REGINCLASS(prog, c, (U8*)s));
	    }
	    break;
	case CANY:
	    REXEC_FBC_SCAN(
	        if (tmp && (!reginfo || regtry(reginfo, &s)))
		    goto got_it;
		else
		    tmp = doevery;
	    );
	    break;

	case EXACTFA:
	    if (UTF_PATTERN || utf8_target) {
		utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
		goto do_exactf_utf8;
	    }
	    fold_array = PL_fold_latin1;    /* Latin1 folds are not affected by */
	    folder = foldEQ_latin1;	    /* /a, except the sharp s one which */
	    goto do_exactf_non_utf8;	    /* isn't dealt with by these */

	case EXACTF:
	    if (utf8_target) {

		/* regcomp.c already folded this if pattern is in UTF-8 */
		utf8_fold_flags = 0;
		goto do_exactf_utf8;
	    }
	    fold_array = PL_fold;
	    folder = foldEQ;
	    goto do_exactf_non_utf8;

	case EXACTFL:
	    if (UTF_PATTERN || utf8_target) {
		utf8_fold_flags = FOLDEQ_UTF8_LOCALE;
		goto do_exactf_utf8;
	    }
	    fold_array = PL_fold_locale;
	    folder = foldEQ_locale;
	    goto do_exactf_non_utf8;

	case EXACTFU_SS:
	    if (UTF_PATTERN) {
		utf8_fold_flags = FOLDEQ_S2_ALREADY_FOLDED;
	    }
	    goto do_exactf_utf8;

	case EXACTFU_TRICKYFOLD:
	case EXACTFU:
	    if (UTF_PATTERN || utf8_target) {
		utf8_fold_flags = (UTF_PATTERN) ? FOLDEQ_S2_ALREADY_FOLDED : 0;
		goto do_exactf_utf8;
	    }

	    /* Any 'ss' in the pattern should have been replaced by regcomp,
	     * so we don't have to worry here about this single special case
	     * in the Latin1 range */
	    fold_array = PL_fold_latin1;
	    folder = foldEQ_latin1;

	    /* FALL THROUGH */

	do_exactf_non_utf8: /* Neither pattern nor string are UTF8, and there
			       are no glitches with fold-length differences
			       between the target string and pattern */

	    /* The idea in the non-utf8 EXACTF* cases is to first find the
	     * first character of the EXACTF* node and then, if necessary,
	     * case-insensitively compare the full text of the node.  c1 is the
	     * first character.  c2 is its fold.  This logic will not work for
	     * Unicode semantics and the german sharp ss, which hence should
	     * not be compiled into a node that gets here. */
	    pat_string = STRING(c);
	    ln  = STR_LEN(c);	/* length to match in octets/bytes */

	    /* We know that we have to match at least 'ln' bytes (which is the
	     * same as characters, since not utf8).  If we have to match 3
	     * characters, and there are only 2 availabe, we know without
	     * trying that it will fail; so don't start a match past the
	     * required minimum number from the far end */
	    e = HOP3c(strend, -((I32)ln), s);

	    if (!reginfo && e < s) {
		e = s;			/* Due to minlen logic of intuit() */
	    }

	    c1 = *pat_string;
	    c2 = fold_array[c1];
	    if (c1 == c2) { /* If char and fold are the same */
		REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
	    }
	    else {
		REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
	    }
	    break;

	do_exactf_utf8:
	{
	    unsigned expansion;


	    /* If one of the operands is in utf8, we can't use the simpler
	     * folding above, due to the fact that many different characters
	     * can have the same fold, or portion of a fold, or different-
	     * length fold */
	    pat_string = STRING(c);
	    ln  = STR_LEN(c);	/* length to match in octets/bytes */
	    pat_end = pat_string + ln;
	    lnc = (UTF_PATTERN) /* length to match in characters */
		    ? utf8_length((U8 *) pat_string, (U8 *) pat_end)
		    : ln;

	    /* We have 'lnc' characters to match in the pattern, but because of
	     * multi-character folding, each character in the target can match
	     * up to 3 characters (Unicode guarantees it will never exceed
	     * this) if it is utf8-encoded; and up to 2 if not (based on the
	     * fact that the Latin 1 folds are already determined, and the
	     * only multi-char fold in that range is the sharp-s folding to
	     * 'ss'.  Thus, a pattern character can match as little as 1/3 of a
	     * string character.  Adjust lnc accordingly, rounding up, so that
	     * if we need to match at least 4+1/3 chars, that really is 5. */
	    expansion = (utf8_target) ? UTF8_MAX_FOLD_CHAR_EXPAND : 2;
	    lnc = (lnc + expansion - 1) / expansion;

	    /* As in the non-UTF8 case, if we have to match 3 characters, and
	     * only 2 are left, it's guaranteed to fail, so don't start a
	     * match that would require us to go beyond the end of the string
	     */
	    e = HOP3c(strend, -((I32)lnc), s);

	    if (!reginfo && e < s) {
		e = s;			/* Due to minlen logic of intuit() */
	    }

	    /* XXX Note that we could recalculate e to stop the loop earlier,
	     * as the worst case expansion above will rarely be met, and as we
	     * go along we would usually find that e moves further to the left.
	     * This would happen only after we reached the point in the loop
	     * where if there were no expansion we should fail.  Unclear if
	     * worth the expense */

	    while (s <= e) {
		char *my_strend= (char *)strend;
		if (foldEQ_utf8_flags(s, &my_strend, 0,  utf8_target,
		      pat_string, NULL, ln, cBOOL(UTF_PATTERN), utf8_fold_flags)
		    && (!reginfo || regtry(reginfo, &s)) )
		{
		    goto got_it;
		}
		s += (utf8_target) ? UTF8SKIP(s) : 1;
	    }
	    break;
	}
	case BOUNDL:
	    PL_reg_flags |= RF_tainted;
	    FBC_BOUND(isALNUM_LC,
		      isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)),
		      isALNUM_LC_utf8((U8*)s));
	    break;
	case NBOUNDL:
	    PL_reg_flags |= RF_tainted;
	    FBC_NBOUND(isALNUM_LC,
		       isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)),
		       isALNUM_LC_utf8((U8*)s));
	    break;
	case BOUND:
	    FBC_BOUND(isWORDCHAR,
		      isALNUM_uni(tmp),
		      cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
	    break;
	case BOUNDA:
	    FBC_BOUND_NOLOAD(isWORDCHAR_A,
			     isWORDCHAR_A(tmp),
			     isWORDCHAR_A((U8*)s));
	    break;
	case NBOUND:
	    FBC_NBOUND(isWORDCHAR,
		       isALNUM_uni(tmp),
		       cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
	    break;
	case NBOUNDA:
	    FBC_NBOUND_NOLOAD(isWORDCHAR_A,
			      isWORDCHAR_A(tmp),
			      isWORDCHAR_A((U8*)s));
	    break;
	case BOUNDU:
	    FBC_BOUND(isWORDCHAR_L1,
		      isALNUM_uni(tmp),
		      cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
	    break;
	case NBOUNDU:
	    FBC_NBOUND(isWORDCHAR_L1,
		       isALNUM_uni(tmp),
		       cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
	    break;
	case ALNUML:
	    REXEC_FBC_CSCAN_TAINT(
		isALNUM_LC_utf8((U8*)s),
		isALNUM_LC(*s)
	    );
	    break;
	case ALNUMU:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_ALNUM(),
		swash_fetch(PL_utf8_alnum,(U8*)s, utf8_target),
                isWORDCHAR_L1((U8) *s)
	    );
	    break;
	case ALNUM:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_ALNUM(),
		swash_fetch(PL_utf8_alnum,(U8*)s, utf8_target),
                isWORDCHAR((U8) *s)
	    );
	    break;
	case ALNUMA:
	    /* Don't need to worry about utf8, as it can match only a single
	     * byte invariant character */
	    REXEC_FBC_CLASS_SCAN( isWORDCHAR_A(*s));
	    break;
	case NALNUMU:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_ALNUM(),
		!swash_fetch(PL_utf8_alnum,(U8*)s, utf8_target),
                ! isWORDCHAR_L1((U8) *s)
	    );
	    break;
	case NALNUM:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_ALNUM(),
		!swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target),
                ! isALNUM(*s)
	    );
	    break;
	case NALNUMA:
	    REXEC_FBC_CSCAN(
		!isWORDCHAR_A(*s),
		!isWORDCHAR_A(*s)
	    );
	    break;
	case NALNUML:
	    REXEC_FBC_CSCAN_TAINT(
		!isALNUM_LC_utf8((U8*)s),
		!isALNUM_LC(*s)
	    );
	    break;
	case SPACEU:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_SPACE(),
		*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target),
                isSPACE_L1((U8) *s)
	    );
	    break;
	case SPACE:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_SPACE(),
		*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target),
                isSPACE((U8) *s)
	    );
	    break;
	case SPACEA:
	    /* Don't need to worry about utf8, as it can match only a single
	     * byte invariant character */
	    REXEC_FBC_CLASS_SCAN( isSPACE_A(*s));
	    break;
	case SPACEL:
	    REXEC_FBC_CSCAN_TAINT(
		isSPACE_LC_utf8((U8*)s),
		isSPACE_LC(*s)
	    );
	    break;
	case NSPACEU:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_SPACE(),
		!( *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target)),
                ! isSPACE_L1((U8) *s)
	    );
	    break;
	case NSPACE:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_SPACE(),
		!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target)),
                ! isSPACE((U8) *s)
	    );
	    break;
	case NSPACEA:
	    REXEC_FBC_CSCAN(
		!isSPACE_A(*s),
		!isSPACE_A(*s)
	    );
	    break;
	case NSPACEL:
	    REXEC_FBC_CSCAN_TAINT(
		!isSPACE_LC_utf8((U8*)s),
		!isSPACE_LC(*s)
	    );
	    break;
	case DIGIT:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_DIGIT(),
		swash_fetch(PL_utf8_digit,(U8*)s, utf8_target),
		isDIGIT(*s)
	    );
	    break;
	case DIGITA:
	    /* Don't need to worry about utf8, as it can match only a single
	     * byte invariant character */
	    REXEC_FBC_CLASS_SCAN( isDIGIT_A(*s));
	    break;
	case DIGITL:
	    REXEC_FBC_CSCAN_TAINT(
		isDIGIT_LC_utf8((U8*)s),
		isDIGIT_LC(*s)
	    );
	    break;
	case NDIGIT:
	    REXEC_FBC_CSCAN_PRELOAD(
		LOAD_UTF8_CHARCLASS_DIGIT(),
		!swash_fetch(PL_utf8_digit,(U8*)s, utf8_target),
		!isDIGIT(*s)
	    );
	    break;
	case NDIGITA:
	    REXEC_FBC_CSCAN(
		!isDIGIT_A(*s),
		!isDIGIT_A(*s)
	    );
	    break;
	case NDIGITL:
	    REXEC_FBC_CSCAN_TAINT(
		!isDIGIT_LC_utf8((U8*)s),
		!isDIGIT_LC(*s)
	    );
	    break;
	case LNBREAK:
	    REXEC_FBC_CSCAN(
		is_LNBREAK_utf8(s),
		is_LNBREAK_latin1(s)
	    );
	    break;
	case VERTWS:
	    REXEC_FBC_CSCAN(
		is_VERTWS_utf8(s),
		is_VERTWS_latin1(s)
	    );
	    break;
	case NVERTWS:
	    REXEC_FBC_CSCAN(
		!is_VERTWS_utf8(s),
		!is_VERTWS_latin1(s)
	    );
	    break;
	case HORIZWS:
	    REXEC_FBC_CSCAN(
		is_HORIZWS_utf8(s),
		is_HORIZWS_latin1(s)
	    );
	    break;
	case NHORIZWS:
	    REXEC_FBC_CSCAN(
		!is_HORIZWS_utf8(s),
		!is_HORIZWS_latin1(s)
	    );	    
	    break;
	case AHOCORASICKC:
	case AHOCORASICK: 
	    {
	        DECL_TRIE_TYPE(c);
                /* what trie are we using right now */
        	reg_ac_data *aho
        	    = (reg_ac_data*)progi->data->data[ ARG( c ) ];
        	reg_trie_data *trie
		    = (reg_trie_data*)progi->data->data[ aho->trie ];
		HV *widecharmap = MUTABLE_HV(progi->data->data[ aho->trie + 1 ]);

		const char *last_start = strend - trie->minlen;
#ifdef DEBUGGING
		const char *real_start = s;
#endif
		STRLEN maxlen = trie->maxlen;
		SV *sv_points;
		U8 **points; /* map of where we were in the input string
		                when reading a given char. For ASCII this
		                is unnecessary overhead as the relationship
		                is always 1:1, but for Unicode, especially
		                case folded Unicode this is not true. */
		U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
		U8 *bitmap=NULL;


                GET_RE_DEBUG_FLAGS_DECL;

                /* We can't just allocate points here. We need to wrap it in
                 * an SV so it gets freed properly if there is a croak while
                 * running the match */
                ENTER;
	        SAVETMPS;
                sv_points=newSV(maxlen * sizeof(U8 *));
                SvCUR_set(sv_points,
                    maxlen * sizeof(U8 *));
                SvPOK_on(sv_points);
                sv_2mortal(sv_points);
                points=(U8**)SvPV_nolen(sv_points );
                if ( trie_type != trie_utf8_fold 
                     && (trie->bitmap || OP(c)==AHOCORASICKC) ) 
                {
                    if (trie->bitmap) 
                        bitmap=(U8*)trie->bitmap;
                    else
                        bitmap=(U8*)ANYOF_BITMAP(c);
                }
                /* this is the Aho-Corasick algorithm modified a touch
                   to include special handling for long "unknown char" 
                   sequences. The basic idea being that we use AC as long
                   as we are dealing with a possible matching char, when
                   we encounter an unknown char (and we have not encountered
                   an accepting state) we scan forward until we find a legal 
                   starting char. 
                   AC matching is basically that of trie matching, except
                   that when we encounter a failing transition, we fall back
                   to the current states "fail state", and try the current char 
                   again, a process we repeat until we reach the root state, 
                   state 1, or a legal transition. If we fail on the root state 
                   then we can either terminate if we have reached an accepting 
                   state previously, or restart the entire process from the beginning 
                   if we have not.

                 */
                while (s <= last_start) {
                    const U32 uniflags = UTF8_ALLOW_DEFAULT;
                    U8 *uc = (U8*)s;
                    U16 charid = 0;
                    U32 base = 1;
                    U32 state = 1;
                    UV uvc = 0;
                    STRLEN len = 0;
                    STRLEN foldlen = 0;
                    U8 *uscan = (U8*)NULL;
                    U8 *leftmost = NULL;
#ifdef DEBUGGING                    
                    U32 accepted_word= 0;
#endif
                    U32 pointpos = 0;

                    while ( state && uc <= (U8*)strend ) {
                        int failed=0;
                        U32 word = aho->states[ state ].wordnum;

                        if( state==1 ) {
                            if ( bitmap ) {
                                DEBUG_TRIE_EXECUTE_r(
                                    if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
                                        dump_exec_pos( (char *)uc, c, strend, real_start, 
                                            (char *)uc, utf8_target );
                                        PerlIO_printf( Perl_debug_log,
                                            " Scanning for legal start char...\n");
                                    }
                                );
				if (utf8_target) {
				    while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
					uc += UTF8SKIP(uc);
				    }
				} else {
				    while ( uc <= (U8*)last_start  && !BITMAP_TEST(bitmap,*uc) ) {
					uc++;
				    }
				}
                                s= (char *)uc;
                            }
                            if (uc >(U8*)last_start) break;
                        }
                                            
                        if ( word ) {
                            U8 *lpos= points[ (pointpos - trie->wordinfo[word].len) % maxlen ];
                            if (!leftmost || lpos < leftmost) {
                                DEBUG_r(accepted_word=word);
                                leftmost= lpos;
                            }
                            if (base==0) break;
                            
                        }
                        points[pointpos++ % maxlen]= uc;
			REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
					     uscan, len, uvc, charid, foldlen,
					     foldbuf, uniflags);
                        DEBUG_TRIE_EXECUTE_r({
                            dump_exec_pos( (char *)uc, c, strend, real_start, 
                                s,   utf8_target );
                            PerlIO_printf(Perl_debug_log,
                                " Charid:%3u CP:%4"UVxf" ",
                                 charid, uvc);
                        });

                        do {
#ifdef DEBUGGING
                            word = aho->states[ state ].wordnum;
#endif
                            base = aho->states[ state ].trans.base;

                            DEBUG_TRIE_EXECUTE_r({
                                if (failed) 
                                    dump_exec_pos( (char *)uc, c, strend, real_start, 
                                        s,   utf8_target );
                                PerlIO_printf( Perl_debug_log,
                                    "%sState: %4"UVxf", word=%"UVxf,
                                    failed ? " Fail transition to " : "",
                                    (UV)state, (UV)word);
                            });
                            if ( base ) {
                                U32 tmp;
				I32 offset;
                                if (charid &&
				     ( ((offset = base + charid
					- 1 - trie->uniquecharcount)) >= 0)
                                     && ((U32)offset < trie->lasttrans)
                                     && trie->trans[offset].check == state
                                     && (tmp=trie->trans[offset].next))
                                {
                                    DEBUG_TRIE_EXECUTE_r(
                                        PerlIO_printf( Perl_debug_log," - legal\n"));
                                    state = tmp;
                                    break;
                                }
                                else {
                                    DEBUG_TRIE_EXECUTE_r(
                                        PerlIO_printf( Perl_debug_log," - fail\n"));
                                    failed = 1;
                                    state = aho->fail[state];
                                }
                            }
                            else {
                                /* we must be accepting here */
                                DEBUG_TRIE_EXECUTE_r(
                                        PerlIO_printf( Perl_debug_log," - accepting\n"));
                                failed = 1;
                                break;
                            }
                        } while(state);
                        uc += len;
                        if (failed) {
                            if (leftmost)
                                break;
                            if (!state) state = 1;
                        }
                    }
                    if ( aho->states[ state ].wordnum ) {
                        U8 *lpos = points[ (pointpos - trie->wordinfo[aho->states[ state ].wordnum].len) % maxlen ];
                        if (!leftmost || lpos < leftmost) {
                            DEBUG_r(accepted_word=aho->states[ state ].wordnum);
                            leftmost = lpos;
                        }
                    }
                    if (leftmost) {
                        s = (char*)leftmost;
                        DEBUG_TRIE_EXECUTE_r({
                            PerlIO_printf( 
                                Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n",
                                (UV)accepted_word, (IV)(s - real_start)
                            );
                        });
                        if (!reginfo || regtry(reginfo, &s)) {
                            FREETMPS;
		            LEAVE;
                            goto got_it;
                        }
                        s = HOPc(s,1);
                        DEBUG_TRIE_EXECUTE_r({
                            PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n");
                        });
                    } else {
                        DEBUG_TRIE_EXECUTE_r(
                            PerlIO_printf( Perl_debug_log,"No match.\n"));
                        break;
                    }
                }
                FREETMPS;
                LEAVE;
	    }
	    break;
	default:
	    Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
	    break;
	}
	return 0;
      got_it:
	return s;
}


/*
 - regexec_flags - match a regexp against a string
 */
I32
Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, register char *strend,
	      char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
/* strend: pointer to null at end of string */
/* strbeg: real beginning of string */
/* minend: end of match must be >=minend after stringarg. */
/* data: May be used for some additional optimizations. 
         Currently its only used, with a U32 cast, for transmitting 
         the ganch offset when doing a /g match. This will change */
/* nosave: For optimizations. */
{
    dVAR;
    struct regexp *const prog = (struct regexp *)SvANY(rx);
    /*register*/ char *s;
    register regnode *c;
    /*register*/ char *startpos = stringarg;
    I32 minlen;		/* must match at least this many chars */
    I32 dontbother = 0;	/* how many characters not to try at end */
    I32 end_shift = 0;			/* Same for the end. */		/* CC */
    I32 scream_pos = -1;		/* Internal iterator of scream. */
    char *scream_olds = NULL;
    const bool utf8_target = cBOOL(DO_UTF8(sv));
    I32 multiline;
    RXi_GET_DECL(prog,progi);
    regmatch_info reginfo;  /* create some info to pass to regtry etc */
    regexp_paren_pair *swap = NULL;
    GET_RE_DEBUG_FLAGS_DECL;

    PERL_ARGS_ASSERT_REGEXEC_FLAGS;
    PERL_UNUSED_ARG(data);

    /* Be paranoid... */
    if (prog == NULL || startpos == NULL) {
	Perl_croak(aTHX_ "NULL regexp parameter");
	return 0;
    }

    multiline = prog->extflags & RXf_PMf_MULTILINE;
    reginfo.prog = rx;	 /* Yes, sorry that this is confusing.  */

    RX_MATCH_UTF8_set(rx, utf8_target);
    DEBUG_EXECUTE_r( 
        debug_start_match(rx, utf8_target, startpos, strend,
        "Matching");
    );

    minlen = prog->minlen;
    
    if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
        DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
			      "String too short [regexec_flags]...\n"));
	goto phooey;
    }

    
    /* Check validity of program. */
    if (UCHARAT(progi->program) != REG_MAGIC) {
	Perl_croak(aTHX_ "corrupted regexp program");
    }

    PL_reg_flags = 0;
    PL_reg_eval_set = 0;
    PL_reg_maxiter = 0;

    if (RX_UTF8(rx))
	PL_reg_flags |= RF_utf8;

    /* Mark beginning of line for ^ and lookbehind. */
    reginfo.bol = startpos; /* XXX not used ??? */
    PL_bostr  = strbeg;
    reginfo.sv = sv;

    /* Mark end of line for $ (and such) */
    PL_regeol = strend;

    /* see how far we have to get to not match where we matched before */
    reginfo.till = startpos+minend;

    /* If there is a "must appear" string, look for it. */
    s = startpos;

    if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */
	MAGIC *mg;
	if (flags & REXEC_IGNOREPOS){	/* Means: check only at start */
	    reginfo.ganch = startpos + prog->gofs;
	    DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
	      "GPOS IGNOREPOS: reginfo.ganch = startpos + %"UVxf"\n",(UV)prog->gofs));
	} else if (sv && SvTYPE(sv) >= SVt_PVMG
		  && SvMAGIC(sv)
		  && (mg = mg_find(sv, PERL_MAGIC_regex_global))
		  && mg->mg_len >= 0) {
	    reginfo.ganch = strbeg + mg->mg_len;	/* Defined pos() */
	    DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
		"GPOS MAGIC: reginfo.ganch = strbeg + %"IVdf"\n",(IV)mg->mg_len));

	    if (prog->extflags & RXf_ANCH_GPOS) {
	        if (s > reginfo.ganch)
		    goto phooey;
		s = reginfo.ganch - prog->gofs;
	        DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
		     "GPOS ANCH_GPOS: s = ganch - %"UVxf"\n",(UV)prog->gofs));
		if (s < strbeg)
		    goto phooey;
	    }
	}
	else if (data) {
	    reginfo.ganch = strbeg + PTR2UV(data);
            DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
		 "GPOS DATA: reginfo.ganch= strbeg + %"UVxf"\n",PTR2UV(data)));

	} else {				/* pos() not defined */
	    reginfo.ganch = strbeg;
            DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
		 "GPOS: reginfo.ganch = strbeg\n"));
	}
    }
    if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) {
        /* We have to be careful. If the previous successful match
           was from this regex we don't want a subsequent partially
           successful match to clobber the old results.
           So when we detect this possibility we add a swap buffer
           to the re, and switch the buffer each match. If we fail
           we switch it back, otherwise we leave it swapped.
        */
        swap = prog->offs;
        /* do we need a save destructor here for eval dies? */
        Newxz(prog->offs, (prog->nparens + 1), regexp_paren_pair);
    }
    if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
	re_scream_pos_data d;

	d.scream_olds = &scream_olds;
	d.scream_pos = &scream_pos;
	s = re_intuit_start(rx, sv, s, strend, flags, &d);
	if (!s) {
	    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
	    goto phooey;	/* not present */
	}
    }



    /* Simplest case:  anchored match need be tried only once. */
    /*  [unless only anchor is BOL and multiline is set] */
    if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) {
	if (s == startpos && regtry(&reginfo, &startpos))
	    goto got_it;
	else if (multiline || (prog->intflags & PREGf_IMPLICIT)
		 || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */
	{
	    char *end;

	    if (minlen)
		dontbother = minlen - 1;
	    end = HOP3c(strend, -dontbother, strbeg) - 1;
	    /* for multiline we only have to try after newlines */
	    if (prog->check_substr || prog->check_utf8) {
                /* because of the goto we can not easily reuse the macros for bifurcating the
                   unicode/non-unicode match modes here like we do elsewhere - demerphq */
                if (utf8_target) {
                    if (s == startpos)
                        goto after_try_utf8;
                    while (1) {
                        if (regtry(&reginfo, &s)) {
                            goto got_it;
                        }
                      after_try_utf8:
                        if (s > end) {
                            goto phooey;
                        }
                        if (prog->extflags & RXf_USE_INTUIT) {
                            s = re_intuit_start(rx, sv, s + UTF8SKIP(s), strend, flags, NULL);
                            if (!s) {
                                goto phooey;
                            }
                        }
                        else {
                            s += UTF8SKIP(s);
                        }
                    }
                } /* end search for check string in unicode */
                else {
                    if (s == startpos) {
                        goto after_try_latin;
                    }
                    while (1) {
                        if (regtry(&reginfo, &s)) {
                            goto got_it;
                        }
                      after_try_latin:
                        if (s > end) {
                            goto phooey;
                        }
                        if (prog->extflags & RXf_USE_INTUIT) {
                            s = re_intuit_start(rx, sv, s + 1, strend, flags, NULL);
                            if (!s) {
                                goto phooey;
                            }
                        }
                        else {
                            s++;
                        }
                    }
                } /* end search for check string in latin*/
	    } /* end search for check string */
	    else { /* search for newline */
		if (s > startpos) {
                    /*XXX: The s-- is almost definitely wrong here under unicode - demeprhq*/
		    s--;
		}
		/* We can use a more efficient search as newlines are the same in unicode as they are in latin */
		while (s <= end) { /* note it could be possible to match at the end of the string */
		    if (*s++ == '\n') {	/* don't need PL_utf8skip here */
			if (regtry(&reginfo, &s))
			    goto got_it;
		    }
		}
	    } /* end search for newline */
	} /* end anchored/multiline check string search */
	goto phooey;
    } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK)) 
    {
        /* the warning about reginfo.ganch being used without initialization
           is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN 
           and we only enter this block when the same bit is set. */
        char *tmp_s = reginfo.ganch - prog->gofs;

	if (tmp_s >= strbeg && regtry(&reginfo, &tmp_s))
	    goto got_it;
	goto phooey;
    }

    /* Messy cases:  unanchored match. */
    if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
	/* we have /x+whatever/ */
	/* it must be a one character string (XXXX Except UTF_PATTERN?) */
	char ch;
#ifdef DEBUGGING
	int did_match = 0;
#endif
	if (!(utf8_target ? prog->anchored_utf8 : prog->anchored_substr))
	    utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
	ch = SvPVX_const(utf8_target ? prog->anchored_utf8 : prog->anchored_substr)[0];

	if (utf8_target) {
	    REXEC_FBC_SCAN(
		if (*s == ch) {
		    DEBUG_EXECUTE_r( did_match = 1 );
		    if (regtry(&reginfo, &s)) goto got_it;
		    s += UTF8SKIP(s);
		    while (s < strend && *s == ch)
			s += UTF8SKIP(s);
		}
	    );
	}
	else {
	    REXEC_FBC_SCAN(
		if (*s == ch) {
		    DEBUG_EXECUTE_r( did_match = 1 );
		    if (regtry(&reginfo, &s)) goto got_it;
		    s++;
		    while (s < strend && *s == ch)
			s++;
		}
	    );
	}
	DEBUG_EXECUTE_r(if (!did_match)
		PerlIO_printf(Perl_debug_log,
                                  "Did not find anchored character...\n")
               );
    }
    else if (prog->anchored_substr != NULL
	      || prog->anchored_utf8 != NULL
	      || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
		  && prog->float_max_offset < strend - s)) {
	SV *must;
	I32 back_max;
	I32 back_min;
	char *last;
	char *last1;		/* Last position checked before */
#ifdef DEBUGGING
	int did_match = 0;
#endif
	if (prog->anchored_substr || prog->anchored_utf8) {
	    if (!(utf8_target ? prog->anchored_utf8 : prog->anchored_substr))
		utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
	    must = utf8_target ? prog->anchored_utf8 : prog->anchored_substr;
	    back_max = back_min = prog->anchored_offset;
	} else {
	    if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
		utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
	    must = utf8_target ? prog->float_utf8 : prog->float_substr;
	    back_max = prog->float_max_offset;
	    back_min = prog->float_min_offset;
	}
	
	    
	if (must == &PL_sv_undef)
	    /* could not downgrade utf8 check substring, so must fail */
	    goto phooey;

        if (back_min<0) {
	    last = strend;
	} else {
            last = HOP3c(strend,	/* Cannot start after this */
        	  -(I32)(CHR_SVLEN(must)
        		 - (SvTAIL(must) != 0) + back_min), strbeg);
        }
	if (s > PL_bostr)
	    last1 = HOPc(s, -1);
	else
	    last1 = s - 1;	/* bogus */

	/* XXXX check_substr already used to find "s", can optimize if
	   check_substr==must. */
	scream_pos = -1;
	dontbother = end_shift;
	strend = HOPc(strend, -dontbother);
	while ( (s <= last) &&
		((flags & REXEC_SCREAM) && SvSCREAM(sv)
		 ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg,
				    end_shift, &scream_pos, 0))
		 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)),
				  (unsigned char*)strend, must,
				  multiline ? FBMrf_MULTILINE : 0))) ) {
	    /* we may be pointing at the wrong string */
	    if ((flags & REXEC_SCREAM) && RXp_MATCH_COPIED(prog))
		s = strbeg + (s - SvPVX_const(sv));
	    DEBUG_EXECUTE_r( did_match = 1 );
	    if (HOPc(s, -back_max) > last1) {
		last1 = HOPc(s, -back_min);
		s = HOPc(s, -back_max);
	    }
	    else {
		char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;

		last1 = HOPc(s, -back_min);
		s = t;
	    }
	    if (utf8_target) {
		while (s <= last1) {
		    if (regtry(&reginfo, &s))
			goto got_it;
		    s += UTF8SKIP(s);
		}
	    }
	    else {
		while (s <= last1) {
		    if (regtry(&reginfo, &s))
			goto got_it;
		    s++;
		}
	    }
	}
	DEBUG_EXECUTE_r(if (!did_match) {
            RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
                SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
            PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n",
			      ((must == prog->anchored_substr || must == prog->anchored_utf8)
			       ? "anchored" : "floating"),
                quoted, RE_SV_TAIL(must));
        });		    
	goto phooey;
    }
    else if ( (c = progi->regstclass) ) {
	if (minlen) {
	    const OPCODE op = OP(progi->regstclass);
	    /* don't bother with what can't match */
	    if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE)
	        strend = HOPc(strend, -(minlen - 1));
	}
	DEBUG_EXECUTE_r({
	    SV * const prop = sv_newmortal();
	    regprop(prog, prop, c);
	    {
		RE_PV_QUOTED_DECL(quoted,utf8_target,PERL_DEBUG_PAD_ZERO(1),
		    s,strend-s,60);
		PerlIO_printf(Perl_debug_log,
		    "Matching stclass %.*s against %s (%d bytes)\n",
		    (int)SvCUR(prop), SvPVX_const(prop),
		     quoted, (int)(strend - s));
	    }
	});
        if (find_byclass(prog, c, s, strend, &reginfo))
	    goto got_it;
	DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n"));
    }
    else {
	dontbother = 0;
	if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
	    /* Trim the end. */
	    char *last= NULL;
	    SV* float_real;

	    if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
		utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
	    float_real = utf8_target ? prog->float_utf8 : prog->float_substr;

	    if ((flags & REXEC_SCREAM) && SvSCREAM(sv)) {
		last = screaminstr(sv, float_real, s - strbeg,
				   end_shift, &scream_pos, 1); /* last one */
		if (!last)
		    last = scream_olds; /* Only one occurrence. */
		/* we may be pointing at the wrong string */
		else if (RXp_MATCH_COPIED(prog))
		    s = strbeg + (s - SvPVX_const(sv));
	    }
	    else {
		STRLEN len;
                const char * const little = SvPV_const(float_real, len);
		if (SvTAIL(float_real)) {
		    /* This means that float_real contains an artificial \n on the end
		     * due to the presence of something like this: /foo$/
		     * where we can match both "foo" and "foo\n" at the end of the string.
		     * So we have to compare the end of the string first against the float_real
		     * without the \n and then against the full float_real with the string.
		     * We have to watch out for cases where the string might be smaller
		     * than the float_real or the float_real without the \n.
		     */
		    char *checkpos= strend - len;
		    DEBUG_OPTIMISE_r(
			PerlIO_printf(Perl_debug_log,
			    "%sChecking for float_real.%s\n",
			    PL_colors[4], PL_colors[5]));
		    if (checkpos + 1 < strbeg) {
			/* can't match, even if we remove the trailing \n string is too short to match */
			DEBUG_EXECUTE_r(
			    PerlIO_printf(Perl_debug_log,
				"%sString shorter than required trailing substring, cannot match.%s\n",
				PL_colors[4], PL_colors[5]));
			goto phooey;
		    } else if (memEQ(checkpos + 1, little, len - 1)) {
			/* can match, the end of the string matches without the "\n" */
			last = checkpos + 1;
		    } else if (checkpos < strbeg) {
			/* cant match, string is too short when the "\n" is included */
			DEBUG_EXECUTE_r(
			    PerlIO_printf(Perl_debug_log,
				"%sString does not contain required trailing substring, cannot match.%s\n",
				PL_colors[4], PL_colors[5]));
			goto phooey;
		    } else if (!multiline) {
			/* non multiline match, so compare with the "\n" at the end of the string */
			if (memEQ(checkpos, little, len)) {
			    last= checkpos;
			} else {
			    DEBUG_EXECUTE_r(
				PerlIO_printf(Perl_debug_log,
				    "%sString does not contain required trailing substring, cannot match.%s\n",
				    PL_colors[4], PL_colors[5]));
			    goto phooey;
			}
		    } else {
			/* multiline match, so we have to search for a place where the full string is located */
			goto find_last;
		    }
		} else {
		  find_last:
		    if (len)
			last = rninstr(s, strend, little, little + len);
		    else
			last = strend;	/* matching "$" */
		}
	    }
	    if (!last) {
		/* at one point this block contained a comment which was probably
		 * incorrect, which said that this was a "should not happen" case.
		 * Even if it was true when it was written I am pretty sure it is
		 * not anymore, so I have removed the comment and replaced it with
		 * this one. Yves */
		DEBUG_EXECUTE_r(
		    PerlIO_printf(Perl_debug_log,
			"String does not contain required substring, cannot match.\n"
	            ));
		goto phooey;
	    }
	    dontbother = strend - last + prog->float_min_offset;
	}
	if (minlen && (dontbother < minlen))
	    dontbother = minlen - 1;
	strend -= dontbother; 		   /* this one's always in bytes! */
	/* We don't know much -- general case. */
	if (utf8_target) {
	    for (;;) {
		if (regtry(&reginfo, &s))
		    goto got_it;
		if (s >= strend)
		    break;
		s += UTF8SKIP(s);
	    };
	}
	else {
	    do {
		if (regtry(&reginfo, &s))
		    goto got_it;
	    } while (s++ < strend);
	}
    }

    /* Failure. */
    goto phooey;

got_it:
    Safefree(swap);
    RX_MATCH_TAINTED_set(rx, PL_reg_flags & RF_tainted);

    if (PL_reg_eval_set)
	restore_pos(aTHX_ prog);
    if (RXp_PAREN_NAMES(prog)) 
        (void)hv_iterinit(RXp_PAREN_NAMES(prog));

    /* make sure $`, $&, $', and $digit will work later */
    if ( !(flags & REXEC_NOT_FIRST) ) {
	RX_MATCH_COPY_FREE(rx);
	if (flags & REXEC_COPY_STR) {
	    const I32 i = PL_regeol - startpos + (stringarg - strbeg);
#ifdef PERL_OLD_COPY_ON_WRITE
	    if ((SvIsCOW(sv)
		 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
		if (DEBUG_C_TEST) {
		    PerlIO_printf(Perl_debug_log,
				  "Copy on write: regexp capture, type %d\n",
				  (int) SvTYPE(sv));
		}
		prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
		prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
		assert (SvPOKp(prog->saved_copy));
	    } else
#endif
	    {
		RX_MATCH_COPIED_on(rx);
		s = savepvn(strbeg, i);
		prog->subbeg = s;
	    }
	    prog->sublen = i;
	}
	else {
	    prog->subbeg = strbeg;
	    prog->sublen = PL_regeol - strbeg;	/* strend may have been modified */
	}
    }

    return 1;

phooey:
    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
			  PL_colors[4], PL_colors[5]));
    if (PL_reg_eval_set)
	restore_pos(aTHX_ prog);
    if (swap) {
        /* we failed :-( roll it back */
        Safefree(prog->offs);
        prog->offs = swap;
    }

    return 0;
}


/*
 - regtry - try match at specific point
 */
STATIC I32			/* 0 failure, 1 success */
S_regtry(pTHX_ regmatch_info *reginfo, char **startpos)
{
    dVAR;
    CHECKPOINT lastcp;
    REGEXP *const rx = reginfo->prog;
    regexp *const prog = (struct regexp *)SvANY(rx);
    RXi_GET_DECL(prog,progi);
    GET_RE_DEBUG_FLAGS_DECL;

    PERL_ARGS_ASSERT_REGTRY;

    reginfo->cutpoint=NULL;

    if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) {
	MAGIC *mg;

	PL_reg_eval_set = RS_init;
	DEBUG_EXECUTE_r(DEBUG_s(
	    PerlIO_printf(Perl_debug_log, "  setting stack tmpbase at %"IVdf"\n",
			  (IV)(PL_stack_sp - PL_stack_base));
	    ));
	SAVESTACK_CXPOS();
	cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
	/* Otherwise OP_NEXTSTATE will free whatever on stack now.  */
	SAVETMPS;
	/* Apparently this is not needed, judging by wantarray. */
	/* SAVEI8(cxstack[cxstack_ix].blk_gimme);
	   cxstack[cxstack_ix].blk_gimme = G_SCALAR; */

	if (reginfo->sv) {
	    /* Make $_ available to executed code. */
	    if (reginfo->sv != DEFSV) {
		SAVE_DEFSV;
		DEFSV_set(reginfo->sv);
	    }
	
	    if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv)
		  && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) {
		/* prepare for quick setting of pos */
#ifdef PERL_OLD_COPY_ON_WRITE
		if (SvIsCOW(reginfo->sv))
		    sv_force_normal_flags(reginfo->sv, 0);
#endif
		mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global,
				 &PL_vtbl_mglob, NULL, 0);
		mg->mg_len = -1;
	    }
	    PL_reg_magic    = mg;
	    PL_reg_oldpos   = mg->mg_len;
	    SAVEDESTRUCTOR_X(restore_pos, prog);
        }
        if (!PL_reg_curpm) {
	    Newxz(PL_reg_curpm, 1, PMOP);
#ifdef USE_ITHREADS
            {
		SV* const repointer = &PL_sv_undef;
                /* this regexp is also owned by the new PL_reg_curpm, which
		   will try to free it.  */
                av_push(PL_regex_padav, repointer);
                PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
                PL_regex_pad = AvARRAY(PL_regex_padav);
            }
#endif      
        }
#ifdef USE_ITHREADS
	/* It seems that non-ithreads works both with and without this code.
	   So for efficiency reasons it seems best not to have the code
	   compiled when it is not needed.  */
	/* This is safe against NULLs: */
	ReREFCNT_dec(PM_GETRE(PL_reg_curpm));
	/* PM_reg_curpm owns a reference to this regexp.  */
	(void)ReREFCNT_inc(rx);
#endif
	PM_SETRE(PL_reg_curpm, rx);
	PL_reg_oldcurpm = PL_curpm;
	PL_curpm = PL_reg_curpm;
	if (RXp_MATCH_COPIED(prog)) {
	    /*  Here is a serious problem: we cannot rewrite subbeg,
		since it may be needed if this match fails.  Thus
		$` inside (?{}) could fail... */
	    PL_reg_oldsaved = prog->subbeg;
	    PL_reg_oldsavedlen = prog->sublen;
#ifdef PERL_OLD_COPY_ON_WRITE
	    PL_nrs = prog->saved_copy;
#endif
	    RXp_MATCH_COPIED_off(prog);
	}
	else
	    PL_reg_oldsaved = NULL;
	prog->subbeg = PL_bostr;
	prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
    }
    DEBUG_EXECUTE_r(PL_reg_starttry = *startpos);
    prog->offs[0].start = *startpos - PL_bostr;
    PL_reginput = *startpos;
    PL_reglastparen = &prog->lastparen;
    PL_reglastcloseparen = &prog->lastcloseparen;
    prog->lastparen = 0;
    prog->lastcloseparen = 0;
    PL_regsize = 0;
    PL_regoffs = prog->offs;
    if (PL_reg_start_tmpl <= prog->nparens) {
	PL_reg_start_tmpl = prog->nparens*3/2 + 3;
        if(PL_reg_start_tmp)
            Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
        else
            Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
    }

    /* XXXX What this code is doing here?!!!  There should be no need
       to do this again and again, PL_reglastparen should take care of
       this!  --ilya*/

    /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
     * Actually, the code in regcppop() (which Ilya may be meaning by
     * PL_reglastparen), is not needed at all by the test suite
     * (op/regexp, op/pat, op/split), but that code is needed otherwise
     * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
     * Meanwhile, this code *is* needed for the
     * above-mentioned test suite tests to succeed.  The common theme
     * on those tests seems to be returning null fields from matches.
     * --jhi updated by dapm */
#if 1
    if (prog->nparens) {
	regexp_paren_pair *pp = PL_regoffs;
	register I32 i;
	for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
	    ++pp;
	    pp->start = -1;
	    pp->end = -1;
	}
    }
#endif
    REGCP_SET(lastcp);
    if (regmatch(reginfo, progi->program + 1)) {
	PL_regoffs[0].end = PL_reginput - PL_bostr;
	return 1;
    }
    if (reginfo->cutpoint)
        *startpos= reginfo->cutpoint;
    REGCP_UNWIND(lastcp);
    return 0;
}


#define sayYES goto yes
#define sayNO goto no
#define sayNO_SILENT goto no_silent

/* we dont use STMT_START/END here because it leads to 
   "unreachable code" warnings, which are bogus, but distracting. */
#define CACHEsayNO \
    if (ST.cache_mask) \
       PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \
    sayNO

/* this is used to determine how far from the left messages like
   'failed...' are printed. It should be set such that messages 
   are inline with the regop output that created them.
*/
#define REPORT_CODE_OFF 32


#define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
#define CHRTEST_VOID   -1000 /* the c1/c2 "next char" test should be skipped */

#define SLAB_FIRST(s) (&(s)->states[0])
#define SLAB_LAST(s)  (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])

/* grab a new slab and return the first slot in it */

STATIC regmatch_state *
S_push_slab(pTHX)
{
#if PERL_VERSION < 9 && !defined(PERL_CORE)
    dMY_CXT;
#endif
    regmatch_slab *s = PL_regmatch_slab->next;
    if (!s) {
	Newx(s, 1, regmatch_slab);
	s->prev = PL_regmatch_slab;
	s->next = NULL;
	PL_regmatch_slab->next = s;
    }
    PL_regmatch_slab = s;
    return SLAB_FIRST(s);
}


/* push a new state then goto it */

#define PUSH_STATE_GOTO(state, node) \
    scan = node; \
    st->resume_state = state; \
    goto push_state;

/* push a new state with success backtracking, then goto it */

#define PUSH_YES_STATE_GOTO(state, node) \
    scan = node; \
    st->resume_state = state; \
    goto push_yes_state;



/*

regmatch() - main matching routine

This is basically one big switch statement in a loop. We execute an op,
set 'next' to point the next op, and continue. If we come to a point which
we may need to backtrack to on failure such as (A|B|C), we push a
backtrack state onto the backtrack stack. On failure, we pop the top
state, and re-enter the loop at the state indicated. If there are no more
states to pop, we return failure.

Sometimes we also need to backtrack on success; for example /A+/, where
after successfully matching one A, we need to go back and try to
match another one; similarly for lookahead assertions: if the assertion
completes successfully, we backtrack to the state just before the assertion
and then carry on.  In these cases, the pushed state is marked as
'backtrack on success too'. This marking is in fact done by a chain of
pointers, each pointing to the previous 'yes' state. On success, we pop to
the nearest yes state, discarding any intermediate failure-only states.
Sometimes a yes state is pushed just to force some cleanup code to be
called at the end of a successful match or submatch; e.g. (??{$re}) uses
it to free the inner regex.

Note that failure backtracking rewinds the cursor position, while
success backtracking leaves it alone.

A pattern is complete when the END op is executed, while a subpattern
such as (?=foo) is complete when the SUCCESS op is executed. Both of these
ops trigger the "pop to last yes state if any, otherwise return true"
behaviour.

A common convention in this function is to use A and B to refer to the two
subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
the subpattern to be matched possibly multiple times, while B is the entire
rest of the pattern. Variable and state names reflect this convention.

The states in the main switch are the union of ops and failure/success of
substates associated with with that op.  For example, IFMATCH is the op
that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
successfully matched A and IFMATCH_A_fail is a state saying that we have
just failed to match A. Resume states always come in pairs. The backtrack
state we push is marked as 'IFMATCH_A', but when that is popped, we resume
at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
on success or failure.

The struct that holds a backtracking state is actually a big union, with
one variant for each major type of op. The variable st points to the
top-most backtrack struct. To make the code clearer, within each
block of code we #define ST to alias the relevant union.

Here's a concrete example of a (vastly oversimplified) IFMATCH
implementation:

    switch (state) {
    ....

#define ST st->u.ifmatch

    case IFMATCH: // we are executing the IFMATCH op, (?=A)B
	ST.foo = ...; // some state we wish to save
	...
	// push a yes backtrack state with a resume value of
	// IFMATCH_A/IFMATCH_A_fail, then continue execution at the
	// first node of A:
	PUSH_YES_STATE_GOTO(IFMATCH_A, A);
	// NOTREACHED

    case IFMATCH_A: // we have successfully executed A; now continue with B
	next = B;
	bar = ST.foo; // do something with the preserved value
	break;

    case IFMATCH_A_fail: // A failed, so the assertion failed
	...;   // do some housekeeping, then ...
	sayNO; // propagate the failure

#undef ST

    ...
    }

For any old-timers reading this who are familiar with the old recursive
approach, the code above is equivalent to:

    case IFMATCH: // we are executing the IFMATCH op, (?=A)B
    {
	int foo = ...
	...
	if (regmatch(A)) {
	    next = B;
	    bar = foo;
	    break;
	}
	...;   // do some housekeeping, then ...
	sayNO; // propagate the failure
    }

The topmost backtrack state, pointed to by st, is usually free. If you
want to claim it, populate any ST.foo fields in it with values you wish to
save, then do one of

	PUSH_STATE_GOTO(resume_state, node);
	PUSH_YES_STATE_GOTO(resume_state, node);

which sets that backtrack state's resume value to 'resume_state', pushes a
new free entry to the top of the backtrack stack, then goes to 'node'.
On backtracking, the free slot is popped, and the saved state becomes the
new free state. An ST.foo field in this new top state can be temporarily
accessed to retrieve values, but once the main loop is re-entered, it
becomes available for reuse.

Note that the depth of the backtrack stack constantly increases during the
left-to-right execution of the pattern, rather than going up and down with
the pattern nesting. For example the stack is at its maximum at Z at the
end of the pattern, rather than at X in the following:

    /(((X)+)+)+....(Y)+....Z/

The only exceptions to this are lookahead/behind assertions and the cut,
(?>A), which pop all the backtrack states associated with A before
continuing.
 
Backtrack state structs are allocated in slabs of about 4K in size.
PL_regmatch_state and st always point to the currently active state,
and PL_regmatch_slab points to the slab currently containing
PL_regmatch_state.  The first time regmatch() is called, the first slab is
allocated, and is never freed until interpreter destruction. When the slab
is full, a new one is allocated and chained to the end. At exit from
regmatch(), slabs allocated since entry are freed.

*/
 

#define DEBUG_STATE_pp(pp)				    \
    DEBUG_STATE_r({					    \
	DUMP_EXEC_POS(locinput, scan, utf8_target);		    \
	PerlIO_printf(Perl_debug_log,			    \
	    "    %*s"pp" %s%s%s%s%s\n",			    \
	    depth*2, "",				    \
	    PL_reg_name[st->resume_state],                     \
	    ((st==yes_state||st==mark_state) ? "[" : ""),   \
	    ((st==yes_state) ? "Y" : ""),                   \
	    ((st==mark_state) ? "M" : ""),                  \
	    ((st==yes_state||st==mark_state) ? "]" : "")    \
	);                                                  \
    });


#define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)

#ifdef DEBUGGING

STATIC void
S_debug_start_match(pTHX_ const REGEXP *prog, const bool utf8_target,
    const char *start, const char *end, const char *blurb)
{
    const bool utf8_pat = RX_UTF8(prog) ? 1 : 0;

    PERL_ARGS_ASSERT_DEBUG_START_MATCH;

    if (!PL_colorset)   
            reginitcolors();    
    {
        RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), 
            RX_PRECOMP_const(prog), RX_PRELEN(prog), 60);   
        
        RE_PV_QUOTED_DECL(s1, utf8_target, PERL_DEBUG_PAD_ZERO(1),
            start, end - start, 60); 
        
        PerlIO_printf(Perl_debug_log, 
            "%s%s REx%s %s against %s\n", 
		       PL_colors[4], blurb, PL_colors[5], s0, s1); 
        
        if (utf8_target||utf8_pat)
            PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
                utf8_pat ? "pattern" : "",
                utf8_pat && utf8_target ? " and " : "",
                utf8_target ? "string" : ""
            ); 
    }
}

STATIC void
S_dump_exec_pos(pTHX_ const char *locinput, 
                      const regnode *scan, 
                      const char *loc_regeol, 
                      const char *loc_bostr, 
                      const char *loc_reg_starttry,
                      const bool utf8_target)
{
    const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
    const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
    int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
    /* The part of the string before starttry has one color
       (pref0_len chars), between starttry and current
       position another one (pref_len - pref0_len chars),
       after the current position the third one.
       We assume that pref0_len <= pref_len, otherwise we
       decrease pref0_len.  */
    int pref_len = (locinput - loc_bostr) > (5 + taill) - l
	? (5 + taill) - l : locinput - loc_bostr;
    int pref0_len;

    PERL_ARGS_ASSERT_DUMP_EXEC_POS;

    while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
	pref_len++;
    pref0_len = pref_len  - (locinput - loc_reg_starttry);
    if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
	l = ( loc_regeol - locinput > (5 + taill) - pref_len
	      ? (5 + taill) - pref_len : loc_regeol - locinput);
    while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
	l--;
    if (pref0_len < 0)
	pref0_len = 0;
    if (pref0_len > pref_len)
	pref0_len = pref_len;
    {
	const int is_uni = (utf8_target && OP(scan) != CANY) ? 1 : 0;

	RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
	    (locinput - pref_len),pref0_len, 60, 4, 5);
	
	RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
		    (locinput - pref_len + pref0_len),
		    pref_len - pref0_len, 60, 2, 3);
	
	RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
		    locinput, loc_regeol - locinput, 10, 0, 1);

	const STRLEN tlen=len0+len1+len2;
	PerlIO_printf(Perl_debug_log,
		    "%4"IVdf" <%.*s%.*s%s%.*s>%*s|",
		    (IV)(locinput - loc_bostr),
		    len0, s0,
		    len1, s1,
		    (docolor ? "" : "> <"),
		    len2, s2,
		    (int)(tlen > 19 ? 0 :  19 - tlen),
		    "");
    }
}

#endif

/* reg_check_named_buff_matched()
 * Checks to see if a named buffer has matched. The data array of 
 * buffer numbers corresponding to the buffer is expected to reside
 * in the regexp->data->data array in the slot stored in the ARG() of
 * node involved. Note that this routine doesn't actually care about the
 * name, that information is not preserved from compilation to execution.
 * Returns the index of the leftmost defined buffer with the given name
 * or 0 if non of the buffers matched.
 */
STATIC I32
S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan)
{
    I32 n;
    RXi_GET_DECL(rex,rexi);
    SV *sv_dat= MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
    I32 *nums=(I32*)SvPVX(sv_dat);

    PERL_ARGS_ASSERT_REG_CHECK_NAMED_BUFF_MATCHED;

    for ( n=0; n<SvIVX(sv_dat); n++ ) {
        if ((I32)*PL_reglastparen >= nums[n] &&
            PL_regoffs[nums[n]].end != -1)
        {
            return nums[n];
        }
    }
    return 0;
}


/* free all slabs above current one  - called during LEAVE_SCOPE */

STATIC void
S_clear_backtrack_stack(pTHX_ void *p)
{
    regmatch_slab *s = PL_regmatch_slab->next;
    PERL_UNUSED_ARG(p);

    if (!s)
	return;
    PL_regmatch_slab->next = NULL;
    while (s) {
	regmatch_slab * const osl = s;
	s = s->next;
	Safefree(osl);
    }
}


#define SETREX(Re1,Re2) \
    if (PL_reg_eval_set) PM_SETRE((PL_reg_curpm), (Re2)); \
    Re1 = (Re2)

STATIC I32			/* 0 failure, 1 success */
S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
{
#if PERL_VERSION < 9 && !defined(PERL_CORE)
    dMY_CXT;
#endif
    dVAR;
    register const bool utf8_target = PL_reg_match_utf8;
    const U32 uniflags = UTF8_ALLOW_DEFAULT;
    REGEXP *rex_sv = reginfo->prog;
    regexp *rex = (struct regexp *)SvANY(rex_sv);
    RXi_GET_DECL(rex,rexi);
    I32	oldsave;
    /* the current state. This is a cached copy of PL_regmatch_state */
    register regmatch_state *st;
    /* cache heavy used fields of st in registers */
    register regnode *scan;
    register regnode *next;
    register U32 n = 0;	/* general value; init to avoid compiler warning */
    register I32 ln = 0; /* len or last;  init to avoid compiler warning */
    register char *locinput = PL_reginput;
    register I32 nextchr;   /* is always set to UCHARAT(locinput) */

    bool result = 0;	    /* return value of S_regmatch */
    int depth = 0;	    /* depth of backtrack stack */
    U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */
    const U32 max_nochange_depth =
        (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ?
        3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH;
    regmatch_state *yes_state = NULL; /* state to pop to on success of
							    subpattern */
    /* mark_state piggy backs on the yes_state logic so that when we unwind 
       the stack on success we can update the mark_state as we go */
    regmatch_state *mark_state = NULL; /* last mark state we have seen */
    regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
    struct regmatch_state  *cur_curlyx = NULL; /* most recent curlyx */
    U32 state_num;
    bool no_final = 0;      /* prevent failure from backtracking? */
    bool do_cutgroup = 0;   /* no_final only until next branch/trie entry */
    char *startpoint = PL_reginput;
    SV *popmark = NULL;     /* are we looking for a mark? */
    SV *sv_commit = NULL;   /* last mark name seen in failure */
    SV *sv_yes_mark = NULL; /* last mark name we have seen 
                               during a successful match */
    U32 lastopen = 0;       /* last open we saw */
    bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;   
    SV* const oreplsv = GvSV(PL_replgv);
    /* these three flags are set by various ops to signal information to
     * the very next op. They have a useful lifetime of exactly one loop
     * iteration, and are not preserved or restored by state pushes/pops
     */
    bool sw = 0;	    /* the condition value in (?(cond)a|b) */
    bool minmod = 0;	    /* the next "{n,m}" is a "{n,m}?" */
    int logical = 0;	    /* the following EVAL is:
				0: (?{...})
				1: (?(?{...})X|Y)
				2: (??{...})
			       or the following IFMATCH/UNLESSM is:
			        false: plain (?=foo)
				true:  used as a condition: (?(?=foo))
			    */
#ifdef DEBUGGING
    GET_RE_DEBUG_FLAGS_DECL;
#endif

    PERL_ARGS_ASSERT_REGMATCH;

    DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({
	    PerlIO_printf(Perl_debug_log,"regmatch start\n");
    }));
    /* on first ever call to regmatch, allocate first slab */
    if (!PL_regmatch_slab) {
	Newx(PL_regmatch_slab, 1, regmatch_slab);
	PL_regmatch_slab->prev = NULL;
	PL_regmatch_slab->next = NULL;
	PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
    }

    oldsave = PL_savestack_ix;
    SAVEDESTRUCTOR_X(S_clear_backtrack_stack, NULL);
    SAVEVPTR(PL_regmatch_slab);
    SAVEVPTR(PL_regmatch_state);

    /* grab next free state slot */
    st = ++PL_regmatch_state;
    if (st >  SLAB_LAST(PL_regmatch_slab))
	st = PL_regmatch_state = S_push_slab(aTHX);

    /* Note that nextchr is a byte even in UTF */
    nextchr = UCHARAT(locinput);
    scan = prog;
    while (scan != NULL) {

        DEBUG_EXECUTE_r( {
	    SV * const prop = sv_newmortal();
	    regnode *rnext=regnext(scan);
	    DUMP_EXEC_POS( locinput, scan, utf8_target );
	    regprop(rex, prop, scan);
            
	    PerlIO_printf(Perl_debug_log,
		    "%3"IVdf":%*s%s(%"IVdf")\n",
		    (IV)(scan - rexi->program), depth*2, "",
		    SvPVX_const(prop),
		    (PL_regkind[OP(scan)] == END || !rnext) ? 
		        0 : (IV)(rnext - rexi->program));
	});

	next = scan + NEXT_OFF(scan);
	if (next == scan)
	    next = NULL;
	state_num = OP(scan);

      reenter_switch:

	assert(PL_reglastparen == &rex->lastparen);
	assert(PL_reglastcloseparen == &rex->lastcloseparen);
	assert(PL_regoffs == rex->offs);

	switch (state_num) {
	case BOL:
	    if (locinput == PL_bostr)
	    {
		/* reginfo->till = reginfo->bol; */
		break;
	    }
	    sayNO;
	case MBOL:
	    if (locinput == PL_bostr ||
		((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
	    {
		break;
	    }
	    sayNO;
	case SBOL:
	    if (locinput == PL_bostr)
		break;
	    sayNO;
	case GPOS:
	    if (locinput == reginfo->ganch)
		break;
	    sayNO;

	case KEEPS:
	    /* update the startpoint */
	    st->u.keeper.val = PL_regoffs[0].start;
	    PL_reginput = locinput;
	    PL_regoffs[0].start = locinput - PL_bostr;
	    PUSH_STATE_GOTO(KEEPS_next, next);
	    /*NOT-REACHED*/
	case KEEPS_next_fail:
	    /* rollback the start point change */
	    PL_regoffs[0].start = st->u.keeper.val;
	    sayNO_SILENT;
	    /*NOT-REACHED*/
	case EOL:
		goto seol;
	case MEOL:
	    if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
		sayNO;
	    break;
	case SEOL:
	  seol:
	    if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
		sayNO;
	    if (PL_regeol - locinput > 1)
		sayNO;
	    break;
	case EOS:
	    if (PL_regeol != locinput)
		sayNO;
	    break;
	case SANY:
	    if (!nextchr && locinput >= PL_regeol)
		sayNO;
 	    if (utf8_target) {
	        locinput += PL_utf8skip[nextchr];
		if (locinput > PL_regeol)
 		    sayNO;
 		nextchr = UCHARAT(locinput);
 	    }
 	    else
 		nextchr = UCHARAT(++locinput);
	    break;
	case CANY:
	    if (!nextchr && locinput >= PL_regeol)
		sayNO;
	    nextchr = UCHARAT(++locinput);
	    break;
	case REG_ANY:
	    if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
		sayNO;
	    if (utf8_target) {
		locinput += PL_utf8skip[nextchr];
		if (locinput > PL_regeol)
		    sayNO;
		nextchr = UCHARAT(locinput);
	    }
	    else
		nextchr = UCHARAT(++locinput);
	    break;

#undef  ST
#define ST st->u.trie
        case TRIEC:
            /* In this case the charclass data is available inline so
               we can fail fast without a lot of extra overhead. 
             */
            if(!ANYOF_BITMAP_TEST(scan, *locinput)) {
                DEBUG_EXECUTE_r(
                    PerlIO_printf(Perl_debug_log,
                              "%*s  %sfailed to match trie start class...%s\n",
                              REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
                );
                sayNO_SILENT;
                /* NOTREACHED */
            }
            /* FALL THROUGH */
	case TRIE:
	    /* the basic plan of execution of the trie is:
	     * At the beginning, run though all the states, and
	     * find the longest-matching word. Also remember the position
	     * of the shortest matching word. For example, this pattern:
	     *    1  2 3 4    5
	     *    ab|a|x|abcd|abc
	     * when matched against the string "abcde", will generate
	     * accept states for all words except 3, with the longest
	     * matching word being 4, and the shortest being 1 (with
	     * the position being after char 1 of the string).
	     *
	     * Then for each matching word, in word order (i.e. 1,2,4,5),
	     * we run the remainder of the pattern; on each try setting
	     * the current position to the character following the word,
	     * returning to try the next word on failure.
	     *
	     * We avoid having to build a list of words at runtime by
	     * using a compile-time structure, wordinfo[].prev, which
	     * gives, for each word, the previous accepting word (if any).
	     * In the case above it would contain the mappings 1->2, 2->0,
	     * 3->0, 4->5, 5->1.  We can use this table to generate, from
	     * the longest word (4 above), a list of all words, by
	     * following the list of prev pointers; this gives us the
	     * unordered list 4,5,1,2. Then given the current word we have
	     * just tried, we can go through the list and find the
	     * next-biggest word to try (so if we just failed on word 2,
	     * the next in the list is 4).
	     *
	     * Since at runtime we don't record the matching position in
	     * the string for each word, we have to work that out for
	     * each word we're about to process. The wordinfo table holds
	     * the character length of each word; given that we recorded
	     * at the start: the position of the shortest word and its
	     * length in chars, we just need to move the pointer the
	     * difference between the two char lengths. Depending on
	     * Unicode status and folding, that's cheap or expensive.
	     *
	     * This algorithm is optimised for the case where are only a
	     * small number of accept states, i.e. 0,1, or maybe 2.
	     * With lots of accepts states, and having to try all of them,
	     * it becomes quadratic on number of accept states to find all
	     * the next words.
	     */

	    {
                /* what type of TRIE am I? (utf8 makes this contextual) */
                DECL_TRIE_TYPE(scan);

                /* what trie are we using right now */
		reg_trie_data * const trie
        	    = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
		HV * widecharmap = MUTABLE_HV(rexi->data->data[ ARG( scan ) + 1 ]);
                U32 state = trie->startstate;

                if (trie->bitmap && !TRIE_BITMAP_TEST(trie,*locinput) ) {
        	    if (trie->states[ state ].wordnum) {
        	         DEBUG_EXECUTE_r(
                            PerlIO_printf(Perl_debug_log,
                        	          "%*s  %smatched empty string...%s\n",
                        	          REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
                        );
			if (!trie->jump)
			    break;
        	    } else {
        	        DEBUG_EXECUTE_r(
                            PerlIO_printf(Perl_debug_log,
                        	          "%*s  %sfailed to match trie start class...%s\n",
                        	          REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
                        );
        	        sayNO_SILENT;
        	   }
                }

            { 
		U8 *uc = ( U8* )locinput;

		STRLEN len = 0;
		STRLEN foldlen = 0;
		U8 *uscan = (U8*)NULL;
		U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
		U32 charcount = 0; /* how many input chars we have matched */
		U32 accepted = 0; /* have we seen any accepting states? */

		ST.B = next;
		ST.jump = trie->jump;
		ST.me = scan;
		ST.firstpos = NULL;
		ST.longfold = FALSE; /* char longer if folded => it's harder */
		ST.nextword = 0;

		/* fully traverse the TRIE; note the position of the
		   shortest accept state and the wordnum of the longest
		   accept state */

		while ( state && uc <= (U8*)PL_regeol ) {
                    U32 base = trie->states[ state ].trans.base;
                    UV uvc = 0;
                    U16 charid = 0;
		    U16 wordnum;
                    wordnum = trie->states[ state ].wordnum;

		    if (wordnum) { /* it's an accept state */
			if (!accepted) {
			    accepted = 1;
			    /* record first match position */
			    if (ST.longfold) {
				ST.firstpos = (U8*)locinput;
				ST.firstchars = 0;
			    }
			    else {
				ST.firstpos = uc;
				ST.firstchars = charcount;
			    }
			}
			if (!ST.nextword || wordnum < ST.nextword)
			    ST.nextword = wordnum;
			ST.topword = wordnum;
		    }

		    DEBUG_TRIE_EXECUTE_r({
		                DUMP_EXEC_POS( (char *)uc, scan, utf8_target );
			        PerlIO_printf( Perl_debug_log,
			            "%*s  %sState: %4"UVxf" Accepted: %c ",
			            2+depth * 2, "", PL_colors[4],
			            (UV)state, (accepted ? 'Y' : 'N'));
		    });

		    /* read a char and goto next state */
		    if ( base ) {
			I32 offset;
			REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
					     uscan, len, uvc, charid, foldlen,
					     foldbuf, uniflags);
			charcount++;
			if (foldlen>0)
			    ST.longfold = TRUE;
			if (charid &&
			     ( ((offset =
			      base + charid - 1 - trie->uniquecharcount)) >= 0)

			     && ((U32)offset < trie->lasttrans)
			     && trie->trans[offset].check == state)
			{
			    state = trie->trans[offset].next;
			}
			else {
			    state = 0;
			}
			uc += len;

		    }
		    else {
			state = 0;
		    }
		    DEBUG_TRIE_EXECUTE_r(
		        PerlIO_printf( Perl_debug_log,
		            "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
		            charid, uvc, (UV)state, PL_colors[5] );
		    );
		}
		if (!accepted)
		   sayNO;

		/* calculate total number of accept states */
		{
		    U16 w = ST.topword;
		    accepted = 0;
		    while (w) {
			w = trie->wordinfo[w].prev;
			accepted++;
		    }
		    ST.accepted = accepted;
		}

		DEBUG_EXECUTE_r(
		    PerlIO_printf( Perl_debug_log,
			"%*s  %sgot %"IVdf" possible matches%s\n",
			REPORT_CODE_OFF + depth * 2, "",
			PL_colors[4], (IV)ST.accepted, PL_colors[5] );
		);
		goto trie_first_try; /* jump into the fail handler */
	    }}
	    /* NOTREACHED */

	case TRIE_next_fail: /* we failed - try next alternative */
            if ( ST.jump) {
                REGCP_UNWIND(ST.cp);
	        for (n = *PL_reglastparen; n > ST.lastparen; n--)
		    PL_regoffs[n].end = -1;
	        *PL_reglastparen = n;
	    }
	    if (!--ST.accepted) {
	        DEBUG_EXECUTE_r({
		    PerlIO_printf( Perl_debug_log,
			"%*s  %sTRIE failed...%s\n",
			REPORT_CODE_OFF+depth*2, "", 
			PL_colors[4],
			PL_colors[5] );
		});
		sayNO_SILENT;
	    }
	    {
		/* Find next-highest word to process.  Note that this code
		 * is O(N^2) per trie run (O(N) per branch), so keep tight */
		register U16 min = 0;
		register U16 word;
		register U16 const nextword = ST.nextword;
		register reg_trie_wordinfo * const wordinfo
		    = ((reg_trie_data*)rexi->data->data[ARG(ST.me)])->wordinfo;
		for (word=ST.topword; word; word=wordinfo[word].prev) {
		    if (word > nextword && (!min || word < min))
			min = word;
		}
		ST.nextword = min;
	    }

          trie_first_try:
            if (do_cutgroup) {
                do_cutgroup = 0;
                no_final = 0;
            }

            if ( ST.jump) {
                ST.lastparen = *PL_reglastparen;
	        REGCP_SET(ST.cp);
            }

	    /* find start char of end of current word */
	    {
		U32 chars; /* how many chars to skip */
		U8 *uc = ST.firstpos;
		reg_trie_data * const trie
		    = (reg_trie_data*)rexi->data->data[ARG(ST.me)];

		assert((trie->wordinfo[ST.nextword].len - trie->prefixlen)
			    >=  ST.firstchars);
		chars = (trie->wordinfo[ST.nextword].len - trie->prefixlen)
			    - ST.firstchars;

		if (ST.longfold) {
		    /* the hard option - fold each char in turn and find
		     * its folded length (which may be different */
		    U8 foldbuf[UTF8_MAXBYTES_CASE + 1];
		    STRLEN foldlen;
		    STRLEN len;
		    UV uvc;
		    U8 *uscan;

		    while (chars) {
			if (utf8_target) {
			    uvc = utf8n_to_uvuni((U8*)uc, UTF8_MAXLEN, &len,
						    uniflags);
			    uc += len;
			}
			else {
			    uvc = *uc;
			    uc++;
			}
			uvc = to_uni_fold(uvc, foldbuf, &foldlen);
			uscan = foldbuf;
			while (foldlen) {
			    if (!--chars)
				break;
			    uvc = utf8n_to_uvuni(uscan, UTF8_MAXLEN, &len,
					    uniflags);
			    uscan += len;
			    foldlen -= len;
			}
		    }
		}
		else {
		    if (utf8_target)
			while (chars--)
			    uc += UTF8SKIP(uc);
		    else
			uc += chars;
		}
		PL_reginput = (char *)uc;
	    }

	    scan = (ST.jump && ST.jump[ST.nextword]) 
			? ST.me + ST.jump[ST.nextword]
			: ST.B;

	    DEBUG_EXECUTE_r({
		PerlIO_printf( Perl_debug_log,
		    "%*s  %sTRIE matched word #%d, continuing%s\n",
		    REPORT_CODE_OFF+depth*2, "", 
		    PL_colors[4],
		    ST.nextword,
		    PL_colors[5]
		    );
	    });

	    if (ST.accepted > 1 || has_cutgroup) {
		PUSH_STATE_GOTO(TRIE_next, scan);
		/* NOTREACHED */
	    }
	    /* only one choice left - just continue */
	    DEBUG_EXECUTE_r({
		AV *const trie_words
		    = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]);
		SV ** const tmp = av_fetch( trie_words,
		    ST.nextword-1, 0 );
		SV *sv= tmp ? sv_newmortal() : NULL;

		PerlIO_printf( Perl_debug_log,
		    "%*s  %sonly one match left, short-circuiting: #%d <%s>%s\n",
		    REPORT_CODE_OFF+depth*2, "", PL_colors[4],
		    ST.nextword,
		    tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
			    PL_colors[0], PL_colors[1],
			    (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)|PERL_PV_ESCAPE_NONASCII
			) 
		    : "not compiled under -Dr",
		    PL_colors[5] );
	    });

	    locinput = PL_reginput;
	    nextchr = UCHARAT(locinput);
	    continue; /* execute rest of RE */
	    /* NOTREACHED */
#undef  ST

	case EXACT: {
	    char *s = STRING(scan);
	    ln = STR_LEN(scan);
	    if (utf8_target != UTF_PATTERN) {
		/* The target and the pattern have differing utf8ness. */
		char *l = locinput;
		const char * const e = s + ln;

		if (utf8_target) {
		    /* The target is utf8, the pattern is not utf8. */
		    while (s < e) {
			STRLEN ulen;
			if (l >= PL_regeol)
			     sayNO;
			if (NATIVE_TO_UNI(*(U8*)s) !=
			    utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
					    uniflags))
			     sayNO;
			l += ulen;
			s ++;
		    }
		}
		else {
		    /* The target is not utf8, the pattern is utf8. */
		    while (s < e) {
			STRLEN ulen;
			if (l >= PL_regeol)
			    sayNO;
			if (NATIVE_TO_UNI(*((U8*)l)) !=
			    utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
					   uniflags))
			    sayNO;
			s += ulen;
			l ++;
		    }
		}
		locinput = l;
		nextchr = UCHARAT(locinput);
		break;
	    }
	    /* The target and the pattern have the same utf8ness. */
	    /* Inline the first character, for speed. */
	    if (UCHARAT(s) != nextchr)
		sayNO;
	    if (PL_regeol - locinput < ln)
		sayNO;
	    if (ln > 1 && memNE(s, locinput, ln))
		sayNO;
	    locinput += ln;
	    nextchr = UCHARAT(locinput);
	    break;
	    }
	case EXACTFL: {
	    re_fold_t folder;
	    const U8 * fold_array;
	    const char * s;
	    U32 fold_utf8_flags;

	    PL_reg_flags |= RF_tainted;
	    folder = foldEQ_locale;
	    fold_array = PL_fold_locale;
	    fold_utf8_flags = FOLDEQ_UTF8_LOCALE;
	    goto do_exactf;

	case EXACTFU_SS:
	case EXACTFU_TRICKYFOLD:
	case EXACTFU:
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    fold_utf8_flags = (UTF_PATTERN) ? FOLDEQ_S1_ALREADY_FOLDED : 0;
	    goto do_exactf;

	case EXACTFA:
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    fold_utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
	    goto do_exactf;

	case EXACTF:
	    folder = foldEQ;
	    fold_array = PL_fold;
	    fold_utf8_flags = 0;

	  do_exactf:
	    s = STRING(scan);
	    ln = STR_LEN(scan);

	    if (utf8_target || UTF_PATTERN || state_num == EXACTFU_SS) {
	      /* Either target or the pattern are utf8, or has the issue where
	       * the fold lengths may differ. */
		const char * const l = locinput;
		char *e = PL_regeol;

		if (! foldEQ_utf8_flags(s, 0,  ln, cBOOL(UTF_PATTERN),
			                l, &e, 0,  utf8_target, fold_utf8_flags))
		{
		    sayNO;
		}
		locinput = e;
		nextchr = UCHARAT(locinput);
		break;
	    }

	    /* Neither the target nor the pattern are utf8 */
	    if (UCHARAT(s) != nextchr &&
		UCHARAT(s) != fold_array[nextchr])
	    {
		sayNO;
	    }
	    if (PL_regeol - locinput < ln)
		sayNO;
	    if (ln > 1 && ! folder(s, locinput, ln))
		sayNO;
	    locinput += ln;
	    nextchr = UCHARAT(locinput);
	    break;
	}

	/* XXX Could improve efficiency by separating these all out using a
	 * macro or in-line function.  At that point regcomp.c would no longer
	 * have to set the FLAGS fields of these */
	case BOUNDL:
	case NBOUNDL:
	    PL_reg_flags |= RF_tainted;
	    /* FALL THROUGH */
	case BOUND:
	case BOUNDU:
	case BOUNDA:
	case NBOUND:
	case NBOUNDU:
	case NBOUNDA:
	    /* was last char in word? */
	    if (utf8_target
		&& FLAGS(scan) != REGEX_ASCII_RESTRICTED_CHARSET
		&& FLAGS(scan) != REGEX_ASCII_MORE_RESTRICTED_CHARSET)
	    {
		if (locinput == PL_bostr)
		    ln = '\n';
		else {
		    const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);

		    ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags);
		}
		if (FLAGS(scan) != REGEX_LOCALE_CHARSET) {
		    ln = isALNUM_uni(ln);
		    LOAD_UTF8_CHARCLASS_ALNUM();
		    n = swash_fetch(PL_utf8_alnum, (U8*)locinput, utf8_target);
		}
		else {
		    ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln));
		    n = isALNUM_LC_utf8((U8*)locinput);
		}
	    }
	    else {

		/* Here the string isn't utf8, or is utf8 and only ascii
		 * characters are to match \w.  In the latter case looking at
		 * the byte just prior to the current one may be just the final
		 * byte of a multi-byte character.  This is ok.  There are two
		 * cases:
		 * 1) it is a single byte character, and then the test is doing
		 *	just what it's supposed to.
		 * 2) it is a multi-byte character, in which case the final
		 *	byte is never mistakable for ASCII, and so the test
		 *	will say it is not a word character, which is the
		 *	correct answer. */
		ln = (locinput != PL_bostr) ?
		    UCHARAT(locinput - 1) : '\n';
		switch (FLAGS(scan)) {
		    case REGEX_UNICODE_CHARSET:
			ln = isWORDCHAR_L1(ln);
			n = isWORDCHAR_L1(nextchr);
			break;
		    case REGEX_LOCALE_CHARSET:
			ln = isALNUM_LC(ln);
			n = isALNUM_LC(nextchr);
			break;
		    case REGEX_DEPENDS_CHARSET:
			ln = isALNUM(ln);
			n = isALNUM(nextchr);
			break;
		    case REGEX_ASCII_RESTRICTED_CHARSET:
		    case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
			ln = isWORDCHAR_A(ln);
			n = isWORDCHAR_A(nextchr);
			break;
		    default:
			Perl_croak(aTHX_ "panic: Unexpected FLAGS %u in op %u", FLAGS(scan), OP(scan));
			break;
		}
	    }
	    /* Note requires that all BOUNDs be lower than all NBOUNDs in
	     * regcomp.sym */
	    if (((!ln) == (!n)) == (OP(scan) < NBOUND))
		    sayNO;
	    break;
	case ANYOFV:
	case ANYOF:
	    if (utf8_target || state_num == ANYOFV) {
	        STRLEN inclasslen = PL_regeol - locinput;
		if (locinput >= PL_regeol)
		    sayNO;

	        if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, utf8_target))
		    sayNO;
		locinput += inclasslen;
		nextchr = UCHARAT(locinput);
		break;
	    }
	    else {
		if (nextchr < 0)
		    nextchr = UCHARAT(locinput);
		if (!nextchr && locinput >= PL_regeol)
		    sayNO;
		if (!REGINCLASS(rex, scan, (U8*)locinput))
		    sayNO;
		nextchr = UCHARAT(++locinput);
		break;
	    }
	    break;
	/* Special char classes - The defines start on line 129 or so */
        CCC_TRY_U(ALNUM,  NALNUM,  isWORDCHAR,
		  ALNUML, NALNUML, isALNUM_LC, isALNUM_LC_utf8,
		  ALNUMU, NALNUMU, isWORDCHAR_L1,
		  ALNUMA, NALNUMA, isWORDCHAR_A,
		  alnum, "a");

        CCC_TRY_U(SPACE,  NSPACE,  isSPACE,
		  SPACEL, NSPACEL, isSPACE_LC, isSPACE_LC_utf8,
		  SPACEU, NSPACEU, isSPACE_L1,
		  SPACEA, NSPACEA, isSPACE_A,
		  space, " ");

        CCC_TRY(DIGIT,  NDIGIT,  isDIGIT,
		DIGITL, NDIGITL, isDIGIT_LC, isDIGIT_LC_utf8,
		DIGITA, NDIGITA, isDIGIT_A,
		digit, "0");

	case CLUMP: /* Match \X: logical Unicode character.  This is defined as
		       a Unicode extended Grapheme Cluster */
	    /* From http://www.unicode.org/reports/tr29 (5.2 version).  An
	      extended Grapheme Cluster is:

	       CR LF
	       | Prepend* Begin Extend*
	       | .

	       Begin is (Hangul-syllable | ! Control)
	       Extend is (Grapheme_Extend | Spacing_Mark)
	       Control is [ GCB_Control CR LF ]

	       The discussion below shows how the code for CLUMP is derived
	       from this regex.  Note that most of these concepts are from
	       property values of the Grapheme Cluster Boundary (GCB) property.
	       No code point can have multiple property values for a given
	       property.  Thus a code point in Prepend can't be in Control, but
	       it must be in !Control.  This is why Control above includes
	       GCB_Control plus CR plus LF.  The latter two are used in the GCB
	       property separately, and so can't be in GCB_Control, even though
	       they logically are controls.  Control is not the same as gc=cc,
	       but includes format and other characters as well.

	       The Unicode definition of Hangul-syllable is:
		   L+
		   | (L* ( ( V | LV ) V* | LVT ) T*)
		   | T+ 
		  )
	       Each of these is a value for the GCB property, and hence must be
	       disjoint, so the order they are tested is immaterial, so the
	       above can safely be changed to
		   T+
		   | L+
		   | (L* ( LVT | ( V | LV ) V*) T*)

	       The last two terms can be combined like this:
		   L* ( L
		        | (( LVT | ( V | LV ) V*) T*))

	       And refactored into this:
		   L* (L | LVT T* | V  V* T* | LV  V* T*)

	       That means that if we have seen any L's at all we can quit
	       there, but if the next character is an LVT, a V, or an LV we
	       should keep going.

	       There is a subtlety with Prepend* which showed up in testing.
	       Note that the Begin, and only the Begin is required in:
	        | Prepend* Begin Extend*
	       Also, Begin contains '! Control'.  A Prepend must be a
	       '!  Control', which means it must also be a Begin.  What it
	       comes down to is that if we match Prepend* and then find no
	       suitable Begin afterwards, that if we backtrack the last
	       Prepend, that one will be a suitable Begin.
	    */

	    if (locinput >= PL_regeol)
		sayNO;
	    if  (! utf8_target) {

		/* Match either CR LF  or '.', as all the other possibilities
		 * require utf8 */
		locinput++;	    /* Match the . or CR */
		if (nextchr == '\r' /* And if it was CR, and the next is LF,
				       match the LF */
		    && locinput < PL_regeol
		    && UCHARAT(locinput) == '\n') locinput++;
	    }
	    else {

		/* Utf8: See if is ( CR LF ); already know that locinput <
		 * PL_regeol, so locinput+1 is in bounds */
		if (nextchr == '\r' && UCHARAT(locinput + 1) == '\n') {
		    locinput += 2;
		}
		else {
		    /* In case have to backtrack to beginning, then match '.' */
		    char *starting = locinput;

		    /* In case have to backtrack the last prepend */
		    char *previous_prepend = 0;

		    LOAD_UTF8_CHARCLASS_GCB();

		    /* Match (prepend)* */
		    while (locinput < PL_regeol
			   && swash_fetch(PL_utf8_X_prepend,
					  (U8*)locinput, utf8_target))
		    {
			previous_prepend = locinput;
			locinput += UTF8SKIP(locinput);
		    }

		    /* As noted above, if we matched a prepend character, but
		     * the next thing won't match, back off the last prepend we
		     * matched, as it is guaranteed to match the begin */
		    if (previous_prepend
			&& (locinput >=  PL_regeol
			    || ! swash_fetch(PL_utf8_X_begin,
					     (U8*)locinput, utf8_target)))
		    {
			locinput = previous_prepend;
		    }

		    /* Note that here we know PL_regeol > locinput, as we
		     * tested that upon input to this switch case, and if we
		     * moved locinput forward, we tested the result just above
		     * and it either passed, or we backed off so that it will
		     * now pass */
		    if (! swash_fetch(PL_utf8_X_begin, (U8*)locinput, utf8_target)) {

			/* Here did not match the required 'Begin' in the
			 * second term.  So just match the very first
			 * character, the '.' of the final term of the regex */
			locinput = starting + UTF8SKIP(starting);
		    } else {

			/* Here is the beginning of a character that can have
			 * an extender.  It is either a hangul syllable, or a
			 * non-control */
			if (swash_fetch(PL_utf8_X_non_hangul,
					(U8*)locinput, utf8_target))
			{

			    /* Here not a Hangul syllable, must be a
			     * ('!  * Control') */
			    locinput += UTF8SKIP(locinput);
			} else {

			    /* Here is a Hangul syllable.  It can be composed
			     * of several individual characters.  One
			     * possibility is T+ */
			    if (swash_fetch(PL_utf8_X_T,
					    (U8*)locinput, utf8_target))
			    {
				while (locinput < PL_regeol
					&& swash_fetch(PL_utf8_X_T,
							(U8*)locinput, utf8_target))
				{
				    locinput += UTF8SKIP(locinput);
				}
			    } else {

				/* Here, not T+, but is a Hangul.  That means
				 * it is one of the others: L, LV, LVT or V,
				 * and matches:
				 * L* (L | LVT T* | V  V* T* | LV  V* T*) */

				/* Match L*           */
				while (locinput < PL_regeol
					&& swash_fetch(PL_utf8_X_L,
							(U8*)locinput, utf8_target))
				{
				    locinput += UTF8SKIP(locinput);
				}

				/* Here, have exhausted L*.  If the next
				 * character is not an LV, LVT nor V, it means
				 * we had to have at least one L, so matches L+
				 * in the original equation, we have a complete
				 * hangul syllable.  Are done. */

				if (locinput < PL_regeol
				    && swash_fetch(PL_utf8_X_LV_LVT_V,
						    (U8*)locinput, utf8_target))
				{

				    /* Otherwise keep going.  Must be LV, LVT
				     * or V.  See if LVT */
				    if (swash_fetch(PL_utf8_X_LVT,
						    (U8*)locinput, utf8_target))
				    {
					locinput += UTF8SKIP(locinput);
				    } else {

					/* Must be  V or LV.  Take it, then
					 * match V*     */
					locinput += UTF8SKIP(locinput);
					while (locinput < PL_regeol
						&& swash_fetch(PL_utf8_X_V,
							 (U8*)locinput, utf8_target))
					{
					    locinput += UTF8SKIP(locinput);
					}
				    }

				    /* And any of LV, LVT, or V can be followed
				     * by T*            */
				    while (locinput < PL_regeol
					   && swash_fetch(PL_utf8_X_T,
							   (U8*)locinput,
							   utf8_target))
				    {
					locinput += UTF8SKIP(locinput);
				    }
				}
			    }
			}

			/* Match any extender */
			while (locinput < PL_regeol
				&& swash_fetch(PL_utf8_X_extend,
						(U8*)locinput, utf8_target))
			{
			    locinput += UTF8SKIP(locinput);
			}
		    }
		}
		if (locinput > PL_regeol) sayNO;
	    }
	    nextchr = UCHARAT(locinput);
	    break;
            
	case NREFFL:
	{   /* The capture buffer cases.  The ones beginning with N for the
	       named buffers just convert to the equivalent numbered and
	       pretend they were called as the corresponding numbered buffer
	       op.  */
	    /* don't initialize these in the declaration, it makes C++
	       unhappy */
	    char *s;
	    char type;
	    re_fold_t folder;
	    const U8 *fold_array;
	    UV utf8_fold_flags;

	    PL_reg_flags |= RF_tainted;
	    folder = foldEQ_locale;
	    fold_array = PL_fold_locale;
	    type = REFFL;
	    utf8_fold_flags = FOLDEQ_UTF8_LOCALE;
	    goto do_nref;

	case NREFFA:
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    type = REFFA;
	    utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
	    goto do_nref;

	case NREFFU:
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    type = REFFU;
	    utf8_fold_flags = 0;
	    goto do_nref;

	case NREFF:
	    folder = foldEQ;
	    fold_array = PL_fold;
	    type = REFF;
	    utf8_fold_flags = 0;
	    goto do_nref;

	case NREF:
	    type = REF;
	    folder = NULL;
	    fold_array = NULL;
	    utf8_fold_flags = 0;
	  do_nref:

	    /* For the named back references, find the corresponding buffer
	     * number */
	    n = reg_check_named_buff_matched(rex,scan);

            if ( ! n ) {
                sayNO;
	    }
	    goto do_nref_ref_common;

	case REFFL:
	    PL_reg_flags |= RF_tainted;
	    folder = foldEQ_locale;
	    fold_array = PL_fold_locale;
	    utf8_fold_flags = FOLDEQ_UTF8_LOCALE;
	    goto do_ref;

	case REFFA:
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
	    goto do_ref;

	case REFFU:
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    utf8_fold_flags = 0;
	    goto do_ref;

	case REFF:
	    folder = foldEQ;
	    fold_array = PL_fold;
	    utf8_fold_flags = 0;
	    goto do_ref;

        case REF:
	    folder = NULL;
	    fold_array = NULL;
	    utf8_fold_flags = 0;

	  do_ref:
	    type = OP(scan);
	    n = ARG(scan);  /* which paren pair */

	  do_nref_ref_common:
	    ln = PL_regoffs[n].start;
	    PL_reg_leftiter = PL_reg_maxiter;		/* Void cache */
	    if (*PL_reglastparen < n || ln == -1)
		sayNO;			/* Do not match unless seen CLOSEn. */
	    if (ln == PL_regoffs[n].end)
		break;

	    s = PL_bostr + ln;
	    if (type != REF	/* REF can do byte comparison */
		&& (utf8_target || type == REFFU))
	    { /* XXX handle REFFL better */
		char * limit = PL_regeol;

		/* This call case insensitively compares the entire buffer
		    * at s, with the current input starting at locinput, but
		    * not going off the end given by PL_regeol, and returns in
		    * limit upon success, how much of the current input was
		    * matched */
		if (! foldEQ_utf8_flags(s, NULL, PL_regoffs[n].end - ln, utf8_target,
				    locinput, &limit, 0, utf8_target, utf8_fold_flags))
		{
		    sayNO;
		}
		locinput = limit;
		nextchr = UCHARAT(locinput);
		break;
	    }

	    /* Not utf8:  Inline the first character, for speed. */
	    if (UCHARAT(s) != nextchr &&
		(type == REF ||
		 UCHARAT(s) != fold_array[nextchr]))
		sayNO;
	    ln = PL_regoffs[n].end - ln;
	    if (locinput + ln > PL_regeol)
		sayNO;
	    if (ln > 1 && (type == REF
			   ? memNE(s, locinput, ln)
			   : ! folder(s, locinput, ln)))
		sayNO;
	    locinput += ln;
	    nextchr = UCHARAT(locinput);
	    break;
	}
	case NOTHING:
	case TAIL:
	    break;
	case BACK:
	    break;

#undef  ST
#define ST st->u.eval
	{
	    SV *ret;
	    REGEXP *re_sv;
            regexp *re;
            regexp_internal *rei;
            regnode *startpoint;

	case GOSTART:
	case GOSUB: /*    /(...(?1))/   /(...(?&foo))/   */
	    if (cur_eval && cur_eval->locinput==locinput) {
                if (cur_eval->u.eval.close_paren == (U32)ARG(scan)) 
                    Perl_croak(aTHX_ "Infinite recursion in regex");
                if ( ++nochange_depth > max_nochange_depth )
                    Perl_croak(aTHX_ 
                        "Pattern subroutine nesting without pos change"
                        " exceeded limit in regex");
            } else {
                nochange_depth = 0;
            }
	    re_sv = rex_sv;
            re = rex;
            rei = rexi;
            (void)ReREFCNT_inc(rex_sv);
            if (OP(scan)==GOSUB) {
                startpoint = scan + ARG2L(scan);
                ST.close_paren = ARG(scan);
            } else {
                startpoint = rei->program+1;
                ST.close_paren = 0;
            }
            goto eval_recurse_doit;
            /* NOTREACHED */
        case EVAL:  /*   /(?{A})B/   /(??{A})B/  and /(?(?{A})X|Y)B/   */        
            if (cur_eval && cur_eval->locinput==locinput) {
		if ( ++nochange_depth > max_nochange_depth )
                    Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex");
            } else {
                nochange_depth = 0;
            }    
	    {
		/* execute the code in the {...} */
		dSP;
		SV ** const before = SP;
		OP_4tree * const oop = PL_op;
		COP * const ocurcop = PL_curcop;
		PAD *old_comppad;
		char *saved_regeol = PL_regeol;
		struct re_save_state saved_state;

		/* To not corrupt the existing regex state while executing the
		 * eval we would normally put it on the save stack, like with
		 * save_re_context. However, re-evals have a weird scoping so we
		 * can't just add ENTER/LEAVE here. With that, things like
		 *
		 *    (?{$a=2})(a(?{local$a=$a+1}))*aak*c(?{$b=$a})
		 *
		 * would break, as they expect the localisation to be unwound
		 * only when the re-engine backtracks through the bit that
		 * localised it.
		 *
		 * What we do instead is just saving the state in a local c
		 * variable.
		 */
		Copy(&PL_reg_state, &saved_state, 1, struct re_save_state);

		n = ARG(scan);
		PL_op = (OP_4tree*)rexi->data->data[n];
		DEBUG_STATE_r( PerlIO_printf(Perl_debug_log, 
		    "  re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
		/* wrap the call in two SAVECOMPPADs. This ensures that
		 * when the save stack is eventually unwound, all the
		 * accumulated SAVEt_CLEARSV's will be processed with
		 * interspersed SAVEt_COMPPAD's to ensure that lexicals
		 * are cleared in the right pad */
		SAVECOMPPAD();
		PAD_SAVE_LOCAL(old_comppad, (PAD*)rexi->data->data[n + 2]);
		PL_regoffs[0].end = PL_reg_magic->mg_len = locinput - PL_bostr;

                if (sv_yes_mark) {
                    SV *sv_mrk = get_sv("REGMARK", 1);
                    sv_setsv(sv_mrk, sv_yes_mark);
                }

		CALLRUNOPS(aTHX);			/* Scalar context. */
		SPAGAIN;
		if (SP == before)
		    ret = &PL_sv_undef;   /* protect against empty (?{}) blocks. */
		else {
		    ret = POPs;
		    PUTBACK;
		}

		Copy(&saved_state, &PL_reg_state, 1, struct re_save_state);

		PL_op = oop;
		SAVECOMPPAD();
		PAD_RESTORE_LOCAL(old_comppad);
		PL_curcop = ocurcop;
		PL_regeol = saved_regeol;
		if (!logical) {
		    /* /(?{...})/ */
		    sv_setsv(save_scalar(PL_replgv), ret);
		    break;
		}
	    }
	    if (logical == 2) { /* Postponed subexpression: /(??{...})/ */
		logical = 0;
		{
		    /* extract RE object from returned value; compiling if
		     * necessary */
		    MAGIC *mg = NULL;
		    REGEXP *rx = NULL;

		    if (SvROK(ret)) {
			SV *const sv = SvRV(ret);

			if (SvTYPE(sv) == SVt_REGEXP) {
			    rx = (REGEXP*) sv;
			} else if (SvSMAGICAL(sv)) {
			    mg = mg_find(sv, PERL_MAGIC_qr);
			    assert(mg);
			}
		    } else if (SvTYPE(ret) == SVt_REGEXP) {
			rx = (REGEXP*) ret;
		    } else if (SvSMAGICAL(ret)) {
			if (SvGMAGICAL(ret)) {
			    /* I don't believe that there is ever qr magic
			       here.  */
			    assert(!mg_find(ret, PERL_MAGIC_qr));
			    sv_unmagic(ret, PERL_MAGIC_qr);
			}
			else {
			    mg = mg_find(ret, PERL_MAGIC_qr);
			    /* testing suggests mg only ends up non-NULL for
			       scalars who were upgraded and compiled in the
			       else block below. In turn, this is only
			       triggered in the "postponed utf8 string" tests
			       in t/op/pat.t  */
			}
		    }

		    if (mg) {
			rx = (REGEXP *) mg->mg_obj; /*XXX:dmq*/
			assert(rx);
		    }
		    if (rx) {
			rx = reg_temp_copy(NULL, rx);
		    }
		    else {
			U32 pm_flags = 0;
			const I32 osize = PL_regsize;

			if (DO_UTF8(ret)) {
			    assert (SvUTF8(ret));
			} else if (SvUTF8(ret)) {
			    /* Not doing UTF-8, despite what the SV says. Is
			       this only if we're trapped in use 'bytes'?  */
			    /* Make a copy of the octet sequence, but without
			       the flag on, as the compiler now honours the
			       SvUTF8 flag on ret.  */
			    STRLEN len;
			    const char *const p = SvPV(ret, len);
			    ret = newSVpvn_flags(p, len, SVs_TEMP);
			}
			rx = CALLREGCOMP(ret, pm_flags);
			if (!(SvFLAGS(ret)
			      & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
				 | SVs_GMG))) {
			    /* This isn't a first class regexp. Instead, it's
			       caching a regexp onto an existing, Perl visible
			       scalar.  */
			    sv_magic(ret, MUTABLE_SV(rx), PERL_MAGIC_qr, 0, 0);
			}
			PL_regsize = osize;
		    }
		    re_sv = rx;
		    re = (struct regexp *)SvANY(rx);
		}
                RXp_MATCH_COPIED_off(re);
                re->subbeg = rex->subbeg;
                re->sublen = rex->sublen;
		rei = RXi_GET(re);
                DEBUG_EXECUTE_r(
                    debug_start_match(re_sv, utf8_target, locinput, PL_regeol,
                        "Matching embedded");
		);		
		startpoint = rei->program + 1;
               	ST.close_paren = 0; /* only used for GOSUB */
               	/* borrowed from regtry */
                if (PL_reg_start_tmpl <= re->nparens) {
                    PL_reg_start_tmpl = re->nparens*3/2 + 3;
                    if(PL_reg_start_tmp)
                        Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
                    else
                        Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
                }               	

        eval_recurse_doit: /* Share code with GOSUB below this line */                		
		/* run the pattern returned from (??{...}) */
		ST.cp = regcppush(0);	/* Save *all* the positions. */
		REGCP_SET(ST.lastcp);
		
		PL_regoffs = re->offs; /* essentially NOOP on GOSUB */
		
		/* see regtry, specifically PL_reglast(?:close)?paren is a pointer! (i dont know why) :dmq */
		PL_reglastparen = &re->lastparen;
		PL_reglastcloseparen = &re->lastcloseparen;
		re->lastparen = 0;
		re->lastcloseparen = 0;

		PL_reginput = locinput;
		PL_regsize = 0;

		/* XXXX This is too dramatic a measure... */
		PL_reg_maxiter = 0;

		ST.toggle_reg_flags = PL_reg_flags;
		if (RX_UTF8(re_sv))
		    PL_reg_flags |= RF_utf8;
		else
		    PL_reg_flags &= ~RF_utf8;
		ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */

		ST.prev_rex = rex_sv;
		ST.prev_curlyx = cur_curlyx;
		SETREX(rex_sv,re_sv);
		rex = re;
		rexi = rei;
		cur_curlyx = NULL;
		ST.B = next;
		ST.prev_eval = cur_eval;
		cur_eval = st;
		/* now continue from first node in postoned RE */
		PUSH_YES_STATE_GOTO(EVAL_AB, startpoint);
		/* NOTREACHED */
	    }
	    /* logical is 1,   /(?(?{...})X|Y)/ */
	    sw = cBOOL(SvTRUE(ret));
	    logical = 0;
	    break;
	}

	case EVAL_AB: /* cleanup after a successful (??{A})B */
	    /* note: this is called twice; first after popping B, then A */
	    PL_reg_flags ^= ST.toggle_reg_flags; 
	    ReREFCNT_dec(rex_sv);
	    SETREX(rex_sv,ST.prev_rex);
	    rex = (struct regexp *)SvANY(rex_sv);
	    rexi = RXi_GET(rex);
	    regcpblow(ST.cp);
	    cur_eval = ST.prev_eval;
	    cur_curlyx = ST.prev_curlyx;

	    /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
	    PL_reglastparen = &rex->lastparen;
	    PL_reglastcloseparen = &rex->lastcloseparen;
	    /* also update PL_regoffs */
	    PL_regoffs = rex->offs;
	    
	    /* XXXX This is too dramatic a measure... */
	    PL_reg_maxiter = 0;
            if ( nochange_depth )
	        nochange_depth--;
	    sayYES;


	case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */
	    /* note: this is called twice; first after popping B, then A */
	    PL_reg_flags ^= ST.toggle_reg_flags; 
	    ReREFCNT_dec(rex_sv);
	    SETREX(rex_sv,ST.prev_rex);
	    rex = (struct regexp *)SvANY(rex_sv);
	    rexi = RXi_GET(rex); 
	    /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
	    PL_reglastparen = &rex->lastparen;
	    PL_reglastcloseparen = &rex->lastcloseparen;

	    PL_reginput = locinput;
	    REGCP_UNWIND(ST.lastcp);
	    regcppop(rex);
	    cur_eval = ST.prev_eval;
	    cur_curlyx = ST.prev_curlyx;
	    /* XXXX This is too dramatic a measure... */
	    PL_reg_maxiter = 0;
	    if ( nochange_depth )
	        nochange_depth--;
	    sayNO_SILENT;
#undef ST

	case OPEN:
	    n = ARG(scan);  /* which paren pair */
	    PL_reg_start_tmp[n] = locinput;
	    if (n > PL_regsize)
		PL_regsize = n;
            lastopen = n;
	    break;
	case CLOSE:
	    n = ARG(scan);  /* which paren pair */
	    PL_regoffs[n].start = PL_reg_start_tmp[n] - PL_bostr;
	    PL_regoffs[n].end = locinput - PL_bostr;
	    /*if (n > PL_regsize)
		PL_regsize = n;*/
	    if (n > *PL_reglastparen)
		*PL_reglastparen = n;
	    *PL_reglastcloseparen = n;
            if (cur_eval && cur_eval->u.eval.close_paren == n) {
	        goto fake_end;
	    }    
	    break;
        case ACCEPT:
            if (ARG(scan)){
                regnode *cursor;
                for (cursor=scan;
                     cursor && OP(cursor)!=END; 
                     cursor=regnext(cursor)) 
                {
                    if ( OP(cursor)==CLOSE ){
                        n = ARG(cursor);
                        if ( n <= lastopen ) {
                            PL_regoffs[n].start
				= PL_reg_start_tmp[n] - PL_bostr;
                            PL_regoffs[n].end = locinput - PL_bostr;
                            /*if (n > PL_regsize)
                            PL_regsize = n;*/
                            if (n > *PL_reglastparen)
                                *PL_reglastparen = n;
                            *PL_reglastcloseparen = n;
                            if ( n == ARG(scan) || (cur_eval &&
                                cur_eval->u.eval.close_paren == n))
                                break;
                        }
                    }
                }
            }
	    goto fake_end;
	    /*NOTREACHED*/	    
	case GROUPP:
	    n = ARG(scan);  /* which paren pair */
	    sw = cBOOL(*PL_reglastparen >= n && PL_regoffs[n].end != -1);
	    break;
	case NGROUPP:
	    /* reg_check_named_buff_matched returns 0 for no match */
	    sw = cBOOL(0 < reg_check_named_buff_matched(rex,scan));
	    break;
        case INSUBP:
            n = ARG(scan);
            sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == n));
            break;
        case DEFINEP:
            sw = 0;
            break;
	case IFTHEN:
	    PL_reg_leftiter = PL_reg_maxiter;		/* Void cache */
	    if (sw)
		next = NEXTOPER(NEXTOPER(scan));
	    else {
		next = scan + ARG(scan);
		if (OP(next) == IFTHEN) /* Fake one. */
		    next = NEXTOPER(NEXTOPER(next));
	    }
	    break;
	case LOGICAL:
	    logical = scan->flags;
	    break;

/*******************************************************************

The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/
pattern, where A and B are subpatterns. (For simple A, CURLYM or
STAR/PLUS/CURLY/CURLYN are used instead.)

A*B is compiled as <CURLYX><A><WHILEM><B>

On entry to the subpattern, CURLYX is called. This pushes a CURLYX
state, which contains the current count, initialised to -1. It also sets
cur_curlyx to point to this state, with any previous value saved in the
state block.

CURLYX then jumps straight to the WHILEM op, rather than executing A,
since the pattern may possibly match zero times (i.e. it's a while {} loop
rather than a do {} while loop).

Each entry to WHILEM represents a successful match of A. The count in the
CURLYX block is incremented, another WHILEM state is pushed, and execution
passes to A or B depending on greediness and the current count.

For example, if matching against the string a1a2a3b (where the aN are
substrings that match /A/), then the match progresses as follows: (the
pushed states are interspersed with the bits of strings matched so far):

    <CURLYX cnt=-1>
    <CURLYX cnt=0><WHILEM>
    <CURLYX cnt=1><WHILEM> a1 <WHILEM>
    <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM>
    <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM>
    <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b

(Contrast this with something like CURLYM, which maintains only a single
backtrack state:

    <CURLYM cnt=0> a1
    a1 <CURLYM cnt=1> a2
    a1 a2 <CURLYM cnt=2> a3
    a1 a2 a3 <CURLYM cnt=3> b
)

Each WHILEM state block marks a point to backtrack to upon partial failure
of A or B, and also contains some minor state data related to that
iteration.  The CURLYX block, pointed to by cur_curlyx, contains the
overall state, such as the count, and pointers to the A and B ops.

This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx
must always point to the *current* CURLYX block, the rules are:

When executing CURLYX, save the old cur_curlyx in the CURLYX state block,
and set cur_curlyx to point the new block.

When popping the CURLYX block after a successful or unsuccessful match,
restore the previous cur_curlyx.

When WHILEM is about to execute B, save the current cur_curlyx, and set it
to the outer one saved in the CURLYX block.

When popping the WHILEM block after a successful or unsuccessful B match,
restore the previous cur_curlyx.

Here's an example for the pattern (AI* BI)*BO
I and O refer to inner and outer, C and W refer to CURLYX and WHILEM:

cur_
curlyx backtrack stack
------ ---------------
NULL   
CO     <CO prev=NULL> <WO>
CI     <CO prev=NULL> <WO> <CI prev=CO> <WI> ai 
CO     <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi 
NULL   <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo

At this point the pattern succeeds, and we work back down the stack to
clean up, restoring as we go:

CO     <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi 
CI     <CO prev=NULL> <WO> <CI prev=CO> <WI> ai 
CO     <CO prev=NULL> <WO>
NULL   

*******************************************************************/

#define ST st->u.curlyx

	case CURLYX:    /* start of /A*B/  (for complex A) */
	{
	    /* No need to save/restore up to this paren */
	    I32 parenfloor = scan->flags;
	    
	    assert(next); /* keep Coverity happy */
	    if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
		next += ARG(next);

	    /* XXXX Probably it is better to teach regpush to support
	       parenfloor > PL_regsize... */
	    if (parenfloor > (I32)*PL_reglastparen)
		parenfloor = *PL_reglastparen; /* Pessimization... */

	    ST.prev_curlyx= cur_curlyx;
	    cur_curlyx = st;
	    ST.cp = PL_savestack_ix;

	    /* these fields contain the state of the current curly.
	     * they are accessed by subsequent WHILEMs */
	    ST.parenfloor = parenfloor;
	    ST.me = scan;
	    ST.B = next;
	    ST.minmod = minmod;
	    minmod = 0;
	    ST.count = -1;	/* this will be updated by WHILEM */
	    ST.lastloc = NULL;  /* this will be updated by WHILEM */

	    PL_reginput = locinput;
	    PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next));
	    /* NOTREACHED */
	}

	case CURLYX_end: /* just finished matching all of A*B */
	    cur_curlyx = ST.prev_curlyx;
	    sayYES;
	    /* NOTREACHED */

	case CURLYX_end_fail: /* just failed to match all of A*B */
	    regcpblow(ST.cp);
	    cur_curlyx = ST.prev_curlyx;
	    sayNO;
	    /* NOTREACHED */


#undef ST
#define ST st->u.whilem

	case WHILEM:     /* just matched an A in /A*B/  (for complex A) */
	{
	    /* see the discussion above about CURLYX/WHILEM */
	    I32 n;
	    int min = ARG1(cur_curlyx->u.curlyx.me);
	    int max = ARG2(cur_curlyx->u.curlyx.me);
	    regnode *A = NEXTOPER(cur_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS;

	    assert(cur_curlyx); /* keep Coverity happy */
	    n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */
	    ST.save_lastloc = cur_curlyx->u.curlyx.lastloc;
	    ST.cache_offset = 0;
	    ST.cache_mask = 0;
	    
	    PL_reginput = locinput;

	    DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
		  "%*s  whilem: matched %ld out of %d..%d\n",
		  REPORT_CODE_OFF+depth*2, "", (long)n, min, max)
	    );

	    /* First just match a string of min A's. */

	    if (n < min) {
		ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
		cur_curlyx->u.curlyx.lastloc = locinput;
		REGCP_SET(ST.lastcp);

		PUSH_STATE_GOTO(WHILEM_A_pre, A);
		/* NOTREACHED */
	    }

	    /* If degenerate A matches "", assume A done. */

	    if (locinput == cur_curlyx->u.curlyx.lastloc) {
		DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
		   "%*s  whilem: empty match detected, trying continuation...\n",
		   REPORT_CODE_OFF+depth*2, "")
		);
		goto do_whilem_B_max;
	    }

	    /* super-linear cache processing */

	    if (scan->flags) {

		if (!PL_reg_maxiter) {
		    /* start the countdown: Postpone detection until we
		     * know the match is not *that* much linear. */
		    PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4);
		    /* possible overflow for long strings and many CURLYX's */
		    if (PL_reg_maxiter < 0)
			PL_reg_maxiter = I32_MAX;
		    PL_reg_leftiter = PL_reg_maxiter;
		}

		if (PL_reg_leftiter-- == 0) {
		    /* initialise cache */
		    const I32 size = (PL_reg_maxiter + 7)/8;
		    if (PL_reg_poscache) {
			if ((I32)PL_reg_poscache_size < size) {
			    Renew(PL_reg_poscache, size, char);
			    PL_reg_poscache_size = size;
			}
			Zero(PL_reg_poscache, size, char);
		    }
		    else {
			PL_reg_poscache_size = size;
			Newxz(PL_reg_poscache, size, char);
		    }
		    DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
      "%swhilem: Detected a super-linear match, switching on caching%s...\n",
			      PL_colors[4], PL_colors[5])
		    );
		}

		if (PL_reg_leftiter < 0) {
		    /* have we already failed at this position? */
		    I32 offset, mask;
		    offset  = (scan->flags & 0xf) - 1
		  		+ (locinput - PL_bostr)  * (scan->flags>>4);
		    mask    = 1 << (offset % 8);
		    offset /= 8;
		    if (PL_reg_poscache[offset] & mask) {
			DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
			    "%*s  whilem: (cache) already tried at this position...\n",
			    REPORT_CODE_OFF+depth*2, "")
			);
			sayNO; /* cache records failure */
		    }
		    ST.cache_offset = offset;
		    ST.cache_mask   = mask;
		}
	    }

	    /* Prefer B over A for minimal matching. */

	    if (cur_curlyx->u.curlyx.minmod) {
		ST.save_curlyx = cur_curlyx;
		cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
		ST.cp = regcppush(ST.save_curlyx->u.curlyx.parenfloor);
		REGCP_SET(ST.lastcp);
		PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B);
		/* NOTREACHED */
	    }

	    /* Prefer A over B for maximal matching. */

	    if (n < max) { /* More greed allowed? */
		ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
		cur_curlyx->u.curlyx.lastloc = locinput;
		REGCP_SET(ST.lastcp);
		PUSH_STATE_GOTO(WHILEM_A_max, A);
		/* NOTREACHED */
	    }
	    goto do_whilem_B_max;
	}
	/* NOTREACHED */

	case WHILEM_B_min: /* just matched B in a minimal match */
	case WHILEM_B_max: /* just matched B in a maximal match */
	    cur_curlyx = ST.save_curlyx;
	    sayYES;
	    /* NOTREACHED */

	case WHILEM_B_max_fail: /* just failed to match B in a maximal match */
	    cur_curlyx = ST.save_curlyx;
	    cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
	    cur_curlyx->u.curlyx.count--;
	    CACHEsayNO;
	    /* NOTREACHED */

	case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
	    /* FALL THROUGH */
	case WHILEM_A_pre_fail: /* just failed to match even minimal A */
	    REGCP_UNWIND(ST.lastcp);
	    regcppop(rex);
	    cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
	    cur_curlyx->u.curlyx.count--;
	    CACHEsayNO;
	    /* NOTREACHED */

	case WHILEM_A_max_fail: /* just failed to match A in a maximal match */
	    REGCP_UNWIND(ST.lastcp);
	    regcppop(rex);	/* Restore some previous $<digit>s? */
	    PL_reginput = locinput;
	    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
		"%*s  whilem: failed, trying continuation...\n",
		REPORT_CODE_OFF+depth*2, "")
	    );
	  do_whilem_B_max:
	    if (cur_curlyx->u.curlyx.count >= REG_INFTY
		&& ckWARN(WARN_REGEXP)
		&& !(PL_reg_flags & RF_warned))
	    {
		PL_reg_flags |= RF_warned;
		Perl_warner(aTHX_ packWARN(WARN_REGEXP),
		     "Complex regular subexpression recursion limit (%d) "
		     "exceeded",
		     REG_INFTY - 1);
	    }

	    /* now try B */
	    ST.save_curlyx = cur_curlyx;
	    cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
	    PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B);
	    /* NOTREACHED */

	case WHILEM_B_min_fail: /* just failed to match B in a minimal match */
	    cur_curlyx = ST.save_curlyx;
	    REGCP_UNWIND(ST.lastcp);
	    regcppop(rex);

	    if (cur_curlyx->u.curlyx.count >= /*max*/ARG2(cur_curlyx->u.curlyx.me)) {
		/* Maximum greed exceeded */
		if (cur_curlyx->u.curlyx.count >= REG_INFTY
		    && ckWARN(WARN_REGEXP)
		    && !(PL_reg_flags & RF_warned))
		{
		    PL_reg_flags |= RF_warned;
		    Perl_warner(aTHX_ packWARN(WARN_REGEXP),
			"Complex regular subexpression recursion "
			"limit (%d) exceeded",
			REG_INFTY - 1);
		}
		cur_curlyx->u.curlyx.count--;
		CACHEsayNO;
	    }

	    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
		"%*s  trying longer...\n", REPORT_CODE_OFF+depth*2, "")
	    );
	    /* Try grabbing another A and see if it helps. */
	    PL_reginput = locinput;
	    cur_curlyx->u.curlyx.lastloc = locinput;
	    ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
	    REGCP_SET(ST.lastcp);
	    PUSH_STATE_GOTO(WHILEM_A_min,
		/*A*/ NEXTOPER(ST.save_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS);
	    /* NOTREACHED */

#undef  ST
#define ST st->u.branch

	case BRANCHJ:	    /*  /(...|A|...)/ with long next pointer */
	    next = scan + ARG(scan);
	    if (next == scan)
		next = NULL;
	    scan = NEXTOPER(scan);
	    /* FALL THROUGH */

	case BRANCH:	    /*  /(...|A|...)/ */
	    scan = NEXTOPER(scan); /* scan now points to inner node */
	    ST.lastparen = *PL_reglastparen;
	    ST.next_branch = next;
	    REGCP_SET(ST.cp);
	    PL_reginput = locinput;

	    /* Now go into the branch */
	    if (has_cutgroup) {
	        PUSH_YES_STATE_GOTO(BRANCH_next, scan);    
	    } else {
	        PUSH_STATE_GOTO(BRANCH_next, scan);
	    }
	    /* NOTREACHED */
        case CUTGROUP:
            PL_reginput = locinput;
            sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL :
                MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
            PUSH_STATE_GOTO(CUTGROUP_next,next);
            /* NOTREACHED */
        case CUTGROUP_next_fail:
            do_cutgroup = 1;
            no_final = 1;
            if (st->u.mark.mark_name)
                sv_commit = st->u.mark.mark_name;
            sayNO;	    
            /* NOTREACHED */
        case BRANCH_next:
            sayYES;
            /* NOTREACHED */
	case BRANCH_next_fail: /* that branch failed; try the next, if any */
	    if (do_cutgroup) {
	        do_cutgroup = 0;
	        no_final = 0;
	    }
	    REGCP_UNWIND(ST.cp);
	    for (n = *PL_reglastparen; n > ST.lastparen; n--)
		PL_regoffs[n].end = -1;
	    *PL_reglastparen = n;
	    /*dmq: *PL_reglastcloseparen = n; */
	    scan = ST.next_branch;
	    /* no more branches? */
	    if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) {
	        DEBUG_EXECUTE_r({
		    PerlIO_printf( Perl_debug_log,
			"%*s  %sBRANCH failed...%s\n",
			REPORT_CODE_OFF+depth*2, "", 
			PL_colors[4],
			PL_colors[5] );
		});
		sayNO_SILENT;
            }
	    continue; /* execute next BRANCH[J] op */
	    /* NOTREACHED */
    
	case MINMOD:
	    minmod = 1;
	    break;

#undef  ST
#define ST st->u.curlym

	case CURLYM:	/* /A{m,n}B/ where A is fixed-length */

	    /* This is an optimisation of CURLYX that enables us to push
	     * only a single backtracking state, no matter how many matches
	     * there are in {m,n}. It relies on the pattern being constant
	     * length, with no parens to influence future backrefs
	     */

	    ST.me = scan;
	    scan = NEXTOPER(scan) + NODE_STEP_REGNODE;

	    /* if paren positive, emulate an OPEN/CLOSE around A */
	    if (ST.me->flags) {
		U32 paren = ST.me->flags;
		if (paren > PL_regsize)
		    PL_regsize = paren;
		if (paren > *PL_reglastparen)
		    *PL_reglastparen = paren;
		scan += NEXT_OFF(scan); /* Skip former OPEN. */
	    }
	    ST.A = scan;
	    ST.B = next;
	    ST.alen = 0;
	    ST.count = 0;
	    ST.minmod = minmod;
	    minmod = 0;
	    ST.c1 = CHRTEST_UNINIT;
	    REGCP_SET(ST.cp);

	    if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */
		goto curlym_do_B;

	  curlym_do_A: /* execute the A in /A{m,n}B/  */
	    PL_reginput = locinput;
	    PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */
	    /* NOTREACHED */

	case CURLYM_A: /* we've just matched an A */
	    locinput = st->locinput;
	    nextchr = UCHARAT(locinput);

	    ST.count++;
	    /* after first match, determine A's length: u.curlym.alen */
	    if (ST.count == 1) {
		if (PL_reg_match_utf8) {
		    char *s = locinput;
		    while (s < PL_reginput) {
			ST.alen++;
			s += UTF8SKIP(s);
		    }
		}
		else {
		    ST.alen = PL_reginput - locinput;
		}
		if (ST.alen == 0)
		    ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me);
	    }
	    DEBUG_EXECUTE_r(
		PerlIO_printf(Perl_debug_log,
			  "%*s  CURLYM now matched %"IVdf" times, len=%"IVdf"...\n",
			  (int)(REPORT_CODE_OFF+(depth*2)), "",
			  (IV) ST.count, (IV)ST.alen)
	    );

	    locinput = PL_reginput;
	                
	    if (cur_eval && cur_eval->u.eval.close_paren && 
	        cur_eval->u.eval.close_paren == (U32)ST.me->flags) 
	        goto fake_end;
	        
	    {
		I32 max = (ST.minmod ? ARG1(ST.me) : ARG2(ST.me));
		if ( max == REG_INFTY || ST.count < max )
		    goto curlym_do_A; /* try to match another A */
	    }
	    goto curlym_do_B; /* try to match B */

	case CURLYM_A_fail: /* just failed to match an A */
	    REGCP_UNWIND(ST.cp);

	    if (ST.minmod || ST.count < ARG1(ST.me) /* min*/ 
	        || (cur_eval && cur_eval->u.eval.close_paren &&
	            cur_eval->u.eval.close_paren == (U32)ST.me->flags))
		sayNO;

	  curlym_do_B: /* execute the B in /A{m,n}B/  */
	    PL_reginput = locinput;
	    if (ST.c1 == CHRTEST_UNINIT) {
		/* calculate c1 and c2 for possible match of 1st char
		 * following curly */
		ST.c1 = ST.c2 = CHRTEST_VOID;
		if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) {
		    regnode *text_node = ST.B;
		    if (! HAS_TEXT(text_node))
			FIND_NEXT_IMPT(text_node);
	            /* this used to be 
	                
	                (HAS_TEXT(text_node) && PL_regkind[OP(text_node)] == EXACT)
	                
	            	But the former is redundant in light of the latter.
	            	
	            	if this changes back then the macro for 
	            	IS_TEXT and friends need to change.
	             */
		    if (PL_regkind[OP(text_node)] == EXACT)
		    {
		        
			ST.c1 = (U8)*STRING(text_node);
			switch (OP(text_node)) {
			    case EXACTF: ST.c2 = PL_fold[ST.c1]; break;
			    case EXACTFA:
			    case EXACTFU_SS:
			    case EXACTFU_TRICKYFOLD:
			    case EXACTFU: ST.c2 = PL_fold_latin1[ST.c1]; break;
			    case EXACTFL: ST.c2 = PL_fold_locale[ST.c1]; break;
			    default: ST.c2 = ST.c1;
			}
		    }
		}
	    }

	    DEBUG_EXECUTE_r(
		PerlIO_printf(Perl_debug_log,
		    "%*s  CURLYM trying tail with matches=%"IVdf"...\n",
		    (int)(REPORT_CODE_OFF+(depth*2)),
		    "", (IV)ST.count)
		);
	    if (ST.c1 != CHRTEST_VOID
		    && UCHARAT(PL_reginput) != ST.c1
		    && UCHARAT(PL_reginput) != ST.c2)
	    {
		/* simulate B failing */
		DEBUG_OPTIMISE_r(
		    PerlIO_printf(Perl_debug_log,
		        "%*s  CURLYM Fast bail c1=%"IVdf" c2=%"IVdf"\n",
		        (int)(REPORT_CODE_OFF+(depth*2)),"",
		        (IV)ST.c1,(IV)ST.c2
		));
		state_num = CURLYM_B_fail;
		goto reenter_switch;
	    }

	    if (ST.me->flags) {
		/* mark current A as captured */
		I32 paren = ST.me->flags;
		if (ST.count) {
		    PL_regoffs[paren].start
			= HOPc(PL_reginput, -ST.alen) - PL_bostr;
		    PL_regoffs[paren].end = PL_reginput - PL_bostr;
		    /*dmq: *PL_reglastcloseparen = paren; */
		}
		else
		    PL_regoffs[paren].end = -1;
		if (cur_eval && cur_eval->u.eval.close_paren &&
		    cur_eval->u.eval.close_paren == (U32)ST.me->flags) 
		{
		    if (ST.count) 
	                goto fake_end;
	            else
	                sayNO;
	        }
	    }
	    
	    PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */
	    /* NOTREACHED */

	case CURLYM_B_fail: /* just failed to match a B */
	    REGCP_UNWIND(ST.cp);
	    if (ST.minmod) {
		I32 max = ARG2(ST.me);
		if (max != REG_INFTY && ST.count == max)
		    sayNO;
		goto curlym_do_A; /* try to match a further A */
	    }
	    /* backtrack one A */
	    if (ST.count == ARG1(ST.me) /* min */)
		sayNO;
	    ST.count--;
	    locinput = HOPc(locinput, -ST.alen);
	    goto curlym_do_B; /* try to match B */

#undef ST
#define ST st->u.curly

#define CURLY_SETPAREN(paren, success) \
    if (paren) { \
	if (success) { \
	    PL_regoffs[paren].start = HOPc(locinput, -1) - PL_bostr; \
	    PL_regoffs[paren].end = locinput - PL_bostr; \
	    *PL_reglastcloseparen = paren; \
	} \
	else \
	    PL_regoffs[paren].end = -1; \
    }

	case STAR:		/*  /A*B/ where A is width 1 */
	    ST.paren = 0;
	    ST.min = 0;
	    ST.max = REG_INFTY;
	    scan = NEXTOPER(scan);
	    goto repeat;
	case PLUS:		/*  /A+B/ where A is width 1 */
	    ST.paren = 0;
	    ST.min = 1;
	    ST.max = REG_INFTY;
	    scan = NEXTOPER(scan);
	    goto repeat;
	case CURLYN:		/*  /(A){m,n}B/ where A is width 1 */
	    ST.paren = scan->flags;	/* Which paren to set */
	    if (ST.paren > PL_regsize)
		PL_regsize = ST.paren;
	    if (ST.paren > *PL_reglastparen)
		*PL_reglastparen = ST.paren;
	    ST.min = ARG1(scan);  /* min to match */
	    ST.max = ARG2(scan);  /* max to match */
	    if (cur_eval && cur_eval->u.eval.close_paren &&
	        cur_eval->u.eval.close_paren == (U32)ST.paren) {
	        ST.min=1;
	        ST.max=1;
	    }
            scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
	    goto repeat;
	case CURLY:		/*  /A{m,n}B/ where A is width 1 */
	    ST.paren = 0;
	    ST.min = ARG1(scan);  /* min to match */
	    ST.max = ARG2(scan);  /* max to match */
	    scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
	  repeat:
	    /*
	    * Lookahead to avoid useless match attempts
	    * when we know what character comes next.
	    *
	    * Used to only do .*x and .*?x, but now it allows
	    * for )'s, ('s and (?{ ... })'s to be in the way
	    * of the quantifier and the EXACT-like node.  -- japhy
	    */

	    if (ST.min > ST.max) /* XXX make this a compile-time check? */
		sayNO;
	    if (HAS_TEXT(next) || JUMPABLE(next)) {
		U8 *s;
		regnode *text_node = next;

		if (! HAS_TEXT(text_node)) 
		    FIND_NEXT_IMPT(text_node);

		if (! HAS_TEXT(text_node))
		    ST.c1 = ST.c2 = CHRTEST_VOID;
		else {
		    if ( PL_regkind[OP(text_node)] != EXACT ) {
			ST.c1 = ST.c2 = CHRTEST_VOID;
			goto assume_ok_easy;
		    }
		    else
			s = (U8*)STRING(text_node);
                    
                    /*  Currently we only get here when 
                        
                        PL_rekind[OP(text_node)] == EXACT
                    
                        if this changes back then the macro for IS_TEXT and 
                        friends need to change. */
		    if (!UTF_PATTERN) {
			ST.c1 = *s;
			switch (OP(text_node)) {
			    case EXACTF: ST.c2 = PL_fold[ST.c1]; break;
			    case EXACTFA:
			    case EXACTFU_SS:
			    case EXACTFU_TRICKYFOLD:
			    case EXACTFU: ST.c2 = PL_fold_latin1[ST.c1]; break;
			    case EXACTFL: ST.c2 = PL_fold_locale[ST.c1]; break;
			    default: ST.c2 = ST.c1; break;
			}
		    }
		    else { /* UTF_PATTERN */
			if (IS_TEXTFU(text_node) || IS_TEXTF(text_node)) {
			     STRLEN ulen1, ulen2;
			     U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
			     U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];

			     to_utf8_lower((U8*)s, tmpbuf1, &ulen1);
			     to_utf8_upper((U8*)s, tmpbuf2, &ulen2);
#ifdef EBCDIC
			     ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0,
						    ckWARN(WARN_UTF8) ?
                                                    0 : UTF8_ALLOW_ANY);
			     ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0,
                                                    ckWARN(WARN_UTF8) ?
                                                    0 : UTF8_ALLOW_ANY);
#else
			     ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0,
						    uniflags);
			     ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0,
						    uniflags);
#endif
			}
			else {
			    ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0,
						     uniflags);
			}
		    }
		}
	    }
	    else
		ST.c1 = ST.c2 = CHRTEST_VOID;
	assume_ok_easy:

	    ST.A = scan;
	    ST.B = next;
	    PL_reginput = locinput;
	    if (minmod) {
		minmod = 0;
		if (ST.min && regrepeat(rex, ST.A, ST.min, depth) < ST.min)
		    sayNO;
		ST.count = ST.min;
		locinput = PL_reginput;
		REGCP_SET(ST.cp);
		if (ST.c1 == CHRTEST_VOID)
		    goto curly_try_B_min;

		ST.oldloc = locinput;

		/* set ST.maxpos to the furthest point along the
		 * string that could possibly match */
		if  (ST.max == REG_INFTY) {
		    ST.maxpos = PL_regeol - 1;
		    if (utf8_target)
			while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos))
			    ST.maxpos--;
		}
		else if (utf8_target) {
		    int m = ST.max - ST.min;
		    for (ST.maxpos = locinput;
			 m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--)
			ST.maxpos += UTF8SKIP(ST.maxpos);
		}
		else {
		    ST.maxpos = locinput + ST.max - ST.min;
		    if (ST.maxpos >= PL_regeol)
			ST.maxpos = PL_regeol - 1;
		}
		goto curly_try_B_min_known;

	    }
	    else {
		ST.count = regrepeat(rex, ST.A, ST.max, depth);
		locinput = PL_reginput;
		if (ST.count < ST.min)
		    sayNO;
		if ((ST.count > ST.min)
		    && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL))
		{
		    /* A{m,n} must come at the end of the string, there's
		     * no point in backing off ... */
		    ST.min = ST.count;
		    /* ...except that $ and \Z can match before *and* after
		       newline at the end.  Consider "\n\n" =~ /\n+\Z\n/.
		       We may back off by one in this case. */
		    if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS)
			ST.min--;
		}
		REGCP_SET(ST.cp);
		goto curly_try_B_max;
	    }
	    /* NOTREACHED */


	case CURLY_B_min_known_fail:
	    /* failed to find B in a non-greedy match where c1,c2 valid */
	    if (ST.paren && ST.count)
		PL_regoffs[ST.paren].end = -1;

	    PL_reginput = locinput;	/* Could be reset... */
	    REGCP_UNWIND(ST.cp);
	    /* Couldn't or didn't -- move forward. */
	    ST.oldloc = locinput;
	    if (utf8_target)
		locinput += UTF8SKIP(locinput);
	    else
		locinput++;
	    ST.count++;
	  curly_try_B_min_known:
	     /* find the next place where 'B' could work, then call B */
	    {
		int n;
		if (utf8_target) {
		    n = (ST.oldloc == locinput) ? 0 : 1;
		    if (ST.c1 == ST.c2) {
			STRLEN len;
			/* set n to utf8_distance(oldloc, locinput) */
			while (locinput <= ST.maxpos &&
			       utf8n_to_uvchr((U8*)locinput,
					      UTF8_MAXBYTES, &len,
					      uniflags) != (UV)ST.c1) {
			    locinput += len;
			    n++;
			}
		    }
		    else {
			/* set n to utf8_distance(oldloc, locinput) */
			while (locinput <= ST.maxpos) {
			    STRLEN len;
			    const UV c = utf8n_to_uvchr((U8*)locinput,
						  UTF8_MAXBYTES, &len,
						  uniflags);
			    if (c == (UV)ST.c1 || c == (UV)ST.c2)
				break;
			    locinput += len;
			    n++;
			}
		    }
		}
		else {
		    if (ST.c1 == ST.c2) {
			while (locinput <= ST.maxpos &&
			       UCHARAT(locinput) != ST.c1)
			    locinput++;
		    }
		    else {
			while (locinput <= ST.maxpos
			       && UCHARAT(locinput) != ST.c1
			       && UCHARAT(locinput) != ST.c2)
			    locinput++;
		    }
		    n = locinput - ST.oldloc;
		}
		if (locinput > ST.maxpos)
		    sayNO;
		/* PL_reginput == oldloc now */
		if (n) {
		    ST.count += n;
		    if (regrepeat(rex, ST.A, n, depth) < n)
			sayNO;
		}
		PL_reginput = locinput;
		CURLY_SETPAREN(ST.paren, ST.count);
		if (cur_eval && cur_eval->u.eval.close_paren && 
		    cur_eval->u.eval.close_paren == (U32)ST.paren) {
		    goto fake_end;
	        }
		PUSH_STATE_GOTO(CURLY_B_min_known, ST.B);
	    }
	    /* NOTREACHED */


	case CURLY_B_min_fail:
	    /* failed to find B in a non-greedy match where c1,c2 invalid */
	    if (ST.paren && ST.count)
		PL_regoffs[ST.paren].end = -1;

	    REGCP_UNWIND(ST.cp);
	    /* failed -- move forward one */
	    PL_reginput = locinput;
	    if (regrepeat(rex, ST.A, 1, depth)) {
		ST.count++;
		locinput = PL_reginput;
		if (ST.count <= ST.max || (ST.max == REG_INFTY &&
			ST.count > 0)) /* count overflow ? */
		{
		  curly_try_B_min:
		    CURLY_SETPAREN(ST.paren, ST.count);
		    if (cur_eval && cur_eval->u.eval.close_paren &&
		        cur_eval->u.eval.close_paren == (U32)ST.paren) {
                        goto fake_end;
                    }
		    PUSH_STATE_GOTO(CURLY_B_min, ST.B);
		}
	    }
	    sayNO;
	    /* NOTREACHED */


	curly_try_B_max:
	    /* a successful greedy match: now try to match B */
            if (cur_eval && cur_eval->u.eval.close_paren &&
                cur_eval->u.eval.close_paren == (U32)ST.paren) {
                goto fake_end;
            }
	    {
		UV c = 0;
		if (ST.c1 != CHRTEST_VOID)
		    c = utf8_target ? utf8n_to_uvchr((U8*)PL_reginput,
					   UTF8_MAXBYTES, 0, uniflags)
				: (UV) UCHARAT(PL_reginput);
		/* If it could work, try it. */
		if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) {
		    CURLY_SETPAREN(ST.paren, ST.count);
		    PUSH_STATE_GOTO(CURLY_B_max, ST.B);
		    /* NOTREACHED */
		}
	    }
	    /* FALL THROUGH */
	case CURLY_B_max_fail:
	    /* failed to find B in a greedy match */
	    if (ST.paren && ST.count)
		PL_regoffs[ST.paren].end = -1;

	    REGCP_UNWIND(ST.cp);
	    /*  back up. */
	    if (--ST.count < ST.min)
		sayNO;
	    PL_reginput = locinput = HOPc(locinput, -1);
	    goto curly_try_B_max;

#undef ST

	case END:
	    fake_end:
	    if (cur_eval) {
		/* we've just finished A in /(??{A})B/; now continue with B */
		I32 tmpix;
		st->u.eval.toggle_reg_flags
			    = cur_eval->u.eval.toggle_reg_flags;
		PL_reg_flags ^= st->u.eval.toggle_reg_flags; 

		st->u.eval.prev_rex = rex_sv;		/* inner */
		SETREX(rex_sv,cur_eval->u.eval.prev_rex);
		rex = (struct regexp *)SvANY(rex_sv);
		rexi = RXi_GET(rex);
		cur_curlyx = cur_eval->u.eval.prev_curlyx;
		(void)ReREFCNT_inc(rex_sv);
		st->u.eval.cp = regcppush(0);	/* Save *all* the positions. */

		/* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
		PL_reglastparen = &rex->lastparen;
		PL_reglastcloseparen = &rex->lastcloseparen;

		REGCP_SET(st->u.eval.lastcp);
		PL_reginput = locinput;

		/* Restore parens of the outer rex without popping the
		 * savestack */
		tmpix = PL_savestack_ix;
		PL_savestack_ix = cur_eval->u.eval.lastcp;
		regcppop(rex);
		PL_savestack_ix = tmpix;

		st->u.eval.prev_eval = cur_eval;
		cur_eval = cur_eval->u.eval.prev_eval;
		DEBUG_EXECUTE_r(
		    PerlIO_printf(Perl_debug_log, "%*s  EVAL trying tail ... %"UVxf"\n",
				      REPORT_CODE_OFF+depth*2, "",PTR2UV(cur_eval)););
                if ( nochange_depth )
	            nochange_depth--;

                PUSH_YES_STATE_GOTO(EVAL_AB,
			st->u.eval.prev_eval->u.eval.B); /* match B */
	    }

	    if (locinput < reginfo->till) {
		DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
				      "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n",
				      PL_colors[4],
				      (long)(locinput - PL_reg_starttry),
				      (long)(reginfo->till - PL_reg_starttry),
				      PL_colors[5]));
               				      
		sayNO_SILENT;		/* Cannot match: too short. */
	    }
	    PL_reginput = locinput;	/* put where regtry can find it */
	    sayYES;			/* Success! */

	case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */
	    DEBUG_EXECUTE_r(
	    PerlIO_printf(Perl_debug_log,
		"%*s  %ssubpattern success...%s\n",
		REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]));
	    PL_reginput = locinput;	/* put where regtry can find it */
	    sayYES;			/* Success! */

#undef  ST
#define ST st->u.ifmatch

	case SUSPEND:	/* (?>A) */
	    ST.wanted = 1;
	    PL_reginput = locinput;
	    goto do_ifmatch;	

	case UNLESSM:	/* -ve lookaround: (?!A), or with flags, (?<!A) */
	    ST.wanted = 0;
	    goto ifmatch_trivial_fail_test;

	case IFMATCH:	/* +ve lookaround: (?=A), or with flags, (?<=A) */
	    ST.wanted = 1;
	  ifmatch_trivial_fail_test:
	    if (scan->flags) {
		char * const s = HOPBACKc(locinput, scan->flags);
		if (!s) {
		    /* trivial fail */
		    if (logical) {
			logical = 0;
			sw = 1 - cBOOL(ST.wanted);
		    }
		    else if (ST.wanted)
			sayNO;
		    next = scan + ARG(scan);
		    if (next == scan)
			next = NULL;
		    break;
		}
		PL_reginput = s;
	    }
	    else
		PL_reginput = locinput;

	  do_ifmatch:
	    ST.me = scan;
	    ST.logical = logical;
	    logical = 0; /* XXX: reset state of logical once it has been saved into ST */
	    
	    /* execute body of (?...A) */
	    PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan)));
	    /* NOTREACHED */

	case IFMATCH_A_fail: /* body of (?...A) failed */
	    ST.wanted = !ST.wanted;
	    /* FALL THROUGH */

	case IFMATCH_A: /* body of (?...A) succeeded */
	    if (ST.logical) {
		sw = cBOOL(ST.wanted);
	    }
	    else if (!ST.wanted)
		sayNO;

	    if (OP(ST.me) == SUSPEND)
		locinput = PL_reginput;
	    else {
		locinput = PL_reginput = st->locinput;
		nextchr = UCHARAT(locinput);
	    }
	    scan = ST.me + ARG(ST.me);
	    if (scan == ST.me)
		scan = NULL;
	    continue; /* execute B */

#undef ST

	case LONGJMP:
	    next = scan + ARG(scan);
	    if (next == scan)
		next = NULL;
	    break;
	case COMMIT:
	    reginfo->cutpoint = PL_regeol;
	    /* FALLTHROUGH */
	case PRUNE:
	    PL_reginput = locinput;
	    if (!scan->flags)
	        sv_yes_mark = sv_commit = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
	    PUSH_STATE_GOTO(COMMIT_next,next);
	    /* NOTREACHED */
	case COMMIT_next_fail:
	    no_final = 1;    
	    /* FALLTHROUGH */	    
	case OPFAIL:
	    sayNO;
	    /* NOTREACHED */

#define ST st->u.mark
        case MARKPOINT:
            ST.prev_mark = mark_state;
            ST.mark_name = sv_commit = sv_yes_mark 
                = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
            mark_state = st;
            ST.mark_loc = PL_reginput = locinput;
            PUSH_YES_STATE_GOTO(MARKPOINT_next,next);
            /* NOTREACHED */
        case MARKPOINT_next:
            mark_state = ST.prev_mark;
            sayYES;
            /* NOTREACHED */
        case MARKPOINT_next_fail:
            if (popmark && sv_eq(ST.mark_name,popmark)) 
            {
                if (ST.mark_loc > startpoint)
	            reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
                popmark = NULL; /* we found our mark */
                sv_commit = ST.mark_name;

                DEBUG_EXECUTE_r({
                        PerlIO_printf(Perl_debug_log,
		            "%*s  %ssetting cutpoint to mark:%"SVf"...%s\n",
		            REPORT_CODE_OFF+depth*2, "", 
		            PL_colors[4], SVfARG(sv_commit), PL_colors[5]);
		});
            }
            mark_state = ST.prev_mark;
            sv_yes_mark = mark_state ? 
                mark_state->u.mark.mark_name : NULL;
            sayNO;
            /* NOTREACHED */
        case SKIP:
            PL_reginput = locinput;
            if (scan->flags) {
                /* (*SKIP) : if we fail we cut here*/
                ST.mark_name = NULL;
                ST.mark_loc = locinput;
                PUSH_STATE_GOTO(SKIP_next,next);    
            } else {
                /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was, 
                   otherwise do nothing.  Meaning we need to scan 
                 */
                regmatch_state *cur = mark_state;
                SV *find = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
                
                while (cur) {
                    if ( sv_eq( cur->u.mark.mark_name, 
                                find ) ) 
                    {
                        ST.mark_name = find;
                        PUSH_STATE_GOTO( SKIP_next, next );
                    }
                    cur = cur->u.mark.prev_mark;
                }
            }    
            /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */
            break;    
	case SKIP_next_fail:
	    if (ST.mark_name) {
	        /* (*CUT:NAME) - Set up to search for the name as we 
	           collapse the stack*/
	        popmark = ST.mark_name;	   
	    } else {
	        /* (*CUT) - No name, we cut here.*/
	        if (ST.mark_loc > startpoint)
	            reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
	        /* but we set sv_commit to latest mark_name if there
	           is one so they can test to see how things lead to this
	           cut */    
                if (mark_state) 
                    sv_commit=mark_state->u.mark.mark_name;	            
            } 
            no_final = 1; 
            sayNO;
            /* NOTREACHED */
#undef ST
        case LNBREAK:
            if ((n=is_LNBREAK(locinput,utf8_target))) {
                locinput += n;
                nextchr = UCHARAT(locinput);
            } else
                sayNO;
            break;

#define CASE_CLASS(nAmE)                              \
        case nAmE:                                    \
	    if (locinput >= PL_regeol)                \
		sayNO;                                \
            if ((n=is_##nAmE(locinput,utf8_target))) {    \
                locinput += n;                        \
                nextchr = UCHARAT(locinput);          \
            } else                                    \
                sayNO;                                \
            break;                                    \
        case N##nAmE:                                 \
	    if (locinput >= PL_regeol)                \
		sayNO;                                \
            if ((n=is_##nAmE(locinput,utf8_target))) {    \
                sayNO;                                \
            } else {                                  \
                locinput += UTF8SKIP(locinput);       \
                nextchr = UCHARAT(locinput);          \
            }                                         \
            break

        CASE_CLASS(VERTWS);
        CASE_CLASS(HORIZWS);
#undef CASE_CLASS

	default:
	    PerlIO_printf(Perl_error_log, "%"UVxf" %d\n",
			  PTR2UV(scan), OP(scan));
	    Perl_croak(aTHX_ "regexp memory corruption");
	    
	} /* end switch */ 

        /* switch break jumps here */
	scan = next; /* prepare to execute the next op and ... */
	continue;    /* ... jump back to the top, reusing st */
	/* NOTREACHED */

      push_yes_state:
	/* push a state that backtracks on success */
	st->u.yes.prev_yes_state = yes_state;
	yes_state = st;
	/* FALL THROUGH */
      push_state:
	/* push a new regex state, then continue at scan  */
	{
	    regmatch_state *newst;

	    DEBUG_STACK_r({
	        regmatch_state *cur = st;
	        regmatch_state *curyes = yes_state;
	        int curd = depth;
	        regmatch_slab *slab = PL_regmatch_slab;
                for (;curd > -1;cur--,curd--) {
                    if (cur < SLAB_FIRST(slab)) {
                	slab = slab->prev;
                	cur = SLAB_LAST(slab);
                    }
                    PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n",
                        REPORT_CODE_OFF + 2 + depth * 2,"",
                        curd, PL_reg_name[cur->resume_state],
                        (curyes == cur) ? "yes" : ""
                    );
                    if (curyes == cur)
	                curyes = cur->u.yes.prev_yes_state;
                }
            } else 
                DEBUG_STATE_pp("push")
            );
	    depth++;
	    st->locinput = locinput;
	    newst = st+1; 
	    if (newst >  SLAB_LAST(PL_regmatch_slab))
		newst = S_push_slab(aTHX);
	    PL_regmatch_state = newst;

	    locinput = PL_reginput;
	    nextchr = UCHARAT(locinput);
	    st = newst;
	    continue;
	    /* NOTREACHED */
	}
    }

    /*
    * We get here only if there's trouble -- normally "case END" is
    * the terminating point.
    */
    Perl_croak(aTHX_ "corrupted regexp pointers");
    /*NOTREACHED*/
    sayNO;

yes:
    if (yes_state) {
	/* we have successfully completed a subexpression, but we must now
	 * pop to the state marked by yes_state and continue from there */
	assert(st != yes_state);
#ifdef DEBUGGING
	while (st != yes_state) {
	    st--;
	    if (st < SLAB_FIRST(PL_regmatch_slab)) {
		PL_regmatch_slab = PL_regmatch_slab->prev;
		st = SLAB_LAST(PL_regmatch_slab);
	    }
	    DEBUG_STATE_r({
	        if (no_final) {
	            DEBUG_STATE_pp("pop (no final)");        
	        } else {
	            DEBUG_STATE_pp("pop (yes)");
	        }
	    });
	    depth--;
	}
#else
	while (yes_state < SLAB_FIRST(PL_regmatch_slab)
	    || yes_state > SLAB_LAST(PL_regmatch_slab))
	{
	    /* not in this slab, pop slab */
	    depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1);
	    PL_regmatch_slab = PL_regmatch_slab->prev;
	    st = SLAB_LAST(PL_regmatch_slab);
	}
	depth -= (st - yes_state);
#endif
	st = yes_state;
	yes_state = st->u.yes.prev_yes_state;
	PL_regmatch_state = st;
        
        if (no_final) {
            locinput= st->locinput;
            nextchr = UCHARAT(locinput);
        }
	state_num = st->resume_state + no_final;
	goto reenter_switch;
    }

    DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
			  PL_colors[4], PL_colors[5]));

    if (PL_reg_eval_set) {
	/* each successfully executed (?{...}) block does the equivalent of
	 *   local $^R = do {...}
	 * When popping the save stack, all these locals would be undone;
	 * bypass this by setting the outermost saved $^R to the latest
	 * value */
	if (oreplsv != GvSV(PL_replgv))
	    sv_setsv(oreplsv, GvSV(PL_replgv));
    }
    result = 1;
    goto final_exit;

no:
    DEBUG_EXECUTE_r(
	PerlIO_printf(Perl_debug_log,
            "%*s  %sfailed...%s\n",
            REPORT_CODE_OFF+depth*2, "", 
            PL_colors[4], PL_colors[5])
	);

no_silent:
    if (no_final) {
        if (yes_state) {
            goto yes;
        } else {
            goto final_exit;
        }
    }    
    if (depth) {
	/* there's a previous state to backtrack to */
	st--;
	if (st < SLAB_FIRST(PL_regmatch_slab)) {
	    PL_regmatch_slab = PL_regmatch_slab->prev;
	    st = SLAB_LAST(PL_regmatch_slab);
	}
	PL_regmatch_state = st;
	locinput= st->locinput;
	nextchr = UCHARAT(locinput);

	DEBUG_STATE_pp("pop");
	depth--;
	if (yes_state == st)
	    yes_state = st->u.yes.prev_yes_state;

	state_num = st->resume_state + 1; /* failure = success + 1 */
	goto reenter_switch;
    }
    result = 0;

  final_exit:
    if (rex->intflags & PREGf_VERBARG_SEEN) {
        SV *sv_err = get_sv("REGERROR", 1);
        SV *sv_mrk = get_sv("REGMARK", 1);
        if (result) {
            sv_commit = &PL_sv_no;
            if (!sv_yes_mark) 
                sv_yes_mark = &PL_sv_yes;
        } else {
            if (!sv_commit) 
                sv_commit = &PL_sv_yes;
            sv_yes_mark = &PL_sv_no;
        }
        sv_setsv(sv_err, sv_commit);
        sv_setsv(sv_mrk, sv_yes_mark);
    }

    /* clean up; in particular, free all slabs above current one */
    LEAVE_SCOPE(oldsave);

    return result;
}

/*
 - regrepeat - repeatedly match something simple, report how many
 */
/*
 * [This routine now assumes that it will only match on things of length 1.
 * That was true before, but now we assume scan - reginput is the count,
 * rather than incrementing count on every character.  [Er, except utf8.]]
 */
STATIC I32
S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max, int depth)
{
    dVAR;
    register char *scan;
    register I32 c;
    register char *loceol = PL_regeol;
    register I32 hardcount = 0;
    register bool utf8_target = PL_reg_match_utf8;
    UV utf8_flags;
#ifndef DEBUGGING
    PERL_UNUSED_ARG(depth);
#endif

    PERL_ARGS_ASSERT_REGREPEAT;

    scan = PL_reginput;
    if (max == REG_INFTY)
	max = I32_MAX;
    else if (max < loceol - scan)
	loceol = scan + max;
    switch (OP(p)) {
    case REG_ANY:
	if (utf8_target) {
	    loceol = PL_regeol;
	    while (scan < loceol && hardcount < max && *scan != '\n') {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && *scan != '\n')
		scan++;
	}
	break;
    case SANY:
        if (utf8_target) {
	    loceol = PL_regeol;
	    while (scan < loceol && hardcount < max) {
	        scan += UTF8SKIP(scan);
		hardcount++;
	    }
	}
	else
	    scan = loceol;
	break;
    case CANY:
	scan = loceol;
	break;
    case EXACT:
	/* To get here, EXACTish nodes must have *byte* length == 1.  That
	 * means they match only characters in the string that can be expressed
	 * as a single byte.  For non-utf8 strings, that means a simple match.
	 * For utf8 strings, the character matched must be an invariant, or
	 * downgradable to a single byte.  The pattern's utf8ness is
	 * irrelevant, as since it's a single byte, it either isn't utf8, or if
	 * it is, it's an invariant */

	c = (U8)*STRING(p);
	assert(! UTF_PATTERN || UNI_IS_INVARIANT(c));

	if (! utf8_target || UNI_IS_INVARIANT(c)) {
	    while (scan < loceol && UCHARAT(scan) == c) {
		scan++;
	    }
	}
	else {

	    /* Here, the string is utf8, and the pattern char is different
	     * in utf8 than not, so can't compare them directly.  Outside the
	     * loop, find the two utf8 bytes that represent c, and then
	     * look for those in sequence in the utf8 string */
	    U8 high = UTF8_TWO_BYTE_HI(c);
	    U8 low = UTF8_TWO_BYTE_LO(c);
	    loceol = PL_regeol;

	    while (hardcount < max
		    && scan + 1 < loceol
		    && UCHARAT(scan) == high
		    && UCHARAT(scan + 1) == low)
	    {
		scan += 2;
		hardcount++;
	    }
	}
	break;
    case EXACTFA:
	utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
	goto do_exactf;

    case EXACTFL:
	PL_reg_flags |= RF_tainted;
	utf8_flags = FOLDEQ_UTF8_LOCALE;
	goto do_exactf;

    case EXACTF:
	    utf8_flags = 0;
	    goto do_exactf;

    case EXACTFU_SS:
    case EXACTFU_TRICKYFOLD:
    case EXACTFU:
	utf8_flags = (UTF_PATTERN) ? FOLDEQ_S2_ALREADY_FOLDED : 0;

	/* The comments for the EXACT case above apply as well to these fold
	 * ones */

    do_exactf:
	c = (U8)*STRING(p);
	assert(! UTF_PATTERN || UNI_IS_INVARIANT(c));

	if (utf8_target || OP(p) == EXACTFU_SS) { /* Use full Unicode fold matching */
	    char *tmpeol = loceol;
	    while (hardcount < max
		    && foldEQ_utf8_flags(scan, &tmpeol, 0, utf8_target,
				   STRING(p), NULL, 1, cBOOL(UTF_PATTERN), utf8_flags))
	    {
		scan = tmpeol;
		tmpeol = loceol;
		hardcount++;
	    }

	    /* XXX Note that the above handles properly the German sharp s in
	     * the pattern matching ss in the string.  But it doesn't handle
	     * properly cases where the string contains say 'LIGATURE ff' and
	     * the pattern is 'f+'.  This would require, say, a new function or
	     * revised interface to foldEQ_utf8(), in which the maximum number
	     * of characters to match could be passed and it would return how
	     * many actually did.  This is just one of many cases where
	     * multi-char folds don't work properly, and so the fix is being
	     * deferred */
	}
	else {
	    U8 folded;

	    /* Here, the string isn't utf8 and c is a single byte; and either
	     * the pattern isn't utf8 or c is an invariant, so its utf8ness
	     * doesn't affect c.  Can just do simple comparisons for exact or
	     * fold matching. */
	    switch (OP(p)) {
		case EXACTF: folded = PL_fold[c]; break;
		case EXACTFA:
		case EXACTFU_TRICKYFOLD:
		case EXACTFU: folded = PL_fold_latin1[c]; break;
		case EXACTFL: folded = PL_fold_locale[c]; break;
		default: Perl_croak(aTHX_ "panic: Unexpected op %u", OP(p));
	    }
	    while (scan < loceol &&
		   (UCHARAT(scan) == c || UCHARAT(scan) == folded))
	    {
		scan++;
	    }
	}
	break;
    case ANYOFV:
    case ANYOF:
	if (utf8_target || OP(p) == ANYOFV) {
	    STRLEN inclasslen;
	    loceol = PL_regeol;
	    inclasslen = loceol - scan;
	    while (hardcount < max
		   && ((inclasslen = loceol - scan) > 0)
		   && reginclass(prog, p, (U8*)scan, &inclasslen, utf8_target))
	    {
		scan += inclasslen;
		hardcount++;
	    }
	} else {
	    while (scan < loceol && REGINCLASS(prog, p, (U8*)scan))
		scan++;
	}
	break;
    case ALNUMU:
	if (utf8_target) {
    utf8_wordchar:
	    loceol = PL_regeol;
	    LOAD_UTF8_CHARCLASS_ALNUM();
	    while (hardcount < max && scan < loceol &&
                   swash_fetch(PL_utf8_alnum, (U8*)scan, utf8_target))
            {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
        } else {
            while (scan < loceol && isWORDCHAR_L1((U8) *scan)) {
                scan++;
            }
	}
	break;
    case ALNUM:
	if (utf8_target)
	    goto utf8_wordchar;
	while (scan < loceol && isALNUM((U8) *scan)) {
	    scan++;
	}
	break;
    case ALNUMA:
	while (scan < loceol && isWORDCHAR_A((U8) *scan)) {
	    scan++;
	}
	break;
    case ALNUML:
	PL_reg_flags |= RF_tainted;
	if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol &&
		   isALNUM_LC_utf8((U8*)scan)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && isALNUM_LC(*scan))
		scan++;
	}
	break;
    case NALNUMU:
	if (utf8_target) {

    utf8_Nwordchar:

	    loceol = PL_regeol;
	    LOAD_UTF8_CHARCLASS_ALNUM();
	    while (hardcount < max && scan < loceol &&
                   ! swash_fetch(PL_utf8_alnum, (U8*)scan, utf8_target))
            {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
        } else {
            while (scan < loceol && ! isWORDCHAR_L1((U8) *scan)) {
                scan++;
            }
	}
	break;
    case NALNUM:
	if (utf8_target)
	    goto utf8_Nwordchar;
	while (scan < loceol && ! isALNUM((U8) *scan)) {
	    scan++;
	}
	break;
    case NALNUMA:
	if (utf8_target) {
	    while (scan < loceol && ! isWORDCHAR_A((U8) *scan)) {
		scan += UTF8SKIP(scan);
	    }
	}
	else {
	    while (scan < loceol && ! isWORDCHAR_A((U8) *scan)) {
		scan++;
	    }
	}
	break;
    case NALNUML:
	PL_reg_flags |= RF_tainted;
	if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol &&
		   !isALNUM_LC_utf8((U8*)scan)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && !isALNUM_LC(*scan))
		scan++;
	}
	break;
    case SPACEU:
	if (utf8_target) {

    utf8_space:

	    loceol = PL_regeol;
	    LOAD_UTF8_CHARCLASS_SPACE();
	    while (hardcount < max && scan < loceol &&
		   (*scan == ' ' ||
                    swash_fetch(PL_utf8_space,(U8*)scan, utf8_target)))
            {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	    break;
	}
	else {
            while (scan < loceol && isSPACE_L1((U8) *scan)) {
                scan++;
            }
	    break;
	}
    case SPACE:
	if (utf8_target)
	    goto utf8_space;

	while (scan < loceol && isSPACE((U8) *scan)) {
	    scan++;
	}
	break;
    case SPACEA:
	while (scan < loceol && isSPACE_A((U8) *scan)) {
	    scan++;
	}
	break;
    case SPACEL:
	PL_reg_flags |= RF_tainted;
	if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol &&
		   isSPACE_LC_utf8((U8*)scan)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && isSPACE_LC(*scan))
		scan++;
	}
	break;
    case NSPACEU:
	if (utf8_target) {

    utf8_Nspace:

	    loceol = PL_regeol;
	    LOAD_UTF8_CHARCLASS_SPACE();
	    while (hardcount < max && scan < loceol &&
		   ! (*scan == ' ' ||
                      swash_fetch(PL_utf8_space,(U8*)scan, utf8_target)))
            {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	    break;
	}
	else {
            while (scan < loceol && ! isSPACE_L1((U8) *scan)) {
                scan++;
            }
	}
	break;
    case NSPACE:
	if (utf8_target)
	    goto utf8_Nspace;

	while (scan < loceol && ! isSPACE((U8) *scan)) {
	    scan++;
	}
	break;
    case NSPACEA:
	if (utf8_target) {
	    while (scan < loceol && ! isSPACE_A((U8) *scan)) {
		scan += UTF8SKIP(scan);
	    }
	}
	else {
	    while (scan < loceol && ! isSPACE_A((U8) *scan)) {
		scan++;
	    }
	}
	break;
    case NSPACEL:
	PL_reg_flags |= RF_tainted;
	if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol &&
		   !isSPACE_LC_utf8((U8*)scan)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && !isSPACE_LC(*scan))
		scan++;
	}
	break;
    case DIGIT:
	if (utf8_target) {
	    loceol = PL_regeol;
	    LOAD_UTF8_CHARCLASS_DIGIT();
	    while (hardcount < max && scan < loceol &&
		   swash_fetch(PL_utf8_digit, (U8*)scan, utf8_target)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && isDIGIT(*scan))
		scan++;
	}
	break;
    case DIGITA:
	while (scan < loceol && isDIGIT_A((U8) *scan)) {
	    scan++;
	}
	break;
    case DIGITL:
	PL_reg_flags |= RF_tainted;
	if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol &&
		   isDIGIT_LC_utf8((U8*)scan)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && isDIGIT_LC(*scan))
		scan++;
	}
	break;
    case NDIGIT:
	if (utf8_target) {
	    loceol = PL_regeol;
	    LOAD_UTF8_CHARCLASS_DIGIT();
	    while (hardcount < max && scan < loceol &&
		   !swash_fetch(PL_utf8_digit, (U8*)scan, utf8_target)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && !isDIGIT(*scan))
		scan++;
	}
	break;
    case NDIGITA:
	if (utf8_target) {
	    while (scan < loceol && ! isDIGIT_A((U8) *scan)) {
		scan += UTF8SKIP(scan);
	    }
	}
	else {
	    while (scan < loceol && ! isDIGIT_A((U8) *scan)) {
		scan++;
	    }
	}
	break;
    case NDIGITL:
	PL_reg_flags |= RF_tainted;
	if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol &&
		   !isDIGIT_LC_utf8((U8*)scan)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && !isDIGIT_LC(*scan))
		scan++;
	}
	break;
    case LNBREAK:
        if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol && (c=is_LNBREAK_utf8(scan))) {
		scan += c;
		hardcount++;
	    }
	} else {
	    /*
	      LNBREAK can match two latin chars, which is ok,
	      because we have a null terminated string, but we
	      have to use hardcount in this situation
	    */
	    while (scan < loceol && (c=is_LNBREAK_latin1(scan)))  {
		scan+=c;
		hardcount++;
	    }
	}	
	break;
    case HORIZWS:
        if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol && (c=is_HORIZWS_utf8(scan))) {
		scan += c;
		hardcount++;
	    }
	} else {
	    while (scan < loceol && is_HORIZWS_latin1(scan)) 
		scan++;		
	}	
	break;
    case NHORIZWS:
        if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol && !is_HORIZWS_utf8(scan)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && !is_HORIZWS_latin1(scan))
		scan++;

	}	
	break;
    case VERTWS:
        if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol && (c=is_VERTWS_utf8(scan))) {
		scan += c;
		hardcount++;
	    }
	} else {
	    while (scan < loceol && is_VERTWS_latin1(scan)) 
		scan++;

	}	
	break;
    case NVERTWS:
        if (utf8_target) {
	    loceol = PL_regeol;
	    while (hardcount < max && scan < loceol && !is_VERTWS_utf8(scan)) {
		scan += UTF8SKIP(scan);
		hardcount++;
	    }
	} else {
	    while (scan < loceol && !is_VERTWS_latin1(scan)) 
		scan++;
          
	}	
	break;

    default:		/* Called on something of 0 width. */
	break;		/* So match right here or not at all. */
    }

    if (hardcount)
	c = hardcount;
    else
	c = scan - PL_reginput;
    PL_reginput = scan;

    DEBUG_r({
	GET_RE_DEBUG_FLAGS_DECL;
	DEBUG_EXECUTE_r({
	    SV * const prop = sv_newmortal();
	    regprop(prog, prop, p);
	    PerlIO_printf(Perl_debug_log,
			"%*s  %s can match %"IVdf" times out of %"IVdf"...\n",
			REPORT_CODE_OFF + depth*2, "", SvPVX_const(prop),(IV)c,(IV)max);
	});
    });

    return(c);
}


#if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION)
/*
- regclass_swash - prepare the utf8 swash.  Wraps the shared core version to
create a copy so that changes the caller makes won't change the shared one
 */
SV *
Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
{
    PERL_ARGS_ASSERT_REGCLASS_SWASH;
    return newSVsv(core_regclass_swash(prog, node, doinit, listsvp, altsvp));
}
#endif

STATIC SV *
S_core_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
{
    /* Returns the swash for the input 'node' in the regex 'prog'.
     * If <doinit> is true, will attempt to create the swash if not already
     *	  done.
     * If <listsvp> is non-null, will return the swash initialization string in
     *	  it.
     * If <altsvp> is non-null, will return the alternates to the regular swash
     *	  in it
     * Tied intimately to how regcomp.c sets up the data structure */

    dVAR;
    SV *sw  = NULL;
    SV *si  = NULL;
    SV *alt = NULL;
    SV*  invlist = NULL;

    RXi_GET_DECL(prog,progi);
    const struct reg_data * const data = prog ? progi->data : NULL;

    PERL_ARGS_ASSERT_CORE_REGCLASS_SWASH;

    assert(ANYOF_NONBITMAP(node));

    if (data && data->count) {
	const U32 n = ARG(node);

	if (data->what[n] == 's') {
	    SV * const rv = MUTABLE_SV(data->data[n]);
	    AV * const av = MUTABLE_AV(SvRV(rv));
	    SV **const ary = AvARRAY(av);
	    bool invlist_has_user_defined_property;
	
	    si = *ary;	/* ary[0] = the string to initialize the swash with */

	    /* Elements 3 and 4 are either both present or both absent. [3] is
	     * any inversion list generated at compile time; [4] indicates if
	     * that inversion list has any user-defined properties in it. */
	    if (av_len(av) >= 3) {
		invlist = ary[3];
		invlist_has_user_defined_property = cBOOL(SvUV(ary[4]));
	    }
	    else {
		invlist = NULL;
		invlist_has_user_defined_property = FALSE;
	    }

	    /* Element [1] is reserved for the set-up swash.  If already there,
	     * return it; if not, create it and store it there */
	    if (SvROK(ary[1])) {
		sw = ary[1];
	    }
	    else if (si && doinit) {

		sw = _core_swash_init("utf8", /* the utf8 package */
				      "", /* nameless */
				      si,
				      1, /* binary */
				      0, /* not from tr/// */
				      FALSE, /* is error if can't find
						property */
				      invlist,
				      invlist_has_user_defined_property);
		(void)av_store(av, 1, sw);
	    }

	    /* Element [2] is for any multi-char folds.  Note that is a
	     * fundamentally flawed design, because can't backtrack and try
	     * again.  See [perl #89774] */
	    if (SvTYPE(ary[2]) == SVt_PVAV) {
	        alt = ary[2];
	    }
	}
    }
	
    if (listsvp) {
	SV* matches_string = newSVpvn("", 0);
	SV** invlistsvp;

	/* Use the swash, if any, which has to have incorporated into it all
	 * possibilities */
	if (   sw
	    && SvROK(sw)
	    && SvTYPE(SvRV(sw)) == SVt_PVHV
	    && (invlistsvp = hv_fetchs(MUTABLE_HV(SvRV(sw)), "INVLIST", FALSE)))
	{
	    invlist = *invlistsvp;
	}
	else if (si && si != &PL_sv_undef) {

	    /* If no swash, use the input nitialization string, if available */
	    sv_catsv(matches_string, si);
	}

	/* Add the inversion list to whatever we have.  This may have come from
	 * the swash, or from an input parameter */
	if (invlist) {
	    sv_catsv(matches_string, _invlist_contents(invlist));
	}
	*listsvp = matches_string;
    }

    if (altsvp)
	*altsvp  = alt;

    return sw;
}

/*
 - reginclass - determine if a character falls into a character class
 
  n is the ANYOF regnode
  p is the target string
  lenp is pointer to the maximum number of bytes of how far to go in p
    (This is assumed wthout checking to always be at least the current
    character's size)
  utf8_target tells whether p is in UTF-8.

  Returns true if matched; false otherwise.  If lenp is not NULL, on return
  from a successful match, the value it points to will be updated to how many
  bytes in p were matched.  If there was no match, the value is undefined,
  possibly changed from the input.

  Note that this can be a synthetic start class, a combination of various
  nodes, so things you think might be mutually exclusive, such as locale,
  aren't.  It can match both locale and non-locale

 */

STATIC bool
S_reginclass(pTHX_ const regexp * const prog, register const regnode * const n, register const U8* const p, STRLEN* lenp, register const bool utf8_target)
{
    dVAR;
    const char flags = ANYOF_FLAGS(n);
    bool match = FALSE;
    UV c = *p;
    STRLEN c_len = 0;
    STRLEN maxlen;

    PERL_ARGS_ASSERT_REGINCLASS;

    /* If c is not already the code point, get it */
    if (utf8_target && !UTF8_IS_INVARIANT(c)) {
	c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &c_len,
		(UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV)
		| UTF8_ALLOW_FFFF | UTF8_CHECK_ONLY);
		/* see [perl #37836] for UTF8_ALLOW_ANYUV; [perl #38293] for
		 * UTF8_ALLOW_FFFF */
	if (c_len == (STRLEN)-1)
	    Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)");
    }
    else {
	c_len = 1;
    }

    /* Use passed in max length, or one character if none passed in or less
     * than one character.  And assume will match just one character.  This is
     * overwritten later if matched more. */
    if (lenp) {
	maxlen = (*lenp > c_len) ? *lenp : c_len;
	*lenp = c_len;

    }
    else {
	maxlen = c_len;
    }

    /* If this character is potentially in the bitmap, check it */
    if (c < 256) {
	if (ANYOF_BITMAP_TEST(n, c))
	    match = TRUE;
	else if (flags & ANYOF_NON_UTF8_LATIN1_ALL
		&& ! utf8_target
		&& ! isASCII(c))
	{
	    match = TRUE;
	}

	else if (flags & ANYOF_LOCALE) {
	    PL_reg_flags |= RF_tainted;

	    if ((flags & ANYOF_LOC_NONBITMAP_FOLD)
		 && ANYOF_BITMAP_TEST(n, PL_fold_locale[c]))
	    {
		match = TRUE;
	    }
	    else if (ANYOF_CLASS_TEST_ANY_SET(n) &&
		     ((ANYOF_CLASS_TEST(n, ANYOF_ALNUM)   &&  isALNUM_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NALNUM)  && !isALNUM_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_SPACE)   &&  isSPACE_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NSPACE)  && !isSPACE_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_DIGIT)   &&  isDIGIT_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT)  && !isDIGIT_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC)  &&  isALNUMC_LC(c)) ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) ||
		      (ANYOF_CLASS_TEST(n, ANYOF_ALPHA)   &&  isALPHA_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NALPHA)  && !isALPHA_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_ASCII)   &&  isASCII_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NASCII)  && !isASCII_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_CNTRL)   &&  isCNTRL_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL)  && !isCNTRL_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_GRAPH)   &&  isGRAPH_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH)  && !isGRAPH_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_LOWER)   &&  isLOWER_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NLOWER)  && !isLOWER_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_PRINT)   &&  isPRINT_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NPRINT)  && !isPRINT_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_PUNCT)   &&  isPUNCT_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT)  && !isPUNCT_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_UPPER)   &&  isUPPER_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NUPPER)  && !isUPPER_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT)  &&  isXDIGIT(c))    ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c))    ||
		      (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC)  &&  isPSXSPC(c))    ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c))    ||
		      (ANYOF_CLASS_TEST(n, ANYOF_BLANK)   &&  isBLANK_LC(c))  ||
		      (ANYOF_CLASS_TEST(n, ANYOF_NBLANK)  && !isBLANK_LC(c))
		     ) /* How's that for a conditional? */
	    ) {
		match = TRUE;
	    }
	}
    }

    /* If the bitmap didn't (or couldn't) match, and something outside the
     * bitmap could match, try that.  Locale nodes specifiy completely the
     * behavior of code points in the bit map (otherwise, a utf8 target would
     * cause them to be treated as Unicode and not locale), except in
     * the very unlikely event when this node is a synthetic start class, which
     * could be a combination of locale and non-locale nodes.  So allow locale
     * to match for the synthetic start class, which will give a false
     * positive that will be resolved when the match is done again as not part
     * of the synthetic start class */
    if (!match) {
	if (utf8_target && (flags & ANYOF_UNICODE_ALL) && c >= 256) {
	    match = TRUE;	/* Everything above 255 matches */
	}
	else if (ANYOF_NONBITMAP(n)
		 && ((flags & ANYOF_NONBITMAP_NON_UTF8)
		     || (utf8_target
		         && (c >=256
			     || (! (flags & ANYOF_LOCALE))
			     || (flags & ANYOF_IS_SYNTHETIC)))))
	{
	    AV *av;
	    SV * const sw = core_regclass_swash(prog, n, TRUE, 0, (SV**)&av);

	    if (sw) {
		U8 * utf8_p;
		if (utf8_target) {
		    utf8_p = (U8 *) p;
		} else {

		    /* Not utf8.  Convert as much of the string as available up
		     * to the limit of how far the (single) character in the
		     * pattern can possibly match (no need to go further).  If
		     * the node is a straight ANYOF or not folding, it can't
		     * match more than one.  Otherwise, It can match up to how
		     * far a single char can fold to.  Since not utf8, each
		     * character is a single byte, so the max it can be in
		     * bytes is the same as the max it can be in characters */
		    STRLEN len = (OP(n) == ANYOF
				  || ! (flags & ANYOF_LOC_NONBITMAP_FOLD))
				  ? 1
				  : (maxlen < UTF8_MAX_FOLD_CHAR_EXPAND)
				    ? maxlen
				    : UTF8_MAX_FOLD_CHAR_EXPAND;
		    utf8_p = bytes_to_utf8(p, &len);
		}

		if (swash_fetch(sw, utf8_p, TRUE))
		    match = TRUE;
		else if (flags & ANYOF_LOC_NONBITMAP_FOLD) {

		    /* Here, we need to test if the fold of the target string
		     * matches.  The non-multi char folds have all been moved to
                     * the compilation phase, and the multi-char folds have
                     * been stored by regcomp into 'av'; we linearly check to
                     * see if any match the target string (folded).   We know
                     * that the originals were each one character, but we don't
                     * currently know how many characters/bytes each folded to,
                     * except we do know that there are small limits imposed by
                     * Unicode.  XXX A performance enhancement would be to have
                     * regcomp.c store the max number of chars/bytes that are
                     * in an av entry, as, say the 0th element.  Even better
                     * would be to have a hash of the few characters that can
                     * start a multi-char fold to the max number of chars of
                     * those folds.
		     *
		     * If there is a match, we will need to advance (if lenp is
		     * specified) the match pointer in the target string.  But
		     * what we are comparing here isn't that string directly,
		     * but its fold, whose length may differ from the original.
		     * As we go along in constructing the fold, therefore, we
		     * create a map so that we know how many bytes in the
		     * source to advance given that we have matched a certain
		     * number of bytes in the fold.  This map is stored in
		     * 'map_fold_len_back'.  Let n mean the number of bytes in
		     * the fold of the first character that we are folding.
		     * Then map_fold_len_back[n] is set to the number of bytes
		     * in that first character.  Similarly let m be the
		     * corresponding number for the second character to be
		     * folded.  Then map_fold_len_back[n+m] is set to the
		     * number of bytes occupied by the first two source
		     * characters. ... */
		    U8 map_fold_len_back[UTF8_MAXBYTES_CASE+1] = { 0 };
		    U8 folded[UTF8_MAXBYTES_CASE+1];
		    STRLEN foldlen = 0; /* num bytes in fold of 1st char */
		    STRLEN total_foldlen = 0; /* num bytes in fold of all
						  chars */

		    if (OP(n) == ANYOF || maxlen == 1 || ! lenp || ! av) {

			/* Here, only need to fold the first char of the target
			 * string.  It the source wasn't utf8, is 1 byte long */
			to_utf8_fold(utf8_p, folded, &foldlen);
			total_foldlen = foldlen;
			map_fold_len_back[foldlen] = (utf8_target)
						     ? UTF8SKIP(utf8_p)
						     : 1;
		    }
		    else {

			/* Here, need to fold more than the first char.  Do so
			 * up to the limits */
			U8* source_ptr = utf8_p;    /* The source for the fold
						       is the regex target
						       string */
			U8* folded_ptr = folded;
			U8* e = utf8_p + maxlen;    /* Can't go beyond last
						       available byte in the
						       target string */
			U8 i;
			for (i = 0;
			     i < UTF8_MAX_FOLD_CHAR_EXPAND && source_ptr < e;
			     i++)
			{

			    /* Fold the next character */
			    U8 this_char_folded[UTF8_MAXBYTES_CASE+1];
			    STRLEN this_char_foldlen;
			    to_utf8_fold(source_ptr,
				         this_char_folded,
					 &this_char_foldlen);

			    /* Bail if it would exceed the byte limit for
			     * folding a single char. */
			    if (this_char_foldlen + folded_ptr - folded >
							    UTF8_MAXBYTES_CASE)
			    {
				break;
			    }

			    /* Add the fold of this character */
			    Copy(this_char_folded,
				 folded_ptr,
				 this_char_foldlen,
				 U8);
			    source_ptr += UTF8SKIP(source_ptr);
			    folded_ptr += this_char_foldlen;
			    total_foldlen = folded_ptr - folded;

			    /* Create map from the number of bytes in the fold
			     * back to the number of bytes in the source.  If
			     * the source isn't utf8, the byte count is just
			     * the number of characters so far */
			    map_fold_len_back[total_foldlen]
						      = (utf8_target)
							? source_ptr - utf8_p
							: i + 1;
			}
			*folded_ptr = '\0';
		    }


		    /* Do the linear search to see if the fold is in the list
		     * of multi-char folds. */
		    if (av) {
		        I32 i;
			for (i = 0; i <= av_len(av); i++) {
			    SV* const sv = *av_fetch(av, i, FALSE);
			    STRLEN len;
			    const char * const s = SvPV_const(sv, len);

			    if (len <= total_foldlen
				&& memEQ(s, (char*)folded, len)

				   /* If 0, means matched a partial char. See
				    * [perl #90536] */
				&& map_fold_len_back[len])
			    {

				/* Advance the target string ptr to account for
				 * this fold, but have to translate from the
				 * folded length to the corresponding source
				 * length. */
				if (lenp) {
				    *lenp = map_fold_len_back[len];
				}
				match = TRUE;
				break;
			    }
			}
		    }
		}

		/* If we allocated a string above, free it */
		if (! utf8_target) Safefree(utf8_p);
	    }
	}
    }

    return (flags & ANYOF_INVERT) ? !match : match;
}

STATIC U8 *
S_reghop3(U8 *s, I32 off, const U8* lim)
{
    /* return the position 'off' UTF-8 characters away from 's', forward if
     * 'off' >= 0, backwards if negative.  But don't go outside of position
     * 'lim', which better be < s  if off < 0 */

    dVAR;

    PERL_ARGS_ASSERT_REGHOP3;

    if (off >= 0) {
	while (off-- && s < lim) {
	    /* XXX could check well-formedness here */
	    s += UTF8SKIP(s);
	}
    }
    else {
        while (off++ && s > lim) {
            s--;
            if (UTF8_IS_CONTINUED(*s)) {
                while (s > lim && UTF8_IS_CONTINUATION(*s))
                    s--;
	    }
            /* XXX could check well-formedness here */
	}
    }
    return s;
}

#ifdef XXX_dmq
/* there are a bunch of places where we use two reghop3's that should
   be replaced with this routine. but since thats not done yet 
   we ifdef it out - dmq
*/
STATIC U8 *
S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim)
{
    dVAR;

    PERL_ARGS_ASSERT_REGHOP4;

    if (off >= 0) {
        while (off-- && s < rlim) {
            /* XXX could check well-formedness here */
            s += UTF8SKIP(s);
        }
    }
    else {
        while (off++ && s > llim) {
            s--;
            if (UTF8_IS_CONTINUED(*s)) {
                while (s > llim && UTF8_IS_CONTINUATION(*s))
                    s--;
            }
            /* XXX could check well-formedness here */
        }
    }
    return s;
}
#endif

STATIC U8 *
S_reghopmaybe3(U8* s, I32 off, const U8* lim)
{
    dVAR;

    PERL_ARGS_ASSERT_REGHOPMAYBE3;

    if (off >= 0) {
	while (off-- && s < lim) {
	    /* XXX could check well-formedness here */
	    s += UTF8SKIP(s);
	}
	if (off >= 0)
	    return NULL;
    }
    else {
        while (off++ && s > lim) {
            s--;
            if (UTF8_IS_CONTINUED(*s)) {
                while (s > lim && UTF8_IS_CONTINUATION(*s))
                    s--;
	    }
            /* XXX could check well-formedness here */
	}
	if (off <= 0)
	    return NULL;
    }
    return s;
}

static void
restore_pos(pTHX_ void *arg)
{
    dVAR;
    regexp * const rex = (regexp *)arg;
    if (PL_reg_eval_set) {
	if (PL_reg_oldsaved) {
	    rex->subbeg = PL_reg_oldsaved;
	    rex->sublen = PL_reg_oldsavedlen;
#ifdef PERL_OLD_COPY_ON_WRITE
	    rex->saved_copy = PL_nrs;
#endif
	    RXp_MATCH_COPIED_on(rex);
	}
	PL_reg_magic->mg_len = PL_reg_oldpos;
	PL_reg_eval_set = 0;
	PL_curpm = PL_reg_oldcurpm;
    }	
}

STATIC void
S_to_utf8_substr(pTHX_ register regexp *prog)
{
    int i = 1;

    PERL_ARGS_ASSERT_TO_UTF8_SUBSTR;

    do {
	if (prog->substrs->data[i].substr
	    && !prog->substrs->data[i].utf8_substr) {
	    SV* const sv = newSVsv(prog->substrs->data[i].substr);
	    prog->substrs->data[i].utf8_substr = sv;
	    sv_utf8_upgrade(sv);
	    if (SvVALID(prog->substrs->data[i].substr)) {
		if (SvTAIL(prog->substrs->data[i].substr)) {
		    /* Trim the trailing \n that fbm_compile added last
		       time.  */
		    SvCUR_set(sv, SvCUR(sv) - 1);
		    /* Whilst this makes the SV technically "invalid" (as its
		       buffer is no longer followed by "\0") when fbm_compile()
		       adds the "\n" back, a "\0" is restored.  */
		    fbm_compile(sv, FBMcf_TAIL);
		} else
		    fbm_compile(sv, 0);
	    }
	    if (prog->substrs->data[i].substr == prog->check_substr)
		prog->check_utf8 = sv;
	}
    } while (i--);
}

STATIC void
S_to_byte_substr(pTHX_ register regexp *prog)
{
    dVAR;
    int i = 1;

    PERL_ARGS_ASSERT_TO_BYTE_SUBSTR;

    do {
	if (prog->substrs->data[i].utf8_substr
	    && !prog->substrs->data[i].substr) {
	    SV* sv = newSVsv(prog->substrs->data[i].utf8_substr);
	    if (sv_utf8_downgrade(sv, TRUE)) {
		if (SvVALID(prog->substrs->data[i].utf8_substr)) {
		    if (SvTAIL(prog->substrs->data[i].utf8_substr)) {
			/* Trim the trailing \n that fbm_compile added last
			   time.  */
			SvCUR_set(sv, SvCUR(sv) - 1);
			fbm_compile(sv, FBMcf_TAIL);
		    } else
			fbm_compile(sv, 0);
		}
	    } else {
		SvREFCNT_dec(sv);
		sv = &PL_sv_undef;
	    }
	    prog->substrs->data[i].substr = sv;
	    if (prog->substrs->data[i].utf8_substr == prog->check_utf8)
		prog->check_substr = sv;
	}
    } while (i--);
}

/*
 * Local variables:
 * c-indentation-style: bsd
 * c-basic-offset: 4
 * indent-tabs-mode: t
 * End:
 *
 * ex: set ts=8 sts=4 sw=4 noet:
 */