sync to 1.9.5: lookup hashes are now static tables

shortening the code, and, according to kristaps@, speeding it up

8 files changed, 84 insertions, 194 deletions

diff --git a/usr.bin/mandoc/libman.h b/usr.bin/mandoc/libman.h
index 70c1e8e4edf..f0a5de14a4b 100644
--- a/usr.bin/mandoc/libman.h
+++ b/usr.bin/mandoc/libman.h
@@ -1,4 +1,4 @@
-/*	$Id: libman.h,v 1.8 2009/08/22 23:17:39 schwarze Exp $ */
+/*	$Id: libman.h,v 1.9 2009/09/21 21:11:36 schwarze Exp $ */
 /*
  * Copyright (c) 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -27,7 +27,6 @@ enum	man_next {
 struct	man {
 	void		*data;
 	struct man_cb	 cb;
-	void		*htab;
 	int		 pflags;
 	int		 flags;
 #define	MAN_HALT	(1 << 0)
@@ -95,9 +94,8 @@ int		  man_body_alloc(struct man *, int, int, int);
 int		  man_elem_alloc(struct man *, int, int, int);
 void		  man_node_free(struct man_node *);
 void		  man_node_freelist(struct man_node *);
-void		 *man_hash_alloc(void);
-int		  man_hash_find(const void *, const char *);
-void		  man_hash_free(void *);
+void		  man_hash_init(void);
+int		  man_hash_find(const char *);
 int		  man_macroend(struct man *);
 int		  man_args(struct man *, int, int *, char *, char **);
 #define	ARGS_ERROR	(-1)
diff --git a/usr.bin/mandoc/libmdoc.h b/usr.bin/mandoc/libmdoc.h
index e8c4ac6abea..f51f9d709a1 100644
--- a/usr.bin/mandoc/libmdoc.h
+++ b/usr.bin/mandoc/libmdoc.h
@@ -1,4 +1,4 @@
-/*	$Id: libmdoc.h,v 1.20 2009/08/22 22:39:55 schwarze Exp $ */
+/*	$Id: libmdoc.h,v 1.21 2009/09/21 21:11:37 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -27,7 +27,6 @@ enum	mdoc_next {
 struct	mdoc {
 	void		 *data;
 	struct mdoc_cb	  cb;
-	void		 *htab;
 	int		  flags;
 #define	MDOC_HALT	 (1 << 0)	/* Error in parse. Halt. */
 #define	MDOC_LITERAL	 (1 << 1)	/* In a literal scope. */
@@ -139,9 +138,8 @@ int		  mdoc_tail_alloc(struct mdoc *, int, int, int);
 int		  mdoc_body_alloc(struct mdoc *, int, int, int);
 void		  mdoc_node_free(struct mdoc_node *);
 void		  mdoc_node_freelist(struct mdoc_node *);
-void		 *mdoc_hash_alloc(void);
-int		  mdoc_hash_find(const void *, const char *);
-void		  mdoc_hash_free(void *);
+void		  mdoc_hash_init(void);
+int		  mdoc_hash_find(const char *);
 int		  mdoc_iscdelim(char);
 int		  mdoc_isdelim(const char *);
 size_t		  mdoc_isescape(const char *);
diff --git a/usr.bin/mandoc/man.c b/usr.bin/mandoc/man.c
index c3888bb684c..d749a73eff3 100644
--- a/usr.bin/mandoc/man.c
+++ b/usr.bin/mandoc/man.c
@@ -1,4 +1,4 @@
-/*	$Id: man.c,v 1.12 2009/09/18 22:46:14 schwarze Exp $ */
+/*	$Id: man.c,v 1.13 2009/09/21 21:11:37 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -101,9 +101,6 @@ man_free(struct man *man)
 {
 
 	man_free1(man);
-
-	if (man->htab)
-		man_hash_free(man->htab);
 	free(man);
 }
 
@@ -121,14 +118,11 @@ man_alloc(void *data, int pflags, const struct man_cb *cb)
 		return(NULL);
 	}
 
+	man_hash_init();
+
 	p->data = data;
 	p->pflags = pflags;
 	(void)memcpy(&p->cb, cb, sizeof(struct man_cb));
-
-	if (NULL == (p->htab = man_hash_alloc())) {
-		free(p);
-		return(NULL);
-	}
 	return(p);
 }
 
@@ -509,7 +503,7 @@ man_pmacro(struct man *m, int ln, char *buf)
 		return(1);
 	}
 	
-	if (MAN_MAX == (c = man_hash_find(m->htab, mac))) {
+	if (MAN_MAX == (c = man_hash_find(mac))) {
 		if ( ! (MAN_IGN_MACRO & m->pflags)) {
 			(void)man_perr(m, ln, ppos, WMACRO);
 			goto err;
diff --git a/usr.bin/mandoc/man_hash.c b/usr.bin/mandoc/man_hash.c
index 2b83f0f56b3..b7d376fa2bf 100644
--- a/usr.bin/mandoc/man_hash.c
+++ b/usr.bin/mandoc/man_hash.c
@@ -1,4 +1,4 @@
-/*	$Id: man_hash.c,v 1.5 2009/08/22 23:17:40 schwarze Exp $ */
+/*	$Id: man_hash.c,v 1.6 2009/09/21 21:11:37 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -15,39 +15,27 @@
  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  */
 #include <assert.h>
+#include <limits.h>
 #include <stdlib.h>
 #include <string.h>
 
 #include "libman.h"
 
+static	u_char		table[26 * 6];
 
-/* ARGUSED */
+/*
+ * XXX - this hash has global scope, so if intended for use as a library
+ * with multiple callers, it will need re-invocation protection.
+ */
 void
-man_hash_free(void *htab)
-{
-
-	free(htab);
-}
-
-
-/* ARGUSED */
-void *
-man_hash_alloc(void)
+man_hash_init(void)
 {
-	int		*htab;
 	int		 i, j, x;
 
-	/* Initialised to -1. */
-
-	htab = malloc(26 * 6 * sizeof(int));
-	if (NULL == htab)
-		return(NULL);
-	for (i = 0; i < 26 * 6; i++)
-		htab[i] = -1;
+	memset(table, UCHAR_MAX, sizeof(table));
 
 	for (i = 0; i < MAN_MAX; i++) {
 		x = man_macronames[i][0];
-
 		assert((x >= 65 && x <= 90) ||
 				(x >= 97 && x <= 122));
 
@@ -55,25 +43,18 @@ man_hash_alloc(void)
 		x *= 6;
 
 		for (j = 0; j < 6; j++)
-			if (-1 == htab[x + j]) {
-				htab[x + j] = i;
+			if (UCHAR_MAX == table[x + j]) {
+				table[x + j] = (u_char)i;
 				break;
 			}
-
 		assert(j < 6);
 	}
-
-	return((void *)htab);
 }
 
-
 int
-man_hash_find(const void *arg, const char *tmp)
+man_hash_find(const char *tmp)
 {
 	int		 x, i, tok;
-	const int	*htab;
-
-	htab = (const int *)arg;
 
 	if (0 == (x = tmp[0]))
 		return(MAN_MAX);
@@ -84,7 +65,7 @@ man_hash_find(const void *arg, const char *tmp)
 	x *= 6;
 
 	for (i = 0; i < 6; i++) {
-		if (-1 == (tok = htab[x + i]))
+		if (UCHAR_MAX == (tok = table[x + i]))
 			return(MAN_MAX);
 		if (0 == strcmp(tmp, man_macronames[tok]))
 			return(tok);
@@ -92,4 +73,3 @@ man_hash_find(const void *arg, const char *tmp)
 
 	return(MAN_MAX);
 }
-
diff --git a/usr.bin/mandoc/mdoc.c b/usr.bin/mandoc/mdoc.c
index da3983aca93..f29b4b0e836 100644
--- a/usr.bin/mandoc/mdoc.c
+++ b/usr.bin/mandoc/mdoc.c
@@ -1,4 +1,4 @@
-/*	$Id: mdoc.c,v 1.26 2009/09/18 22:37:05 schwarze Exp $ */
+/*	$Id: mdoc.c,v 1.27 2009/09/21 21:11:37 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -226,8 +226,6 @@ mdoc_free(struct mdoc *mdoc)
 {
 
 	mdoc_free1(mdoc);
-	if (mdoc->htab)
-		mdoc_hash_free(mdoc->htab);
 	free(mdoc);
 }
 
@@ -245,13 +243,12 @@ mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb)
 	if (cb)
 		(void)memcpy(&p->cb, cb, sizeof(struct mdoc_cb));
 
+	mdoc_hash_init();
+
 	p->data = data;
 	p->pflags = pflags;
 
-	if (NULL == (p->htab = mdoc_hash_alloc())) {
-		free(p);
-		return(NULL);
-	} else if (mdoc_alloc1(p))
+	if (mdoc_alloc1(p))
 		return(p);
 
 	free(p);
@@ -721,7 +718,7 @@ parsemacro(struct mdoc *m, int ln, char *buf)
 		return(1);
 	} 
 	
-	if (MDOC_MAX == (c = mdoc_hash_find(m->htab, mac))) {
+	if (MDOC_MAX == (c = mdoc_hash_find(mac))) {
 		if ( ! macrowarn(m, ln, mac))
 			goto err;
 		return(1);
diff --git a/usr.bin/mandoc/mdoc_action.c b/usr.bin/mandoc/mdoc_action.c
index 258b8bb059c..b19180a3936 100644
--- a/usr.bin/mandoc/mdoc_action.c
+++ b/usr.bin/mandoc/mdoc_action.c
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_action.c,v 1.20 2009/08/22 22:50:17 schwarze Exp $ */
+/*	$Id: mdoc_action.c,v 1.21 2009/09/21 21:11:37 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -53,7 +53,7 @@ static	int	  post_tilde(POST_ARGS);
 static	int	  pre_bd(PRE_ARGS);
 static	int	  pre_dl(PRE_ARGS);
 
-const	struct actions mdoc_actions[MDOC_MAX] = {
+static	const struct actions mdoc_actions[MDOC_MAX] = {
 	{ NULL, NULL }, /* Ap */
 	{ NULL, post_dd }, /* Dd */ 
 	{ NULL, post_dt }, /* Dt */ 
@@ -613,7 +613,7 @@ post_bl_width(POST_ARGS)
 
 	if (0 == strcmp(p, "Ds"))
 		width = 6;
-	else if (MDOC_MAX == (tok = mdoc_hash_find(m->htab, p)))
+	else if (MDOC_MAX == (tok = mdoc_hash_find(p)))
 		return(1);
 	else if (0 == (width = mdoc_macro2len(tok))) 
 		return(mdoc_nwarn(m, n, ENOWIDTH));
diff --git a/usr.bin/mandoc/mdoc_hash.c b/usr.bin/mandoc/mdoc_hash.c
index b188e340bf7..4af63bea72c 100644
--- a/usr.bin/mandoc/mdoc_hash.c
+++ b/usr.bin/mandoc/mdoc_hash.c
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_hash.c,v 1.5 2009/08/09 18:01:15 schwarze Exp $ */
+/*	$Id: mdoc_hash.c,v 1.6 2009/09/21 21:11:37 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -18,149 +18,71 @@
 
 #include <assert.h>
 #include <ctype.h>
+#include <limits.h>
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
 
 #include "libmdoc.h"
 
-#define	ADJUST_MAJOR(x) 					\
-	do if (37 == (x))					\
-		(x) = 0; 		/* %   -> 00 */		\
-	else if (91 > (x)) 					\
-		(x) -= 64; 		/* A-Z -> 01 - 26 */	\
-	else 							\
-		(x) -= 70;		/* a-z -> 27 - 52 */	\
-	while (/*CONSTCOND*/0)
-
-#define ADJUST_MINOR(y)						\
-	do if (49 == (y))					\
-		(y) = 0;		/* 1   -> 00 */		\
-	else if (91 > (y))					\
-		(y) -= 65;		/* A-Z -> 00 - 25 */	\
-	else 							\
-		(y) -= 97;		/* a-z -> 00 - 25 */	\
-	while (/*CONSTCOND*/0)
-
-#define INDEX(maj, min) 					\
-	((maj) * 26 * 3) + ((min) * 3)
-
-#define	SLOTCMP(slot, val)					\
-	(mdoc_macronames[(slot)][0] == (val)[0] && 		\
-	 mdoc_macronames[(slot)][1] == (val)[1] && 		\
-	 (0 == (val)[2] || 					\
-	  mdoc_macronames[(slot)][2] == (val)[2]))
-
+static	u_char		 table[27 * 12];
 
+/*
+ * XXX - this hash has global scope, so if intended for use as a library
+ * with multiple callers, it will need re-invocation protection.
+ */
 void
-mdoc_hash_free(void *htab)
+mdoc_hash_init(void)
 {
+	int		 i, j, major;
+	const char	*p;
 
-	free(htab);
-}
-
-
-
-void *
-mdoc_hash_alloc(void)
-{
-	int		  i, major, minor, ind;
-	const void	**htab;
-
-	htab = calloc(26 * 3 * 52, sizeof(struct mdoc_macro *));
-	if (NULL == htab) 
-		return(NULL);
+	memset(table, UCHAR_MAX, sizeof(table));
 
 	for (i = 0; i < MDOC_MAX; i++) {
-		major = mdoc_macronames[i][0];
-		assert(isalpha((u_char)major) || 37 == major);
-
-		ADJUST_MAJOR(major);
-
-		minor = mdoc_macronames[i][1];
-		assert(isalpha((u_char)minor) || 49 == minor);
-
-		ADJUST_MINOR(minor);
+		p = mdoc_macronames[i];
 
-		ind = INDEX(major, minor);
+		if (isalpha((u_char)p[1]))
+			major = 12 * (tolower((u_char)p[1]) - 97);
+		else
+			major = 12 * 26;
 
-		if (NULL == htab[ind]) {
-			htab[ind] = &mdoc_macros[i];
-			continue;
-		}
+		for (j = 0; j < 12; j++)
+			if (UCHAR_MAX == table[major + j]) {
+				table[major + j] = (u_char)i;
+				break;
+			}
 
-		if (NULL == htab[++ind]) {
-			htab[ind] = &mdoc_macros[i];
-			continue;
-		}
-
-		assert(NULL == htab[++ind]);
-		htab[ind] = &mdoc_macros[i];
+		assert(j < 12);
 	}
-
-	return((void *)htab);
 }
 
-
 int
-mdoc_hash_find(const void *arg, const char *tmp)
+mdoc_hash_find(const char *p)
 {
-	int		  major, minor, ind, slot;
-	const void	**htab;
-
-	htab = /* LINTED */
-		(const void **)arg;
+	int		  major, i, j;
 
-	if (0 == (major = tmp[0]))
+	if (0 == p[0])
 		return(MDOC_MAX);
-	if (0 == (minor = tmp[1]))
+	if ( ! isalpha((u_char)p[0]) && '%' != p[0])
 		return(MDOC_MAX);
 
-	if (tmp[2] && tmp[3])
-		return(MDOC_MAX);
-
-	if (37 != major && ! isalpha((u_char)major))
-		return(MDOC_MAX);
-	if (49 != minor && ! isalpha((u_char)minor))
+	if (isalpha((u_char)p[1]))
+		major = 12 * (tolower((u_char)p[1]) - 97);
+	else if ('1' == p[1])
+		major = 12 * 26;
+	else 
 		return(MDOC_MAX);
 
-	ADJUST_MAJOR(major);
-	ADJUST_MINOR(minor);
-
-	ind = INDEX(major, minor);
-
-	if (ind < 0 || ind >= 26 * 3 * 52)
+	if (p[2] && p[3])
 		return(MDOC_MAX);
 
-	if (htab[ind]) {
-		slot = htab[ind] - /* LINTED */
-			(void *)mdoc_macros;
-		assert(0 == (size_t)slot % sizeof(struct mdoc_macro));
-		slot /= sizeof(struct mdoc_macro);
-		if (SLOTCMP(slot, tmp))
-			return(slot);
-		ind++;
-	}
-
-	if (htab[ind]) {
-		slot = htab[ind] - /* LINTED */
-			(void *)mdoc_macros;
-		assert(0 == (size_t)slot % sizeof(struct mdoc_macro));
-		slot /= sizeof(struct mdoc_macro);
-		if (SLOTCMP(slot, tmp))
-			return(slot);
-		ind++;
+	for (j = 0; j < 12; j++) {
+		if (UCHAR_MAX == (i = table[major + j]))
+			break;
+		if (0 == strcmp(p, mdoc_macronames[i]))
+			return(i);
 	}
 
-	if (NULL == htab[ind]) 
-		return(MDOC_MAX);
-	slot = htab[ind] - /* LINTED */
-		(void *)mdoc_macros;
-	assert(0 == (size_t)slot % sizeof(struct mdoc_macro));
-	slot /= sizeof(struct mdoc_macro);
-	if (SLOTCMP(slot, tmp))
-		return(slot);
-
 	return(MDOC_MAX);
 }
-
diff --git a/usr.bin/mandoc/mdoc_macro.c b/usr.bin/mandoc/mdoc_macro.c
index a3989b9dbde..c0781dca46e 100644
--- a/usr.bin/mandoc/mdoc_macro.c
+++ b/usr.bin/mandoc/mdoc_macro.c
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_macro.c,v 1.23 2009/08/22 22:39:55 schwarze Exp $ */
+/*	$Id: mdoc_macro.c,v 1.24 2009/09/21 21:11:37 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -46,8 +46,8 @@ static	int	  rew_sub(enum mdoc_type, struct mdoc *,
 static	int	  rew_last(struct mdoc *, 
 			const struct mdoc_node *);
 static	int	  append_delims(struct mdoc *, int, int *, char *);
-static	int	  lookup(struct mdoc *, int, const char *);
-static	int	  lookup_raw(struct mdoc *, const char *);
+static	int	  lookup(int, const char *);
+static	int	  lookup_raw(const char *);
 static	int	  swarn(struct mdoc *, enum mdoc_type, int, int, 
 			const struct mdoc_node *);
 
@@ -258,12 +258,13 @@ mdoc_macroend(struct mdoc *m)
  * Look up a macro from within a subsequent context.
  */
 static int
-lookup(struct mdoc *mdoc, int from, const char *p)
+lookup(int from, const char *p)
 {
+	/* FIXME: make -diag lists be un-PARSED. */
 
 	if ( ! (MDOC_PARSED & mdoc_macros[from].flags))
 		return(MDOC_MAX);
-	return(lookup_raw(mdoc, p));
+	return(lookup_raw(p));
 }
 
 
@@ -271,11 +272,11 @@ lookup(struct mdoc *mdoc, int from, const char *p)
  * Lookup a macro following the initial line macro.
  */
 static int
-lookup_raw(struct mdoc *mdoc, const char *p)
+lookup_raw(const char *p)
 {
 	int		 res;
 
-	if (MDOC_MAX == (res = mdoc_hash_find(mdoc->htab, p)))
+	if (MDOC_MAX == (res = mdoc_hash_find(p)))
 		return(MDOC_MAX);
 	if (MDOC_CALLABLE & mdoc_macros[res].flags)
 		return(res);
@@ -669,7 +670,7 @@ blk_exp_close(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == c)
 			break;
 
-		if (MDOC_MAX != (c = lookup(m, tok, p))) {
+		if (MDOC_MAX != (c = lookup(tok, p))) {
 			if ( ! flushed) {
 				if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
 					return(0);
@@ -753,7 +754,7 @@ in_line(MACRO_PROT_ARGS)
 
 		/* Quoted words shouldn't be looked-up. */
 
-		c = ARGS_QWORD == w ? MDOC_MAX : lookup(m, tok, p);
+		c = ARGS_QWORD == w ? MDOC_MAX : lookup(tok, p);
 
 		/* 
 		 * In this case, we've located a submacro and must
@@ -932,7 +933,7 @@ blk_full(MACRO_PROT_ARGS)
 			continue;
 		}
 
-		if (MDOC_MAX == (c = lookup(m, tok, p))) {
+		if (MDOC_MAX == (c = lookup(tok, p))) {
 			if ( ! mdoc_word_alloc(m, line, lastarg, p))
 				return(0);
 			continue;
@@ -995,7 +996,7 @@ blk_part_imp(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == c)
 			break;
 
-		if (MDOC_MAX == (c = lookup(m, tok, p))) {
+		if (MDOC_MAX == (c = lookup(tok, p))) {
 			if ( ! mdoc_word_alloc(m, line, la, p))
 				return(0);
 			continue;
@@ -1098,7 +1099,7 @@ blk_part_exp(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == c)
 			break;
 
-		if (MDOC_MAX != (c = lookup(m, tok, p))) {
+		if (MDOC_MAX != (c = lookup(tok, p))) {
 			if ( ! flushed) {
 				if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
 					return(0);
@@ -1209,7 +1210,7 @@ in_line_argn(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == c)
 			break;
 
-		if (MDOC_MAX != (c = lookup(m, tok, p))) {
+		if (MDOC_MAX != (c = lookup(tok, p))) {
 			if ( ! flushed && ! rew_elem(m, tok))
 				return(0);
 			flushed = 1;
@@ -1284,7 +1285,7 @@ in_line_eoln(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == w)
 			break;
 
-		c = ARGS_QWORD == w ? MDOC_MAX : lookup(m, tok, p);
+		c = ARGS_QWORD == w ? MDOC_MAX : lookup(tok, p);
 
 		if (MDOC_MAX != c) {
 			if ( ! rew_elem(m, tok))
@@ -1333,7 +1334,7 @@ phrase(struct mdoc *m, int line, int ppos, char *buf)
 		if (ARGS_EOLN == w)
 			break;
 
-		c = ARGS_QWORD == w ? MDOC_MAX : lookup_raw(m, p);
+		c = ARGS_QWORD == w ? MDOC_MAX : lookup_raw(p);
 
 		if (MDOC_MAX != c) {
 			if ( ! mdoc_macro(m, c, line, la, &pos, buf))