rework opencvs so that we can deal with binary files. previously we assumed all files were ascii,

which broke things in real-world usage.  now a checkout of src should work, albeit using lots of
memory and cpu.  fixing this is the next step.

testing by many.
ok & some input joris@

6 files changed, 87 insertions, 69 deletions

diff --git a/usr.bin/cvs/annotate.c b/usr.bin/cvs/annotate.c
index 5e26f7a425f..50a2440fca4 100644
--- a/usr.bin/cvs/annotate.c
+++ b/usr.bin/cvs/annotate.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: annotate.c,v 1.33 2007/01/11 02:35:55 joris Exp $	*/
+/*	$OpenBSD: annotate.c,v 1.34 2007/01/11 17:44:18 niallo Exp $	*/
 /*
  * Copyright (c) 2006 Xavier Santolaria <xsa@openbsd.org>
  *
@@ -112,6 +112,7 @@ cvs_annotate_local(struct cvs_file *cf)
 	BUF *b;
 	RCSNUM *ann_rev;
 	char *content;
+	size_t len;
 
 	ann_rev = NULL;
 
@@ -136,8 +137,9 @@ cvs_annotate_local(struct cvs_file *cf)
 	b = rcs_getrev(cf->file_rcs, ann_rev);
 	cvs_buf_putc(b, '\0');
 
+	len = cvs_buf_len(b);
 	content = cvs_buf_release(b);
-	if ((lines = cvs_splitlines(content)) == NULL)
+	if ((lines = cvs_splitlines(content, len)) == NULL)
 		fatal("cvs_annotate_local: cvs_splitlines failed");
 
         xfree(content);
diff --git a/usr.bin/cvs/diff3.c b/usr.bin/cvs/diff3.c
index e6df284e4ad..8dc97f1f011 100644
--- a/usr.bin/cvs/diff3.c
+++ b/usr.bin/cvs/diff3.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: diff3.c,v 1.29 2006/11/10 08:32:37 xsa Exp $	*/
+/*	$OpenBSD: diff3.c,v 1.30 2007/01/11 17:44:18 niallo Exp $	*/
 
 /*
  * Copyright (C) Caldera International Inc.  2001-2002.
@@ -72,7 +72,7 @@ static const char copyright[] =
 
 #ifndef lint
 static const char rcsid[] =
-    "$OpenBSD: diff3.c,v 1.29 2006/11/10 08:32:37 xsa Exp $";
+    "$OpenBSD: diff3.c,v 1.30 2007/01/11 17:44:18 niallo Exp $";
 #endif /* not lint */
 
 #include "includes.h"
@@ -163,6 +163,7 @@ cvs_diff3(RCSFILE *rf, char *workfile, int workfd, RCSNUM *rev1,
 	char *argv[5], r1[16], r2[16];
 	char *dp13, *dp23, *path1, *path2, *path3;
 	BUF *b1, *b2, *b3, *d1, *d2, *diffb;
+	size_t dlen, plen;
 
 	b1 = b2 = b3 = d1 = d2 = diffb = NULL;
 
@@ -230,11 +231,13 @@ cvs_diff3(RCSFILE *rf, char *workfile, int workfd, RCSNUM *rev1,
 	cvs_buf_putc(diffb, '\0');
 	cvs_buf_putc(b1, '\0');
 
+	plen = cvs_buf_len(diffb);
 	patch = cvs_buf_release(diffb);
+	dlen = cvs_buf_len(b1);
 	data = cvs_buf_release(b1);
 	diffb = b1 = NULL;
 
-	if ((diffb = cvs_patchfile(data, patch, ed_patch_lines)) == NULL)
+	if ((diffb = cvs_patchfile(data, dlen, patch, plen, ed_patch_lines)) == NULL)
 		goto out;
 
 	if (verbose == 1 && diff3_conflicts != 0) {
@@ -319,6 +322,7 @@ ed_patch_lines(struct cvs_lines *dlines, struct cvs_lines *plines)
 	char op, *ep;
 	struct cvs_line *sort, *lp, *dlp, *ndlp, *insert_after;
 	int start, end, i, lineno;
+	u_char tmp;
 
 	dlp = TAILQ_FIRST(&(dlines->l_lines));
 	lp = TAILQ_FIRST(&(plines->l_lines));
@@ -326,10 +330,17 @@ ed_patch_lines(struct cvs_lines *dlines, struct cvs_lines *plines)
 	end = 0;
 	for (lp = TAILQ_NEXT(lp, l_list); lp != NULL;
 	    lp = TAILQ_NEXT(lp, l_list)) {
-		if (lp->l_line[0] == '\0')
-			fatal("ed_patch_lines");
+		/* Skip blank lines */
+		if (lp->l_len < 2)
+			continue;
+		/* NUL-terminate line buffer for strtol() safety. */
+		tmp = lp->l_line[lp->l_len - 1];
+		lp->l_line[lp->l_len - 1] = '\0';
+		/* len - 1 is NUL terminator so we use len - 2 for 'op' */
 		op = lp->l_line[strlen(lp->l_line) - 1];
 		start = (int)strtol(lp->l_line, &ep, 10);
+		/* Restore the last byte of the buffer */
+		lp->l_line[lp->l_len - 1] = tmp;
 		if (op == 'a') {
 			if (start > dlines->l_nblines ||
 			    start < 0 || *ep != 'a')
@@ -386,7 +397,7 @@ ed_patch_lines(struct cvs_lines *dlines, struct cvs_lines *plines)
 				if (lp == NULL)
 					fatal("ed_patch_lines");
 
-				if (!strcmp(lp->l_line, "."))
+				if (!memcmp(lp->l_line, ".", 1))
 					break;
 
 				TAILQ_REMOVE(&(plines->l_lines), lp, l_list);
diff --git a/usr.bin/cvs/rcs.c b/usr.bin/cvs/rcs.c
index 65a4a812c50..7084d97fd56 100644
--- a/usr.bin/cvs/rcs.c
+++ b/usr.bin/cvs/rcs.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: rcs.c,v 1.192 2006/12/21 15:03:15 niallo Exp $	*/
+/*	$OpenBSD: rcs.c,v 1.193 2007/01/11 17:44:18 niallo Exp $	*/
 /*
  * Copyright (c) 2004 Jean-Francois Brousseau <jfb@openbsd.org>
  * All rights reserved.
@@ -493,11 +493,6 @@ rcs_write(RCSFILE *rfp)
 		fputs("@\ntext\n@", fp);
 		if (rdp->rd_text != NULL) {
 			rcs_strprint(rdp->rd_text, rdp->rd_tlen, fp);
-
-			if (rdp->rd_tlen != 0) {
-				if (rdp->rd_text[rdp->rd_tlen-1] != '\n')
-					fputc('\n', fp);
-			}
 		}
 		fputs("@\n", fp);
 	}
@@ -1078,6 +1073,7 @@ rcs_patch_lines(struct cvs_lines *dlines, struct cvs_lines *plines)
 	char op, *ep;
 	struct cvs_line *lp, *dlp, *ndlp;
 	int i, lineno, nbln;
+	u_char tmp;
 
 	dlp = TAILQ_FIRST(&(dlines->l_lines));
 	lp = TAILQ_FIRST(&(plines->l_lines));
@@ -1085,15 +1081,21 @@ rcs_patch_lines(struct cvs_lines *dlines, struct cvs_lines *plines)
 	/* skip first bogus line */
 	for (lp = TAILQ_NEXT(lp, l_list); lp != NULL;
 	    lp = TAILQ_NEXT(lp, l_list)) {
+		if (lp->l_len < 2)
+			fatal("line too short, RCS patch seems broken");
 		op = *(lp->l_line);
+		/* NUL-terminate line buffer for strtol() safety. */
+		tmp = lp->l_line[lp->l_len - 1];
+		lp->l_line[lp->l_len - 1] = '\0';
 		lineno = (int)strtol((lp->l_line + 1), &ep, 10);
-		if (lineno > dlines->l_nblines || lineno < 0 ||
-		    *ep != ' ')
-			fatal("invalid line specification in RCS patch: %s",
-			    ep);
+		if (lineno - 1 > dlines->l_nblines || lineno < 0) {
+			fatal("invalid line specification in RCS patch");
+		}
 		ep++;
 		nbln = (int)strtol(ep, &ep, 10);
-		if (nbln < 0 || *ep != '\0')
+		/* Restore the last byte of the buffer */
+		lp->l_line[lp->l_len - 1] = tmp;
+		if (nbln < 0)
 			fatal("invalid line number specification in RCS patch");
 
 		/* find the appropriate line */
@@ -1170,13 +1172,13 @@ rcs_patch_lines(struct cvs_lines *dlines, struct cvs_lines *plines)
 BUF*
 rcs_getrev(RCSFILE *rfp, RCSNUM *frev)
 {
-	size_t i;
+	size_t i, dlen, plen;
 	int done, nextroot, found;
 	BUF *rcsbuf;
 	RCSNUM *tnum, *bnum;
 	struct rcs_branch *brp;
 	struct rcs_delta *hrdp, *trdp, *rdp;
-	char *data;
+	u_char *data, *patch;
 
 	if ((hrdp = rcs_findrev(rfp, rfp->rf_head)) == NULL)
 		fatal("rcs_getrev: no HEAD revision");
@@ -1223,13 +1225,15 @@ again:
 			}
 		}
 
-		cvs_buf_putc(rcsbuf, '\0');
+		plen = rdp->rd_tlen;
+		dlen = cvs_buf_len(rcsbuf);
+		patch = rdp->rd_text;
 		data = cvs_buf_release(rcsbuf);
-
-		rcsbuf = cvs_patchfile(data, rdp->rd_text, rcs_patch_lines);
+		rcsbuf = cvs_patchfile(data, dlen, patch, plen,
+		    rcs_patch_lines);
+		xfree(data);
 		if (rcsbuf == NULL)
 			fatal("rcs_getrev: failed to apply rcsdiff");
-		xfree(data);
 
 		if (!rcsnum_differ(rdp->rd_num, bnum))
 			break;
@@ -2102,11 +2106,15 @@ rcs_parse_deltatext(RCSFILE *rfp)
 		return (-1);
 	}
 
-	rdp->rd_text = xmalloc(RCS_TOKLEN(rfp) + 1);
-	if (strlcpy(rdp->rd_text, RCS_TOKSTR(rfp), (RCS_TOKLEN(rfp) + 1)) >=
-	    RCS_TOKLEN(rfp) + 1)
-		fatal("rcs_parse_deltatext: strlcpy");
-	rdp->rd_tlen = RCS_TOKLEN(rfp);
+	if (RCS_TOKLEN(rfp) == 0) {
+		rdp->rd_text = xmalloc(1);
+		rdp->rd_text[0] = '\0';
+		rdp->rd_tlen = 0;
+	} else {
+		rdp->rd_text = xmalloc(RCS_TOKLEN(rfp));
+		memcpy(rdp->rd_text, RCS_TOKSTR(rfp), RCS_TOKLEN(rfp));
+		rdp->rd_tlen = RCS_TOKLEN(rfp);
+	}
 
 	return (1);
 }
diff --git a/usr.bin/cvs/update.c b/usr.bin/cvs/update.c
index afe1aae7124..19108f9f73d 100644
--- a/usr.bin/cvs/update.c
+++ b/usr.bin/cvs/update.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: update.c,v 1.82 2007/01/11 02:35:55 joris Exp $	*/
+/*	$OpenBSD: update.c,v 1.83 2007/01/11 17:44:18 niallo Exp $	*/
 /*
  * Copyright (c) 2006 Joris Vink <joris@openbsd.org>
  *
@@ -419,6 +419,7 @@ update_has_conflict_markers(struct cvs_file *cf)
 	char *content;
 	struct cvs_line *lp;
 	struct cvs_lines *lines;
+	size_t len;
 
 	cvs_log(LP_TRACE, "update_has_conflict_markers(%s)", cf->file_path);
 
@@ -430,12 +431,11 @@ update_has_conflict_markers(struct cvs_file *cf)
 		    cf->file_path);
 
 	cvs_buf_putc(bp, '\0');
+	len = cvs_buf_len(bp);
 	content = cvs_buf_release(bp);
-	if ((lines = cvs_splitlines(content)) == NULL)
+	if ((lines = cvs_splitlines(content, len)) == NULL)
 		fatal("update_has_conflict_markers: failed to split lines");
 
-	xfree(content);
-
 	conflict = 0;
 	TAILQ_FOREACH(lp, &(lines->l_lines), l_list) {
 		if (lp->l_line == NULL)
@@ -453,5 +453,6 @@ update_has_conflict_markers(struct cvs_file *cf)
 	}
 
 	cvs_freelines(lines);
+	xfree(content);
 	return (conflict);
 }
diff --git a/usr.bin/cvs/util.c b/usr.bin/cvs/util.c
index 11c21f59780..34e1de8d0ac 100644
--- a/usr.bin/cvs/util.c
+++ b/usr.bin/cvs/util.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: util.c,v 1.96 2007/01/07 01:53:12 joris Exp $	*/
+/*	$OpenBSD: util.c,v 1.97 2007/01/11 17:44:18 niallo Exp $	*/
 /*
  * Copyright (c) 2004 Jean-Francois Brousseau <jfb@openbsd.org>
  * Copyright (c) 2005, 2006 Joris Vink <joris@openbsd.org>
@@ -738,32 +738,33 @@ cvs_mkpath(const char *path)
  * Split the contents of a file into a list of lines.
  */
 struct cvs_lines *
-cvs_splitlines(const char *fcont)
+cvs_splitlines(const u_char *data, size_t len)
 {
-	char *dcp;
+	u_char *p, *c;
+	size_t i, tlen;
 	struct cvs_lines *lines;
 	struct cvs_line *lp;
 
 	lines = xmalloc(sizeof(*lines));
+	memset(lines, 0, sizeof(*lines));
 	TAILQ_INIT(&(lines->l_lines));
-	lines->l_nblines = 0;
-	lines->l_data = xstrdup(fcont);
 
 	lp = xmalloc(sizeof(*lp));
-	lp->l_line = NULL;
-	lp->l_lineno = 0;
+	memset(lp, 0, sizeof(*lp));
 	TAILQ_INSERT_TAIL(&(lines->l_lines), lp, l_list);
 
-	for (dcp = lines->l_data; *dcp != '\0';) {
-		lp = xmalloc(sizeof(*lp));
-		lp->l_line = dcp;
-		lp->l_lineno = ++(lines->l_nblines);
-		TAILQ_INSERT_TAIL(&(lines->l_lines), lp, l_list);
-
-		dcp = strchr(dcp, '\n');
-		if (dcp == NULL)
-			break;
-		*(dcp++) = '\0';
+	p = c = data;
+	for (i = 0; i < len; i++) {
+		if (*p == '\n' || (i == len - 1)) {
+			tlen = p - c + 1;
+			lp = xmalloc(sizeof(*lp));
+			lp->l_line = c;
+			lp->l_len = tlen;
+			lp->l_lineno = ++(lines->l_nblines);
+			TAILQ_INSERT_TAIL(&(lines->l_lines), lp, l_list);
+			c = p + 1;
+		}
+		p++;
 	}
 
 	return (lines);
@@ -779,26 +780,21 @@ cvs_freelines(struct cvs_lines *lines)
 		xfree(lp);
 	}
 
-	xfree(lines->l_data);
 	xfree(lines);
 }
 
 BUF *
-cvs_patchfile(const char *data, const char *patch,
+cvs_patchfile(const u_char *data, size_t dlen, const u_char *patch, size_t plen,
     int (*p)(struct cvs_lines *, struct cvs_lines *))
 {
 	struct cvs_lines *dlines, *plines;
 	struct cvs_line *lp;
-	size_t len;
-	int lineno;
 	BUF *res;
 
-	len = strlen(data);
-
-	if ((dlines = cvs_splitlines(data)) == NULL)
+	if ((dlines = cvs_splitlines(data, dlen)) == NULL)
 		return (NULL);
 
-	if ((plines = cvs_splitlines(patch)) == NULL)
+	if ((plines = cvs_splitlines(patch, plen)) == NULL)
 		return (NULL);
 
 	if (p(dlines, plines) < 0) {
@@ -807,12 +803,11 @@ cvs_patchfile(const char *data, const char *patch,
 		return (NULL);
 	}
 
-	lineno = 0;
-	res = cvs_buf_alloc(len, BUF_AUTOEXT);
+	res = cvs_buf_alloc(1024, BUF_AUTOEXT);
 	TAILQ_FOREACH(lp, &dlines->l_lines, l_list) {
-		if (lineno != 0)
-			cvs_buf_fappend(res, "%s\n", lp->l_line);
-		lineno++;
+		if (lp->l_line == NULL)
+			continue;
+		cvs_buf_append(res, lp->l_line, lp->l_len);
 	}
 
 	cvs_freelines(dlines);
diff --git a/usr.bin/cvs/util.h b/usr.bin/cvs/util.h
index c7e89dd859e..1441e6fce2b 100644
--- a/usr.bin/cvs/util.h
+++ b/usr.bin/cvs/util.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: util.h,v 1.12 2007/01/07 01:53:12 joris Exp $	*/
+/*	$OpenBSD: util.h,v 1.13 2007/01/11 17:44:18 niallo Exp $	*/
 /*
  * Copyright (c) 2006 Niall O'Higgins <niallo@openbsd.org>
  * All rights reserved.
@@ -50,6 +50,7 @@ u_int	  cvs_revision_select(RCSFILE *, char *);
 
 struct cvs_line {
 	char			*l_line;
+	size_t			 l_len;
 	int			 l_lineno;
 	TAILQ_ENTRY(cvs_line)	 l_list;
 };
@@ -58,7 +59,6 @@ TAILQ_HEAD(cvs_tqh, cvs_line);
 
 struct cvs_lines {
 	int		l_nblines;
-	char		*l_data;
 	struct cvs_tqh	l_lines;
 };
 
@@ -67,9 +67,10 @@ struct cvs_argvector {
 	char **argv;
 };
 
-BUF			*cvs_patchfile(const char *, const char *,
-			    int (*p)(struct cvs_lines *, struct cvs_lines *));
-struct cvs_lines	*cvs_splitlines(const char *);
+BUF			*cvs_patchfile(const u_char *, size_t, const u_char *,
+			    size_t, int (*p)(struct cvs_lines *,
+			    struct cvs_lines *));
+struct cvs_lines	*cvs_splitlines(const u_char *, size_t);
 void			cvs_freelines(struct cvs_lines *);
 struct cvs_argvector	*cvs_strsplit(char *, const char *);
 void			cvs_argv_destroy(struct cvs_argvector *);