===================================================================
RCS file: /cvs/mandoc/roff.c,v
retrieving revision 1.154
retrieving revision 1.166
diff -u -p -r1.154 -r1.166
--- mandoc/roff.c	2011/07/27 07:09:41	1.154
+++ mandoc/roff.c	2011/07/29 09:19:48	1.166
@@ -1,4 +1,4 @@
-/*	$Id: roff.c,v 1.154 2011/07/27 07:09:41 kristaps Exp $ */
+/*	$Id: roff.c,v 1.166 2011/07/29 09:19:48 kristaps Exp $ */
 /*
  * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
@@ -77,23 +77,32 @@ enum	roffrule {
  * Registers are assumed to be unsigned ints for now.
  */
 struct	reg {
-	int		  set; /* whether set or not */
-	unsigned int	  u; /* unsigned integer */
+	int		 set; /* whether set or not */
+	unsigned int	 u; /* unsigned integer */
 };
 
 struct	roffstr {
-	char		*name; /* key of symbol */
-	char		*string; /* current value */
-	struct roffstr	*next; /* next in list */
+	char		*p;
+	size_t		 sz;
 };
 
+/*
+ * A key-value string pair with lengths.
+ */
+struct	roffkv {
+	struct roffstr	 key;
+	struct roffstr	 val;
+	struct roffkv	*next; /* next in list */
+};
+
 struct	roff {
 	struct mparse	*parse; /* parse point */
 	struct roffnode	*last; /* leaf of stack */
 	enum roffrule	 rstack[RSTACK_MAX]; /* stack of !`ie' rules */
 	int		 rstackpos; /* position in rstack */
 	struct reg	 regs[REG__MAX];
-	struct roffstr	*first_string; /* user-defined strings & macros */
+	struct roffkv	*strtab; /* user-defined strings & macros */
+	struct roffkv	*chrtab; /* user-defined characters */
 	const char	*current_string; /* value of last called user macro */
 	struct tbl_node	*first_tbl; /* first table parsed */
 	struct tbl_node	*last_tbl; /* last table parsed */
@@ -143,6 +152,12 @@ struct	predef {
 #define	PREDEF(__name, __str) \
 	{ (__name), (__str) },
 
+static	enum rofft	 roffhash_find(const char *, size_t);
+static	void		 roffhash_init(void);
+static	void		 roffnode_cleanscope(struct roff *);
+static	void		 roffnode_pop(struct roff *);
+static	void		 roffnode_push(struct roff *, enum rofft,
+				const char *, int, int);
 static	enum rofferr	 roff_block(ROFF_ARGS);
 static	enum rofferr	 roff_block_text(ROFF_ARGS);
 static	enum rofferr	 roff_block_sub(ROFF_ARGS);
@@ -153,18 +168,26 @@ static	enum rofferr	 roff_cond_text(ROFF_ARGS);
 static	enum rofferr	 roff_cond_sub(ROFF_ARGS);
 static	enum rofferr	 roff_ds(ROFF_ARGS);
 static	enum roffrule	 roff_evalcond(const char *, int *);
-static	void		 roff_freestr(struct roff *);
+static	void		 roff_free1(struct roff *);
+static	void		 roff_freestr(struct roffkv **);
 static	char		*roff_getname(struct roff *, char **, int, int);
 static	const char	*roff_getstrn(const struct roff *, 
 				const char *, size_t);
 static	enum rofferr	 roff_line_ignore(ROFF_ARGS);
 static	enum rofferr	 roff_nr(ROFF_ARGS);
+static	void		 roff_openeqn(struct roff *, const char *, 
+				int, int, const char *);
+static	enum rofft	 roff_parse(struct roff *, const char *, int *);
+static	enum rofferr	 roff_parsetext(char *);
 static	void		 roff_res(struct roff *, 
 				char **, size_t *, int, int);
 static	enum rofferr	 roff_rm(ROFF_ARGS);
 static	void		 roff_setstr(struct roff *,
 				const char *, const char *, int);
+static	void		 roff_setstrn(struct roffkv **, const char *, 
+				size_t, const char *, size_t, int);
 static	enum rofferr	 roff_so(ROFF_ARGS);
+static	enum rofferr	 roff_tr(ROFF_ARGS);
 static	enum rofferr	 roff_TE(ROFF_ARGS);
 static	enum rofferr	 roff_TS(ROFF_ARGS);
 static	enum rofferr	 roff_EQ(ROFF_ARGS);
@@ -172,7 +195,7 @@ static	enum rofferr	 roff_EN(ROFF_ARGS);
 static	enum rofferr	 roff_T_(ROFF_ARGS);
 static	enum rofferr	 roff_userdef(ROFF_ARGS);
 
-/* See roff_hash_find() */
+/* See roffhash_find() */
 
 #define	ASCII_HI	 126
 #define	ASCII_LO	 33
@@ -203,7 +226,7 @@ static	struct roffmac	 roffs[ROFF_MAX] = {
 	{ "rm", roff_rm, NULL, NULL, 0, NULL },
 	{ "so", roff_so, NULL, NULL, 0, NULL },
 	{ "ta", roff_line_ignore, NULL, NULL, 0, NULL },
-	{ "tr", roff_line_ignore, NULL, NULL, 0, NULL },
+	{ "tr", roff_tr, NULL, NULL, 0, NULL },
 	{ "TS", roff_TS, NULL, NULL, 0, NULL },
 	{ "TE", roff_TE, NULL, NULL, 0, NULL },
 	{ "T&", roff_T_, NULL, NULL, 0, NULL },
@@ -220,20 +243,11 @@ static	const struct predef predefs[PREDEFS_MAX] = {
 #include "predefs.in"
 };
 
-static	void		 roff_free1(struct roff *);
-static	enum rofft	 roff_hash_find(const char *, size_t);
-static	void		 roff_hash_init(void);
-static	void		 roffnode_cleanscope(struct roff *);
-static	void		 roffnode_push(struct roff *, enum rofft,
-				const char *, int, int);
-static	void		 roffnode_pop(struct roff *);
-static	enum rofft	 roff_parse(struct roff *, const char *, int *);
-
-/* See roff_hash_find() */
+/* See roffhash_find() */
 #define	ROFF_HASH(p)	(p[0] - ASCII_LO)
 
 static void
-roff_hash_init(void)
+roffhash_init(void)
 {
 	struct roffmac	 *n;
 	int		  buc, i;
@@ -258,7 +272,7 @@ roff_hash_init(void)
  * the nil-terminated string name could be found.
  */
 static enum rofft
-roff_hash_find(const char *p, size_t s)
+roffhash_find(const char *p, size_t s)
 {
 	int		 buc;
 	struct roffmac	*n;
@@ -350,7 +364,8 @@ roff_free1(struct roff *r)
 	while (r->last)
 		roffnode_pop(r);
 
-	roff_freestr(r);
+	roff_freestr(&r->strtab);
+	roff_freestr(&r->chrtab);
 }
 
 
@@ -387,7 +402,7 @@ roff_alloc(struct mparse *parse)
 	r->parse = parse;
 	r->rstackpos = -1;
 	
-	roff_hash_init();
+	roffhash_init();
 
 	for (i = 0; i < PREDEFS_MAX; i++) 
 		roff_setstr(r, predefs[i].name, predefs[i].str, 0);
@@ -494,7 +509,7 @@ again:
 
 		/* Replace the escape sequence by the string. */
 
-		pos += (stesc - *bufp);
+		pos = stesc - *bufp;
 
 		nsz = *szp + strlen(res) + 1;
 		n = mandoc_malloc(nsz);
@@ -517,6 +532,7 @@ again:
 static enum rofferr
 roff_parsetext(char *p)
 {
+	char		 l, r;
 	size_t		 sz;
 	const char	*start;
 	enum mandoc_esc	 esc;
@@ -527,6 +543,9 @@ roff_parsetext(char *p)
 		sz = strcspn(p, "-\\");
 		p += sz;
 
+		if ('\0' == *p)
+			break;
+
 		if ('\\' == *p) {
 			/* Skip over escapes. */
 			p++;
@@ -534,11 +553,22 @@ roff_parsetext(char *p)
 				((const char **)&p, NULL, NULL);
 			if (ESCAPE_ERROR == esc)
 				break;
-		} else if ('-' == *p) {
-			if (mandoc_hyph(start, p))
-				*p = ASCII_HYPH;
+			continue;
+		} else if (p == start) {
 			p++;
+			continue;
 		}
+
+		l = *(p - 1);
+		r = *(p + 1);
+		if ('\\' != l &&
+				'\t' != r && '\t' != l &&
+				' ' != r && ' ' != l &&
+				'-' != r && '-' != l &&
+				! isdigit((unsigned char)l) &&
+				! isdigit((unsigned char)r))
+			*p = ASCII_HYPH;
+		p++;
 	}
 
 	return(ROFF_CONT);
@@ -668,7 +698,7 @@ roff_parse(struct roff *r, const char *buf, int *pos)
 	maclen = strcspn(mac + 1, " \\\t\0") + 1;
 
 	t = (r->current_string = roff_getstrn(r, mac, maclen))
-	    ? ROFF_USERDEF : roff_hash_find(mac, maclen);
+	    ? ROFF_USERDEF : roffhash_find(mac, maclen);
 
 	*pos += (int)maclen;
 
@@ -1247,14 +1277,16 @@ roff_T_(ROFF_ARGS)
 	return(ROFF_IGN);
 }
 
-int
+#if 0
+static int
 roff_closeeqn(struct roff *r)
 {
 
 	return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
 }
+#endif
 
-void
+static void
 roff_openeqn(struct roff *r, const char *name, int line, 
 		int offs, const char *buf)
 {
@@ -1319,6 +1351,59 @@ roff_TS(ROFF_ARGS)
 
 /* ARGSUSED */
 static enum rofferr
+roff_tr(ROFF_ARGS)
+{
+	const char	*p, *first, *second;
+	size_t		 fsz, ssz;
+	enum mandoc_esc	 esc;
+
+	p = *bufp + pos;
+
+	if ('\0' == *p) {
+		mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
+		return(ROFF_IGN);
+	}
+
+	while ('\0' != *p) {
+		fsz = ssz = 1;
+
+		first = p++;
+		if ('\\' == *first) {
+			esc = mandoc_escape(&p, NULL, NULL);
+			if (ESCAPE_ERROR == esc) {
+				mandoc_msg
+					(MANDOCERR_BADESCAPE, r->parse, 
+					 ln, (int)(p - *bufp), NULL);
+				return(ROFF_IGN);
+			}
+			fsz = (size_t)(p - first);
+		}
+
+		second = p++;
+		if ('\\' == *second) {
+			esc = mandoc_escape(&p, NULL, NULL);
+			if (ESCAPE_ERROR == esc) {
+				mandoc_msg
+					(MANDOCERR_BADESCAPE, r->parse, 
+					 ln, (int)(p - *bufp), NULL);
+				return(ROFF_IGN);
+			}
+			ssz = (size_t)(p - second);
+		} else if ('\0' == *second) {
+			mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, 
+					ln, (int)(p - *bufp), NULL);
+			second = " ";
+			p--;
+		}
+
+		roff_setstrn(&r->chrtab, first, fsz, second, ssz, 0);
+	}
+
+	return(ROFF_IGN);
+}
+
+/* ARGSUSED */
+static enum rofferr
 roff_so(ROFF_ARGS)
 {
 	char *name;
@@ -1440,26 +1525,40 @@ static void
 roff_setstr(struct roff *r, const char *name, const char *string,
 	int multiline)
 {
-	struct roffstr	 *n;
-	char		 *c;
-	size_t		  oldch, newch;
 
+	roff_setstrn(&r->strtab, name, strlen(name), string,
+			string ? strlen(string) : 0, multiline);
+}
+
+static void
+roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
+		const char *string, size_t stringsz, int multiline)
+{
+	struct roffkv	*n;
+	char		*c;
+	int		 i;
+	size_t		 oldch, newch;
+
 	/* Search for an existing string with the same name. */
-	n = r->first_string;
-	while (n && strcmp(name, n->name))
+	n = *r;
+
+	while (n && strcmp(name, n->key.p))
 		n = n->next;
 
 	if (NULL == n) {
 		/* Create a new string table entry. */
-		n = mandoc_malloc(sizeof(struct roffstr));
-		n->name = mandoc_strdup(name);
-		n->string = NULL;
-		n->next = r->first_string;
-		r->first_string = n;
+		n = mandoc_malloc(sizeof(struct roffkv));
+		n->key.p = mandoc_strndup(name, namesz);
+		n->key.sz = namesz;
+		n->val.p = NULL;
+		n->val.sz = 0;
+		n->next = *r;
+		*r = n;
 	} else if (0 == multiline) {
 		/* In multiline mode, append; else replace. */
-		free(n->string);
-		n->string = NULL;
+		free(n->val.p);
+		n->val.p = NULL;
+		n->val.sz = 0;
 	}
 
 	if (NULL == string)
@@ -1469,61 +1568,66 @@ roff_setstr(struct roff *r, const char *name, const ch
 	 * One additional byte for the '\n' in multiline mode,
 	 * and one for the terminating '\0'.
 	 */
-	newch = strlen(string) + (multiline ? 2u : 1u);
-	if (NULL == n->string) {
-		n->string = mandoc_malloc(newch);
-		*n->string = '\0';
+	newch = stringsz + (multiline ? 2u : 1u);
+
+	if (NULL == n->val.p) {
+		n->val.p = mandoc_malloc(newch);
+		*n->val.p = '\0';
 		oldch = 0;
 	} else {
-		oldch = strlen(n->string);
-		n->string = mandoc_realloc(n->string, oldch + newch);
+		oldch = n->val.sz;
+		n->val.p = mandoc_realloc(n->val.p, oldch + newch);
 	}
 
 	/* Skip existing content in the destination buffer. */
-	c = n->string + (int)oldch;
+	c = n->val.p + (int)oldch;
 
 	/* Append new content to the destination buffer. */
-	while (*string) {
+	i = 0;
+	while (i < (int)stringsz) {
 		/*
 		 * Rudimentary roff copy mode:
 		 * Handle escaped backslashes.
 		 */
-		if ('\\' == *string && '\\' == *(string + 1))
-			string++;
-		*c++ = *string++;
+		if ('\\' == string[i] && '\\' == string[i + 1])
+			i++;
+		*c++ = string[i++];
 	}
 
 	/* Append terminating bytes. */
 	if (multiline)
 		*c++ = '\n';
+
 	*c = '\0';
+	n->val.sz = (int)(c - n->val.p);
 }
 
 static const char *
 roff_getstrn(const struct roff *r, const char *name, size_t len)
 {
-	const struct roffstr *n;
+	const struct roffkv *n;
 
-	n = r->first_string;
-	while (n && (strncmp(name, n->name, len) || '\0' != n->name[(int)len]))
-		n = n->next;
+	for (n = r->strtab; n; n = n->next)
+		if (0 == strncmp(name, n->key.p, len) && 
+				'\0' == n->key.p[(int)len])
+			return(n->val.p);
 
-	return(n ? n->string : NULL);
+	return(NULL);
 }
 
 static void
-roff_freestr(struct roff *r)
+roff_freestr(struct roffkv **r)
 {
-	struct roffstr	 *n, *nn;
+	struct roffkv	 *n, *nn;
 
-	for (n = r->first_string; n; n = nn) {
-		free(n->name);
-		free(n->string);
+	for (n = *r; n; n = nn) {
+		free(n->key.p);
+		free(n->val.p);
 		nn = n->next;
 		free(n);
 	}
 
-	r->first_string = NULL;
+	*r = NULL;
 }
 
 const struct tbl_span *
@@ -1545,4 +1649,90 @@ roff_eqndelim(const struct roff *r)
 {
 
 	return('\0');
+}
+
+/*
+ * Duplicate an input string, making the appropriate character
+ * conversations (as stipulated by `tr') along the way.
+ * Returns a heap-allocated string with all the replacements made.
+ */
+char *
+roff_strdup(const struct roff *r, const char *p)
+{
+	const struct roffkv *cp;
+	char		*res;
+	const char	*pp;
+	size_t		 ssz, sz;
+	enum mandoc_esc	 esc;
+
+	if (NULL == r->chrtab)
+		return(mandoc_strdup(p));
+	else if ('\0' == *p)
+		return(mandoc_strdup(""));
+
+	/*
+	 * Step through each character looking for term matches
+	 * (remember that a `tr' can be invoked with an escape, which is
+	 * a glyph but the escape is multi-character).
+	 * We only do this if the character hash has been initialised
+	 * and the string is >0 length.
+	 */
+
+	res = NULL;
+	ssz = 0;
+
+	while ('\0' != *p) {
+		/* Search for term matches. */
+		for (cp = r->chrtab; cp; cp = cp->next)
+			if (0 == strncmp(p, cp->key.p, cp->key.sz))
+				break;
+
+		if (NULL != cp) {
+			/*
+			 * A match has been found.
+			 * Append the match to the array and move
+			 * forward by its keysize.
+			 */
+			res = mandoc_realloc
+				(res, ssz + cp->val.sz + 1);
+			memcpy(res + ssz, cp->val.p, cp->val.sz);
+			ssz += cp->val.sz;
+			p += (int)cp->key.sz;
+			continue;
+		}
+
+		if ('\\' == *p) {
+			/*
+			 * Handle escapes carefully: we need to copy
+			 * over just the escape itself, or else we might
+			 * do replacements within the escape itself.
+			 * Make sure to pass along the bogus string.
+			 */
+			pp = p++;
+			esc = mandoc_escape(&p, NULL, NULL);
+			if (ESCAPE_ERROR == esc) {
+				sz = strlen(pp);
+				res = mandoc_realloc(res, ssz + sz + 1);
+				memcpy(res + ssz, pp, sz);
+				break;
+			}
+			/* 
+			 * We bail out on bad escapes. 
+			 * No need to warn: we already did so when
+			 * roff_res() was called.
+			 */
+			sz = (int)(p - pp);
+			res = mandoc_realloc(res, ssz + sz + 1);
+			memcpy(res + ssz, pp, sz);
+			ssz += sz;
+			continue;
+		}
+
+		/* Just append the charater. */
+		res = mandoc_realloc(res, ssz + 2);
+		res[ssz++] = *p++;
+	}
+
+	res[(int)ssz] = '\0';
+	return(res);
 }