===================================================================
RCS file: /cvs/mandoc/man.c,v
retrieving revision 1.30
retrieving revision 1.53
diff -u -p -r1.30 -r1.53
--- mandoc/man.c	2009/08/19 09:14:50	1.30
+++ mandoc/man.c	2010/03/23 21:50:43	1.53
@@ -1,4 +1,4 @@
-/*	$Id: man.c,v 1.30 2009/08/19 09:14:50 kristaps Exp $ */
+/*	$Id: man.c,v 1.53 2010/03/23 21:50:43 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -14,6 +14,12 @@
  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  */
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <sys/types.h>
+
 #include <assert.h>
 #include <ctype.h>
 #include <stdarg.h>
@@ -22,13 +28,14 @@
 #include <string.h>
 
 #include "libman.h"
+#include "libmandoc.h"
 
 const	char *const __man_merrnames[WERRMAX] = {		 
 	"invalid character", /* WNPRINT */
-	"system: malloc error", /* WNMEM */
 	"invalid manual section", /* WMSEC */
 	"invalid date format", /* WDATE */
 	"scope of prior line violated", /* WLNSCOPE */
+	"over-zealous prior line scope violation", /* WLNSCOPE2 */
 	"trailing whitespace", /* WTSPACE */
 	"unterminated quoted parameter", /* WTQUOTE */
 	"document has no body", /* WNODATA */
@@ -38,10 +45,11 @@ const	char *const __man_merrnames[WERRMAX] = {		 
 	"expected block head arguments", /* WHEADARGS */
 	"expected block body arguments", /* WBODYARGS */
 	"expected empty block head", /* WNHEADARGS */
-	"unknown macro", /* WMACRO */
 	"ill-formed macro", /* WMACROFORM */
 	"scope open on exit", /* WEXITSCOPE */
-	"no scope context" /* WNOSCOPE */
+	"no scope context", /* WNOSCOPE */
+	"literal context already open", /* WOLITERAL */
+	"no literal context open" /* WNLITERAL */
 };
 
 const	char *const __man_macronames[MAN_MAX] = {		 
@@ -52,19 +60,23 @@ const	char *const __man_macronames[MAN_MAX] = {		 
 	"R",		"B",		"I",		"IR",
 	"RI",		"na",		"i",		"sp",
 	"nf",		"fi",		"r",		"RE",
-	"RS"
+	"RS",		"DT",		"UC",		"PD",
+	"Sp",		"Vb",		"Ve",
 	};
 
 const	char * const *man_macronames = __man_macronames;
 
 static	struct man_node	*man_node_alloc(int, int, 
-				enum man_type, int);
+				enum man_type, enum mant);
 static	int		 man_node_append(struct man *, 
 				struct man_node *);
 static	int		 man_ptext(struct man *, int, char *);
 static	int		 man_pmacro(struct man *, int, char *);
 static	void		 man_free1(struct man *);
-static	int		 man_alloc1(struct man *);
+static	void		 man_alloc1(struct man *);
+static	int		 pstring(struct man *, int, int, 
+				const char *, size_t);
+static	int		 macrowarn(struct man *, int, const char *);
 
 
 const struct man_node *
@@ -83,12 +95,12 @@ man_meta(const struct man *m)
 }
 
 
-int
+void
 man_reset(struct man *man)
 {
 
 	man_free1(man);
-	return(man_alloc1(man));
+	man_alloc1(man);
 }
 
 
@@ -97,9 +109,6 @@ man_free(struct man *man)
 {
 
 	man_free1(man);
-
-	if (man->htab)
-		man_hash_free(man->htab);
 	free(man);
 }
 
@@ -109,22 +118,16 @@ man_alloc(void *data, int pflags, const struct man_cb 
 {
 	struct man	*p;
 
-	if (NULL == (p = calloc(1, sizeof(struct man))))
-		return(NULL);
+	p = mandoc_calloc(1, sizeof(struct man));
 
-	if ( ! man_alloc1(p)) {
-		free(p);
-		return(NULL);
-	}
+	if (cb)
+		memcpy(&p->cb, cb, sizeof(struct man_cb));
 
+	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);
-	}
+	man_alloc1(p);
 	return(p);
 }
 
@@ -167,19 +170,16 @@ man_free1(struct man *man)
 }
 
 
-static int
+static void
 man_alloc1(struct man *m)
 {
 
-	bzero(&m->meta, sizeof(struct man_meta));
+	memset(&m->meta, 0, sizeof(struct man_meta));
 	m->flags = 0;
-	m->last = calloc(1, sizeof(struct man_node));
-	if (NULL == m->last)
-		return(0);
+	m->last = mandoc_calloc(1, sizeof(struct man_node));
 	m->first = m->last;
 	m->last->type = MAN_ROOT;
 	m->next = MAN_NEXT_CHILD;
-	return(1);
 }
 
 
@@ -242,14 +242,11 @@ man_node_append(struct man *man, struct man_node *p)
 
 
 static struct man_node *
-man_node_alloc(int line, int pos, enum man_type type, int tok)
+man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 {
 	struct man_node *p;
 
-	p = calloc(1, sizeof(struct man_node));
-	if (NULL == p)
-		return(NULL);
-
+	p = mandoc_calloc(1, sizeof(struct man_node));
 	p->line = line;
 	p->pos = pos;
 	p->type = type;
@@ -259,13 +256,11 @@ man_node_alloc(int line, int pos, enum man_type type, 
 
 
 int
-man_elem_alloc(struct man *m, int line, int pos, int tok)
+man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 {
 	struct man_node *p;
 
 	p = man_node_alloc(line, pos, MAN_ELEM, tok);
-	if (NULL == p)
-		return(0);
 	if ( ! man_node_append(m, p))
 		return(0);
 	m->next = MAN_NEXT_CHILD;
@@ -274,13 +269,11 @@ man_elem_alloc(struct man *m, int line, int pos, int t
 
 
 int
-man_head_alloc(struct man *m, int line, int pos, int tok)
+man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 {
 	struct man_node *p;
 
 	p = man_node_alloc(line, pos, MAN_HEAD, tok);
-	if (NULL == p)
-		return(0);
 	if ( ! man_node_append(m, p))
 		return(0);
 	m->next = MAN_NEXT_CHILD;
@@ -289,13 +282,11 @@ man_head_alloc(struct man *m, int line, int pos, int t
 
 
 int
-man_body_alloc(struct man *m, int line, int pos, int tok)
+man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 {
 	struct man_node *p;
 
 	p = man_node_alloc(line, pos, MAN_BODY, tok);
-	if (NULL == p)
-		return(0);
 	if ( ! man_node_append(m, p))
 		return(0);
 	m->next = MAN_NEXT_CHILD;
@@ -304,13 +295,11 @@ man_body_alloc(struct man *m, int line, int pos, int t
 
 
 int
-man_block_alloc(struct man *m, int line, int pos, int tok)
+man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 {
 	struct man_node *p;
 
 	p = man_node_alloc(line, pos, MAN_BLOCK, tok);
-	if (NULL == p)
-		return(0);
 	if ( ! man_node_append(m, p))
 		return(0);
 	m->next = MAN_NEXT_CHILD;
@@ -318,23 +307,35 @@ man_block_alloc(struct man *m, int line, int pos, int 
 }
 
 
-int
-man_word_alloc(struct man *m, int line, int pos, const char *word)
+static int
+pstring(struct man *m, int line, int pos, 
+		const char *p, size_t len)
 {
-	struct man_node	*p;
+	struct man_node	*n;
+	size_t		 sv;
 
-	p = man_node_alloc(line, pos, MAN_TEXT, -1);
-	if (NULL == p)
+	n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
+	n->string = mandoc_malloc(len + 1);
+	sv = strlcpy(n->string, p, len + 1);
+
+	/* Prohibit truncation. */
+	assert(sv < len + 1);
+
+	if ( ! man_node_append(m, n))
 		return(0);
-	if (NULL == (p->string = strdup(word)))
-		return(0);
-	if ( ! man_node_append(m, p))
-		return(0);
 	m->next = MAN_NEXT_SIBLING;
 	return(1);
 }
 
 
+int
+man_word_alloc(struct man *m, int line, int pos, const char *word)
+{
+
+	return(pstring(m, line, pos, word, strlen(word)));
+}
+
+
 void
 man_node_free(struct man_node *p)
 {
@@ -350,37 +351,93 @@ man_node_free(struct man_node *p)
 void
 man_node_freelist(struct man_node *p)
 {
+	struct man_node	*n;
 
 	if (p->child)
 		man_node_freelist(p->child);
-	if (p->next)
-		man_node_freelist(p->next);
-
 	assert(0 == p->nchild);
+	n = p->next;
 	man_node_free(p);
+	if (n)
+		man_node_freelist(n);
 }
 
 
 static int
 man_ptext(struct man *m, int line, char *buf)
 {
+	int		 i, j;
+	char		 sv;
 
-	/* First allocate word. */
+	/* Literal free-form text whitespace is preserved. */
 
-	/* FIXME: dechunk words! */
+	if (MAN_LITERAL & m->flags) {
+		if ( ! man_word_alloc(m, line, 0, buf))
+			return(0);
+		goto descope;
+	}
 
-	if ( ! man_word_alloc(m, line, 0, buf))
+	/* First de-chunk and allocate words. */
+
+	for (i = 0; ' ' == buf[i]; i++)
+		/* Skip leading whitespace. */ ;
+
+	if ('\0' == buf[i]) {
+		/* Trailing whitespace? */
+		if (i && ' ' == buf[i - 1])
+			if ( ! man_pwarn(m, line, i - 1, WTSPACE))
+				return(0);
+		if ( ! pstring(m, line, 0, &buf[i], 0))
+			return(0);
+		goto descope;
+	}
+
+	for (j = i; buf[i]; i++) {
+		if (' ' != buf[i])
+			continue;
+
+		/* Escaped whitespace. */
+		if (i && ' ' == buf[i] && '\\' == buf[i - 1])
+			continue;
+
+		sv = buf[i];
+		buf[i++] = '\0';
+
+		if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
+			return(0);
+
+		/* Trailing whitespace?  Check at overwritten byte. */
+
+		if (' ' == sv && '\0' == buf[i])
+			if ( ! man_pwarn(m, line, i - 1, WTSPACE))
+				return(0);
+
+		for ( ; ' ' == buf[i]; i++)
+			/* Skip trailing whitespace. */ ;
+
+		j = i;
+
+		/* Trailing whitespace? */
+
+		if (' ' == buf[i - 1] && '\0' == buf[i])
+			if ( ! man_pwarn(m, line, i - 1, WTSPACE))
+				return(0);
+
+		if ('\0' == buf[i])
+			break;
+	}
+
+	if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 		return(0);
-	m->next = MAN_NEXT_SIBLING;
 
+descope:
+
 	/*
 	 * Co-ordinate what happens with having a next-line scope open:
 	 * first close out the element scope (if applicable), then close
 	 * out the block scope (also if applicable).
 	 */
 
-	/* XXX - this should be in man_action.c. */
-
 	if (MAN_ELINE & m->flags) {
 		m->flags &= ~MAN_ELINE;
 		if ( ! man_unscope(m, m->last->parent))
@@ -397,18 +454,32 @@ man_ptext(struct man *m, int line, char *buf)
 }
 
 
+static int
+macrowarn(struct man *m, int ln, const char *buf)
+{
+	if ( ! (MAN_IGN_MACRO & m->pflags))
+		return(man_verr(m, ln, 0, 
+				"unknown macro: %s%s", 
+				buf, strlen(buf) > 3 ? "..." : ""));
+	return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
+				buf, strlen(buf) > 3 ? "..." : ""));
+}
+
+
 int
 man_pmacro(struct man *m, int ln, char *buf)
 {
-	int		  i, j, c, ppos, fl;
-	char		  mac[5];
+	int		 i, j, ppos, fl;
+	enum mant	 tok;
+	char		 mac[5];
+	struct man_node	*n;
 
 	/* Comments and empties are quickly ignored. */
 
 	fl = m->flags;
 
-	if (0 == buf[1])
-		goto out;
+	if ('\0' == buf[1])
+		return(1);
 
 	i = 1;
 
@@ -416,7 +487,7 @@ man_pmacro(struct man *m, int ln, char *buf)
 		i++;
 		while (buf[i] && ' ' == buf[i])
 			i++;
-		if (0 == buf[i])
+		if ('\0' == buf[i])
 			goto out;
 	}
 
@@ -425,13 +496,19 @@ man_pmacro(struct man *m, int ln, char *buf)
 	/* Copy the first word into a nil-terminated buffer. */
 
 	for (j = 0; j < 4; j++, i++) {
-		if (0 == (mac[j] = buf[i]))
+		if ('\0' == (mac[j] = buf[i]))
 			break;
 		else if (' ' == buf[i])
 			break;
+
+		/* Check for invalid characters. */
+
+		if (isgraph((u_char)buf[i]))
+			continue;
+		return(man_perr(m, ln, i, WNPRINT));
 	}
 
-	mac[j] = 0;
+	mac[j] = '\0';
 
 	if (j == 4 || j < 1) {
 		if ( ! (MAN_IGN_MACRO & m->pflags)) {
@@ -443,13 +520,9 @@ man_pmacro(struct man *m, int ln, char *buf)
 		return(1);
 	}
 	
-	if (MAN_MAX == (c = man_hash_find(m->htab, mac))) {
-		if ( ! (MAN_IGN_MACRO & m->pflags)) {
-			(void)man_perr(m, ln, ppos, WMACRO);
+	if (MAN_MAX == (tok = man_hash_find(mac))) {
+		if ( ! macrowarn(m, ln, mac))
 			goto err;
-		} 
-		if ( ! man_pwarn(m, ln, ppos, WMACRO))
-			goto err;
 		return(1);
 	}
 
@@ -458,17 +531,76 @@ man_pmacro(struct man *m, int ln, char *buf)
 	while (buf[i] && ' ' == buf[i])
 		i++;
 
+	/* Trailing whitespace? */
+
+	if ('\0' == buf[i] && ' ' == buf[i - 1])
+		if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
+			goto err;
+
+	/* 
+	 * Remove prior ELINE macro, as it's being clobbering by a new
+	 * macro.  Note that NSCOPED macros do not close out ELINE
+	 * macros---they don't print text---so we let those slip by.
+	 */
+
+	if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
+			m->flags & MAN_ELINE) {
+		assert(MAN_TEXT != m->last->type);
+
+		/*
+		 * This occurs in the following construction:
+		 *   .B
+		 *   .br
+		 *   .B
+		 *   .br
+		 *   I hate man macros.
+		 * Flat-out disallow this madness.
+		 */
+		if (MAN_NSCOPED & man_macros[m->last->tok].flags)
+			return(man_perr(m, ln, ppos, WLNSCOPE));
+
+		n = m->last;
+
+		assert(n);
+		assert(NULL == n->child);
+		assert(0 == n->nchild);
+
+		if ( ! man_nwarn(m, n, WLNSCOPE))
+			return(0);
+
+		man_node_unlink(m, n);
+		man_node_free(n);
+		m->flags &= ~MAN_ELINE;
+	}
+
 	/* Begin recursive parse sequence. */
 
-	assert(man_macros[c].fp);
+	assert(man_macros[tok].fp);
 
-	if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf))
+	if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 		goto err;
 
 out:
-	if ( ! (MAN_BLINE & fl))
+	/* 
+	 * We weren't in a block-line scope when entering the
+	 * above-parsed macro, so return.
+	 */
+
+	if ( ! (MAN_BLINE & fl)) {
+		m->flags &= ~MAN_ILINE; 
 		return(1);
+	}
 
+	/*
+	 * If we're in a block scope, then allow this macro to slip by
+	 * without closing scope around it.
+	 */
+
+	if (MAN_ILINE & m->flags) {
+		m->flags &= ~MAN_ILINE;
+		return(1);
+	}
+
 	/* 
 	 * If we've opened a new next-line element scope, then return
 	 * now, as the next line will close out the block scope.
@@ -479,8 +611,6 @@ out:
 
 	/* Close out the block scope opened in the prior line.  */
 
-	/* XXX - this should be in man_action.c. */
-
 	assert(MAN_BLINE & m->flags);
 	m->flags &= ~MAN_BLINE;
 
@@ -539,4 +669,29 @@ man_err(struct man *m, int line, int pos, int iserr, e
 		return(man_verr(m, line, pos, p));
 
 	return(man_vwarn(m, line, pos, p));
+}
+
+
+void
+man_node_unlink(struct man *m, struct man_node *n)
+{
+
+	if (n->prev) {
+		n->prev->next = n->next;
+		if (m->last == n) {
+			assert(NULL == n->next);
+			m->last = n->prev;
+			m->next = MAN_NEXT_SIBLING;
+		}
+	} else {
+		n->parent->child = n->next;
+		if (m->last == n) {
+			assert(NULL == n->next);
+			m->last = n->parent;
+			m->next = MAN_NEXT_CHILD;
+		}
+	}
+
+	if (n->next)
+		n->next->prev = n->prev;
 }