===================================================================
RCS file: /cvs/mandoc/mdoc.c,v
retrieving revision 1.49
retrieving revision 1.62
diff -u -p -r1.49 -r1.62
--- mandoc/mdoc.c	2009/03/01 23:14:15	1.49
+++ mandoc/mdoc.c	2009/03/12 02:57:36	1.62
@@ -1,4 +1,4 @@
-/* $Id: mdoc.c,v 1.49 2009/03/01 23:14:15 kristaps Exp $ */
+/* $Id: mdoc.c,v 1.62 2009/03/12 02:57:36 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -32,18 +32,15 @@
  * in macro.c and validate.c.
  */
 
-static	struct mdoc_arg	 *argdup(size_t, const struct mdoc_arg *);
-static	void		  argfree(size_t, struct mdoc_arg *);
-static	void	  	  argcpy(struct mdoc_arg *, 
-				const struct mdoc_arg *);
-
 static	struct mdoc_node *mdoc_node_alloc(const struct mdoc *);
 static	int		  mdoc_node_append(struct mdoc *, 
 				struct mdoc_node *);
-static	void		  mdoc_elem_free(struct mdoc_elem *);
-static	void		  mdoc_text_free(struct mdoc_text *);
 
+static	int		  parsetext(struct mdoc *, int, char *);
+static	int		  parsemacro(struct mdoc *, int, char *);
+static	int		  macrowarn(struct mdoc *, int, const char *);
 
+
 const	char *const __mdoc_macronames[MDOC_MAX] = {		 
 	"\\\"",		"Dd",		"Dt",		"Os",
 	"Sh",		"Ss",		"Pp",		"D1",
@@ -74,7 +71,9 @@ const	char *const __mdoc_macronames[MDOC_MAX] = {		 
 	"Tn",		"Ux",		"Xc",		"Xo",
 	"Fo",		"Fc",		"Oo",		"Oc",
 	"Bk",		"Ek",		"Bt",		"Hf",
-	"Fr",		"Ud",
+	"Fr",		"Ud",		"Lb",		"Ap",
+	"Lp",		"Lk",		"Mt",		"Brq",
+	"Bro",		"Brc"
 	};
 
 const	char *const __mdoc_argnames[MDOC_ARG_MAX] = {		 
@@ -85,20 +84,8 @@ const	char *const __mdoc_argnames[MDOC_ARG_MAX] = {		 
 	"tag",			"diag",			"hang",		 
 	"ohang",		"inset",		"column",	 
 	"width",		"compact",		"std",	 
-	"p1003.1-88",		"p1003.1-90",		"p1003.1-96",
-	"p1003.1-2001",		"p1003.1-2004",		"p1003.1",
-	"p1003.1b",		"p1003.1b-93",		"p1003.1c-95",
-	"p1003.1g-2000",	"p1003.2-92",		"p1387.2-95",
-	"p1003.2",		"p1387.2",		"isoC-90",
-	"isoC-amd1",		"isoC-tcor1",		"isoC-tcor2",
-	"isoC-99",		"ansiC",		"ansiC-89",
-	"ansiC-99",		"ieee754",		"iso8802-3",
-	"xpg3",			"xpg4",			"xpg4.2",
-	"xpg4.3",		"xbd5",			"xcu5",
-	"xsh5",			"xns5",			"xns5.2d2.0",
-	"xcurses4.2",		"susv2",		"susv3",
-	"svid4",		"filled",		"words",
-	"emphasis",		"symbolic",
+	"filled",		"words",		"emphasis",
+	"symbolic"
 	};
 
 const	char * const *mdoc_macronames = __mdoc_macronames;
@@ -135,13 +122,17 @@ mdoc_free(struct mdoc *mdoc)
 		free(mdoc->meta.os);
 	if (mdoc->meta.name)
 		free(mdoc->meta.name);
+	if (mdoc->meta.arch)
+		free(mdoc->meta.arch);
+	if (mdoc->meta.vol)
+		free(mdoc->meta.vol);
 
 	free(mdoc);
 }
 
 
 struct mdoc *
-mdoc_alloc(void *data, const struct mdoc_cb *cb)
+mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb)
 {
 	struct mdoc	*p;
 
@@ -154,7 +145,7 @@ mdoc_alloc(void *data, const struct mdoc_cb *cb)
 	p->last = xcalloc(1, sizeof(struct mdoc_node));
 	p->last->type = MDOC_ROOT;
 	p->first = p->last;
-
+	p->pflags = pflags;
 	p->next = MDOC_NEXT_CHILD;
 	p->htab = mdoc_tokhash_alloc();
 
@@ -180,65 +171,21 @@ mdoc_endparse(struct mdoc *mdoc)
 }
 
 
+/*
+ * Main parse routine.  Parses a single line -- really just hands off to
+ * the macro or text parser.
+ */
 int
-mdoc_parseln(struct mdoc *mdoc, int line, char *buf)
+mdoc_parseln(struct mdoc *m, int ln, char *buf)
 {
-	int		  c, i;
-	char		  tmp[5];
 
-	if (MDOC_HALT & mdoc->flags)
-		return(0);
+	/* If in error-mode, then we parse no more. */
 
-	mdoc->linetok = 0;
-
-	/*
-	 * FIXME: should puke on whitespace in non-literal displays.
-	 */
-
-	if ('.' != *buf) {
-		if (SEC_PROLOGUE == mdoc->lastnamed)
-			return(mdoc_perr(mdoc, line, 0, 
-				"no text in document prologue"));
-		if ( ! mdoc_word_alloc(mdoc, line, 0, buf))
-			return(0);
-		mdoc->next = MDOC_NEXT_SIBLING;
-		return(1);
-	}
-
-	if (buf[1] && '\\' == buf[1])
-		if (buf[2] && '\"' == buf[2])
-			return(1);
-
-	i = 1;
-	while (buf[i] && ! isspace((int)buf[i]) && i < (int)sizeof(tmp))
-		i++;
-
-	if (i == (int)sizeof(tmp)) {
-		mdoc->flags |= MDOC_HALT;
-		return(mdoc_perr(mdoc, line, 1, "unknown macro"));
-	} else if (i <= 2) {
-		mdoc->flags |= MDOC_HALT;
-		return(mdoc_perr(mdoc, line, 1, "unknown macro"));
-	}
-
-	i--;
-
-	(void)memcpy(tmp, buf + 1, (size_t)i);
-	tmp[i++] = 0;
-
-	if (MDOC_MAX == (c = mdoc_find(mdoc, tmp))) {
-		mdoc->flags |= MDOC_HALT;
-		return(mdoc_perr(mdoc, line, 1, "unknown macro"));
-	}
-
-	while (buf[i] && isspace((int)buf[i]))
-		i++;
-
-	if ( ! mdoc_macro(mdoc, c, line, 1, &i, buf)) {
-		mdoc->flags |= MDOC_HALT;
+	if (MDOC_HALT & m->flags)
 		return(0);
-	}
-	return(1);
+
+	return('.' == *buf ? parsemacro(m, ln, buf) :
+			parsetext(m, ln, buf));
 }
 
 
@@ -293,82 +240,46 @@ mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, 
 
 
 int
-mdoc_macro(struct mdoc *mdoc, int tok, 
-		int ln, int ppos, int *pos, char *buf)
+mdoc_macro(struct mdoc *m, int tok, 
+		int ln, int pp, int *pos, char *buf)
 {
 
-	assert(mdoc_macros[tok].fp);
+	/* FIXME - these should happen during validation. */
 
 	if (MDOC_PROLOGUE & mdoc_macros[tok].flags && 
-			SEC_PROLOGUE != mdoc->lastnamed)
-		return(mdoc_perr(mdoc, ln, ppos, "macro disallowed in document body"));
+			SEC_PROLOGUE != m->lastnamed)
+		return(mdoc_perr(m, ln, pp, 
+				"disallowed in document body"));
+
 	if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) && 
-			SEC_PROLOGUE == mdoc->lastnamed)
-		return(mdoc_perr(mdoc, ln, ppos, "macro disallowed in document prologue"));
-	if (1 != ppos && ! (MDOC_CALLABLE & mdoc_macros[tok].flags))
-		return(mdoc_perr(mdoc, ln, ppos, "macro not callable"));
-	return((*mdoc_macros[tok].fp)(mdoc, tok, ln, ppos, pos, buf));
+			SEC_PROLOGUE == m->lastnamed)
+		return(mdoc_perr(m, ln, pp, 
+				"disallowed in prologue"));
+
+	if (1 != pp && ! (MDOC_CALLABLE & mdoc_macros[tok].flags))
+		return(mdoc_perr(m, ln, pp, "not callable"));
+
+	return((*mdoc_macros[tok].fp)(m, tok, ln, pp, pos, buf));
 }
 
 
 static int
 mdoc_node_append(struct mdoc *mdoc, struct mdoc_node *p)
 {
-	const char	 *nn, *nt, *on, *ot, *act;
 
 	assert(mdoc->last);
 	assert(mdoc->first);
 	assert(MDOC_ROOT != p->type);
 
-	/* See if we exceed the suggest line-max. */
-
-	switch (p->type) {
-	case (MDOC_TEXT):
-		/* FALLTHROUGH */
-	case (MDOC_ELEM):
-		/* FALLTHROUGH */
-	case (MDOC_BLOCK):
-		mdoc->linetok++;
-		break;
-	default:
-		break;
-	}
-
-	/* This sort-of works (re-opening of text macros...). */
-	if (mdoc->linetok > MDOC_LINEARG_SOFTMAX) 
-		if ( ! mdoc_nwarn(mdoc, p, WARN_COMPAT, 
-					"suggested %d tokens per line exceeded (has %d)",
-					MDOC_LINEARG_SOFTMAX, mdoc->linetok))
-			return(0);
-
-	if (MDOC_TEXT == mdoc->last->type)
-		on = "<text>";
-	else if (MDOC_ROOT == mdoc->last->type)
-		on = "<root>";
-	else
-		on = mdoc_macronames[mdoc->last->tok];
-
-	if (MDOC_TEXT == p->type)
-		nn = "<text>";
-	else if (MDOC_ROOT == p->type)
-		nn = "<root>";
-	else
-		nn = mdoc_macronames[p->tok];
-
-	ot = mdoc_type2a(mdoc->last->type);
-	nt = mdoc_type2a(p->type);
-
 	switch (mdoc->next) {
 	case (MDOC_NEXT_SIBLING):
 		mdoc->last->next = p;
 		p->prev = mdoc->last;
 		p->parent = mdoc->last->parent;
-		act = "sibling";
 		break;
 	case (MDOC_NEXT_CHILD):
 		mdoc->last->child = p;
 		p->parent = mdoc->last;
-		act = "child";
 		break;
 	default:
 		abort();
@@ -381,23 +292,21 @@ mdoc_node_append(struct mdoc *mdoc, struct mdoc_node *
 	switch (p->type) {
 	case (MDOC_HEAD):
 		assert(MDOC_BLOCK == p->parent->type);
-		p->parent->data.block.head = p;
+		p->parent->head = p;
 		break;
 	case (MDOC_TAIL):
 		assert(MDOC_BLOCK == p->parent->type);
-		p->parent->data.block.tail = p;
+		p->parent->tail = p;
 		break;
 	case (MDOC_BODY):
 		assert(MDOC_BLOCK == p->parent->type);
-		p->parent->data.block.body = p;
+		p->parent->body = p;
 		break;
 	default:
 		break;
 	}
 
 	mdoc->last = p;
-	mdoc_msg(mdoc, "parse: %s `%s' %s of %s `%s'", 
-			nt, nn, act, ot, on);
 	return(1);
 }
 
@@ -486,7 +395,7 @@ mdoc_root_alloc(struct mdoc *mdoc)
 
 int
 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos, 
-		int tok, size_t argsz, const struct mdoc_arg *args)
+		int tok, struct mdoc_arg *args)
 {
 	struct mdoc_node *p;
 
@@ -496,16 +405,18 @@ mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 	p->line = line;
 	p->type = MDOC_BLOCK;
 	p->tok = tok;
-	p->data.block.argc = argsz;
-	p->data.block.argv = argdup(argsz, args);
+	p->args = args;
 
+	if (args)
+		(args->refcnt)++;
+
 	return(mdoc_node_append(mdoc, p));
 }
 
 
 int
 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos, 
-		int tok, size_t argsz, const struct mdoc_arg *args)
+		int tok, struct mdoc_arg *args)
 {
 	struct mdoc_node *p;
 
@@ -515,9 +426,11 @@ mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos, 
 	p->pos = pos;
 	p->type = MDOC_ELEM;
 	p->tok = tok;
-	p->data.elem.argc = argsz;
-	p->data.elem.argv = argdup(argsz, args);
+	p->args = args;
 
+	if (args)
+		(args->refcnt)++;
+
 	return(mdoc_node_append(mdoc, p));
 }
 
@@ -533,77 +446,20 @@ mdoc_word_alloc(struct mdoc *mdoc, 
 	p->line = line;
 	p->pos = pos;
 	p->type = MDOC_TEXT;
-	p->data.text.string = xstrdup(word);
+	p->string = xstrdup(word);
 
 	return(mdoc_node_append(mdoc, p));
 }
 
 
-static void
-argfree(size_t sz, struct mdoc_arg *p)
-{
-	int		 i, j;
-
-	if (0 == sz)
-		return;
-
-	assert(p);
-	/* LINTED */
-	for (i = 0; i < (int)sz; i++)
-		if (p[i].sz > 0) {
-			assert(p[i].value);
-			/* LINTED */
-			for (j = 0; j < (int)p[i].sz; j++)
-				free(p[i].value[j]);
-			free(p[i].value);
-		}
-	free(p);
-}
-
-
-static void
-mdoc_elem_free(struct mdoc_elem *p)
-{
-
-	argfree(p->argc, p->argv);
-}
-
-
-static void
-mdoc_block_free(struct mdoc_block *p)
-{
-
-	argfree(p->argc, p->argv);
-}
-
-
-static void
-mdoc_text_free(struct mdoc_text *p)
-{
-
-	if (p->string)
-		free(p->string);
-}
-
-
 void
 mdoc_node_free(struct mdoc_node *p)
 {
 
-	switch (p->type) {
-	case (MDOC_TEXT):
-		mdoc_text_free(&p->data.text);
-		break;
-	case (MDOC_ELEM):
-		mdoc_elem_free(&p->data.elem);
-		break;
-	case (MDOC_BLOCK):
-		mdoc_block_free(&p->data.block);
-		break;
-	default:
-		break;
-	}
-
+	if (p->string)
+		free(p->string);
+	if (p->args)
+		mdoc_argv_free(p->args);
 	free(p);
 }
 
@@ -621,67 +477,92 @@ mdoc_node_freelist(struct mdoc_node *p)
 }
 
 
-int
-mdoc_find(const struct mdoc *mdoc, const char *key)
+/*
+ * Parse free-form text, that is, a line that does not begin with the
+ * control character.
+ */
+static int
+parsetext(struct mdoc *mdoc, int line, char *buf)
 {
 
-	return(mdoc_tokhash_find(mdoc->htab, key));
+	if (SEC_PROLOGUE == mdoc->lastnamed)
+		return(mdoc_perr(mdoc, line, 0,
+			"text disallowed in prologue"));
+
+	if ( ! mdoc_word_alloc(mdoc, line, 0, buf))
+		return(0);
+
+	mdoc->next = MDOC_NEXT_SIBLING;
+	return(1);
 }
 
 
-static void
-argcpy(struct mdoc_arg *dst, const struct mdoc_arg *src)
+static int
+macrowarn(struct mdoc *m, int ln, const char *buf)
 {
-	int		 i;
-
-	dst->line = src->line;
-	dst->pos = src->pos;
-	dst->arg = src->arg;
-	if (0 == (dst->sz = src->sz))
-		return;
-	dst->value = xcalloc(dst->sz, sizeof(char *));
-	for (i = 0; i < (int)dst->sz; i++)
-		dst->value[i] = xstrdup(src->value[i]);
+	if ( ! (MDOC_IGN_MACRO & m->pflags))
+		return(mdoc_perr(m, ln, 1, "unknown macro: %s%s", 
+				buf, strlen(buf) > 3 ? "..." : ""));
+	return(mdoc_pwarn(m, ln, 1, WARN_SYNTAX,
+				"unknown macro: %s%s",
+				buf, strlen(buf) > 3 ? "..." : ""));
 }
 
 
-static struct mdoc_arg *
-argdup(size_t argsz, const struct mdoc_arg *args)
+
+/*
+ * Parse a macro line, that is, a line beginning with the control
+ * character.
+ */
+int
+parsemacro(struct mdoc *m, int ln, char *buf)
 {
-	struct mdoc_arg	*pp;
-	int		 i;
+	int		  i, c;
+	char		  mac[5];
 
-	if (0 == argsz)
-		return(NULL);
+	/* Comments are quickly ignored. */
 
-	pp = xcalloc((size_t)argsz, sizeof(struct mdoc_arg));
-	for (i = 0; i < (int)argsz; i++)
-		argcpy(&pp[i], &args[i]);
+	if (buf[1] && '\\' == buf[1])
+		if (buf[2] && '\"' == buf[2])
+			return(1);
 
-	return(pp);
-}
+	/* Copy the first word into a nil-terminated buffer. */
 
+	for (i = 1; i < 5; i++) {
+		if (0 == (mac[i - 1] = buf[i]))
+			break;
+		else if (isspace((unsigned char)buf[i]))
+			break;
+	}
 
-/* FIXME: deprecate. */
-char *
-mdoc_node2a(struct mdoc_node *node)
-{
-	static char	 buf[64];
+	mac[i - 1] = 0;
 
-	assert(node);
+	if (i == 5 || i <= 2) {
+		if ( ! macrowarn(m, ln, mac))
+			goto err;
+		return(1);
+	} 
+	
+	if (MDOC_MAX == (c = mdoc_tokhash_find(m->htab, mac))) {
+		if ( ! macrowarn(m, ln, mac))
+			goto err;
+		return(1);
+	}
 
-	buf[0] = 0;
-	(void)xstrlcat(buf, mdoc_type2a(node->type), 64);
-	if (MDOC_ROOT == node->type)
-		return(buf);
-	(void)xstrlcat(buf, " `", 64);
-	if (MDOC_TEXT == node->type)
-		(void)xstrlcat(buf, node->data.text.string, 64);
-	else
-		(void)xstrlcat(buf, mdoc_macronames[node->tok], 64);
-	(void)xstrlcat(buf, "'", 64);
+	/* The macro is sane.  Jump to the next word. */
 
-	return(buf);
-}
+	while (buf[i] && isspace((unsigned char)buf[i]))
+		i++;
 
+	/* Begin recursive parse sequence. */
 
+	if ( ! mdoc_macro(m, c, ln, 1, &i, buf)) 
+		goto err;
+
+	return(1);
+
+err:	/* Error out. */
+
+	m->flags |= MDOC_HALT;
+	return(0);
+}