===================================================================
RCS file: /cvs/mandoc/mdoc.c,v
retrieving revision 1.78
retrieving revision 1.89
diff -u -p -r1.78 -r1.89
--- mandoc/mdoc.c	2009/06/15 09:55:43	1.78
+++ mandoc/mdoc.c	2009/07/07 09:29:15	1.89
@@ -1,4 +1,4 @@
-/*	$Id: mdoc.c,v 1.78 2009/06/15 09:55:43 kristaps Exp $ */
+/*	$Id: mdoc.c,v 1.89 2009/07/07 09:29:15 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -23,18 +23,66 @@
 
 #include "libmdoc.h"
 
-enum	merr {
-	ENOCALL,
-	EBODYPROL,
-	EPROLBODY,
-	ESPACE,
-	ETEXTPROL,
-	ENOBLANK,
-	EMALLOC
+const	char *const __mdoc_merrnames[MERRMAX] = {		 
+	"trailing whitespace", /* ETAILWS */
+	"empty last list column", /* ECOLEMPTY */
+	"argument-like parameter", /* EARGVPARM */
+	"unexpected quoted parameter", /* EQUOTPARM */
+	"unterminated quoted parameter", /* EQUOTTERM */
+	"system: malloc error", /* EMALLOC */
+	"argument parameter suggested", /* EARGVAL */
+	"macro not callable", /* ENOCALL */
+	"macro disallowed in prologue", /* EBODYPROL */
+	"macro disallowed in body", /* EPROLBODY */
+	"text disallowed in prologue", /* ETEXTPROL */
+	"blank line disallowed", /* ENOBLANK */
+	"text parameter too long", /* ETOOLONG */
+	"invalid escape sequence", /* EESCAPE */
+	"invalid character", /* EPRINT */
+	"document has no body", /* ENODAT */
+	"document has no prologue", /* ENOPROLOGUE */
+	"expected line arguments", /* ELINE */
+	"invalid AT&T argument", /* EATT */
+	"default name not yet set", /* ENAME */
+	"missing list type", /* ELISTTYPE */
+	"missing display type", /* EDISPTYPE */
+	"too many display types", /* EMULTIDISP */
+	"too many list types", /* EMULTILIST */
+	"NAME section must be first", /* ESECNAME */
+	"badly-formed NAME section", /* ENAMESECINC */
+	"argument repeated", /* EARGREP */
+	"expected boolean parameter", /* EBOOL */
+	"inconsistent column syntax", /* ECOLMIS */
+	"nested display invalid", /* ENESTDISP */
+	"width argument missing", /* EMISSWIDTH */
+	"invalid section for this manual section", /* EWRONGMSEC */
+	"section out of conventional order", /* ESECOOO */
+	"section repeated", /* ESECREP */
+	"invalid standard argument", /* EBADSTAND */
+	"multi-line arguments discouraged", /* ENOMULTILINE */
+	"multi-line arguments suggested", /* EMULTILINE */
+	"line arguments discouraged", /* ENOLINE */
+	"prologue macro out of conventional order", /* EPROLOOO */
+	"prologue macro repeated", /* EPROLREP */
+	"invalid manual section", /* EBADMSEC */
+	"invalid section", /* EBADSEC */
+	"invalid font mode", /* EFONT */
+	"invalid date syntax", /* EBADDATE */
+	"invalid number format", /* ENUMFMT */
+	"superfluous width argument", /* ENOWIDTH */
+	"system: utsname error", /* EUTSNAME */
+	"obsolete macro", /* EOBS */
+	"macro-like parameter", /* EMACPARM */
+	"end-of-line scope violation", /* EIMPBRK */
+	"empty macro ignored", /* EIGNE */
+	"unclosed explicit scope", /* EOPEN */
+	"unterminated quoted phrase", /* EQUOTPHR */
+	"closure macro without prior context", /* ENOCTX */
+	"invalid whitespace after control character", /* ESPACE */
 };
 
 const	char *const __mdoc_macronames[MDOC_MAX] = {		 
-	"\\\"",		"Dd",		"Dt",		"Os",
+	"Ap",		"Dd",		"Dt",		"Os",
 	"Sh",		"Ss",		"Pp",		"D1",
 	"Dl",		"Bd",		"Ed",		"Bl",
 	"El",		"It",		"Ad",		"An",
@@ -63,12 +111,12 @@ const	char *const __mdoc_macronames[MDOC_MAX] = {		 
 	"Tn",		"Ux",		"Xc",		"Xo",
 	"Fo",		"Fc",		"Oo",		"Oc",
 	"Bk",		"Ek",		"Bt",		"Hf",
-	"Fr",		"Ud",		"Lb",		"Ap",
-	"Lp",		"Lk",		"Mt",		"Brq",
+	"Fr",		"Ud",		"Lb",		"Lp",
+	"Lk",		"Mt",		"Brq",		"Bro",
 	/* LINTED */
-	"Bro",		"Brc",		"\%C",		"Es",
+	"Brc",		"\%C",		"Es",		"En",
 	/* LINTED */
-	"En",		"Dx",		"\%Q"
+	"Dx",		"\%Q"
 	};
 
 const	char *const __mdoc_argnames[MDOC_ARG_MAX] = {		 
@@ -95,13 +143,8 @@ static	int		  node_append(struct mdoc *, 
 static	int		  parsetext(struct mdoc *, int, char *);
 static	int		  parsemacro(struct mdoc *, int, char *);
 static	int		  macrowarn(struct mdoc *, int, const char *);
-static	int		  perr(struct mdoc *, int, int, enum merr);
 
-#define verr(m, t) perr((m), (m)->last->line, (m)->last->pos, (t))
 
-/*
- * Get the first (root) node of the parse tree.
- */
 const struct mdoc_node *
 mdoc_node(const struct mdoc *m)
 {
@@ -118,6 +161,9 @@ mdoc_meta(const struct mdoc *m)
 }
 
 
+/*
+ * Frees volatile resources (parse tree, meta-data, fields).
+ */
 static void
 mdoc_free1(struct mdoc *mdoc)
 {
@@ -137,13 +183,16 @@ mdoc_free1(struct mdoc *mdoc)
 }
 
 
+/*
+ * Allocate all volatile resources (parse tree, meta-data, fields).
+ */
 static int
 mdoc_alloc1(struct mdoc *mdoc)
 {
 
 	bzero(&mdoc->meta, sizeof(struct mdoc_meta));
 	mdoc->flags = 0;
-	mdoc->lastnamed = mdoc->lastsec = 0;
+	mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
 	mdoc->last = calloc(1, sizeof(struct mdoc_node));
 	if (NULL == mdoc->last)
 		return(0);
@@ -156,9 +205,10 @@ mdoc_alloc1(struct mdoc *mdoc)
 
 
 /*
- * Free up all resources contributed by a parse:  the node tree,
- * meta-data and so on.  Then reallocate the root node for another
- * parse.
+ * Free up volatile resources (see mdoc_free1()) then re-initialises the
+ * data with mdoc_alloc1().  After invocation, parse data has been reset
+ * and the parser is ready for re-invocation on a new tree; however,
+ * cross-parse non-volatile data is kept intact.
  */
 int
 mdoc_reset(struct mdoc *mdoc)
@@ -170,7 +220,8 @@ mdoc_reset(struct mdoc *mdoc)
 
 
 /*
- * Completely free up all resources.
+ * Completely free up all volatile and non-volatile parse resources.
+ * After invocation, the pointer is no longer usable.
  */
 void
 mdoc_free(struct mdoc *mdoc)
@@ -183,6 +234,9 @@ mdoc_free(struct mdoc *mdoc)
 }
 
 
+/*
+ * Allocate volatile and non-volatile parse resources.  
+ */
 struct mdoc *
 mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb)
 {
@@ -209,7 +263,7 @@ mdoc_alloc(void *data, int pflags, const struct mdoc_c
 
 /*
  * Climb back up the parse tree, validating open scopes.  Mostly calls
- * through to macro_end in macro.c.
+ * through to macro_end() in macro.c.
  */
 int
 mdoc_endparse(struct mdoc *m)
@@ -226,14 +280,12 @@ mdoc_endparse(struct mdoc *m)
 
 /*
  * Main parse routine.  Parses a single line -- really just hands off to
- * the macro or text parser.
+ * the macro (parsemacro()) or text parser (parsetext()).
  */
 int
 mdoc_parseln(struct mdoc *m, int ln, char *buf)
 {
 
-	/* If in error-mode, then we parse no more. */
-
 	if (MDOC_HALT & m->flags)
 		return(0);
 
@@ -242,22 +294,6 @@ mdoc_parseln(struct mdoc *m, int ln, char *buf)
 }
 
 
-void
-mdoc_vmsg(struct mdoc *mdoc, int ln, int pos, const char *fmt, ...)
-{
-	char		  buf[256];
-	va_list		  ap;
-
-	if (NULL == mdoc->cb.mdoc_msg)
-		return;
-
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
-	(*mdoc->cb.mdoc_msg)(mdoc->data, ln, pos, buf);
-}
-
-
 int
 mdoc_verr(struct mdoc *mdoc, int ln, int pos, 
 		const char *fmt, ...)
@@ -271,13 +307,13 @@ mdoc_verr(struct mdoc *mdoc, int ln, int pos, 
 	va_start(ap, fmt);
 	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 	va_end(ap);
+
 	return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf));
 }
 
 
 int
-mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, 
-		enum mdoc_warn type, const char *fmt, ...)
+mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, const char *fmt, ...)
 {
 	char		 buf[256];
 	va_list		 ap;
@@ -288,199 +324,46 @@ mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, 
 	va_start(ap, fmt);
 	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 	va_end(ap);
-	return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, type, buf));
-}
 
-
-int
-mdoc_nwarn(struct mdoc *mdoc, const struct mdoc_node *node, enum mdoc_warn type,
-		const char *fmt, ...)
-{
-	char		 buf[256];
-	va_list		 ap;
-
-	if (NULL == mdoc->cb.mdoc_warn)
-		return(0);
-
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
-	return((*mdoc->cb.mdoc_warn)(mdoc->data, node->line, node->pos, type,
-	    buf));
+	return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, buf));
 }
 
-int
-mdoc_nerr(struct mdoc *mdoc, const struct mdoc_node *node, const char *fmt, ...)
-{
-	char		 buf[256];
-	va_list		 ap;
 
-	if (NULL == mdoc->cb.mdoc_err)
-		return(0);
-
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
-	return((*mdoc->cb.mdoc_err)(mdoc->data, node->line, node->pos, buf));
-}
-
-
 int
-mdoc_warn(struct mdoc *mdoc, enum mdoc_warn type, const char *fmt, ...)
+mdoc_err(struct mdoc *m, int line, int pos, int iserr, enum merr type)
 {
-	char		 buf[256];
-	va_list		 ap;
+	const char	*p;
 
-	if (NULL == mdoc->cb.mdoc_warn)
-		return(0);
+	p = __mdoc_merrnames[(int)type];
+	assert(p);
 
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
-	return((*mdoc->cb.mdoc_warn)(mdoc->data, mdoc->last->line,
-	    mdoc->last->pos, type, buf));
-}
+	if (iserr)
+		return(mdoc_verr(m, line, pos, p));
 
-
-int
-mdoc_err(struct mdoc *mdoc, const char *fmt, ...)
-{
-	char		 buf[256];
-	va_list		 ap;
-
-	if (NULL == mdoc->cb.mdoc_err)
-		return(0);
-
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
-	return((*mdoc->cb.mdoc_err)(mdoc->data, mdoc->last->line,
-	    mdoc->last->pos, buf));
+	return(mdoc_vwarn(m, line, pos, p));
 }
 
 
-void
-mdoc_msg(struct mdoc *mdoc, const char *fmt, ...)
-{
-	char		  buf[256];
-	va_list		  ap;
-
-	if (NULL == mdoc->cb.mdoc_msg)
-		return;
-
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
-	(*mdoc->cb.mdoc_msg)(mdoc->data, mdoc->last->line, mdoc->last->pos,
-	    buf);
-}
-
-
-void
-mdoc_pmsg(struct mdoc *mdoc, int line, int pos, const char *fmt, ...)
-{
-	char		  buf[256];
-	va_list		  ap;
-
-	if (NULL == mdoc->cb.mdoc_msg)
-		return;
-
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
-	(*mdoc->cb.mdoc_msg)(mdoc->data, line, pos, buf);
-}
-
-
 int
-mdoc_pwarn(struct mdoc *mdoc, int line, int pos, enum mdoc_warn type,
-		const char *fmt, ...)
-{
-	char		 buf[256];
-	va_list		 ap;
-
-	if (NULL == mdoc->cb.mdoc_warn)
-		return(0);
-
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
-	return((*mdoc->cb.mdoc_warn)(mdoc->data, line, pos, type, buf));
-}
-
-int
-mdoc_perr(struct mdoc *mdoc, int line, int pos, const char *fmt, ...)
-{
-	char		 buf[256];
-	va_list		 ap;
-
-	if (NULL == mdoc->cb.mdoc_err)
-		return(0);
-
-	va_start(ap, fmt);
-	(void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
-	va_end(ap);
-	return((*mdoc->cb.mdoc_err)(mdoc->data, line, pos, buf));
-}
-
-
-int
 mdoc_macro(struct mdoc *m, int tok, 
 		int ln, int pp, int *pos, char *buf)
 {
 
-	/* FIXME - these should happen during validation. */
-
 	if (MDOC_PROLOGUE & mdoc_macros[tok].flags && 
-			SEC_PROLOGUE != m->lastnamed)
-		return(perr(m, ln, pp, EPROLBODY));
-
+			MDOC_PBODY & m->flags)
+		return(mdoc_perr(m, ln, pp, EPROLBODY));
 	if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) && 
-			SEC_PROLOGUE == m->lastnamed)
-		return(perr(m, ln, pp, EBODYPROL));
+			! (MDOC_PBODY & m->flags))
+		return(mdoc_perr(m, ln, pp, EBODYPROL));
 
 	if (1 != pp && ! (MDOC_CALLABLE & mdoc_macros[tok].flags))
-		return(perr(m, ln, pp, ENOCALL));
+		return(mdoc_perr(m, ln, pp, ENOCALL));
 
 	return((*mdoc_macros[tok].fp)(m, tok, ln, pp, pos, buf));
 }
 
 
 static int
-perr(struct mdoc *m, int line, int pos, enum merr type)
-{
-	char		*p;
-
-	p = NULL;
-	switch (type) {
-	case (ENOCALL):
-		p = "not callable";
-		break;
-	case (EPROLBODY):
-		p = "macro disallowed in document body";
-		break;
-	case (EBODYPROL):
-		p = "macro disallowed in document prologue";
-		break;
-	case (EMALLOC):
-		p = "memory exhausted";
-		break;
-	case (ETEXTPROL):
-		p = "text disallowed in document prologue";
-		break;
-	case (ENOBLANK):
-		p = "blank lines disallowed in non-literal contexts";
-		break;
-	case (ESPACE):
-		p = "whitespace disallowed after delimiter";
-		break;
-	}
-	assert(p);
-	return(mdoc_perr(m, line, pos, p));
-}
-
-
-static int
 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 {
 
@@ -503,6 +386,8 @@ node_append(struct mdoc *mdoc, struct mdoc_node *p)
 		/* NOTREACHED */
 	}
 
+	p->parent->nchild++;
+
 	if ( ! mdoc_valid_pre(mdoc, p))
 		return(0);
 	if ( ! mdoc_action_pre(mdoc, p))
@@ -549,7 +434,7 @@ node_alloc(struct mdoc *mdoc, int line, 
 	struct mdoc_node *p;
 
 	if (NULL == (p = calloc(1, sizeof(struct mdoc_node)))) {
-		(void)verr(mdoc, EMALLOC);
+		(void)mdoc_nerr(mdoc, mdoc->last, EMALLOC);
 		return(NULL);
 	}
 
@@ -645,9 +530,10 @@ mdoc_word_alloc(struct mdoc *mdoc, 
 	if (NULL == p)
 		return(0);
 	if (NULL == (p->string = strdup(word))) {
-		(void)verr(mdoc, EMALLOC);
+		(void)mdoc_nerr(mdoc, mdoc->last, EMALLOC);
 		return(0);
 	}
+
 	return(node_append(mdoc, p));
 }
 
@@ -656,6 +542,8 @@ void
 mdoc_node_free(struct mdoc_node *p)
 {
 
+	if (p->parent)
+		p->parent->nchild--;
 	if (p->string)
 		free(p->string);
 	if (p->args)
@@ -673,6 +561,7 @@ mdoc_node_freelist(struct mdoc_node *p)
 	if (p->next)
 		mdoc_node_freelist(p->next);
 
+	assert(0 == p->nchild);
 	mdoc_node_free(p);
 }
 
@@ -685,11 +574,11 @@ static int
 parsetext(struct mdoc *m, int line, char *buf)
 {
 
-	if (SEC_PROLOGUE == m->lastnamed)
-		return(perr(m, line, 0, ETEXTPROL));
+	if (SEC_NONE == m->lastnamed)
+		return(mdoc_perr(m, line, 0, ETEXTPROL));
 
 	if (0 == buf[0] && ! (MDOC_LITERAL & m->flags))
-		return(perr(m, line, 0, ENOBLANK));
+		return(mdoc_perr(m, line, 0, ENOBLANK));
 
 	if ( ! mdoc_word_alloc(m, line, 0, buf))
 		return(0);
@@ -703,16 +592,14 @@ static int
 macrowarn(struct mdoc *m, int ln, const char *buf)
 {
 	if ( ! (MDOC_IGN_MACRO & m->pflags))
-		return(mdoc_perr(m, ln, 1, 
+		return(mdoc_verr(m, ln, 1, 
 				"unknown macro: %s%s", 
 				buf, strlen(buf) > 3 ? "..." : ""));
-	return(mdoc_pwarn(m, ln, 1, WARN_SYNTAX,
-				"unknown macro: %s%s",
+	return(mdoc_vwarn(m, ln, 1, "unknown macro: %s%s",
 				buf, strlen(buf) > 3 ? "..." : ""));
 }
 
 
-
 /*
  * Parse a macro line, that is, a line beginning with the control
  * character.
@@ -723,7 +610,7 @@ parsemacro(struct mdoc *m, int ln, char *buf)
 	int		  i, c;
 	char		  mac[5];
 
-	/* Comments and empties are quickly ignored. */
+	/* Empty lines are ignored. */
 
 	if (0 == buf[1])
 		return(1);
@@ -734,12 +621,8 @@ parsemacro(struct mdoc *m, int ln, char *buf)
 			i++;
 		if (0 == buf[i])
 			return(1);
-		return(perr(m, ln, 1, ESPACE));
+		return(mdoc_perr(m, ln, 1, ESPACE));
 	}
-
-	if (buf[1] && '\\' == buf[1])
-		if (buf[2] && '\"' == buf[2])
-			return(1);
 
 	/* Copy the first word into a nil-terminated buffer. */