===================================================================
RCS file: /cvs/mandoc/Attic/validate.c,v
retrieving revision 1.1
retrieving revision 1.12
diff -u -p -r1.1 -r1.12
--- mandoc/Attic/validate.c	2008/11/29 14:14:21	1.1
+++ mandoc/Attic/validate.c	2009/01/02 09:29:06	1.12
@@ -1,4 +1,4 @@
-/* $Id: validate.c,v 1.1 2008/11/29 14:14:21 kristaps Exp $ */
+/* $Id: validate.c,v 1.12 2009/01/02 09:29:06 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -16,350 +16,411 @@
  * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
  * PERFORMANCE OF THIS SOFTWARE.
  */
-#include <sys/param.h>
-
 #include <assert.h>
-#include <ctype.h>
-#include <err.h>
-#include <stdio.h>
 #include <stdlib.h>
-#include <string.h>
 
-#include "libmdocml.h"
 #include "private.h"
 
-#ifdef	__linux__ /* FIXME */
-#define	strlcat		strncat
-#endif
+/* FIXME: `.St' can only have one argument set. */
 
-struct	md_valid {
-	const struct md_args	*args;
-	const struct md_rbuf	*rbuf;
-	struct md_mbuf	*mbuf;
-	struct rofftree	*tree;
+typedef int	(*v_args_sz)(struct mdoc *, int, int, int);
+typedef int	(*v_args)(struct mdoc *, int, int, int, 
+			const char *[], int, const struct mdoc_arg *);
+typedef int	(*v_post)(struct mdoc *, int, int);
 
-	size_t		 indent;
-	size_t		 pos;
+static	int	  need_head_child(struct mdoc *, int, int);
+static	int	  no_head_child(struct mdoc *, int, int);
 
-	int		 flags;
-#define	MD_LITERAL	(1 << 0)
+static	int	  assert_eq0(struct mdoc *, int, int, int);
+static	int	  assert_le1(struct mdoc *, int, int, int);
+static	int	  need_eq0(struct mdoc *, int, int, int);
+static	int	  need_eq1(struct mdoc *, int, int, int);
+static	int	  need_ge1(struct mdoc *, int, int, int);
+static	int	  need_le2(struct mdoc *, int, int, int);
+static	int	  want_eq0(struct mdoc *, int, int, int);
+static	int	  want_ge1(struct mdoc *, int, int, int);
+
+static	int	  args_ref(struct mdoc *, int, int, int, 
+			const char *[], int, const struct mdoc_arg *);
+static	int	  args_bool(struct mdoc *, int, int, int, 
+			const char *[], int, const struct mdoc_arg *);
+static	int	  args_Sh(struct mdoc *, int, int, int, 
+			const char *[], int, const struct mdoc_arg *);
+static	int	  args_blocknest(struct mdoc *, int, int, int, 
+			const char *[], int, const struct mdoc_arg *);
+static	int	  args_At(struct mdoc *, int, int, int, 
+			const char *[], int, const struct mdoc_arg *);
+static	int	  args_Xr(struct mdoc *, int, int, int, 
+			const char *[], int, const struct mdoc_arg *);
+
+struct	valids {
+	v_args_sz sz;
+	v_args	  args;
+	v_post	  post;
 };
 
-static	void		 roffmsg(void *arg, enum roffmsg, 
-				const char *, const char *, char *);
-static	int		 roffhead(void *);
-static	int		 rofftail(void *);
-static	int		 roffin(void *, int, int *, char **);
-static	int		 roffdata(void *, char *);
-static	int		 roffout(void *, int);
-static	int		 roffblkin(void *, int);
-static	int		 roffblkout(void *, int);
-static	int		 roffspecial(void *, int);
 
-static	int		 mbuf_newline(struct md_valid *);
-static	int		 mbuf_indent(struct md_valid *);
-static	int		 mbuf_data(struct md_valid *, char *);
+/* 
+ * FIXME: have arrays of function pointers in case we want multiple
+ * check callbacks per macro.
+ */
 
+const	struct valids mdoc_valids[MDOC_MAX] = {
+	{ NULL, NULL, NULL }, /* \" */
+	{ NULL, NULL, NULL }, /* Dd */ /* TODO */
+	{ NULL, NULL, NULL }, /* Dt */ /* TODO */
+	{ NULL, NULL, NULL }, /* Os */ /* TODO */
+	{ NULL, args_Sh, NULL }, /* Sh */
+	{ NULL, NULL, NULL }, /* Ss */ 
+	{ want_eq0, NULL, NULL }, /* Pp */ 
+	{ NULL, args_blocknest, need_head_child }, /* D1 */
+	{ NULL, args_blocknest, need_head_child }, /* Dl */
+	{ NULL, args_blocknest, NULL }, /* Bd */
+	{ NULL, NULL, no_head_child }, /* Ed */
+	{ NULL, NULL, NULL }, /* Bl */
+	{ NULL, NULL, no_head_child }, /* El */
+	{ NULL, NULL, NULL }, /* It */
+	{ need_ge1, NULL, NULL }, /* Ad */ 
+	{ NULL, NULL, NULL }, /* An */ 
+	{ NULL, NULL, NULL }, /* Ar */
+	{ need_ge1, NULL, NULL }, /* Cd */
+	{ NULL, NULL, NULL }, /* Cm */
+	{ need_ge1, NULL, NULL }, /* Dv */ 
+	{ need_ge1, NULL, NULL }, /* Er */ 
+	{ need_ge1, NULL, NULL }, /* Ev */ 
+	{ NULL, NULL, NULL }, /* Ex */
+	{ need_ge1, NULL, NULL }, /* Fa */ 
+	{ NULL, NULL, NULL }, /* Fd */ 
+	{ NULL, NULL, NULL }, /* Fl */
+	{ need_ge1, NULL, NULL }, /* Fn */ 
+	{ want_ge1, NULL, NULL }, /* Ft */ 
+	{ need_ge1, NULL, NULL }, /* Ic */ 
+	{ need_eq1, NULL, NULL }, /* In */ 
+	{ want_ge1, NULL, NULL }, /* Li */
+	{ want_ge1, NULL, NULL }, /* Nd */ 
+	{ NULL, NULL, NULL }, /* Nm */ 
+	{ NULL, NULL, NULL }, /* Op */
+	{ NULL, NULL, NULL }, /* Ot */
+	{ want_ge1, NULL, NULL }, /* Pa */
+	{ NULL, NULL, NULL }, /* Rv */
+	{ NULL, NULL, NULL }, /* St */
+	{ need_ge1, NULL, NULL }, /* Va */
+	{ need_ge1, NULL, NULL }, /* Vt */ 
+	{ need_le2, args_Xr, NULL }, /* Xr */
+	{ need_ge1, args_ref, NULL }, /* %A */
+	{ need_ge1, args_ref, NULL }, /* %B */
+	{ need_ge1, args_ref, NULL }, /* %D */
+	{ need_ge1, args_ref, NULL }, /* %I */
+	{ need_ge1, args_ref, NULL }, /* %J */
+	{ need_ge1, args_ref, NULL }, /* %N */
+	{ need_ge1, args_ref, NULL }, /* %O */
+	{ need_ge1, args_ref, NULL }, /* %P */
+	{ need_ge1, args_ref, NULL }, /* %R */
+	{ need_ge1, args_ref, NULL }, /* %T */
+	{ need_ge1, args_ref, NULL }, /* %V */
+	{ NULL, NULL, NULL }, /* Ac */
+	{ NULL, NULL, NULL }, /* Ao */
+	{ NULL, NULL, NULL }, /* Aq */
+	{ need_le2, args_At, NULL }, /* At */
+	{ NULL, NULL, NULL }, /* Bc */
+	{ NULL, NULL, NULL }, /* Bf */ 
+	{ NULL, NULL, NULL }, /* Bo */
+	{ NULL, NULL, NULL }, /* Bq */
+	{ assert_le1, NULL, NULL }, /* Bsx */
+	{ assert_le1, NULL, NULL }, /* Bx */
+	{ need_eq1, args_bool, NULL }, /* Db */
+	{ NULL, NULL, NULL }, /* Dc */
+	{ NULL, NULL, NULL }, /* Do */
+	{ NULL, NULL, NULL }, /* Dq */
+	{ NULL, NULL, NULL }, /* Ec */
+	{ NULL, NULL, NULL }, /* Ef */
+	{ need_ge1, NULL, NULL }, /* Em */ 
+	{ NULL, NULL, NULL }, /* Eo */
+	{ assert_le1, NULL, NULL }, /* Fx */
+	{ want_ge1, NULL, NULL }, /* Ms */
+	{ NULL, NULL, NULL }, /* No */
+	{ NULL, NULL, NULL }, /* Ns */
+	{ assert_le1, NULL, NULL }, /* Nx */
+	{ assert_le1, NULL, NULL }, /* Ox */
+	{ NULL, NULL, NULL }, /* Pc */
+	{ NULL, NULL, NULL }, /* Pf */
+	{ NULL, NULL, NULL }, /* Po */
+	{ NULL, NULL, NULL }, /* Pq */
+	{ NULL, NULL, NULL }, /* Qc */
+	{ NULL, NULL, NULL }, /* Ql */
+	{ NULL, NULL, NULL }, /* Qo */
+	{ NULL, NULL, NULL }, /* Qq */
+	{ NULL, NULL, NULL }, /* Re */
+	{ NULL, NULL, NULL }, /* Rs */
+	{ NULL, NULL, NULL }, /* Sc */
+	{ NULL, NULL, NULL }, /* So */
+	{ NULL, NULL, NULL }, /* Sq */
+	{ need_eq1, args_bool, NULL }, /* Sm */
+	{ need_ge1, NULL, NULL }, /* Sx */
+	{ need_ge1, NULL, NULL }, /* Sy */
+	{ want_ge1, NULL, NULL }, /* Tn */
+	{ assert_eq0, NULL, NULL }, /* Ux */
+	{ NULL, NULL, NULL }, /* Xc */
+	{ NULL, NULL, NULL }, /* Xo */
+	{ NULL, NULL, NULL }, /* Fo */ 
+	{ NULL, NULL, NULL }, /* Fc */ 
+	{ NULL, NULL, NULL }, /* Oo */
+	{ NULL, NULL, NULL }, /* Oc */
+	{ NULL, NULL, NULL }, /* Bk */
+	{ NULL, NULL, NULL }, /* Ek */
+	{ need_eq0, NULL, NULL }, /* Bt */
+	{ need_eq1, NULL, NULL }, /* Hf */
+	{ NULL, NULL, NULL }, /* Fr */
+	{ need_eq0, NULL, NULL }, /* Ud */
+};
 
+
 static int
-mbuf_indent(struct md_valid *p)
+need_le2(struct mdoc *mdoc, int tok, int pos, int sz)
 {
-	size_t		 i;
+	if (sz <= 2)
+		return(1);
+	return(mdoc_err(mdoc, tok, pos, ERR_ARGS_LE2));
+}
 
-	assert(0 == p->pos);
 
-	for (i = 0; i < MIN(p->indent, 4); i++)
-		if ( ! md_buf_putstring(p->mbuf, "    "))
-			return(0);
-
-	p->pos = i * 4;
-	return(1);
+static int
+want_ge1(struct mdoc *mdoc, int tok, int pos, int sz)
+{
+	if (sz > 0)
+		return(1);
+	return(mdoc_warn(mdoc, tok, pos, WARN_ARGS_GE1));
 }
 
 
 static int
-mbuf_atnewline(struct md_valid *p)
+want_eq0(struct mdoc *mdoc, int tok, int pos, int sz)
 {
-
-	return(p->pos == MIN(4, p->indent));
+	if (sz == 0)
+		return(1);
+	return(mdoc_warn(mdoc, tok, pos, WARN_ARGS_EQ0));
 }
 
 
 static int
-mbuf_newline(struct md_valid *p)
+need_eq0(struct mdoc *mdoc, int tok, int pos, int sz)
 {
-
-	if (mbuf_atnewline(p))
+	if (sz == 0)
 		return(1);
-	if ( ! md_buf_putchar(p->mbuf, '\n'))
-		return(0);
-
-	p->pos = 0;
-	return(mbuf_indent(p));
+	return(mdoc_err(mdoc, tok, pos, ERR_ARGS_EQ0));
 }
 
 
 static int
-mbuf_data(struct md_valid *p, char *buf)
+assert_le1(struct mdoc *mdoc, int tok, int pos, int sz)
 {
-	size_t		 sz;
-	char		*bufp;
 
-	assert(p->mbuf);
-	assert(0 != p->indent);
-
-	if (MD_LITERAL & p->flags)
-		return(md_buf_putstring(p->mbuf, buf));
-
-	if (0 == p->pos)
-		mbuf_indent(p);
-
-	/*
-	 * Indent if we're at the beginning of a line.  Don't indent
-	 * more than 16 or so characters.
-	 */
-
-	while (*buf) {
-		while (*buf && isspace(*buf))
-			buf++;
-
-		if (0 == *buf)
-			break;
-
-		bufp = buf;
-		while (*buf && ! isspace(*buf))
-			buf++;
-
-		if (0 != *buf)
-			*buf++ = 0;
-
-		/* Process word. */
-
-		sz = strlen(bufp);
-		
-		if (sz + p->pos < 72) {
-			if ( ! md_buf_putstring(p->mbuf, bufp))
-				return(0);
-
-			/* FIXME: check punctuation. */
-
-			if ( ! md_buf_putchar(p->mbuf, ' '))
-				return(0);
-			p->pos += sz + 1;
-			continue;
-		}
-
-		if ( ! mbuf_newline(p))
-			return(0);
-
-		if ( ! md_buf_putstring(p->mbuf, bufp))
-			return(0);
-
-		/* FIXME: check punctuation. */
-
-		if ( ! md_buf_putchar(p->mbuf, ' '))
-			return(0);
-		p->pos += sz + 1;
-	}
-
+	assert(sz <= 1);
 	return(1);
 }
 
 
-int
-md_line_valid(void *arg, char *buf)
+static int
+assert_eq0(struct mdoc *mdoc, int tok, int pos, int sz)
 {
-	struct md_valid	*p;
 
-	p = (struct md_valid *)arg;
-	return(roff_engine(p->tree, buf));
+	assert(sz == 0);
+	return(1);
 }
 
 
-int
-md_exit_valid(void *data, int flush)
+static int
+need_eq1(struct mdoc *mdoc, int tok, int pos, int sz)
 {
-	int		 c;
-	struct md_valid	*p;
-
-	p = (struct md_valid *)data;
-	c = roff_free(p->tree, flush);
-	free(p);
-
-	return(c);
+	if (sz == 1)
+		return(1);
+	return(mdoc_err(mdoc, tok, pos, ERR_ARGS_EQ1));
 }
 
 
-void *
-md_init_valid(const struct md_args *args,
-		struct md_mbuf *mbuf, const struct md_rbuf *rbuf)
+static int
+need_ge1(struct mdoc *mdoc, int tok, int pos, int sz)
 {
-	struct roffcb	 cb;
-	struct md_valid	*p;
-
-	cb.roffhead = roffhead;
-	cb.rofftail = rofftail;
-	cb.roffin = roffin;
-	cb.roffout = roffout;
-	cb.roffblkin = roffblkin;
-	cb.roffblkout = roffblkout;
-	cb.roffspecial = roffspecial;
-	cb.roffmsg = roffmsg;
-	cb.roffdata = roffdata;
-
-	if (NULL == (p = calloc(1, sizeof(struct md_valid))))
-		err(1, "malloc");
-
-	p->args = args;
-	p->mbuf = mbuf;
-	p->rbuf = rbuf;
-
-	assert(mbuf);
-
-	if (NULL == (p->tree = roff_alloc(&cb, p))) {
-		free(p);
-		return(NULL);
-	}
-
-	return(p);
+	if (sz > 0)
+		return(1);
+	return(mdoc_err(mdoc, tok, pos, ERR_ARGS_GE1));
 }
 
 
-/* ARGSUSED */
 static int
-roffhead(void *arg)
+no_head_child(struct mdoc *mdoc, int tok, int pos)
 {
 
+	/* TODO */
 	return(1);
 }
 
 
 static int
-rofftail(void *arg)
+need_head_child(struct mdoc *mdoc, int tok, int pos)
 {
-	struct md_valid	*p;
+	struct mdoc_node *n;
 
-	assert(arg);
-	p = (struct md_valid *)arg;
+	assert(mdoc->last);
+	n = mdoc->last;
 
-	if (mbuf_atnewline(p))
-		return(1);
+	assert(MDOC_BLOCK == n->type);
+	assert(n->child);
 
-	return(md_buf_putchar(p->mbuf, '\n'));
+	n = n->child;
+
+	if (MDOC_HEAD != n->type) 
+		return(mdoc_err(mdoc, tok, pos, ERR_CHILD_HEAD));
+	if (n->child)
+		return(1);
+	return(mdoc_err(mdoc, tok, pos, ERR_CHILD_HEAD));
 }
 
 
 static int
-roffspecial(void *arg, int tok)
+args_Sh(struct mdoc *mdoc, int tok, int pos, 
+		int sz, const char *args[], 
+		int argc, const struct mdoc_arg *argv)
 {
+	enum mdoc_sec	 sec;
 
+	sec = mdoc_atosec((size_t)sz, args);
+	if (SEC_CUSTOM != sec && sec < mdoc->sec_lastn)
+		if ( ! mdoc_warn(mdoc, tok, pos, WARN_SEC_OO))
+			return(0);
+	if (SEC_BODY == mdoc->sec_last && SEC_NAME != sec)
+		return(mdoc_err(mdoc, tok, pos, ERR_SEC_NAME));
+
 	return(1);
 }
 
 
 static int
-roffblkin(void *arg, int tok)
+args_bool(struct mdoc *mdoc, int tok, int pos, 
+		int sz, const char *args[], 
+		int argc, const struct mdoc_arg *argv)
 {
-	struct md_valid	*p;
 
-	assert(arg);
-	p = (struct md_valid *)arg;
-
-	if ( ! mbuf_atnewline(p)) {
-		if ( ! md_buf_putchar(p->mbuf, '\n'))
-			return(0);
-		p->pos = 0;
-		if ( ! mbuf_indent(p))
-			return(0);
-	}
-
-	if ( ! md_buf_putstring(p->mbuf, toknames[tok]))
-		return(0);
-
-	if ( ! md_buf_putchar(p->mbuf, '\n'))
-		return(0);
-
-	p->pos = 0;
-	p->indent++;
-
-	return(mbuf_indent(p));
+	assert(1 == sz);
+	if (xstrcmp(args[0], "on"))
+		return(1);	
+	if (xstrcmp(args[0], "off"))
+		return(1);
+	return(mdoc_err(mdoc, tok, pos, ERR_SYNTAX_ARGBAD));
 }
 
 
 static int
-roffblkout(void *arg, int tok)
+args_ref(struct mdoc *mdoc, int tok, int pos, 
+		int sz, const char *args[], 
+		int argc, const struct mdoc_arg *argv)
 {
-	struct md_valid	*p;
+	struct mdoc_node *n;
 
-	assert(arg);
-	p = (struct md_valid *)arg;
+	assert(mdoc->last);
+	for (n = mdoc->last ; n; n = n->parent) {
+		if (MDOC_BLOCK != n->type)
+			continue;
+		if (MDOC_Rs != n->data.block.tok)
+			break;
+		return(1);
+	}
 
-	if ( ! md_buf_putchar(p->mbuf, '\n'))
-		return(0);
-
-	p->pos = 0;
-	p->indent--;
-
-	return(mbuf_indent(p));
+	return(mdoc_err(mdoc, tok, pos, ERR_SCOPE_NOCTX));
 }
 
 
 static int
-roffin(void *arg, int tok, int *argcp, char **argvp)
+args_Xr(struct mdoc *mdoc, int tok, int pos, 
+		int sz, const char *args[], 
+		int argc, const struct mdoc_arg *argv)
 {
 
+	if (1 == sz)
+		return(1);
+	if (0 == sz)
+		return(mdoc_err(mdoc, tok, pos, ERR_ARGS_GE1));
+
+	if (MSEC_DEFAULT == mdoc_atomsec(args[1]))
+		return(mdoc_err(mdoc, tok, pos, ERR_SYNTAX_ARGFORM));
 	return(1);
 }
 
 
 static int
-roffout(void *arg, int tok)
+args_At(struct mdoc *mdoc, int tok, int pos, 
+		int sz, const char *args[], 
+		int argc, const struct mdoc_arg *argv)
 {
+	int		 i;
 
+	if (0 == sz)
+		return(1);
+
+	i = 0;
+	if (ATT_DEFAULT == mdoc_atoatt(args[i]))
+		i++;
+
+	for ( ; i < sz; i++) {
+		if ( ! mdoc_isdelim(args[i]))
+			continue;
+		return(mdoc_err(mdoc, tok, pos, ERR_SYNTAX_NOPUNCT));
+	}
 	return(1);
 }
 
 
-
-static void
-roffmsg(void *arg, enum roffmsg lvl, 
-		const char *buf, const char *pos, char *msg)
+static int
+args_blocknest(struct mdoc *mdoc, int tok, int pos, 
+		int sz, const char *args[], 
+		int argc, const struct mdoc_arg *argv)
 {
-	char		*level;
-	struct md_valid	*p;
+	struct mdoc_node *node;
 
-	assert(arg);
-	p = (struct md_valid *)arg;
+	/*
+	 * We can't be nested within any other block displays (or really
+	 * any other kind of display, although Bd is the only multi-line
+	 * one that will show up). 
+	 */
+	assert(mdoc->last);
 
-	switch (lvl) {
-	case (ROFF_WARN):
-		if ( ! (MD_WARN_ALL & p->args->warnings))
-			return;
-		level = "warning";
+	/* LINTED */
+	for (node = mdoc->last; node; node = node->parent) {
+		if (node->type != MDOC_BLOCK)
+			continue;
+		if (node->data.block.tok != MDOC_Bd)
+			continue;
 		break;
-	case (ROFF_ERROR):
-		level = "error";
-		break;
-	default:
-		abort();
 	}
-	
-	if (pos)
-		(void)fprintf(stderr, "%s:%zu: %s: %s\n", 
-				p->rbuf->name, p->rbuf->line, level, msg);
-	else
-		(void)fprintf(stderr, "%s: %s: %s\n", 
-				p->rbuf->name, level, msg);
+	if (NULL == node)
+		return(1);
+	return(mdoc_err(mdoc, tok, pos, ERR_SCOPE_NONEST));
+}
 
+
+int
+mdoc_valid_pre(struct mdoc *mdoc, int tok, int pos, 
+		int sz, const char *args[], 
+		int argc, const struct mdoc_arg *argv)
+{
+
+	assert(tok < MDOC_MAX);
+	if (mdoc_valids[tok].sz) 
+		if ( ! (*mdoc_valids[tok].sz)(mdoc, tok, pos, sz))
+			return(0);
+
+	if (NULL == mdoc_valids[tok].args) 
+		return(1);
+	return((*mdoc_valids[tok].args)(mdoc, 
+			tok, pos, sz, args, argc, argv));
 }
 
 
-static int
-roffdata(void *arg, char *buf)
+int
+mdoc_valid_post(struct mdoc *mdoc, int tok, int pos)
 {
-	struct md_valid	*p;
 
-	assert(arg);
-	p = (struct md_valid *)arg;
-	return(mbuf_data(p, buf));
+	if (NULL == mdoc_valids[tok].post)
+		return(1);
+	return((*mdoc_valids[tok].post)(mdoc, tok, pos));
 }