Considerable improvements across the board.

6 files changed, 243 insertions, 208 deletions

diff --git a/Makefile b/Makefile
index 3d2c3e48..93d27a3d 100644
--- a/Makefile
+++ b/Makefile
@@ -1,4 +1,4 @@
-CFLAGS += -W -Wall -g
+CFLAGS += -W -Wall -Wno-unused-parameter -g
 LINTFLAGS += -c -e -f -u
 
 LNS	= mdocml.ln html4_strict.ln dummy.ln libmdocml.ln
diff --git a/html4_strict.c b/html4_strict.c
index 98934973..f46b3373 100644
--- a/html4_strict.c
+++ b/html4_strict.c
@@ -27,72 +27,78 @@
 #include "libmdocml.h"
 #include "private.h"
 
-enum	roffd {
-	ROFF_ENTER = 0,
-	ROFF_EXIT
+enum	roffd { 
+	ROFF_ENTER = 0, 
+	ROFF_EXIT 
 };
 
-enum	rofftype {
-	ROFF_NONE = 0,
-	ROFF_LAYOUT
+enum	rofftype { 
+	ROFF_TITLE, 
+	ROFF_COMMENT, 
+	ROFF_TEXT, 
+	ROFF_LAYOUT 
 };
 
-struct rofftree;
-
 #define	ROFFCALL_ARGS	  const struct md_args *arg,		\
 			  struct md_mbuf *out,			\
 			  const struct md_rbuf *in,		\
 			  const char *buf, size_t sz,		\
 			  size_t pos, enum roffd type,		\
 			  struct rofftree *tree
-typedef	int		(*roffcall)(ROFFCALL_ARGS);
 
-static	int		  roff_Dd(ROFFCALL_ARGS);
-static	int		  roff_Dt(ROFFCALL_ARGS);
-static	int		  roff_Os(ROFFCALL_ARGS);
-static	int		  roff_Sh(ROFFCALL_ARGS);
+struct	rofftree;
 
 struct	rofftok {
-	char		  id;
-#define	ROFF___	 	  0
-#define	ROFF_Dd		  1
-#define	ROFF_Dt		  2
-#define	ROFF_Os		  3
-#define	ROFF_Sh		  4
-#define	ROFF_Max	  5
+	int		  id;
 	char		  name[2];
-	roffcall	  cb;
+	int		(*cb)(ROFFCALL_ARGS);
 	enum rofftype	  type;
 	int		  flags;
-#define	ROFF_NESTED	 (1 << 0)
-};
-
-static const struct rofftok tokens[ROFF_Max] = {
-	{ ROFF___,    "\\\"",	   NULL, ROFF_NONE,	0 },
-	{ ROFF_Dd,	"Dd",	roff_Dd, ROFF_NONE,	0 },
-	{ ROFF_Dt,	"Dt",	roff_Dt, ROFF_NONE,	0 },
-	{ ROFF_Os,	"Os",	roff_Os, ROFF_LAYOUT,	0 },
-	{ ROFF_Sh,	"Sh",	roff_Sh, ROFF_LAYOUT,	0 },
+#define	ROFF_NESTED	 (1 << 0) /* FIXME: test. */
+#define	ROFF_PARSED	 (1 << 1) /* FIXME: test. */
+#define	ROFF_CALLABLE	 (1 << 2) /* FIXME: test. */
 };
 
 struct	roffnode {
 	int		  tok;	
 	struct roffnode	 *parent;
-	/* TODO: line number at acquisition. */
+	size_t		  line;
 };
 
-struct rofftree {
+struct	rofftree {
 	struct roffnode	 *last;
 	time_t		  date;
 	char		  title[256];
 	char		  section[256];
 	char		  volume[256];
 	int		  state;
-#define	ROFF_PRELUDE_Os	 (1 << 1)
-#define	ROFF_PRELUDE_Dt	 (1 << 2)
-#define	ROFF_PRELUDE_Dd	 (1 << 3)
+#define	ROFF_PRELUDE	 (1 << 1)
+#define	ROFF_PRELUDE_Os	 (1 << 2)
+#define	ROFF_PRELUDE_Dt	 (1 << 3)
+#define	ROFF_PRELUDE_Dd	 (1 << 4)
+#define	ROFF_BODY	 (1 << 5)
 };
 
+#define	ROFF___	 	  0
+#define	ROFF_Dd		  1
+#define	ROFF_Dt		  2
+#define	ROFF_Os		  3
+#define	ROFF_Sh		  4
+#define	ROFF_An		  5
+#define	ROFF_Li		  6
+#define	ROFF_Max	  7
+
+static	int		  roff_Dd(ROFFCALL_ARGS);
+static	int		  roff_Dt(ROFFCALL_ARGS);
+static	int		  roff_Os(ROFFCALL_ARGS);
+static	int		  roff_Sh(ROFFCALL_ARGS);
+static	int		  roff_An(ROFFCALL_ARGS);
+static	int		  roff_Li(ROFFCALL_ARGS);
+
+static	struct roffnode	 *roffnode_new(int, size_t, 
+				struct rofftree *);
+static	void		  roffnode_free(int, struct rofftree *);
+
 static	int		  rofffind(const char *);
 static	int 		  roffparse(const struct md_args *, 
 				struct md_mbuf *,
@@ -108,25 +114,44 @@ static	void		  dbg_enter(const struct md_args *, int);
 static	void		  dbg_leave(const struct md_args *, int);
 
 
+static const struct rofftok tokens[ROFF_Max] = 
+{
+{ ROFF___, "\\\"",     NULL, ROFF_COMMENT,			  0 },
+{ ROFF_Dd,   "Dd",  roff_Dd, ROFF_TITLE,			  0 },
+{ ROFF_Dt,   "Dt",  roff_Dt, ROFF_TITLE,			  0 },
+{ ROFF_Os,   "Os",  roff_Os, ROFF_TITLE,			  0 },
+{ ROFF_Sh,   "Sh",  roff_Sh, ROFF_LAYOUT,			  0 },
+{ ROFF_An,   "An",  roff_An, ROFF_TEXT,	ROFF_PARSED 		    },
+{ ROFF_Li,   "Li",  roff_Li, ROFF_TEXT,	ROFF_PARSED | ROFF_CALLABLE },
+};
+
+
 int
 md_exit_html4_strict(const struct md_args *args, struct md_mbuf *out, 
-		const struct md_rbuf *in, void *data)
+		const struct md_rbuf *in, int error, void *data)
 {
 	struct rofftree	*tree;
-	int		 error;
 
 	assert(args);
 	assert(data);
 	tree = (struct rofftree *)data;
-	error = 0;
 
+	if (-1 == error)
+		out = NULL;
+
+	/* LINTED */
 	while (tree->last)
-		if ( ! (*tokens[tree->last->tok].cb)
-				(args, error ? NULL : out, in, NULL,
-				 0, 0, ROFF_EXIT, tree))
-			error = 1;
+		if ( ! (*tokens[tree->last->tok].cb)(args, out, in, 
+					NULL, 0, 0, ROFF_EXIT, tree))
+			out = NULL;
+
+	if (out && (ROFF_PRELUDE & tree->state)) {
+		warnx("%s: prelude never finished", in->name);
+		error = 1;
+	}
 
 	free(tree);
+
 	return(error ? 0 : 1);
 }
 
@@ -149,6 +174,8 @@ md_init_html4_strict(const struct md_args *args, struct md_mbuf *out,
 		return(0);
 	}
 
+	tree->state = ROFF_PRELUDE;
+
 	*data = tree;
 	return(1);
 }
@@ -229,18 +256,53 @@ roffparse(const struct md_args *args, struct md_mbuf *out,
 	 */
 
 	if (3 > sz) {
-		warnx("%s: malformed input (line %zu, col 1)", 
+		warnx("%s: malformed line (line %zu)", 
 				in->name, in->line);
 		return(0);
 	} else if (ROFF_Max == (tokid = rofffind(buf + 1))) {
-		warnx("%s: unknown token `%c%c' (line %zu, col 1)",
+		warnx("%s: unknown line token `%c%c' (line %zu)",
 				in->name, *(buf + 1), 
 				*(buf + 2), in->line);
 		return(0);
-	} else if (NULL == tokens[tokid].cb) 
-		return(1); /* Skip token. */
+	} 
 
-	pos = 3;
+	/* Domain cross-contamination (and sanity) checks. */
+
+	switch (tokens[tokid].type) {
+	case (ROFF_TITLE):
+		if (ROFF_PRELUDE & tree->state) {
+			assert( ! (ROFF_BODY & tree->state));
+			break;
+		}
+		assert(ROFF_BODY & tree->state);
+		warnx("%s: prelude token `%s' in body (line %zu)",
+				in->name, tokens[tokid].name, in->line);
+		return(0);
+	case (ROFF_LAYOUT):
+		/* FALLTHROUGH */
+	case (ROFF_TEXT):
+		if (ROFF_BODY & tree->state) {
+			assert( ! (ROFF_PRELUDE & tree->state));
+			break;
+		}
+		assert(ROFF_PRELUDE & tree->state);
+		warnx("%s: text token `%s' in prelude (line %zu)",
+				in->name, tokens[tokid].name, in->line);
+		return(0);
+	default:
+		return(1);
+	}
+
+	/*
+	 * Text-domain checks.
+	 */
+
+	if (ROFF_TEXT == tokens[tokid].type &&
+			! (ROFF_PARSED & tokens[tokid].flags)) {
+		warnx("%s: text token `%s' not callable (line %zu)",
+				in->name, tokens[tokid].name, in->line);
+		return(0);
+	}
 
 	/*
 	 * If this is a non-nestable layout token and we're below a
@@ -253,6 +315,7 @@ roffparse(const struct md_args *args, struct md_mbuf *out,
 	 */
 
 	node = NULL;
+	pos = 3;
 
 	if (ROFF_LAYOUT == tokens[tokid].type && 
 			! (ROFF_NESTED & tokens[tokid].flags)) {
@@ -264,14 +327,16 @@ roffparse(const struct md_args *args, struct md_mbuf *out,
 
 			if ( ! (ROFF_NESTED & tokens[node->tok].flags))
 				continue;
-			warnx("%s: scope of %s broken by %s "
-					"(line %zu, col %zu)",
-					in->name, tokens[tokid].name,
+			warnx("%s: scope of %s (line %zu) broken by "
+					"%s (line %zu)", in->name, 
+					tokens[tokid].name,
+					node->line, 
 					tokens[node->tok].name,
-					in->line, pos);
+					in->line);
 			return(0);
 		}
 	}
+
 	if (node) {
 		assert(ROFF_LAYOUT == tokens[tokid].type);
 		assert( ! (ROFF_NESTED & tokens[tokid].flags));
@@ -279,6 +344,7 @@ roffparse(const struct md_args *args, struct md_mbuf *out,
 
 		/* Clear up to last scoped token. */
 
+		/* LINTED */
 		do {
 			t = tree->last->tok;
 			if ( ! (*tokens[tree->last->tok].cb)
@@ -302,84 +368,134 @@ rofffind(const char *name)
 
 	assert(name);
 	/* FIXME: use a table, this is slow but ok for now. */
+
+	/* LINTED */
 	for (i = 0; i < ROFF_Max; i++)
+		/* LINTED */
 		if (0 == strncmp(name, tokens[i].name, 2))
-			return(i);
+			return((int)i);
 	
 	return(ROFF_Max);
 }
 
 
-/* ARGUSED */
-static	int
-roff_Dd(ROFFCALL_ARGS)
+static struct roffnode *
+roffnode_new(int tokid, size_t line, struct rofftree *tree)
 {
+	struct roffnode	*p;
+	
+	if (NULL == (p = malloc(sizeof(struct roffnode)))) {
+		warn("malloc");
+		return(NULL);
+	}
 
-	assert(in);
-	assert(tree);
-	assert(arg);
-	assert(out);
-	assert(buf);
-	assert(sz > 0);
-	assert(pos > 0);
-	assert(type == ROFF_ENTER);
+	p->line = line;
+	p->tok = tokid;
+	p->parent = tree->last;
+	tree->last = p;
+	return(p);
+}
 
-	if (tree->last) {
-		warnx("%s: superfluous prelude (line %zu, col %zu)",
-				in->name, in->line, pos);
-		return(0);
-	}
 
-	if (0 != tree->state) {
-		warnx("%s: bad manual prelude (line %zu, col %zu)",
-				in->name, in->line, pos);
-		return(1);
-	}
+static void
+roffnode_free(int tokid, struct rofftree *tree)
+{
+	struct roffnode	*p;
 
-	/* TODO: parse date from buffer. */
+	assert(tree->last);
+	assert(tree->last->tok == tokid);
 
-	tree->date = time(NULL);
-	tree->state |= ROFF_PRELUDE_Dd;
+	p = tree->last;
+	tree->last = tree->last->parent;
+	free(p);
+}
 
-	(void)printf("Dd\n");
 
-	return(1);
+static int dbg_lvl = 0; /* FIXME: de-globalise. */
+
+
+static void
+dbg_enter(const struct md_args *args, int tokid)
+{
+	int		 i;
+
+	assert(args);
+	if ( ! (args->dbg & MD_DBG_TREE))
+		return;
+
+	assert(tokid >= 0 && tokid <= ROFF_Max);
+
+	/* LINTED */
+	for (i = 0; i < dbg_lvl; i++)
+		(void)printf("    ");
+
+	(void)printf("%s\n", tokens[tokid].name);
+
+	if (ROFF_LAYOUT == tokens[tokid].type)
+		dbg_lvl++;
 }
 
 
-static	int
-roff_Dt(ROFFCALL_ARGS)
+static void
+dbg_leave(const struct md_args *args, int tokid)
 {
+	int		 i;
 
-	assert(in);
-	assert(tree);
-	assert(arg);
-	assert(out);
-	assert(buf);
-	assert(sz > 0);
-	assert(pos > 0);
-	assert(type == ROFF_ENTER);
+	assert(args);
+	if ( ! (args->dbg & MD_DBG_TREE))
+		return;
+	if (ROFF_LAYOUT != tokens[tokid].type)
+		return;
 
-	if (tree->last) {
-		warnx("%s: superfluous prelude (line %zu, col %zu)",
-				in->name, in->line, pos);
+	assert(tokid >= 0 && tokid <= ROFF_Max);
+	assert(dbg_lvl > 0);
+
+	dbg_lvl--;
+
+	/* LINTED */
+	for (i = 0; i < dbg_lvl; i++) 
+		(void)printf("    ");
+
+	(void)printf("%s\n", tokens[tokid].name);
+}
+
+
+static	int
+roff_Dd(ROFFCALL_ARGS)
+{
+
+	assert(ROFF_PRELUDE & tree->state);
+	if (ROFF_PRELUDE_Dt & tree->state ||
+			ROFF_PRELUDE_Dd & tree->state) {
+		warnx("%s: bad prelude ordering (line %zu)",
+				in->name, in->line);
 		return(0);
 	}
 
+	assert(NULL == tree->last);
+	tree->state |= ROFF_PRELUDE_Dd;
+
+	dbg_enter(arg, ROFF_Dd);
+	return(1);
+}
+
+
+static	int
+roff_Dt(ROFFCALL_ARGS)
+{
+
+	assert(ROFF_PRELUDE & tree->state);
 	if ( ! (ROFF_PRELUDE_Dd & tree->state) ||
-			(ROFF_PRELUDE_Os & tree->state) ||
 			(ROFF_PRELUDE_Dt & tree->state)) {
-		warnx("%s: bad manual prelude (line %zu, col %zu)",
-				in->name, in->line, pos);
-		return(1);
+		warnx("%s: bad prelude ordering (line %zu)",
+				in->name, in->line);
+		return(0);
 	}
 
-	/* TODO: parse titles from buffer. */
-
+	assert(NULL == tree->last);
 	tree->state |= ROFF_PRELUDE_Dt;
 
-	(void)printf("Dt\n");
-
+	dbg_enter(arg, ROFF_Dt);
 	return(1);
 }
 
@@ -387,146 +503,63 @@ roff_Dt(ROFFCALL_ARGS)
 static	int
 roff_Os(ROFFCALL_ARGS)
 {
-	struct roffnode	*node;
-
-	assert(arg);
-	assert(tree);
-	assert(in);
 
 	if (ROFF_EXIT == type) {
-		assert(tree->last);
-		assert(tree->last->tok == ROFF_Os);
-
-		/* TODO: flush out ML footer. */
-
-		node = tree->last;
-		tree->last = node->parent;
-		free(node);
-
+		roffnode_free(ROFF_Os, tree);
 		dbg_leave(arg, ROFF_Os);
-
 		return(1);
 	} 
 
-	assert(out);
-	assert(buf);
-	assert(sz > 0);
-	assert(pos > 0);
-
-	if (tree->last) {
-		warnx("%s: superfluous prelude (line %zu, col %zu)",
-				in->name, in->line, pos);
-		return(0);
-	}
-
-	if ((ROFF_PRELUDE_Os & tree->state) ||
-			! (ROFF_PRELUDE_Dt & tree->state) ||
+	assert(ROFF_PRELUDE & tree->state);
+	if ( ! (ROFF_PRELUDE_Dt & tree->state) ||
 			! (ROFF_PRELUDE_Dd & tree->state)) {
-		warnx("%s: bad manual prelude (line %zu, col %zu)",
-				in->name, in->line, pos);
-		return(1);
+		warnx("%s: bad prelude ordering (line %zu)",
+				in->name, in->line);
+		return(0);
 	}
 
-	node = malloc(sizeof(struct roffnode));
-	if (NULL == node) {
-		warn("malloc");
+	assert(NULL == tree->last);
+	if (NULL == roffnode_new(ROFF_Os, in->line, tree))
 		return(0);
-	}
-	node->tok = ROFF_Os;
-	node->parent = NULL;
 
 	tree->state |= ROFF_PRELUDE_Os;
-	tree->last = node;
+	tree->state &= ~ROFF_PRELUDE;
+	tree->state |= ROFF_BODY;
 
 	dbg_enter(arg, ROFF_Os);
-
 	return(1);
 }
 
 
-static	int
+static int
 roff_Sh(ROFFCALL_ARGS)
 {
-	struct roffnode	*node;
-
-	assert(arg);
-	assert(tree);
-	assert(tree->last);
-	assert(in);
 
 	if (ROFF_EXIT == type) {
-		assert(tree->last->tok == ROFF_Sh);
-
-		node = tree->last;
-		tree->last = node->parent;
-		free(node);
-
+		roffnode_free(ROFF_Sh, tree);
 		dbg_leave(arg, ROFF_Sh);
-
 		return(1);
 	} 
 
-	assert(out);
-	assert(buf);
-	assert(sz > 0);
-	assert(pos > 0);
-
-	node = malloc(sizeof(struct roffnode));
-	if (NULL == node) {
-		warn("malloc");
+	if (NULL == roffnode_new(ROFF_Sh, in->line, tree))
 		return(0);
-	}
-	node->tok = ROFF_Sh;
-	node->parent = tree->last;
-
-	tree->last = node;
 
 	dbg_enter(arg, ROFF_Sh);
-
 	return(1);
 }
 
 
-static int dbg_lvl = 0; /* FIXME: de-globalise. */
-
-
-static void
-dbg_enter(const struct md_args *args, int tokid)
+static int
+roff_Li(ROFFCALL_ARGS) 
 {
-	int		 i;
-
-	assert(args);
-	if ( ! (args->dbg & MD_DBG_TREE))
-		return;
 
-	assert(tokid >= 0 && tokid <= ROFF_Max);
-
-	for (i = 0; i < dbg_lvl; i++)
-		(void)printf(" ");
-
-	(void)printf("%s\n", tokens[tokid].name);
-
-	if (ROFF_LAYOUT == tokens[tokid].type)
-		dbg_lvl++;
+	return(1);
 }
 
 
-static void
-dbg_leave(const struct md_args *args, int tokid)
+static int
+roff_An(ROFFCALL_ARGS) 
 {
-	int		 i;
 
-	assert(args);
-	if ( ! (args->dbg & MD_DBG_TREE))
-		return;
-
-	assert(tokid >= 0 && tokid <= ROFF_Max);
-	assert(dbg_lvl > 0);
-
-	dbg_lvl--;
-	for (i = 0; i < dbg_lvl; i++) 
-		(void)printf(" ");
-
-	(void)printf("%s\n", tokens[tokid].name);
+	return(1);
 }
-
diff --git a/libmdocml.c b/libmdocml.c
index 7317b6d9..a38678d0 100644
--- a/libmdocml.c
+++ b/libmdocml.c
@@ -135,7 +135,7 @@ md_run_leave(const struct md_args *args, struct md_mbuf *mbuf,
 	/* Run exiters. */
 	switch (args->type) {
 	case (MD_HTML4_STRICT):
-		if ( ! md_exit_html4_strict(args, mbuf, rbuf, data))
+		if ( ! md_exit_html4_strict(args, mbuf, rbuf, c, data))
 			return(-1);
 		break;
 	case (MD_DUMMY):
@@ -235,8 +235,7 @@ md_run(const struct md_args *args,
 	/* Run initialisers. */
 	switch (args->type) {
 	case (MD_HTML4_STRICT):
-		if ( ! md_init_html4_strict
-				(args, &mbuf, &rbuf, &data))
+		if ( ! md_init_html4_strict(args, &mbuf, &rbuf, &data))
 			return(-1);
 		break;
 	case (MD_DUMMY):
diff --git a/libmdocml.h b/libmdocml.h
index 6b96094d..c25602fa 100644
--- a/libmdocml.h
+++ b/libmdocml.h
@@ -43,7 +43,7 @@ struct	md_args {
 	union md_params	 params;/* Parameters for parser. */
 	enum md_type	 type;	/* Type of parser. */
 	int		 dbg;	/* Debug level. */
-#define	MD_DBG_TREE	(1 << 0)
+#define	MD_DBG_TREE	(1 << 0)/* Print the parse tree to stdout. */
 };
 
 struct	md_buf {
diff --git a/mdocml.c b/mdocml.c
index a9fe064a..6bd6d8cf 100644
--- a/mdocml.c
+++ b/mdocml.c
@@ -167,6 +167,9 @@ begin_bufs(const struct md_args *args,
 	if (-1 == fstat(in->fd, &stin)) {
 		warn("%s", in->name);
 		return(1);
+	} else if (0 == stin.st_size) {
+		warnx("%s: empty file", in->name);
+		return(1);
 	} else if (-1 == fstat(out->fd, &stout)) {
 		warn("%s", out->name);
 		return(1);
diff --git a/private.h b/private.h
index 934061c6..8231037f 100644
--- a/private.h
+++ b/private.h
@@ -40,7 +40,7 @@ __BEGIN_DECLS
 typedef	int	(*md_init)(const struct md_args *, struct md_mbuf *,
 			const struct md_rbuf *, void **);
 typedef	int	(*md_exit)(const struct md_args *, struct md_mbuf *, 
-			const struct md_rbuf *, void *);
+			const struct md_rbuf *, int, void *);
 typedef	int	(*md_line)(const struct md_args *, 
 			struct md_mbuf *, const struct md_rbuf *, 
 			const char *, size_t, void *);
@@ -53,7 +53,7 @@ int		  md_init_html4_strict(const struct md_args *,
 			void **);
 int		  md_exit_html4_strict(const struct md_args *,
 			struct md_mbuf *, const struct md_rbuf *, 
-			void *);
+			int, void *);
 
 int		  md_line_dummy(const struct md_args *,
 			struct md_mbuf *, const struct md_rbuf *,