Removed tree-writing capability (for now, only validates and exits).

4 files changed, 157 insertions, 129 deletions

diff --git a/Makefile b/Makefile
index c559c725..e46e1d7d 100644
--- a/Makefile
+++ b/Makefile
@@ -3,7 +3,7 @@ VERSION	= 1.2.0
 CFLAGS += -W -Wall -Wno-unused-parameter -g 
 
 LIBLNS	= macro.ln mdoc.ln hash.ln strings.ln xstd.ln argv.ln \
-	  validate.ln action.ln
+	  validate.ln action.ln 
 
 BINLNS	= tree.ln mdocml.ln
 
@@ -14,14 +14,14 @@ LLNS	= llib-llibmdoc.ln llib-lmdocml.ln
 LIBS	= libmdoc.a
 
 LIBOBJS	= macro.o mdoc.o hash.o strings.o xstd.o argv.o \
-	  validate.o action.o
+	  validate.o action.o 
 
 BINOBJS	= tree.o mdocml.o
 
 OBJS	= $(LIBOBJS) $(BINOBJS)
 
 SRCS	= macro.c mdoc.c mdocml.c hash.c strings.c xstd.c argv.c \
-	  validate.c action.c tree.c
+	  validate.c action.c tree.c 
 
 HEADS	= mdoc.h private.h
 
diff --git a/mdoc.c b/mdoc.c
index 473f42b5..33372194 100644
--- a/mdoc.c
+++ b/mdoc.c
@@ -212,12 +212,14 @@ 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(void);
 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 *);
 
 
+
 const struct mdoc_node *
 mdoc_node(struct mdoc *mdoc)
 {
@@ -510,6 +512,14 @@ mdoc_node_append(struct mdoc *mdoc, struct mdoc_node *p)
 }
 
 
+static struct mdoc_node *
+mdoc_node_alloc(void)
+{
+
+	return(xcalloc(1, sizeof(struct mdoc_node)));
+}
+
+
 int
 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 {
@@ -518,7 +528,7 @@ mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 	assert(mdoc->first);
 	assert(mdoc->last);
 
-	p = xcalloc(1, sizeof(struct mdoc_node));
+	p = mdoc_node_alloc();
 
 	p->line = line;
 	p->pos = pos;
@@ -537,7 +547,7 @@ mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 	assert(mdoc->first);
 	assert(mdoc->last);
 
-	p = xcalloc(1, sizeof(struct mdoc_node));
+	p = mdoc_node_alloc();
 
 	p->line = line;
 	p->pos = pos;
@@ -556,7 +566,7 @@ mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 	assert(mdoc->first);
 	assert(mdoc->last);
 
-	p = xcalloc(1, sizeof(struct mdoc_node));
+	p = mdoc_node_alloc();
 
 	p->line = line;
 	p->pos = pos;
@@ -572,7 +582,7 @@ mdoc_root_alloc(struct mdoc *mdoc)
 {
 	struct mdoc_node *p;
 
-	p = xcalloc(1, sizeof(struct mdoc_node));
+	p = mdoc_node_alloc();
 
 	p->type = MDOC_ROOT;
 
@@ -586,7 +596,7 @@ mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 {
 	struct mdoc_node *p;
 
-	p = xcalloc(1, sizeof(struct mdoc_node));
+	p = mdoc_node_alloc();
 
 	p->pos = pos;
 	p->line = line;
@@ -605,7 +615,7 @@ mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 {
 	struct mdoc_node *p;
 
-	p = xcalloc(1, sizeof(struct mdoc_node));
+	p = mdoc_node_alloc();
 
 	p->line = line;
 	p->pos = pos;
@@ -624,7 +634,8 @@ mdoc_word_alloc(struct mdoc *mdoc,
 {
 	struct mdoc_node *p;
 
-	p = xcalloc(1, sizeof(struct mdoc_node));
+	p = mdoc_node_alloc();
+
 	p->line = line;
 	p->pos = pos;
 	p->type = MDOC_TEXT;
diff --git a/mdoc.h b/mdoc.h
index a232128b..7d6da91b 100644
--- a/mdoc.h
+++ b/mdoc.h
@@ -195,9 +195,10 @@
 #define	MDOC_Symbolic	 61
 #define	MDOC_ARG_MAX	 62
 
+/* Warnings are either syntax or groff-compatibility. */
 enum	mdoc_warn {
-	WARN_SYNTAX,		/* Syntax warn (at line/col). */
-	WARN_COMPAT		/* Groff compat warn (at line/col). */
+	WARN_SYNTAX,
+	WARN_COMPAT
 };
 
 /* Possible values for the `At' macro. */
@@ -226,19 +227,6 @@ struct	mdoc_arg {
 	char		**value;
 };
 
-/*
- * Simplified grammar of syntax tree:
- *
- * MDOC_ROOT: root of tree
- * MDOC_TEXT: free-form text
- * MDOC_ELEM: elem [args] MDOC_TEXT...
- * MDOC_BLOCK, MDOC_HEAD, MDOC_BODY, MDOC_TAIL:
- *   MDOC_BLOCK:
- *     MDOC_HEAD [args] (MDOC_TEXT|MDOC_ELEM|MDOC_BLOCK)...
- *     MDOC_BODY (MDOC_TEXT|MDOC_ELEM|MDOC_BLOCK)...
- *     MDOC_TAIL (optional) (MDOC_TEXT|MDOC_ELEM|MDOC_BLOCK)...
- */
-
 /* Type of a syntax node. */
 enum	mdoc_type {
 	MDOC_TEXT,
@@ -340,7 +328,7 @@ enum	mdoc_arch {
 	ARCH_zaurus
 };
 
-/* Meta-information from prologue. */
+/* Information from prologue. */
 struct	mdoc_meta {
 	enum mdoc_msec	  msec;
 	enum mdoc_vol	  vol;
@@ -351,10 +339,12 @@ struct	mdoc_meta {
 	char		 *name;
 };
 
+/* Text-only node. */
 struct	mdoc_text {
 	char		 *string;
 };
 
+/* Block (scoped) node. */
 struct	mdoc_block {
 	size_t		  argc;
 	struct mdoc_arg	 *argv;
@@ -363,6 +353,7 @@ struct	mdoc_block {
 	struct mdoc_node *tail;
 };
 
+/* In-line element node. */
 struct	mdoc_elem {
 	size_t		  sz;
 	char		**args;
@@ -370,13 +361,14 @@ struct	mdoc_elem {
 	struct mdoc_arg	 *argv;
 };
 
+/* Typed nodes of an AST node. */
 union	mdoc_data {
 	struct mdoc_text  text;
 	struct mdoc_elem  elem;
 	struct mdoc_block block;
 };
 
-/* Syntax node in parse tree. */
+/* Node in AST. */
 struct	mdoc_node {
 	struct mdoc_node *parent;
 	struct mdoc_node *child;
@@ -401,7 +393,10 @@ struct	mdoc_cb {
 			enum mdoc_warn, const char *);
 };
 
+/* Global table of macro names (`Bd', `Ed', etc.). */
 extern	const char *const *mdoc_macronames;
+
+/* Global table of argument names (`column', `tag', etc.). */
 extern	const char *const *mdoc_argnames;
 
 __BEGIN_DECLS
@@ -414,13 +409,13 @@ void	 	  mdoc_free(struct mdoc *);
 /* Allocate a new parser instance. */
 struct	mdoc	 *mdoc_alloc(void *data, const struct mdoc_cb *);
 
-/* Parse a single line (boolean retval). */
+/* Parse a single line in a stream (boolean retval). */
 int	 	  mdoc_parseln(struct mdoc *, int, char *buf);
 
-/* Get result first node. */
+/* Get result first node (after mdoc_endparse!). */
 const struct mdoc_node *mdoc_node(struct mdoc *);
 
-/* Get result meta-information. */
+/* Get result meta-information (after mdoc_endparse!). */
 const struct mdoc_meta *mdoc_meta(struct mdoc *);
 
 /* Signal end of parse sequence (boolean retval). */
diff --git a/mdocml.c b/mdocml.c
index 87884080..f76ee0b5 100644
--- a/mdocml.c
+++ b/mdocml.c
@@ -30,15 +30,8 @@
 
 #include "mdoc.h"
 
-#define	xfprintf	(void)fprintf
-
 #define	MD_LINE_SZ	(256)		/* Max input line size. */
 
-/* TODO: have a struct for each transformer. */
-
-typedef	int		(*mdocprint)(const struct mdoc_node *);
-
-
 struct	md_parse {
 	int		  warn;		/* Warning flags. */
 #define	MD_WARN_SYNTAX	 (1 << 0)	/* Show syntax warnings. */
@@ -49,17 +42,17 @@ struct	md_parse {
 	struct mdoc	 *mdoc;		/* Active parser. */
 	char		 *buf;		/* Input buffer. */
 	u_long		  bufsz;	/* Input buffer size. */
-	char		 *name;		/* Input file name. */
-	int		  fd;		/* Input file desc. */
-	mdocprint	  print;	/* Node-print function. */
+	char		 *in;		/* Input file name. */
+	int		  fdin;		/* Input file desc. */
 };
 
 extern	char	 	 *__progname;
 
-extern	int		  treeprint(const struct mdoc_node *);
-
 static	void		  usage(void);
 
+static	int		  parse_opts(struct md_parse *, int, char *[]);
+static	int		  parse_subopts(struct md_parse *, char *);
+
 static	int		  parse_begin(struct md_parse *);
 static	int		  parse_leave(struct md_parse *, int);
 static	int		  io_begin(struct md_parse *);
@@ -79,69 +72,12 @@ extern	int		  getsubopt(char **, char *const *, char **);
 int
 main(int argc, char *argv[])
 {
-	int		 c;
 	struct md_parse	 parser;
-	char		*opts, *v, *filter;
-#define ALL     	 0
-#define COMPAT     	 1
-#define SYNTAX     	 2
-#define ERROR     	 3
-	char		*toks[] = { "all", "compat", "syntax", 
-				    "error", NULL };
-
-	extern char	*optarg;
-	extern int	 optind;
-
-	filter = NULL;
 
 	(void)memset(&parser, 0, sizeof(struct md_parse));
 
-	while (-1 != (c = getopt(argc, argv, "f:vW:")))
-		switch (c) {
-		case ('f'):
-			filter = optarg;
-			break;
-		case ('v'):
-			parser.dbg++;
-			break;
-		case ('W'):
-			opts = optarg;
-			while (*opts) 
-				switch (getsubopt(&opts, toks, &v)) {
-				case (ALL):
-					parser.warn |= MD_WARN_ALL;
-					break;
-				case (COMPAT):
-					parser.warn |= MD_WARN_COMPAT;
-					break;
-				case (SYNTAX):
-					parser.warn |= MD_WARN_SYNTAX;
-					break;
-				case (ERROR):
-					parser.warn |= MD_WARN_ERR;
-					break;
-				default:
-					usage();
-					return(1);
-				}
-			break;
-		default:
-			usage();
-			return(1);
-		}
-
-	argv += optind;
-	argc -= optind;
-
-	parser.name = "-";
-	if (1 == argc)
-		parser.name = *argv++;
-
-	if (filter) {
-		if (0 == strcmp(filter, "tree"))
-			parser.print = treeprint;
-	}
-
+	if ( ! parse_opts(&parser, argc, argv))
+		return(EXIT_FAILURE);
 	if ( ! io_begin(&parser))
 		return(EXIT_FAILURE);
 
@@ -153,11 +89,11 @@ static int
 io_leave(struct md_parse *p, int code)
 {
 
-	if (-1 == p->fd || STDIN_FILENO == p->fd)
+	if (-1 == p->fdin || STDIN_FILENO == p->fdin)
 		return(code);
 
-	if (-1 == close(p->fd)) {
-		warn("%s", p->name);
+	if (-1 == close(p->fdin)) {
+		warn("%s", p->in);
 		code = 0;
 	}
 	return(code);
@@ -165,13 +101,82 @@ io_leave(struct md_parse *p, int code)
 
 
 static int
+parse_subopts(struct md_parse *p, char *arg)
+{
+	char		*v;
+	char		*toks[] = { "all", "compat", 
+				"syntax", "error", NULL };
+
+	/* 
+	 * Future -Wxxx levels and so on should be here.  For now we
+	 * only recognise syntax and compat warnings as categories,
+	 * beyond the usually "all" and "error" (make warn error out).
+	 */
+
+	while (*arg) 
+		switch (getsubopt(&arg, toks, &v)) {
+		case (0):
+			p->warn |= MD_WARN_ALL;
+			break;
+		case (1):
+			p->warn |= MD_WARN_COMPAT;
+			break;
+		case (2):
+			p->warn |= MD_WARN_SYNTAX;
+			break;
+		case (3):
+			p->warn |= MD_WARN_ERR;
+			break;
+		default:
+			usage();
+			return(0);
+		}
+
+	return(1);
+}
+
+
+static int
+parse_opts(struct md_parse *p, int argc, char *argv[])
+{
+	int		 c;
+
+	extern char	*optarg;
+	extern int	 optind;
+
+	p->in = "-";
+
+	while (-1 != (c = getopt(argc, argv, "vW:")))
+		switch (c) {
+		case ('v'):
+			p->dbg++;
+			break;
+		case ('W'):
+			if ( ! parse_subopts(p, optarg))
+				return(0);
+			break;
+		default:
+			usage();
+			return(0);
+		}
+
+	argv += optind;
+	if (0 == (argc -= optind))
+		return(1);
+
+	p->in = *argv++;
+	return(1);
+}
+
+
+static int
 io_begin(struct md_parse *p)
 {
 
-	p->fd = STDIN_FILENO;
-	if (0 != strncmp(p->name, "-", 1))
-		if (-1 == (p->fd = open(p->name, O_RDONLY, 0))) {
-			warn("%s", p->name);
+	p->fdin = STDIN_FILENO;
+	if (0 != strncmp(p->in, "-", 1))
+		if (-1 == (p->fdin = open(p->in, O_RDONLY, 0))) {
+			warn("%s", p->in);
 			return(io_leave(p, 0));
 		}
 
@@ -194,11 +199,17 @@ buf_begin(struct md_parse *p)
 {
 	struct stat	 st;
 
-	if (-1 == fstat(p->fd, &st)) {
-		warn("%s", p->name);
-		return(1);
+	if (-1 == fstat(p->fdin, &st)) {
+		warn("%s", p->in);
+		return(0);
 	} 
 
+	/*
+	 * Try to intuit the fastest way of sucking down buffered data
+	 * by using either the block buffer size or the hard-coded one.
+	 * This is inspired by bin/cat.c.
+	 */
+
 	p->bufsz = MAX(st.st_blksize, BUFSIZ);
 
 	if (NULL == (p->buf = malloc(p->bufsz))) {
@@ -213,18 +224,22 @@ buf_begin(struct md_parse *p)
 static int
 parse_leave(struct md_parse *p, int code)
 {
-	const struct mdoc_node *n;
 
 	if (NULL == p->mdoc)
 		return(code);
 
 	if ( ! mdoc_endparse(p->mdoc))
 		code = 0;
-	if (p->print && (n = mdoc_node(p->mdoc)))
-		(*p->print)(n);
 
-	mdoc_free(p->mdoc);
+#if 0
+	/* TODO */
+	if (code && ! mdoc_write(p->out, mdoc_node(p->mdoc))) {
+		warnx("%s: write error", p->out);
+		code = 0;
+	}
+#endif
 
+	mdoc_free(p->mdoc);
 	return(code);
 }
 
@@ -245,9 +260,19 @@ parse_begin(struct md_parse *p)
 	if (NULL == (p->mdoc = mdoc_alloc(p, &cb)))
 		return(parse_leave(p, 0));
 
+	/*
+	 * This is a little more complicated than fgets.  TODO: have
+	 * some benchmarks that show it's faster (note that I want to
+	 * check many, many manuals simultaneously, so speed is
+	 * important).  Fill a buffer (sized to the block size) with a
+	 * single read, then parse \n-terminated lines into a line
+	 * buffer, which is passed to the parser.  Hard-code the line
+	 * buffer to a particular size -- a reasonable assumption.
+	 */
+
 	for (lnn = 1, pos = 0; ; ) {
-		if (-1 == (sz = read(p->fd, p->buf, p->bufsz))) {
-			warn("%s", p->name);
+		if (-1 == (sz = read(p->fdin, p->buf, p->bufsz))) {
+			warn("%s", p->in);
 			return(parse_leave(p, 0));
 		} else if (0 == sz) 
 			break;
@@ -258,8 +283,7 @@ parse_begin(struct md_parse *p)
 					line[(int)pos++] = p->buf[(int)i];
 					continue;
 				}
-				warnx("%s: line %d too long", 
-						p->name, lnn);
+				warnx("%s: line %d too long", p->in, lnn);
 				return(parse_leave(p, 0));
 			}
 	
@@ -283,8 +307,8 @@ msg_err(void *arg, int line, int col, const char *msg)
 
 	p = (struct md_parse *)arg;
 
-	xfprintf(stderr, "%s:%d: error: %s (column %d)\n", 
-			p->name, line, msg, col);
+	warnx("%s:%d: error: %s (column %d)", 
+			p->in, line, msg, col);
 	return(0);
 }
 
@@ -299,8 +323,8 @@ msg_msg(void *arg, int line, int col, const char *msg)
 	if (0 == p->dbg)
 		return;
 
-	xfprintf(stderr, "%s:%d: debug: %s (column %d)\n", 
-			p->name, line, msg, col);
+	warnx("%s:%d: debug: %s (column %d)", 
+			p->in, line, msg, col);
 }
 
 
@@ -323,14 +347,13 @@ msg_warn(void *arg, int line, int col,
 		return(1);
 	}
 
-	xfprintf(stderr, "%s:%d: warning: %s (column %d)\n", 
-			p->name, line, msg, col);
+	warnx("%s:%d: warning: %s (column %d)", 
+			p->in, line, msg, col);
 
 	if ( ! (p->warn & MD_WARN_ERR))
 		return(1);
 
-	xfprintf(stderr, "%s: considering warnings as errors\n", 
-			__progname);
+	warnx("%s: considering warnings as errors", __progname);
 	return(0);
 }
 
@@ -339,7 +362,6 @@ static void
 usage(void)
 {
 
-	xfprintf(stderr, "usage: %s [-v] [-Wwarn...] [-ffilter] "
-			"[infile]\n", __progname);
+	warnx("usage: %s [-v] [-Wwarn...] [infile]", __progname);
 }