===================================================================
RCS file: /cvs/mandoc/Attic/mdocml.c,v
retrieving revision 1.26
retrieving revision 1.53
diff -u -p -r1.26 -r1.53
--- mandoc/Attic/mdocml.c	2008/12/28 21:25:09	1.26
+++ mandoc/Attic/mdocml.c	2009/02/20 23:35:36	1.53
@@ -1,4 +1,4 @@
-/* $Id: mdocml.c,v 1.26 2008/12/28 21:25:09 kristaps Exp $ */
+	/* $Id: mdocml.c,v 1.53 2009/02/20 23:35:36 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -30,86 +30,54 @@
 
 #include "mdoc.h"
 
-#define	MD_LINE_SZ	(256)
+#define	MD_LINE_SZ	(256)		/* Max input line size. */
 
 struct	md_parse {
-	int		 warn;
-#define	MD_WARN_ALL	(1 << 0)
-#define	MD_WARN_ERR	(1 << 1)
-	int		 dbg;
-	struct mdoc	*mdoc;
-	char		*buf;
-	u_long		 bufsz;
-	char		*name;
-	int		 fd;
-	int		 lnn;
-	char		*line;
+	int		  warn;		/* Warning flags. */
+#define	MD_WARN_SYNTAX	 (1 << 0)	/* Show syntax warnings. */
+#define	MD_WARN_COMPAT	 (1 << 1)	/* Show compat warnings. */
+#define	MD_WARN_ALL	 (0x03)		/* Show all warnings. */
+#define	MD_WARN_ERR	 (1 << 2)	/* Make warnings->errors. */
+	int		  dbg;		/* Debug level. */
+	struct mdoc	 *mdoc;		/* Active parser. */
+	char		 *buf;		/* Input buffer. */
+	u_long		  bufsz;	/* Input buffer size. */
+	char		 *in;		/* Input file name. */
+	int		  fdin;		/* Input file desc. */
 };
 
-static	void		 usage(void);
+extern	char	 	 *__progname;
 
-static	int		 parse_begin(struct md_parse *);
-static	int		 parse_leave(struct md_parse *, int);
-static	int		 io_begin(struct md_parse *);
-static	int		 io_leave(struct md_parse *, int);
-static	int		 buf_begin(struct md_parse *);
-static	int		 buf_leave(struct md_parse *, int);
+static	void		  usage(void);
 
-static	int		 msg_err(void *, int, int, enum mdoc_err);
-static	int		 msg_warn(void *, int, int, enum mdoc_warn);
-static	void		 msg_msg(void *, int, const char *);
+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 *);
+static	int		  io_leave(struct md_parse *, int);
+static	int		  buf_begin(struct md_parse *);
+static	int		  buf_leave(struct md_parse *, int);
+
+static	void		  msg_msg(void *, int, int, const char *);
+static	int		  msg_err(void *, int, int, const char *);
+static	int		  msg_warn(void *, int, int, 
+				enum mdoc_warn, const char *);
+
 #ifdef __linux__
-extern	int		 getsubopt(char **, char *const *, char **);
+extern	int		  getsubopt(char **, char *const *, char **);
 #endif
 
 int
 main(int argc, char *argv[])
 {
-	int		 c;
 	struct md_parse	 parser;
-	char		*opts, *v;
-#define ALL     	 0
-#define ERROR     	 1
-	char		*toks[] = { "all", "error", NULL };
 
-	extern char	*optarg;
-	extern int	 optind;
-
 	(void)memset(&parser, 0, sizeof(struct md_parse));
 
-	while (-1 != (c = getopt(argc, argv, "vW:")))
-		switch (c) {
-		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 (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 ( ! parse_opts(&parser, argc, argv))
+		return(EXIT_FAILURE);
 	if ( ! io_begin(&parser))
 		return(EXIT_FAILURE);
 
@@ -121,11 +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);
@@ -133,13 +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));
 		}
 
@@ -162,11 +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))) {
@@ -178,88 +221,27 @@ buf_begin(struct md_parse *p)
 }
 
 
-static void
-print_node(const struct mdoc_node *n, int indent)
+static int
+parse_leave(struct md_parse *p, int code)
 {
-	const char	 *p, *t;
-	int		  i, j;
-	size_t		  argc, sz;
-	char		**params;
-	struct mdoc_arg	 *argv;
+	extern int termprint(const struct mdoc_node *, 
+			const struct mdoc_meta *);
+	/*extern int treeprint(const struct mdoc_node *, 
+			const struct mdoc_meta *);*/
 
-	argv = NULL;
-	argc = 0;
-	params = NULL;
-	sz = 0;
+	if (NULL == p->mdoc)
+		return(code);
 
-	switch (n->type) {
-	case (MDOC_TEXT):
-		assert(NULL == n->child);
-		p = n->data.text.string;
-		t = "text";
-		break;
-	case (MDOC_BODY):
-		p = mdoc_macronames[n->data.body.tok];
-		t = "block-body";
-		break;
-	case (MDOC_HEAD):
-		p = mdoc_macronames[n->data.head.tok];
-		t = "block-head";
-		params = n->data.head.args;
-		sz = n->data.head.sz;
-		break;
-	case (MDOC_ELEM):
-		assert(NULL == n->child);
-		p = mdoc_macronames[n->data.elem.tok];
-		t = "element";
-		argv = n->data.elem.argv;
-		argc = n->data.elem.argc;
-		params = n->data.elem.args;
-		sz = n->data.elem.sz;
-		break;
-	case (MDOC_BLOCK):
-		p = mdoc_macronames[n->data.block.tok];
-		t = "block";
-		argv = n->data.block.argv;
-		argc = n->data.block.argc;
-		break;
-	default:
-		abort();
-		/* NOTREACHED */
-	}
+	if ( ! mdoc_endparse(p->mdoc))
+		code = 0;
 
-	for (i = 0; i < indent; i++)
-		(void)printf("    ");
-	(void)printf("%s (%s)", p, t);
+	/* TODO */
+	if (code && ! termprint(mdoc_node(p->mdoc), mdoc_meta(p->mdoc)))
+		code = 0;
+	/*if (code && ! treeprint(mdoc_node(p->mdoc), mdoc_meta(p->mdoc)))
+		code = 0;*/
 
-	for (i = 0; i < (int)argc; i++) {
-		(void)printf(" -%s", mdoc_argnames[argv[i].arg]);
-		for (j = 0; j < (int)argv[i].sz; j++)
-			(void)printf(" \"%s\"", argv[i].value[j]);
-	}
-
-	for (i = 0; i < (int)sz; i++)
-		(void)printf(" \"%s\"", params[i]);
-
-	(void)printf("\n");
-
-	if (n->child)
-		print_node(n->child, indent + 1);
-	if (n->next)
-		print_node(n->next, indent);
-}
-
-
-static int
-parse_leave(struct md_parse *p, int code)
-{
-	const struct mdoc_node *n;
-
-	if (p->mdoc) {
-		if ((n = mdoc_result(p->mdoc)))
-			print_node(n, 0);
-		mdoc_free(p->mdoc);
-	}
+	mdoc_free(p->mdoc);
 	return(code);
 }
 
@@ -269,8 +251,9 @@ parse_begin(struct md_parse *p)
 {
 	ssize_t		 sz, i;
 	size_t		 pos;
-	char		 line[256], sv[256];
+	char		 line[MD_LINE_SZ];
 	struct mdoc_cb	 cb;
+	int		 lnn;
 
 	cb.mdoc_err = msg_err;
 	cb.mdoc_warn = msg_warn;
@@ -279,12 +262,19 @@ parse_begin(struct md_parse *p)
 	if (NULL == (p->mdoc = mdoc_alloc(p, &cb)))
 		return(parse_leave(p, 0));
 
-	p->lnn = 1;
-	p->line = sv;
+	/*
+	 * 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 (pos = 0; ; ) {
-		if (-1 == (sz = read(p->fd, p->buf, p->bufsz))) {
-			warn("%s", p->name);
+	for (lnn = 1, pos = 0; ; ) {
+		if (-1 == (sz = read(p->fdin, p->buf, p->bufsz))) {
+			warn("%s", p->in);
 			return(parse_leave(p, 0));
 		} else if (0 == sz) 
 			break;
@@ -292,21 +282,18 @@ parse_begin(struct md_parse *p)
 		for (i = 0; i < sz; i++) {
 			if ('\n' != p->buf[i]) {
 				if (pos < sizeof(line)) {
-					sv[(int)pos] = p->buf[(int)i];
-					line[(int)pos++] = 
-						p->buf[(int)i];
+					line[(int)pos++] = p->buf[(int)i];
 					continue;
 				}
-				warnx("%s: line %d too long", 
-						p->name, p->lnn);
+				warnx("%s: line %d too long", p->in, lnn);
 				return(parse_leave(p, 0));
 			}
 	
-			line[(int)pos] = sv[(int)pos] = 0;
-			if ( ! mdoc_parseln(p->mdoc, line))
+			line[(int)pos] = 0;
+			if ( ! mdoc_parseln(p->mdoc, lnn, line))
 				return(parse_leave(p, 0));
 
-			p->lnn++;
+			lnn++;
 			pos = 0;
 		}
 	}
@@ -316,218 +303,67 @@ parse_begin(struct md_parse *p)
 
 
 static int
-msg_err(void *arg, int tok, int col, enum mdoc_err type)
+msg_err(void *arg, int line, int col, const char *msg)
 {
-	char		 *fmt, *lit;
 	struct md_parse	 *p;
-	int		  i;
 
 	p = (struct md_parse *)arg;
 
-	fmt = lit = NULL;
-
-	switch (type) {
-	case (ERR_SYNTAX_QUOTE):
-		lit = "syntax: disallowed argument quotation";
-		break;
-	case (ERR_SYNTAX_UNQUOTE):
-		lit = "syntax: unterminated quotation";
-		break;
-	case (ERR_SYNTAX_WS):
-		lit = "syntax: whitespace in argument";
-		break;
-	case (ERR_SYNTAX_ARGFORM):
-		fmt = "syntax: macro `%s' arguments malformed";
-		break;
-	case (ERR_SYNTAX_ARG):
-		fmt = "syntax: unknown argument for macro `%s'";
-		break;
-	case (ERR_SCOPE_BREAK):
-		/* Which scope is broken? */
-		fmt = "scope: macro `%s' breaks prior explicit scope";
-		break;
-	case (ERR_SCOPE_NOCTX):
-		fmt = "scope: closure macro `%s' has no context";
-		break;
-	case (ERR_SCOPE_NONEST):
-		fmt = "scope: macro `%s' may not be nested in the current context";
-		break;
-	case (ERR_MACRO_NOTSUP):
-		fmt = "macro `%s' not supported";
-		break;
-	case (ERR_MACRO_NOTCALL):
-		fmt = "macro `%s' not callable";
-		break;
-	case (ERR_SEC_PROLOGUE):
-		fmt = "macro `%s' cannot be called in the prologue";
-		break;
-	case (ERR_SEC_NPROLOGUE):
-		fmt = "macro `%s' called outside of prologue";
-		break;
-	case (ERR_ARGS_GE1):
-		fmt = "macro `%s' expects one or more arguments";
-		break;
-	case (ERR_ARGS_MANY):
-		fmt = "macro `%s' has too many arguments";
-		break;
-	case (ERR_SEC_PROLOGUE_OO):
-		fmt = "prologue macro `%s' is out-of-order";
-		break;
-	case (ERR_SEC_PROLOGUE_REP):
-		fmt = "prologue macro `%s' repeated";
-		break;
-	case (ERR_SEC_NAME):
-		lit = "`NAME' section must be first";
-		break;
-	case (ERR_SYNTAX_ARGVAL):
-		lit = "syntax: expected value for macro argument";
-		break;
-	case (ERR_SYNTAX_ARGBAD):
-		lit = "syntax: invalid value for macro argument";
-		break;
-	case (ERR_SYNTAX_ARGMANY):
-		lit = "syntax: too many values for macro argument";
-		break;
-	default:
-		abort();
-		/* NOTREACHED */
-	}
-
-	if (fmt) {
-		(void)fprintf(stderr, "%s:%d: error: ",
-				p->name, p->lnn);
-		(void)fprintf(stderr, fmt, mdoc_macronames[tok]);
-	} else
-		(void)fprintf(stderr, "%s:%d: error: %s",
-				p->name, p->lnn, lit);
-
-	if (p->dbg < 1) {
-		if (-1 != col)
-			(void)fprintf(stderr, " (column %d)\n", col);
-		return(0);
-	} else if (-1 == col) {
-		(void)fprintf(stderr, "\nFrom: %s", p->line);
-		return(0);
-	}
-
-	(void)fprintf(stderr, "\nFrom: %s\n      ", p->line);
-	for (i = 0; i < col; i++)
-		(void)fprintf(stderr, " ");
-	(void)fprintf(stderr, "^\n");
-
+	warnx("%s:%d: error: %s (column %d)", 
+			p->in, line, msg, col);
 	return(0);
 }
 
 
 static void
-msg_msg(void *arg, int col, const char *msg)
+msg_msg(void *arg, int line, int col, const char *msg)
 {
 	struct md_parse	 *p;
-	int		  i;
 
 	p = (struct md_parse *)arg;
 
-	if (p->dbg < 2)
+	if (0 == p->dbg)
 		return;
 
-	(void)printf("%s:%d: %s", p->name, p->lnn, msg);
-
-	if (p->dbg < 3) {
-		if (-1 != col)
-			(void)printf(" (column %d)\n", col);
-		return;
-	} else if (-1 == col) {
-		(void)printf("\nFrom %s\n", p->line);
-		return;
-	}
-
-	(void)printf("\nFrom: %s\n      ", p->line);
-	for (i = 0; i < col; i++)
-		(void)printf(" ");
-	(void)printf("^\n");
+	warnx("%s:%d: debug: %s (column %d)", 
+			p->in, line, msg, col);
 }
 
 
 static int
-msg_warn(void *arg, int tok, int col, enum mdoc_warn type)
+msg_warn(void *arg, int line, int col, 
+		enum mdoc_warn type, const char *msg)
 {
-	char		 *fmt, *lit;
 	struct md_parse	 *p;
-	int		  i;
-	extern char	 *__progname;
 
 	p = (struct md_parse *)arg;
 
-	if ( ! (p->warn & MD_WARN_ALL))
-		return(1);
-
-	fmt = lit = NULL;
-
 	switch (type) {
-	case (WARN_SYNTAX_WS_EOLN):
-		lit = "syntax: whitespace at end-of-line";
-		break;
-	case (WARN_SYNTAX_QUOTED):
-		lit = "syntax: quotation mark starting string";
-		break;
-	case (WARN_SYNTAX_MACLIKE):
-		lit = "syntax: macro-like argument";
-		break;
-	case (WARN_SYNTAX_ARGLIKE):
-		lit = "syntax: argument-like value";
-		break;
-	case (WARN_SEC_OO):
-		lit = "section is out of conventional order";
-		break;
-	case (WARN_ARGS_GE1):
-		fmt = "macro `%s' suggests one or more arguments";
-		break;
-	case (WARN_ARGS_EQ0):
-		fmt = "macro `%s' suggests zero arguments";
-		break;
-	case (WARN_IGN_AFTER_BLK):
-		fmt = "ignore: macro `%s' ignored after block macro";
-		break;
-	case (WARN_IGN_BEFORE_BLK):
-		fmt = "ignore: macro before block macro `%s' ignored";
-		break;
-	default:
-		abort();
-		/* NOTREACHED */
+	case (WARN_COMPAT):
+		if (p->warn & MD_WARN_COMPAT)
+			break;
+		return(1);
+	case (WARN_SYNTAX):
+		if (p->warn & MD_WARN_SYNTAX)
+			break;
+		return(1);
 	}
 
-	if (fmt) {
-		(void)fprintf(stderr, "%s:%d: warning: ",
-				p->name, p->lnn);
-		(void)fprintf(stderr, fmt, mdoc_macronames[tok]);
-	} else
-		(void)fprintf(stderr, "%s:%d: warning: %s",
-				p->name, p->lnn, lit);
+	warnx("%s:%d: warning: %s (column %d)", 
+			p->in, line, msg, col);
 
-	if (p->dbg >= 1) {
-		(void)fprintf(stderr, "\nFrom: %s\n      ", p->line);
-		for (i = 0; i < col; i++)
-			(void)fprintf(stderr, " ");
-		(void)fprintf(stderr, "^\n");
-	} else
-		(void)fprintf(stderr, " (column %d)\n", col);
+	if ( ! (p->warn & MD_WARN_ERR))
+		return(1);
 
-	if (p->warn & MD_WARN_ERR) {
-		(void)fprintf(stderr, "%s: considering warnings as "
-				"errors\n", __progname);
-		return(0);
-	}
-
-	return(1);
+	warnx("%s: considering warnings as errors", __progname);
+	return(0);
 }
 
 
 static void
 usage(void)
 {
-	extern char	*__progname;
 
-	(void)fprintf(stderr, "usage: %s [-v] [-Wwarn...] [infile]\n",
-			__progname);
+	warnx("usage: %s [-v] [-Wwarn...] [infile]", __progname);
 }