===================================================================
RCS file: /cvs/mandoc/Attic/mdocml.c,v
retrieving revision 1.3
retrieving revision 1.54
diff -u -p -r1.3 -r1.54
--- mandoc/Attic/mdocml.c	2008/11/22 17:14:32	1.3
+++ mandoc/Attic/mdocml.c	2009/02/21 15:34:46	1.54
@@ -1,4 +1,4 @@
-/* $Id: mdocml.c,v 1.3 2008/11/22 17:14:32 kristaps Exp $ */
+	/* $Id: mdocml.c,v 1.54 2009/02/21 15:34:46 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -16,335 +16,361 @@
  * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
  * PERFORMANCE OF THIS SOFTWARE.
  */
-#include <sys/param.h>
 #include <sys/stat.h>
+#include <sys/param.h>
 
 #include <assert.h>
-#include <err.h>
 #include <fcntl.h>
+#include <err.h>
 #include <getopt.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
 
-#include "libmdocml.h"
+#include "mdoc.h"
 
-#define	BUFFER_IN_DEF	BUFSIZ
-#define	BUFFER_OUT_DEF	BUFSIZ
-#define	BUFFER_LINE	BUFSIZ
+#define	MD_LINE_SZ	(256)		/* Max input line size. */
 
-struct md_rbuf {
-	int		 fd;
-	const char	*name;
-	char		*buf;
-	size_t		 bufsz;
-	size_t		 line;
-};
+typedef	int		(*md_print)(const struct mdoc_node *,
+				const struct mdoc_meta *);
 
-struct md_mbuf {
-	int		 fd;
-	const char	*name;
-	char		*buf;
-	size_t		 bufsz;
-	size_t		 pos;
+struct	md_parse {
+	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. */
+	md_print	  fp;
 };
 
-static void		 usage(void);
+extern	char	 	 *__progname;
 
-static int		 md_begin(const char *, const char *);
-static int		 md_begin_io(const char *, const char *);
-static int		 md_begin_bufs(struct md_mbuf *, struct md_rbuf *);
-static int		 md_run(struct md_mbuf *, struct md_rbuf *);
-static int		 md_line(struct md_mbuf *, const struct md_rbuf *,
-				const char *, size_t);
+static	void		  usage(void);
 
-static ssize_t		 md_buf_fill(struct md_rbuf *);
-static int		 md_buf_flush(struct md_mbuf *);
+static	int		  parse_opts(struct md_parse *, int, char *[]);
+static	int		  parse_subopts(struct md_parse *, char *);
 
-static int		 md_buf_putchar(struct md_mbuf *, char);
-static int		 md_buf_puts(struct md_mbuf *, 
-				const char *, size_t);
+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 **);
+#endif
+
 int
 main(int argc, char *argv[])
 {
-	int		 c;
-	char		*out, *in;
+	struct md_parse	 parser;
 
-	extern char	*optarg;
-	extern int	 optind;
+	(void)memset(&parser, 0, sizeof(struct md_parse));
 
-	out = NULL;
-	
-	while (-1 != (c = getopt(argc, argv, "o:")))
-		switch (c) {
-		case ('o'):
-			out = optarg;
-			break;
-		default:
-			usage();
-			return(1);
-		}
+	if ( ! parse_opts(&parser, argc, argv))
+		return(EXIT_FAILURE);
+	if ( ! io_begin(&parser))
+		return(EXIT_FAILURE);
 
-	argv += optind;
-	if (1 != (argc -= optind)) {
-		usage();
-		return(1);
-	}
+	return(EXIT_SUCCESS);
+}
 
-	argc--;
-	in = *argv++;
 
-	return(md_begin(out, in));
+static int
+io_leave(struct md_parse *p, int code)
+{
+
+	if (-1 == p->fdin || STDIN_FILENO == p->fdin)
+		return(code);
+
+	if (-1 == close(p->fdin)) {
+		warn("%s", p->in);
+		code = 0;
+	}
+	return(code);
 }
 
 
 static int
-md_begin(const char *out, const char *in)
+parse_subopts(struct md_parse *p, char *arg)
 {
-	char		 buf[MAXPATHLEN];
+	char		*v;
+	char		*toks[] = { "all", "compat", 
+				"syntax", "error", NULL };
 
-	assert(in);
-	if (out)
-		return(md_begin_io(out, in));
+	/* 
+	 * 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).
+	 */
 
-	if (strlcpy(buf, in, MAXPATHLEN) >= MAXPATHLEN)
-		warnx("output filename too long");
-	else if (strlcat(buf, ".html", MAXPATHLEN) >= MAXPATHLEN)
-		warnx("output filename too long");
-	else 
-		return(md_begin_io(buf, in));
+	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
-md_begin_io(const char *out, const char *in)
+parse_opts(struct md_parse *p, int argc, char *argv[])
 {
 	int		 c;
-	struct md_rbuf	 fin;
-	struct md_mbuf	 fout;
 
-	assert(out);
-	assert(in);
+	extern char	*optarg;
+	extern int	 optind;
 
-	/* TODO: accept "-" as both input and output. */
+	extern int termprint(const struct mdoc_node *, 
+			const struct mdoc_meta *);
+	extern int treeprint(const struct mdoc_node *, 
+			const struct mdoc_meta *);
 
-	fin.name = in;
+	p->in = "-";
 
-	if (-1 == (fin.fd = open(fin.name, O_RDONLY, 0))) {
-		warn("%s", fin.name);
+	while (-1 != (c = getopt(argc, argv, "f:vW:")))
+		switch (c) {
+		case ('f'):
+			if (0 == strcmp(optarg, "tree")) {
+				p->fp = treeprint;
+				break;
+			} else if (0 == strcmp(optarg, "term")) {
+				p->fp = termprint;
+				break;
+			}
+			warnx("unknown filter: %s", optarg);
+			return(0);
+		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);
-	}
 
-	fout.name = out;
+	p->in = *argv++;
+	return(1);
+}
 
-	fout.fd = open(fout.name, O_WRONLY | O_CREAT | O_TRUNC, 0644);
-	if (-1 == fout.fd) {
-		warn("%s", fout.name);
-		if (-1 == close(fin.fd))
-			warn("%s", fin.name);
-		return(1);
-	}
 
-	c = md_begin_bufs(&fout, &fin);
+static int
+io_begin(struct md_parse *p)
+{
 
-	if (-1 == close(fin.fd)) {
-		warn("%s", in);
-		c = 1;
-	}
-	if (-1 == close(fout.fd)) {
-		warn("%s", out);
-		c = 1;
-	}
+	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));
+		}
 
-	return(c);
+	return(io_leave(p, buf_begin(p)));
 }
 
 
 static int
-md_begin_bufs(struct md_mbuf *out, struct md_rbuf *in)
+buf_leave(struct md_parse *p, int code)
 {
-	struct stat	 stin, stout;
-	int		 c;
 
-	assert(in);
-	assert(out);
+	if (p->buf)
+		free(p->buf);
+	return(code);
+}
 
-	if (-1 == fstat(in->fd, &stin)) {
-		warn("%s", in->name);
-		return(1);
-	} else if (-1 == fstat(out->fd, &stout)) {
-		warn("%s", out->name);
-		return(1);
-	}
 
-	in->bufsz = MAX(stin.st_blksize, BUFFER_IN_DEF);
+static int
+buf_begin(struct md_parse *p)
+{
+	struct stat	 st;
 
-	out->bufsz = MAX(stout.st_blksize, BUFFER_OUT_DEF);
+	if (-1 == fstat(p->fdin, &st)) {
+		warn("%s", p->in);
+		return(0);
+	} 
 
-	if (NULL == (in->buf = malloc(in->bufsz))) {
+	/*
+	 * 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))) {
 		warn("malloc");
-		return(1);
-	} else if (NULL == (out->buf = malloc(out->bufsz))) {
-		warn("malloc");
-		free(in->buf);
-		return(1);
+		return(buf_leave(p, 0));
 	}
 
-	c = md_run(out, in);
-
-	free(in->buf);
-	free(out->buf);
-
-	return(c);
+	return(buf_leave(p, parse_begin(p)));
 }
 
 
-static ssize_t
-md_buf_fill(struct md_rbuf *in)
+static int
+parse_leave(struct md_parse *p, int code)
 {
-	ssize_t		 ssz;
+	md_print	 fp;
 
-	assert(in);
-	assert(in->buf);
-	assert(in->bufsz > 0);
-	assert(in->name);
+	if (NULL == p->mdoc)
+		return(code);
 
-	if (-1 == (ssz = read(in->fd, in->buf, in->bufsz))) 
-		warn("%s", in->name);
+	if ( ! mdoc_endparse(p->mdoc))
+		code = 0;
 
-	return(ssz);
+	if (code && (fp = p->fp)) {
+		if ( ! (*fp)(mdoc_node(p->mdoc), mdoc_meta(p->mdoc)))
+			code = 0;
+	}
+
+	mdoc_free(p->mdoc);
+	return(code);
 }
 
 
 static int
-md_run(struct md_mbuf *out, struct md_rbuf *in)
+parse_begin(struct md_parse *p)
 {
 	ssize_t		 sz, i;
-	char		 line[BUFFER_LINE];
 	size_t		 pos;
+	char		 line[MD_LINE_SZ];
+	struct mdoc_cb	 cb;
+	int		 lnn;
 
-	assert(in);
-	assert(out); 
+	cb.mdoc_err = msg_err;
+	cb.mdoc_warn = msg_warn;
+	cb.mdoc_msg = msg_msg;
 
-	out->pos = 0;
-	in->line = 1;
+	if (NULL == (p->mdoc = mdoc_alloc(p, &cb)))
+		return(parse_leave(p, 0));
 
-	/* LINTED */
-	for (pos = 0; ; ) {
-		if (-1 == (sz = md_buf_fill(in)))
-			return(1);
-		else if (0 == sz)
+	/*
+	 * 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->fdin, p->buf, p->bufsz))) {
+			warn("%s", p->in);
+			return(parse_leave(p, 0));
+		} else if (0 == sz) 
 			break;
 
 		for (i = 0; i < sz; i++) {
-			if ('\n' == in->buf[i]) {
-				if (md_line(out, in, line, pos))
-					return(1);
-				in->line++;
-				pos = 0;
-				continue;
+			if ('\n' != p->buf[i]) {
+				if (pos < sizeof(line)) {
+					line[(int)pos++] = p->buf[(int)i];
+					continue;
+				}
+				warnx("%s: line %d too long", p->in, lnn);
+				return(parse_leave(p, 0));
 			}
+	
+			line[(int)pos] = 0;
+			if ( ! mdoc_parseln(p->mdoc, lnn, line))
+				return(parse_leave(p, 0));
 
-			if (pos < BUFFER_LINE) {
-				/* LINTED */
-				line[pos++] = in->buf[i];
-				continue;
-			}
-
-			warnx("%s: line %zu too long",
-					in->name, in->line);
-			return(1);
+			lnn++;
+			pos = 0;
 		}
 	}
 
-	if (0 != pos && md_line(out, in, line, pos))
-		return(1);
-
-	return(md_buf_flush(out) ? 0 : 1);
+	return(parse_leave(p, 1));
 }
 
 
 static int
-md_buf_flush(struct md_mbuf *buf)
+msg_err(void *arg, int line, int col, const char *msg)
 {
-	ssize_t		 sz;
+	struct md_parse	 *p;
 
-	assert(buf);
-	assert(buf->buf);
-	assert(buf->name);
+	p = (struct md_parse *)arg;
 
-	if (0 == buf->pos)
-		return(1);
-
-	sz = write(buf->fd, buf->buf, buf->pos);
-
-	if (-1 == sz) {
-		warn("%s", buf->name);
-		return(0);
-	} else if ((size_t)sz != buf->pos) {
-		warnx("%s: short write", buf->name);
-		return(0);
-	}
-
-	buf->pos = 0;
-	return(1);
+	warnx("%s:%d: error: %s (column %d)", 
+			p->in, line, msg, col);
+	return(0);
 }
 
 
-static int
-md_buf_putchar(struct md_mbuf *buf, char c)
+static void
+msg_msg(void *arg, int line, int col, const char *msg)
 {
-	return(md_buf_puts(buf, &c, 1));
-}
+	struct md_parse	 *p;
 
+	p = (struct md_parse *)arg;
 
-static int
-md_buf_puts(struct md_mbuf *buf, const char *p, size_t sz)
-{
-	size_t		 ssz;
+	if (0 == p->dbg)
+		return;
 
-	assert(p);
-	assert(buf);
-	assert(buf->buf);
-
-	while (buf->pos + sz > buf->bufsz) {
-		ssz = buf->bufsz - buf->pos;
-		(void)memcpy(buf->buf + buf->pos, p, ssz);
-		p += ssz;
-		sz -= ssz;
-		buf->pos += ssz;
-
-		if ( ! md_buf_flush(buf))
-			return(0);
-	}
-
-	(void)memcpy(buf->buf + buf->pos, p, sz);
-	buf->pos += sz;
-	return(1);
+	warnx("%s:%d: debug: %s (column %d)", 
+			p->in, line, msg, col);
 }
 
 
 static int
-md_line(struct md_mbuf *out, const struct md_rbuf *in,
-		const char *buf, size_t sz)
+msg_warn(void *arg, int line, int col, 
+		enum mdoc_warn type, const char *msg)
 {
+	struct md_parse	 *p;
 
-	/* FIXME: this is just a placeholder function. */
+	p = (struct md_parse *)arg;
 
-	assert(buf);
-	assert(out);
-	assert(in);
-
-	if ( ! md_buf_puts(out, buf, sz))
+	switch (type) {
+	case (WARN_COMPAT):
+		if (p->warn & MD_WARN_COMPAT)
+			break;
 		return(1);
-	if ( ! md_buf_putchar(out, '\n'))
+	case (WARN_SYNTAX):
+		if (p->warn & MD_WARN_SYNTAX)
+			break;
 		return(1);
+	}
 
+	warnx("%s:%d: warning: %s (column %d)", 
+			p->in, line, msg, col);
+
+	if ( ! (p->warn & MD_WARN_ERR))
+		return(1);
+
+	warnx("%s: considering warnings as errors", __progname);
 	return(0);
 }
 
@@ -352,7 +378,8 @@ md_line(struct md_mbuf *out, const struct md_rbuf *in,
 static void
 usage(void)
 {
-	extern char	*__progname;
 
-	(void)printf("usage: %s [-o outfile] infile\n", __progname);
+	warnx("usage: %s [-ffilter] [-v] [-Wwarn...] [infile]", 
+			__progname);
 }
+