Purged all old files in favour of new design. Selective reintegration.

1 files changed, 309 insertions, 152 deletions

diff --git a/mdocml.c b/mdocml.c
index 590d3119..a84a1572 100644
--- a/mdocml.c
+++ b/mdocml.c
@@ -16,39 +16,48 @@
  * 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	 /* See begin_bufs. */
-#define	BUFFER_OUT_DEF	BUFSIZ	 /* See begin_bufs. */
+#define	MD_LINE_SZ	(256)
 
-#ifdef	DEBUG
-#define	CSS		"mdocml.css"
-#else
-#define	CSS		"/usr/local/share/mdocml/mdocml.css"
-#endif
+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;
+};
 
 static	void		 usage(void);
 
-static	int		 begin_io(const struct md_args *, 
-				char *, char *);
-static	int		 leave_io(const struct md_buf *, 
-				const struct md_buf *, int);
-static	int		 begin_bufs(const struct md_args *,
-				struct md_buf *, struct md_buf *);
-static int		 leave_bufs(const struct md_buf *, 
-				const struct md_buf *, int);
+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	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 *);
 
 #ifdef __linux__
 extern	int		 getsubopt(char **, char *const *, char **);
@@ -58,8 +67,8 @@ int
 main(int argc, char *argv[])
 {
 	int		 c;
-	char		*out, *in, *opts, *v;
-	struct md_args	 args;
+	struct md_parse	 parser;
+	char		*opts, *v;
 #define ALL     	 0
 #define ERROR     	 1
 	char		*toks[] = { "all", "error", NULL };
@@ -67,49 +76,22 @@ main(int argc, char *argv[])
 	extern char	*optarg;
 	extern int	 optind;
 
-	out = in = NULL;
+	(void)memset(&parser, 0, sizeof(struct md_parse));
 
-	(void)memset(&args, 0, sizeof(struct md_args));
-
-	args.type = MD_NOOP;
-
-	while (-1 != (c = getopt(argc, argv, "c:ef:o:vW:")))
+	while (-1 != (c = getopt(argc, argv, "vW:")))
 		switch (c) {
-		case ('c'):
-			if (args.type != MD_HTML)
-				errx(1, "-c only valid for -fhtml");
-			args.params.html.css = optarg;
-			break;
-		case ('e'):
-			if (args.type != MD_HTML)
-				errx(1, "-e only valid for -fhtml");
-			args.params.html.flags |= HTML_CSS_EMBED;
-			break;
-		case ('f'):
-			if (0 == strcmp(optarg, "html"))
-				args.type = MD_HTML;
-			else if (0 == strcmp(optarg, "xml"))
-				args.type = MD_XML;
-			else if (0 == strcmp(optarg, "noop"))
-				args.type = MD_NOOP;
-			else
-				errx(1, "invalid filter type");
-			break;
-		case ('o'):
-			out = optarg;
-			break;
 		case ('v'):
-			args.verbosity++;
+			parser.dbg++;
 			break;
 		case ('W'):
 			opts = optarg;
 			while (*opts) 
 				switch (getsubopt(&opts, toks, &v)) {
 				case (ALL):
-					args.warnings |= MD_WARN_ALL;
+					parser.warn |= MD_WARN_ALL;
 					break;
 				case (ERROR):
-					args.warnings |= MD_WARN_ERROR;
+					parser.warn |= MD_WARN_ERR;
 					break;
 				default:
 					usage();
@@ -121,148 +103,324 @@ main(int argc, char *argv[])
 			return(1);
 		}
 
-	if (MD_HTML == args.type)
-		if (NULL == args.params.html.css) 
-			args.params.html.css = CSS;
-
 	argv += optind;
 	argc -= optind;
 
+	parser.name = "-";
 	if (1 == argc)
-		in = *argv++;
+		parser.name = *argv++;
+
+	if ( ! io_begin(&parser))
+		return(EXIT_FAILURE);
 
-	return(begin_io(&args, out ? out : "-", in ? in : "-"));
+	return(EXIT_SUCCESS);
 }
 
 
-/* 
- * Close out file descriptors opened in begin_io.  If the descriptor
- * refers to stdin/stdout, then do nothing.
- */
 static int
-leave_io(const struct md_buf *out, 
-		const struct md_buf *in, int c)
+io_leave(struct md_parse *p, int code)
 {
-	assert(out);
-	assert(in);
 
-	if (-1 != in->fd && -1 == close(in->fd)) {
-		assert(in->name);
-		warn("%s", in->name);
-		c = 1;
-	}
-	if (-1 != out->fd && STDOUT_FILENO != out->fd &&
-			-1 == close(out->fd)) {
-		assert(out->name);
-		warn("%s", out->name);
-		c = 1;
+	if (-1 == p->fd || STDIN_FILENO == p->fd)
+		return(code);
+
+	if (-1 == close(p->fd)) {
+		warn("%s", p->name);
+		code = 0;
 	}
-	if (1 == c && STDOUT_FILENO != out->fd)
-		if (-1 == unlink(out->name))
-			warn("%s", out->name);
+	return(code);
+}
+
+
+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);
+			return(io_leave(p, 0));
+		}
 
-	return(c);
+	return(io_leave(p, buf_begin(p)));
 }
 
 
-/*
- * Open file descriptors or assign stdin/stdout, if dictated by the "-"
- * token instead of a filename.
- */
 static int
-begin_io(const struct md_args *args, char *out, char *in)
+buf_leave(struct md_parse *p, int code)
 {
-	struct md_buf	 fi;
-	struct md_buf	 fo;
 
-#define	FI_FL	O_RDONLY
-#define	FO_FL	O_WRONLY|O_CREAT|O_TRUNC
+	if (p->buf)
+		free(p->buf);
+	return(code);
+}
 
-	assert(args);
-	assert(out);
-	assert(in);
 
-	bzero(&fi, sizeof(struct md_buf));
-	bzero(&fo, sizeof(struct md_buf));
+static int
+buf_begin(struct md_parse *p)
+{
+	struct stat	 st;
 
-	fi.fd = STDIN_FILENO;
-	fo.fd = STDOUT_FILENO;
+	if (-1 == fstat(p->fd, &st)) {
+		warn("%s", p->name);
+		return(1);
+	} 
 
-	fi.name = in;
-	fo.name = out;
+	p->bufsz = MAX(st.st_blksize, BUFSIZ);
 
-	if (0 != strncmp(fi.name, "-", 1))
-		if (-1 == (fi.fd = open(fi.name, FI_FL, 0))) {
-			warn("%s", fi.name);
-			return(leave_io(&fo, &fi, 1));
-		}
+	if (NULL == (p->buf = malloc(p->bufsz))) {
+		warn("malloc");
+		return(buf_leave(p, 0));
+	}
+
+	return(buf_leave(p, parse_begin(p)));
+}
+
+
+static void
+print_node(const struct mdoc_node *n, int indent)
+{
+	const char	*p, *t;
+	int		 i;
+
+	switch (n->type) {
+	case (MDOC_TEXT):
+		assert(NULL == n->child);
+		p = "<text>";
+		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";
+		break;
+	case (MDOC_ELEM):
+		assert(NULL == n->child);
+		p = mdoc_macronames[n->data.elem.tok];
+		t = "element";
+		break;
+	case (MDOC_BLOCK):
+		p = mdoc_macronames[n->data.block.tok];
+		t = "block";
+		break;
+	}
+
+	for (i = 0; i < indent; i++)
+		(void)printf("    ");
+	(void)printf("%s (%s)\n", p, t);
+
+	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 (0 != strncmp(fo.name, "-", 1)) 
-		if (-1 == (fo.fd = open(fo.name, FO_FL, 0644))) {
-			warn("%s", fo.name);
-			return(leave_io(&fo, &fi, 1));
+	if (p->mdoc) {
+		if ((n = mdoc_result(p->mdoc)))
+			print_node(n, 0);
+		mdoc_free(p->mdoc);
+	}
+	return(code);
+}
+
+
+static int
+parse_begin(struct md_parse *p)
+{
+	ssize_t		 sz, i;
+	size_t		 pos;
+	char		 line[256], sv[256];
+	struct mdoc_cb	 cb;
+
+	cb.mdoc_err = msg_err;
+	cb.mdoc_warn = msg_warn;
+	cb.mdoc_msg = msg_msg;
+
+	if (NULL == (p->mdoc = mdoc_alloc(p, &cb)))
+		return(parse_leave(p, 0));
+
+	p->lnn = 1;
+	p->line = sv;
+
+	for (pos = 0; ; ) {
+		if (-1 == (sz = read(p->fd, p->buf, p->bufsz))) {
+			warn("%s", p->name);
+			return(parse_leave(p, 0));
+		} else if (0 == sz) 
+			break;
+
+		for (i = 0; i < sz; i++) {
+			if ('\n' != p->buf[i]) {
+				if (pos < sizeof(line)) {
+					/* LINTED */
+					sv[pos] = p->buf[i];
+					line[pos++] = p->buf[i];
+					continue;
+				}
+				warnx("%s: line %d too long", 
+						p->name, p->lnn);
+				return(parse_leave(p, 0));
+			}
+	
+			line[(int)pos] = sv[(int)pos] = 0;
+			if ( ! mdoc_parseln(p->mdoc, line))
+				return(parse_leave(p, 0));
+
+			p->lnn++;
+			pos = 0;
 		}
+	}
 
-	return(leave_io(&fo, &fi, begin_bufs(args, &fo, &fi)));
+	return(parse_leave(p, 1));
 }
 
 
-/*
- * Free buffers allocated in begin_bufs.
- */
 static int
-leave_bufs(const struct md_buf *out, 
-		const struct md_buf *in, int c)
+msg_err(void *arg, int tok, int col, enum mdoc_err type)
+{
+	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: unterminated quotation";
+		break;
+	case (ERR_SYNTAX_WS):
+		lit = "syntax: whitespace in argument";
+		break;
+	case (ERR_SCOPE_BREAK):
+		/* Which scope is broken? */
+		fmt = "macro `%s' breaks prior explicit scope";
+		break;
+	case (ERR_MACRO_NOTSUP):
+		fmt = "macro `%s' not supported";
+		break;
+	case (ERR_MACRO_NOTCALL):
+		fmt = "macro `%s' not callable";
+		break;
+	case (ERR_ARGS_GE1):
+		fmt = "macro `%s' expects one or more arguments";
+		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) {
+		(void)fprintf(stderr, " (column %d)\n", col);
+		return(0);
+	}
+
+	(void)fprintf(stderr, "\nFrom: %s\n      ", p->line);
+	for (i = 0; i < col; i++)
+		(void)fprintf(stderr, " ");
+	(void)fprintf(stderr, "^\n");
+
+	return(0);
+}
+
+
+static void
+msg_msg(void *arg, int col, const char *msg)
 {
-	assert(out);
-	assert(in);
-	if (out->buf)
-		free(out->buf);
-	if (in->buf)
-		free(in->buf);
-	return(c);
+	struct md_parse	 *p;
+	int		  i;
+
+	p = (struct md_parse *)arg;
+
+	if (p->dbg < 2)
+		return;
+
+	(void)printf("%s:%d: %s", p->name, p->lnn, msg);
+
+	if (p->dbg < 3) {
+		(void)printf(" (column %d)\n", col);
+		return;
+	}
+
+	(void)printf("\nFrom: %s\n      ", p->line);
+	for (i = 0; i < col; i++)
+		(void)printf(" ");
+	(void)printf("^\n");
 }
 
 
-/*
- * Allocate buffers to the maximum of either the input file's blocksize
- * or BUFFER_IN_DEF/BUFFER_OUT_DEF, which should be around BUFSIZE.
- */
 static int
-begin_bufs(const struct md_args *args, 
-		struct md_buf *out, struct md_buf *in)
+msg_warn(void *arg, int tok, int col, enum mdoc_warn type)
 {
-	struct stat	 stin, stout;
-	int		 c;
+	char		 *fmt, *lit;
+	struct md_parse	 *p;
+	int		  i;
+	extern char	 *__progname;
 
-	assert(args);
-	assert(in);
-	assert(out);
+	p = (struct md_parse *)arg;
 
-	if (-1 == fstat(in->fd, &stin)) {
-		warn("%s", in->name);
-		return(1);
-	} else if (STDIN_FILENO != in->fd && 0 == stin.st_size) {
-		warnx("%s: empty file", in->name);
+	if ( ! (p->warn & MD_WARN_ALL))
 		return(1);
-	} else if (-1 == fstat(out->fd, &stout)) {
-		warn("%s", out->name);
-		return(1);
-	}
 
-	in->bufsz = MAX(stin.st_blksize, BUFFER_IN_DEF);
-	out->bufsz = MAX(stout.st_blksize, BUFFER_OUT_DEF);
+	fmt = lit = NULL;
+
+	switch (type) {
+	case (WARN_SYNTAX_WS_EOLN):
+		lit = "syntax: whitespace at end-of-line";
+		break;
+	case (WARN_SYNTAX_MACLIKE):
+		lit = "syntax: macro-like argument";
+		break;
+	case (WARN_ARGS_GE1):
+		fmt = "macro `%s' suggests one or more arguments";
+		break;
+	default:
+		abort();
+		/* NOTREACHED */
+	}
 
-	if (NULL == (in->buf = malloc(in->bufsz))) {
-		warn("malloc");
-		return(leave_bufs(out, in, 1));
-	} else if (NULL == (out->buf = malloc(out->bufsz))) {
-		warn("malloc");
-		return(leave_bufs(out, in, 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);
+
+	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) {
+		(void)fprintf(stderr, "%s: considering warnings as "
+				"errors\n", __progname);
+		return(0);
 	}
 
-	c = md_run(args, out, in);
-	return(leave_bufs(out, in, -1 == c ? 1 : 0));
+	return(1);
 }
 
 
@@ -271,8 +429,7 @@ usage(void)
 {
 	extern char	*__progname;
 
-	(void)fprintf(stderr, "usage: %s [-v] [-Wwarn...]  "
-			"[-f filter] [-o outfile] [infile]\n", 
+	(void)fprintf(stderr, "usage: %s [-v] [-Wwarn...] [infile]\n",
 			__progname);
 }