===================================================================
RCS file: /cvs/mandoc/main.c,v
retrieving revision 1.68
retrieving revision 1.75
diff -u -p -r1.68 -r1.75
--- mandoc/main.c	2010/05/15 09:46:31	1.68
+++ mandoc/main.c	2010/05/16 00:04:46	1.75
@@ -1,4 +1,4 @@
-/*	$Id: main.c,v 1.68 2010/05/15 09:46:31 joerg Exp $ */
+/*	$Id: main.c,v 1.75 2010/05/16 00:04:46 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -29,8 +29,10 @@
 #include <string.h>
 #include <unistd.h>
 
+#include "mandoc.h"
 #include "mdoc.h"
 #include "man.h"
+#include "roff.h"
 #include "main.h"
 
 #define	UNCONST(a)	((void *)(uintptr_t)(const void *)(a))
@@ -70,6 +72,7 @@ struct	curparse {
 	const char	 *file;		/* Current parse. */
 	int		  fd;		/* Current parse. */
 	int		  wflags;
+	/* FIXME: set by max error */
 #define	WARN_WALL	 (1 << 0)	/* All-warnings mask. */
 #define	WARN_WERR	 (1 << 2)	/* Warnings->errors. */
 	int		  fflags;
@@ -82,6 +85,7 @@ struct	curparse {
 	enum intt	  inttype;	/* Input parsers... */
 	struct man	 *man;
 	struct mdoc	 *mdoc;
+	struct roff	 *roff;
 	enum outt	  outtype;	/* Output devices... */
 	out_mdoc	  outmdoc;
 	out_man	  	  outman;
@@ -95,9 +99,12 @@ static	void		  ffile(const char *, struct curparse *);
 static	int		  foptions(int *, char *);
 static	struct man	 *man_init(struct curparse *);
 static	struct mdoc	 *mdoc_init(struct curparse *);
-static	int		  merr(void *, int, int, const char *);
+static	struct roff	 *roff_init(struct curparse *);
+static	int		  merr(void *, int, int, const char *); /* DEPRECATED */
 static	int		  moptions(enum intt *, char *);
-static	int		  mwarn(void *, int, int, const char *);
+static	int		  mwarn(void *, int, int, const char *); /* DEPRECATED */
+static	int		  mmsg(enum mandocerr, void *, 
+				int, int, const char *);
 static	int		  pset(const char *, int, struct curparse *,
 				struct man **, struct mdoc **);
 static	int		  toptions(struct curparse *, char *);
@@ -177,6 +184,12 @@ main(int argc, char *argv[])
 
 	if (curp.outfree)
 		(*curp.outfree)(curp.outdata);
+	if (curp.mdoc)
+		mdoc_free(curp.mdoc);
+	if (curp.man)
+		man_free(curp.man);
+	if (curp.roff)
+		roff_free(curp.roff);
 
 	return((with_warning || with_error) ? 
 			EXIT_FAILURE :  EXIT_SUCCESS);
@@ -225,6 +238,14 @@ man_init(struct curparse *curp)
 }
 
 
+static struct roff *
+roff_init(struct curparse *curp)
+{
+
+	return(roff_alloc(mmsg, curp));
+}
+
+
 static struct mdoc *
 mdoc_init(struct curparse *curp)
 {
@@ -316,7 +337,7 @@ read_whole_file(struct curparse *curp, struct buf *fb,
 			return(0);
 		}
 		*with_mmap = 1;
-		fb->sz = st.st_size;
+		fb->sz = (size_t)st.st_size;
 		fb->buf = mmap(NULL, fb->sz, PROT_READ, 
 				MAP_FILE, curp->fd, 0);
 		if (fb->buf != MAP_FAILED)
@@ -342,7 +363,7 @@ read_whole_file(struct curparse *curp, struct buf *fb,
 			if (! resize_buf(fb, 65536))
 				break;
 		}
-		ssz = read(curp->fd, fb->buf + off, fb->sz - off);
+		ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
 		if (ssz == 0) {
 			fb->sz = off;
 			return(1);
@@ -351,7 +372,7 @@ read_whole_file(struct curparse *curp, struct buf *fb,
 			perror(curp->file);
 			break;
 		}
-		off += ssz;
+		off += (size_t)ssz;
 	}
 
 	free(fb->buf);
@@ -365,87 +386,103 @@ static void
 fdesc(struct curparse *curp)
 {
 	struct buf	 ln, blk;
-	int		 j, i, pos, lnn, comment, with_mmap;
+	int		 i, pos, lnn, lnn_start, with_mmap;
+	enum rofferr	 re;
 	struct man	*man;
 	struct mdoc	*mdoc;
+	struct roff	*roff;
 
 	man = NULL;
 	mdoc = NULL;
+	roff = NULL;
 	memset(&ln, 0, sizeof(struct buf));
 
 	/*
-	 * Two buffers: ln and buf.  buf is the input buffer optimised
-	 * here for each file's block size.  ln is a line buffer.  Both
-	 * growable, hence passed in by ptr-ptr.
+	 * Two buffers: ln and buf.  buf is the input file and may be
+	 * memory mapped.  ln is a line buffer and grows on-demand.
 	 */
 
-	if (!read_whole_file(curp, &blk, &with_mmap))
+	if ( ! read_whole_file(curp, &blk, &with_mmap))
 		return;
 
-	/* Fill buf with file blocksize. */
+	if (NULL == curp->roff) 
+		curp->roff = roff_init(curp);
+	if (NULL == (roff = curp->roff))
+		goto bailout;
 
-	for (i = lnn = pos = comment = 0; i < (int)blk.sz; ++i) {
-		if (pos >= (int)ln.sz) {
-			if (! resize_buf(&ln, 256))
-				goto bailout;
-		}
-
-		if ('\n' != blk.buf[i]) {
-			if (comment)
+	for (i = 0, lnn = 1; i < (int)blk.sz;) {
+		pos = 0;
+		lnn_start = lnn;
+		while (i < (int)blk.sz) {
+			if ('\n' == blk.buf[i]) {
+				++i;
+				++lnn;
+				break;
+			}
+			/* Trailing backslash is like a plain character. */
+			if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
+				if (pos >= (int)ln.sz)
+					if (! resize_buf(&ln, 256))
+						goto bailout;
+				ln.buf[pos++] = blk.buf[i++];
 				continue;
-			ln.buf[pos++] = blk.buf[i];
-
-			/* Handle in-line `\"' comments. */
-
-			if (1 == pos || '\"' != ln.buf[pos - 1])
-				continue;
-
-			for (j = pos - 2; j >= 0; j--)
-				if ('\\' != ln.buf[j])
-					break;
-
-			if ( ! ((pos - 2 - j) % 2))
-				continue;
-
-			comment = 1;
-			pos -= 2;
-			for (; pos > 0; --pos) {
-				if (ln.buf[pos - 1] != ' ')
-					break;
-				if (pos > 2 && ln.buf[pos - 2] == '\\')
-					break;
 			}
-			continue;
-		} 
-
-		/* Handle escaped `\\n' newlines. */
-
-		if (pos > 0 && 0 == comment && '\\' == ln.buf[pos - 1]) {
-			for (j = pos - 1; j >= 0; j--)
-				if ('\\' != ln.buf[j])
-					break;
-			if ( ! ((pos - j) % 2)) {
-				pos--;
-				lnn++;
+			/* Found an escape and at least one other character. */
+			if ('\n' == blk.buf[i + 1]) {
+				/* Escaped newlines are skipped over */
+				i += 2;
+				++lnn;
 				continue;
 			}
+			if ('"' == blk.buf[i + 1]) {
+				i += 2;
+				/* Comment, skip to end of line */
+				for (; i < (int)blk.sz; ++i) {
+					if ('\n' == blk.buf[i]) {
+						++i;
+						++lnn;
+						break;
+					}
+				}
+				/* Backout trailing whitespaces */
+				for (; pos > 0; --pos) {
+					if (ln.buf[pos - 1] != ' ')
+						break;
+					if (pos > 2 && ln.buf[pos - 2] == '\\')
+						break;
+				}
+				break;
+			}
+			/* Some other escape sequence, copy and continue. */
+			if (pos + 1 >= (int)ln.sz)
+				if (! resize_buf(&ln, 256))
+					goto bailout;
+
+			ln.buf[pos++] = blk.buf[i++];
+			ln.buf[pos++] = blk.buf[i++];
 		}
 
-		ln.buf[pos] = 0;
-		lnn++;
+ 		if (pos >= (int)ln.sz)
+			if (! resize_buf(&ln, 256))
+				goto bailout;
+		ln.buf[pos] = '\0';
 
+		re = roff_parseln(roff, lnn_start, &ln.buf, &ln.sz);
+		if (ROFF_IGN == re)
+			continue;
+		else if (ROFF_ERR == re)
+			goto bailout;
+
 		/* If unset, assign parser in pset(). */
 
 		if ( ! (man || mdoc) && ! pset(ln.buf, pos, curp, &man, &mdoc))
 			goto bailout;
 
-		pos = comment = 0;
-
 		/* Pass down into parsers. */
 
-		if (man && ! man_parseln(man, lnn, ln.buf))
+		if (man && ! man_parseln(man, lnn_start, ln.buf))
 			goto bailout;
-		if (mdoc && ! mdoc_parseln(mdoc, lnn, ln.buf))
+		if (mdoc && ! mdoc_parseln(mdoc, lnn_start, ln.buf))
 			goto bailout;
 	}
 
@@ -460,6 +497,8 @@ fdesc(struct curparse *curp)
 		goto bailout;
 	if (man && ! man_endparse(man))
 		goto bailout;
+	if (roff && ! roff_endparse(roff))
+		goto bailout;
 
 	/* If unset, allocate output dev now (if applicable). */
 
@@ -484,7 +523,7 @@ fdesc(struct curparse *curp)
 		case (OUTT_LINT):
 			break;
 		default:
-			curp->outdata = ascii_alloc();
+			curp->outdata = ascii_alloc(80);
 			curp->outman = terminal_man;
 			curp->outmdoc = terminal_mdoc;
 			curp->outfree = terminal_free;
@@ -500,20 +539,19 @@ fdesc(struct curparse *curp)
 		(*curp->outmdoc)(curp->outdata, mdoc);
 
  cleanup:
-	if (curp->mdoc) {
-		mdoc_free(curp->mdoc);
-		curp->mdoc = NULL;
-	}
-	if (curp->man) {
-		man_free(curp->man);
-		curp->man = NULL;
-	}
+	if (mdoc)
+		mdoc_reset(mdoc);
+	if (man)
+		man_reset(man);
+	if (roff)
+		roff_reset(roff);
 	if (ln.buf)
 		free(ln.buf);
 	if (with_mmap)
 		munmap(blk.buf, blk.sz);
 	else
 		free(blk.buf);
+
 	return;
 
  bailout:
@@ -536,7 +574,7 @@ pset(const char *buf, int pos, struct curparse *curp,
 	 * default to -man, which is more lenient.
 	 */
 
-	if (buf[0] == '.') {
+	if ('.' == buf[0] || '\'' == buf[0]) {
 		for (i = 1; buf[i]; i++)
 			if (' ' != buf[i] && '\t' != buf[i])
 				break;
@@ -735,3 +773,34 @@ mwarn(void *arg, int line, int col, const char *msg)
 	return(1);
 }
 
+static	const char * const	mandocerrs[MANDOCERR_MAX] = {
+	"ok",
+	"multi-line scope open on exit",
+	"request for scope closure when no matching scope is open",
+	"line arguments will be lost",
+	"memory exhausted"
+};
+
+/*
+ * XXX: this is experimental code that will eventually become the
+ * generic means of covering all warnings and errors!
+ */
+/* ARGSUSED */
+static int
+mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
+{
+#if 0
+	struct curparse *cp;
+
+	cp = (struct curparse *)arg;
+
+	fprintf(stderr, "%s:%d:%d: %s", cp->file, 
+			ln, col + 1, mandocerrs[t]);
+
+	if (msg)
+		fprintf(stderr, ": %s", msg);
+
+	fputc('\n', stderr);
+#endif
+	return(1);
+}