In theory, Kristaps never intended to write a roff parser,

but in practice, most real legacy man(7)uals are using so much
low level roff that we can't really get away without at least
partially handling some roff instructions.

As doing this in man(7) only has become messy and as even some
mdoc(7) pages need it, start a minimal partial roff preprocessor.
As a first step, move handling of .am[i], .de[i] and .ig there.
Do not use the roff preprocessor for new manuals!

Now that we have three main parser libraries - roff, man and mdoc -
each one having its own error handling is becoming messy, too.
Thus, start unifying message handling in one central place,
introducing a new generic function mmsg().

coded by kristaps@

1 files changed, 85 insertions, 18 deletions

diff --git a/usr.bin/mandoc/main.c b/usr.bin/mandoc/main.c
index 63b3019e083..1ddb5844941 100644
--- a/usr.bin/mandoc/main.c
+++ b/usr.bin/mandoc/main.c
@@ -1,4 +1,4 @@
-/*	$Id: main.c,v 1.29 2010/05/15 22:00:22 schwarze Exp $ */
+/*	$Id: main.c,v 1.30 2010/05/16 00:54:03 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -26,8 +26,10 @@
 #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))
@@ -59,6 +61,7 @@ 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;
@@ -71,6 +74,7 @@ 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;
@@ -84,9 +88,12 @@ 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 *);
@@ -166,6 +173,12 @@ 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);
@@ -214,6 +227,14 @@ 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)
 {
@@ -305,7 +326,7 @@ read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
 			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)
@@ -331,7 +352,7 @@ read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
 			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);
@@ -340,7 +361,7 @@ read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
 			perror(curp->file);
 			break;
 		}
-		off += ssz;
+		off += (size_t)ssz;
 	}
 
 	free(fb->buf);
@@ -355,11 +376,14 @@ fdesc(struct curparse *curp)
 {
 	struct buf	 ln, blk;
 	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));
 
 	/*
@@ -367,9 +391,14 @@ fdesc(struct curparse *curp)
 	 * 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;
 
+	if (NULL == curp->roff) 
+		curp->roff = roff_init(curp);
+	if (NULL == (roff = curp->roff))
+		goto bailout;
+
 	for (i = 0, lnn = 1; i < (int)blk.sz;) {
 		pos = 0;
 		lnn_start = lnn;
@@ -425,7 +454,13 @@ fdesc(struct curparse *curp)
  		if (pos >= (int)ln.sz)
 			if (! resize_buf(&ln, 256))
 				goto bailout;
-		ln.buf[pos] = 0;
+		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(). */
 
@@ -434,9 +469,9 @@ fdesc(struct curparse *curp)
 
 		/* 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;
 	}
 
@@ -451,6 +486,8 @@ 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). */
 
@@ -491,20 +528,19 @@ 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:
@@ -726,3 +762,34 @@ 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);
+}