Refactor, no functional change: Remove the parse point from struct buf.

Some functions need multiple parse points, some none at all,
and it varies whether any of them need to be passed around.
So better pass them as a separate argument, and only when needed.

3 files changed, 48 insertions, 53 deletions

diff --git a/libmandoc.h b/libmandoc.h
index c01840b1..b4eacb75 100644
--- a/libmandoc.h
+++ b/libmandoc.h
@@ -1,4 +1,4 @@
-/*	$Id: libmandoc.h,v 1.46 2014/10/28 17:36:19 schwarze Exp $ */
+/*	$Id: libmandoc.h,v 1.47 2014/11/01 04:08:43 schwarze Exp $ */
 /*
  * Copyright (c) 2009, 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
@@ -33,7 +33,6 @@ enum	rofferr {
 struct	buf {
 	char	*buf;
 	size_t	 sz;
-	size_t	 offs;
 };
 
 __BEGIN_DECLS
@@ -72,8 +71,9 @@ int		 man_endparse(struct man *);
 int		 man_addspan(struct man *, const struct tbl_span *);
 int		 man_addeqn(struct man *, const struct eqn *);
 
-int		 preconv_cue(const struct buf *);
-int		 preconv_encode(struct buf *, struct buf *, int *);
+int		 preconv_cue(const struct buf *, size_t);
+int		 preconv_encode(struct buf *, size_t *,
+			struct buf *, size_t *, int *);
 
 void		 roff_free(struct roff *);
 struct roff	*roff_alloc(struct mparse *, const struct mchars *, int);
diff --git a/preconv.c b/preconv.c
index 3213c8ad..780eb70b 100644
--- a/preconv.c
+++ b/preconv.c
@@ -1,4 +1,4 @@
-/*	$Id: preconv.c,v 1.10 2014/10/26 18:22:51 schwarze Exp $ */
+/*	$Id: preconv.c,v 1.11 2014/11/01 04:08:43 schwarze Exp $ */
 /*
  * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2014 Ingo Schwarze <schwarze@openbsd.org>
@@ -25,7 +25,8 @@
 #include "libmandoc.h"
 
 int
-preconv_encode(struct buf *ib, struct buf *ob, int *filenc)
+preconv_encode(struct buf *ib, size_t *ii, struct buf *ob, size_t *oi,
+    int *filenc)
 {
 	size_t		 i;
 	const long	 one = 1L;
@@ -45,7 +46,7 @@ preconv_encode(struct buf *ib, struct buf *ob, int *filenc)
 	if ( ! (*((const char *)(&one))))
 		be = 1;
 
-	for (i = ib->offs; i < ib->sz; i++) {
+	for (i = *ii; i < ib->sz; i++) {
 		cu = ib->buf[i];
 		if (state) {
 			if ( ! (cu & 128) || (cu & 64)) {
@@ -79,11 +80,11 @@ preconv_encode(struct buf *ib, struct buf *ob, int *filenc)
 					(accum << 24);
 
 			if (accum < 0x80)
-				ob->buf[ob->offs++] = accum;
+				ob->buf[(*oi)++] = accum;
 			else
-				ob->offs += snprintf(ob->buf + ob->offs,
+				*oi += snprintf(ob->buf + *oi,
 				    11, "\\[u%.4X]", accum);
-			ib->offs = i + 1;
+			*ii = i + 1;
 			*filenc &= ~MPARSE_LATIN1;
 			return(1);
 		} else {
@@ -134,21 +135,21 @@ latin:
 	if ( ! (*filenc & MPARSE_LATIN1))
 		return(0);
 
-	ob->offs += snprintf(ob->buf + ob->offs, 11,
-	    "\\[u%.4X]", (unsigned char)ib->buf[ib->offs++]);
+	*oi += snprintf(ob->buf + *oi, 11,
+	    "\\[u%.4X]", (unsigned char)ib->buf[(*ii)++]);
 
 	*filenc &= ~MPARSE_UTF8;
 	return(1);
 }
 
 int
-preconv_cue(const struct buf *b)
+preconv_cue(const struct buf *b, size_t offset)
 {
 	const char	*ln, *eoln, *eoph;
 	size_t		 sz, phsz;
 
-	ln = b->buf + b->offs;
-	sz = b->sz - b->offs;
+	ln = b->buf + offset;
+	sz = b->sz - offset;
 
 	/* Look for the end-of-line. */
 
diff --git a/read.c b/read.c
index 6c558817..22070231 100644
--- a/read.c
+++ b/read.c
@@ -1,4 +1,4 @@
-/*	$Id: read.c,v 1.94 2014/10/28 17:36:19 schwarze Exp $ */
+/*	$Id: read.c,v 1.95 2014/11/01 04:08:43 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
@@ -68,7 +68,7 @@ struct	mparse {
 
 static	void	  choose_parser(struct mparse *);
 static	void	  resize_buf(struct buf *, size_t);
-static	void	  mparse_buf_r(struct mparse *, struct buf, int);
+static	void	  mparse_buf_r(struct mparse *, struct buf, size_t, int);
 static	int	  read_whole_file(struct mparse *, const char *, int,
 				struct buf *, int *);
 static	void	  mparse_end(struct mparse *);
@@ -303,27 +303,29 @@ choose_parser(struct mparse *curp)
 }
 
 /*
- * Main parse routine for an opened file.  This is called for each
- * opened file and simply loops around the full input file, possibly
- * nesting (i.e., with `so').
+ * Main parse routine for a buffer.
+ * It assumes encoding and line numbering are already set up.
+ * It can recurse directly (for invocations of user-defined
+ * macros, inline equations, and input line traps)
+ * and indirectly (for .so file inclusion).
  */
 static void
-mparse_buf_r(struct mparse *curp, struct buf blk, int start)
+mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
 {
 	const struct tbl_span	*span;
 	struct buf	 ln;
+	size_t		 pos; /* byte number in the ln buffer */
 	enum rofferr	 rr;
-	int		 i, of, rc;
-	int		 pos; /* byte number in the ln buffer */
+	int		 of, rc;
 	int		 lnn; /* line number in the real file */
 	unsigned char	 c;
 
-	memset(&ln, 0, sizeof(struct buf));
+	memset(&ln, 0, sizeof(ln));
 
 	lnn = curp->line;
 	pos = 0;
 
-	for (i = blk.offs; i < (int)blk.sz; ) {
+	while (i < blk.sz) {
 		if (0 == pos && '\0' == blk.buf[i])
 			break;
 
@@ -333,13 +335,11 @@ mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 
 			if (lnn < 3 &&
 			    curp->filenc & MPARSE_UTF8 &&
-			    curp->filenc & MPARSE_LATIN1) {
-				blk.offs = i;
-				curp->filenc = preconv_cue(&blk);
-			}
+			    curp->filenc & MPARSE_LATIN1)
+				curp->filenc = preconv_cue(&blk, i);
 		}
 
-		while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
+		while (i < blk.sz && (start || blk.buf[i] != '\0')) {
 
 			/*
 			 * When finding an unescaped newline character,
@@ -347,7 +347,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 			 * Skip a preceding carriage return, if any.
 			 */
 
-			if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
+			if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
 			    '\n' == blk.buf[i + 1])
 				++i;
 			if ('\n' == blk.buf[i]) {
@@ -361,7 +361,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 			 * case of 11 bytes: "\\[u10ffff]\0"
 			 */
 
-			if (pos + 11 > (int)ln.sz)
+			if (pos + 11 > ln.sz)
 				resize_buf(&ln, 256);
 
 			/*
@@ -370,13 +370,8 @@ mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 
 			c = blk.buf[i];
 			if (c & 0x80) {
-				blk.offs = i;
-				ln.offs = pos;
-				if (curp->filenc && preconv_encode(
-				    &blk, &ln, &curp->filenc)) {
-					pos = ln.offs;
-					i = blk.offs;
-				} else {
+				if ( ! (curp->filenc && preconv_encode(
+				    &blk, &i, &ln, &pos, &curp->filenc))) {
 					mandoc_vmsg(MANDOCERR_BADCHAR,
 					    curp, curp->line, pos,
 					    "0x%x", c);
@@ -400,7 +395,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 
 			/* Trailing backslash = a plain char. */
 
-			if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
+			if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
 				ln.buf[pos++] = blk.buf[i++];
 				continue;
 			}
@@ -412,7 +407,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 			 * skip that one as well.
 			 */
 
-			if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
+			if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
 			    '\n' == blk.buf[i + 2])
 				++i;
 			if ('\n' == blk.buf[i + 1]) {
@@ -424,7 +419,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 			if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
 				i += 2;
 				/* Comment, skip to end of line */
-				for (; i < (int)blk.sz; ++i) {
+				for (; i < blk.sz; ++i) {
 					if ('\n' == blk.buf[i]) {
 						++i;
 						++lnn;
@@ -461,7 +456,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 			ln.buf[pos++] = blk.buf[i++];
 		}
 
-		if (pos >= (int)ln.sz)
+		if (pos >= ln.sz)
 			resize_buf(&ln, 256);
 
 		ln.buf[pos] = '\0';
@@ -504,14 +499,14 @@ rerun:
 		switch (rr) {
 		case ROFF_REPARSE:
 			if (REPARSE_LIMIT >= ++curp->reparse_count)
-				mparse_buf_r(curp, ln, 0);
+				mparse_buf_r(curp, ln, of, 0);
 			else
 				mandoc_msg(MANDOCERR_ROFFLOOP, curp,
 				    curp->line, pos, NULL);
 			pos = 0;
 			continue;
 		case ROFF_APPEND:
-			pos = (int)strlen(ln.buf);
+			pos = strlen(ln.buf);
 			continue;
 		case ROFF_RERUN:
 			goto rerun;
@@ -522,8 +517,8 @@ rerun:
 			assert(MANDOCLEVEL_FATAL <= curp->file_status);
 			break;
 		case ROFF_SO:
-			if (0 == (MPARSE_SO & curp->options) &&
-			    (i >= (int)blk.sz || '\0' == blk.buf[i])) {
+			if ( ! (curp->options & MPARSE_SO) &&
+			    (i >= blk.sz || blk.buf[i] == '\0')) {
 				curp->sodest = mandoc_strdup(ln.buf + of);
 				free(ln.buf);
 				return;
@@ -650,7 +645,6 @@ read_whole_file(struct mparse *curp, const char *file, int fd,
 			return(0);
 		}
 		*with_mmap = 1;
-		fb->offs = 0;
 		fb->sz = (size_t)st.st_size;
 		fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
 		if (fb->buf != MAP_FAILED)
@@ -682,7 +676,6 @@ read_whole_file(struct mparse *curp, const char *file, int fd,
 		ssz = read(fd, fb->buf + (int)off, fb->sz - off);
 		if (ssz == 0) {
 			fb->sz = off;
-			fb->offs = 0;
 			return(1);
 		}
 		if (ssz == -1) {
@@ -739,6 +732,7 @@ mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
 {
 	struct buf	*svprimary;
 	const char	*svfile;
+	size_t		 offset;
 	static int	 recursion_depth;
 
 	if (64 < recursion_depth) {
@@ -759,11 +753,12 @@ mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
 	    (unsigned char)blk.buf[0] == 0xef &&
 	    (unsigned char)blk.buf[1] == 0xbb &&
 	    (unsigned char)blk.buf[2] == 0xbf) {
-		blk.offs = 3;
+		offset = 3;
 		curp->filenc &= ~MPARSE_LATIN1;
-	}
+	} else
+		offset = 0;
 
-	mparse_buf_r(curp, blk, 1);
+	mparse_buf_r(curp, blk, offset, 1);
 
 	if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
 		mparse_end(curp);
@@ -780,7 +775,6 @@ mparse_readmem(struct mparse *curp, const void *buf, size_t len,
 
 	blk.buf = UNCONST(buf);
 	blk.sz = len;
-	blk.offs = 0;
 
 	mparse_parse_buffer(curp, blk, file);
 	return(curp->file_status);