Make the stored "cat"/"mdoc"/"man" strings just be c/d/a single-character

bytes.  This cuts down a little in index size and allows for cleaner
extraction of information.

6 files changed, 56 insertions, 44 deletions

diff --git a/apropos_db.c b/apropos_db.c
index fd657558..0d2b0728 100644
--- a/apropos_db.c
+++ b/apropos_db.c
@@ -1,4 +1,4 @@
-/*	$Id: apropos_db.c,v 1.23 2011/12/10 21:46:59 kristaps Exp $ */
+/*	$Id: apropos_db.c,v 1.24 2011/12/16 12:06:35 kristaps Exp $ */
 /*
  * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org>
@@ -365,6 +365,7 @@ index_read(const DBT *key, const DBT *val, int index,
 {
 	size_t		 left;
 	char		*np, *cp;
+	char		 type;
 
 #define	INDEX_BREAD(_dst) \
 	do { \
@@ -375,13 +376,23 @@ index_read(const DBT *key, const DBT *val, int index,
 		cp = np + 1; \
 	} while (/* CONSTCOND */ 0)
 
-	left = val->size;
-	cp = (char *)val->data;
+	if (0 == (left = val->size))
+		return(0);
 
+	cp = val->data;
 	rec->res.rec = *(recno_t *)key->data;
 	rec->res.volume = index;
 
-	INDEX_BREAD(rec->res.type);
+	if ('d' == (type = *cp++))
+		rec->res.type = RESTYPE_MDOC;
+	else if ('a' == type)
+		rec->res.type = RESTYPE_MAN;
+	else if ('c' == type)
+		rec->res.type = RESTYPE_CAT;
+	else
+		return(0);
+
+	left--;
 	INDEX_BREAD(rec->res.file);
 	INDEX_BREAD(rec->res.cat);
 	INDEX_BREAD(rec->res.title);
@@ -581,7 +592,6 @@ static void
 recfree(struct rec *rec)
 {
 
-	free(rec->res.type);
 	free(rec->res.file);
 	free(rec->res.cat);
 	free(rec->res.title);
diff --git a/apropos_db.h b/apropos_db.h
index be48b063..2af91206 100644
--- a/apropos_db.h
+++ b/apropos_db.h
@@ -1,4 +1,4 @@
-/*	$Id: apropos_db.h,v 1.10 2011/11/27 23:11:37 schwarze Exp $ */
+/*	$Id: apropos_db.h,v 1.11 2011/12/16 12:06:35 kristaps Exp $ */
 /*
  * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -17,8 +17,14 @@
 #ifndef APROPOS_H
 #define APROPOS_H
 
+enum	restype {
+	RESTYPE_MAN, /* man(7) file */
+	RESTYPE_MDOC, /* mdoc(7) file */
+	RESTYPE_CAT /* pre-formatted file */
+};
+
 struct	res {
-	char		*type; /* file type: mdoc, man or cat */
+	enum restype	 type; /* input file type */
 	char		*file; /* file in file-system */
 	char		*cat; /* category (3p, 3, etc.) */
 	char		*title; /* title (FOO, etc.) */
diff --git a/catman.c b/catman.c
index 54971293..2ac27e5a 100644
--- a/catman.c
+++ b/catman.c
@@ -1,4 +1,4 @@
-/*	$Id: catman.c,v 1.6 2011/12/16 08:04:34 kristaps Exp $ */
+/*	$Id: catman.c,v 1.7 2011/12/16 12:06:35 kristaps Exp $ */
 /*
  * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -204,7 +204,7 @@ indexhtml(char *src, size_t ssz, char *dst, size_t dsz)
 	DBT		 key, val;
 	int		 c, rc;
 	unsigned int	 fl;
-	const char	*f, *cp;
+	const char	*f;
 	char		*d;
 	char		 fname[MAXPATHLEN];
 	pid_t		 pid;
@@ -223,14 +223,15 @@ indexhtml(char *src, size_t ssz, char *dst, size_t dsz)
 	fl = R_FIRST;
 	while (0 == (c = (*idx->seq)(idx, &key, &val, fl))) {
 		fl = R_NEXT;
-		cp = (const char *)val.data;
+		/*
+		 * If the record is zero-length, then it's unassigned.
+		 * Skip past these.
+		 */
 		if (0 == val.size)
 			continue;
-		if (NULL == (f = memchr(cp, '\0', val.size)))
-			break;
-		if (++f - cp >= (int)val.size)
-			break;
-		if (NULL == memchr(f, '\0', val.size - (f - cp)))
+
+		f = (const char *)val.data + 1;
+		if (NULL == memchr(f, '\0', val.size - 1))
 			break;
 
 		src[(int)ssz] = dst[(int)dsz] = '\0';
diff --git a/cgi.c b/cgi.c
index 002855d7..515b1162 100644
--- a/cgi.c
+++ b/cgi.c
@@ -1,4 +1,4 @@
-/*	$Id: cgi.c,v 1.34 2011/12/16 08:04:34 kristaps Exp $ */
+/*	$Id: cgi.c,v 1.35 2011/12/16 12:06:35 kristaps Exp $ */
 /*
  * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -752,8 +752,8 @@ pg_show(const struct req *req, char *path)
 	size_t		 sz;
 	char		*sub;
 	char		 file[MAXPATHLEN];
-	const char	*fn, *cp;
-	int		 rc;
+	const char	*cp;
+	int		 rc, catm;
 	unsigned int	 vol, rec, mr;
 	DB		*idx;
 	DBT		 key, val;
@@ -824,21 +824,21 @@ pg_show(const struct req *req, char *path)
 	if (0 != (rc = (*idx->get)(idx, &key, &val, 0))) {
 		rc < 0 ? resp_baddb() : resp_error400();
 		goto out;
-	} 
+	} else if (0 == val.size) {
+		resp_baddb();
+		goto out;
+	}
 
 	cp = (char *)val.data;
+	catm = 'c' == *cp++;
 
-	if (NULL == (fn = memchr(cp, '\0', val.size)))
-		resp_baddb();
-	else if (++fn - cp >= (int)val.size)
-		resp_baddb();
-	else if (NULL == memchr(fn, '\0', val.size - (fn - cp)))
+	if (NULL == memchr(cp, '\0', val.size - 1)) 
 		resp_baddb();
 	else {
  		file[(int)sz] = '\0';
  		strlcat(file, "/", MAXPATHLEN);
- 		strlcat(file, fn, MAXPATHLEN);
-		if (0 == strcmp(cp, "cat"))
+ 		strlcat(file, cp, MAXPATHLEN);
+		if (catm) 
 			catman(req, file);
 		else
 			format(req, file);
diff --git a/mandocdb.8 b/mandocdb.8
index a19b0083..38b928fd 100644
--- a/mandocdb.8
+++ b/mandocdb.8
@@ -1,4 +1,4 @@
-.\"	$Id: mandocdb.8,v 1.13 2011/12/16 08:04:34 kristaps Exp $
+.\"	$Id: mandocdb.8,v 1.14 2011/12/16 12:06:35 kristaps Exp $
 .\"
 .\" Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
 .\"
@@ -110,14 +110,13 @@ database with record values consisting of
 .Pp
 .Bl -enum -compact
 .It
-the string
-.Cm mdoc ,
-.Cm man ,
+the character
+.Cm d ,
+.Cm a ,
 or
-.Cm cat
+.Cm c
 to indicate the file type
 .Po
-file in
 .Xr mdoc 7 ,
 .Xr man 7 ,
 and post-formatted, respectively
@@ -137,9 +136,7 @@ and the description.
 .Pp
 Each of the above is NUL-terminated.
 .Pp
-Both the manual section and description may be zero-length if the record
-is unassigned.
-Entries are sequentially-numbered, but the filenames are unordered.
+If the record value is zero-length, it is unassigned.
 .Ss Keyword Database
 The keyword database,
 .Pa mandoc.db ,
diff --git a/mandocdb.c b/mandocdb.c
index d9acf3e8..24758a0e 100644
--- a/mandocdb.c
+++ b/mandocdb.c
@@ -1,4 +1,4 @@
-/*	$Id: mandocdb.c,v 1.35 2011/12/16 08:04:34 kristaps Exp $ */
+/*	$Id: mandocdb.c,v 1.36 2011/12/16 12:06:35 kristaps Exp $ */
 /*
  * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org>
@@ -528,6 +528,7 @@ index_merge(const struct of *of, struct mparse *mp,
 	size_t		 sv;
 	unsigned	 seq;
 	struct db_val	 vbuf;
+	char		 type;
 
 	for (rec = 0; of; of = of->next) {
 		fn = of->fname;
@@ -608,7 +609,8 @@ index_merge(const struct of *of, struct mparse *mp,
 		 */
 
 		dbuf->len = 0;
-		buf_append(dbuf, mdoc ? "mdoc" : (man ? "man" : "cat"));
+		type = mdoc ? 'd' : (man ? 'a' : 'c');
+		buf_appendb(dbuf, &type, 1);
 		buf_appendb(dbuf, fn, strlen(fn) + 1);
 		buf_appendb(dbuf, msec, strlen(msec) + 1);
 		buf_appendb(dbuf, mtitle, strlen(mtitle) + 1);
@@ -696,7 +698,7 @@ index_prune(const struct of *ofile, DB *db, const char *dbf,
 		recno_t **recs, size_t *recsz, size_t *reccur)
 {
 	const struct of	*of;
-	const char	*fn, *cp;
+	const char	*fn;
 	struct db_val	*vbuf;
 	unsigned	 seq, sseq;
 	DBT		 key, val;
@@ -707,7 +709,6 @@ index_prune(const struct of *ofile, DB *db, const char *dbf,
 	while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) {
 		seq = R_NEXT;
 		*maxrec = *(recno_t *)key.data;
-		cp = val.data;
 
 		/* Deleted records are zero-sized.  Skip them. */
 
@@ -721,11 +722,8 @@ index_prune(const struct of *ofile, DB *db, const char *dbf,
 		 * Failing any of these, we go into our error handler.
 		 */
 
-		if (NULL == (fn = memchr(cp, '\0', val.size)))
-			break;
-		if (++fn - cp >= (int)val.size)
-			break;
-		if (NULL == memchr(fn, '\0', val.size - (fn - cp)))
+		fn = (char *)val.data + 1;
+		if (NULL == memchr(fn, '\0', val.size - 1))
 			break;
 
 		/*