===================================================================
RCS file: /cvs/texi2mdoc/util.c,v
retrieving revision 1.29
retrieving revision 1.30
diff -u -p -r1.29 -r1.30
--- texi2mdoc/util.c	2015/03/07 11:53:21	1.29
+++ texi2mdoc/util.c	2015/03/11 12:51:41	1.30
@@ -1,4 +1,4 @@
-/*	$Id: util.c,v 1.29 2015/03/07 11:53:21 kristaps Exp $ */
+/*	$Id: util.c,v 1.30 2015/03/11 12:51:41 kristaps Exp $ */
 /*
  * Copyright (c) 2015 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -103,7 +103,121 @@ texivaluefree(struct texivalue *p)
 	free(p->value);
 }
 
+static void
+texidex_free(struct texidex *p)
+{
+	size_t	 i;
+
+	for (i = 0; i < p->indexsz; i++)
+		free(p->index[i]);
+
+	free(p->index);
+	free(p->name);
+	p->index = NULL;
+	p->indexsz = 0;
+}
+
 /*
+ * Add the text beginning at "index" and of "sz" bytes to the index
+ * named "tok" with name size "toksz".
+ * This will also output the necessary mdoc(7) to construct the index.
+ */
+void
+texindex(struct texi *p, const char *tok,
+	size_t toksz, const char *index, size_t sz)
+{
+	size_t	 i;
+#ifdef HAVE_INDEX
+	char	*cp;
+#endif
+
+	if (0 == sz) {
+		texiwarn(p, "zero-length index entry");
+		return;
+	}
+
+	/* Look for the index.  (Must be found.) */
+	for (i = 0; i < p->indexsz; i++) {
+		if (strlen(p->indexs[i].name) != toksz)
+			continue;
+		if (strncmp(p->indexs[i].name, tok, toksz))
+			continue;
+		break;
+	}
+
+	assert(i < p->indexsz);
+	/* Reallocate index's terms. */
+	p->indexs[i].index = realloc
+		(p->indexs[i].index,
+		 (p->indexs[i].indexsz + 1) *
+		 sizeof(char *));
+	if (NULL == p->indexs[i].index)
+		texiabort(p, NULL);
+
+	/* Add term to term array. */
+	p->indexs[i].index[p->indexs[i].indexsz] = 
+		malloc(sz + 1);
+	if (NULL == p->indexs[i].index[p->indexs[i].indexsz])
+		texiabort(p, NULL);
+	memcpy(p->indexs[i].index[p->indexs[i].indexsz],
+		index, sz);
+	p->indexs[i].index[p->indexs[i].indexsz][sz] = '\0';
+	
+	/* Output mdoc(7) for index. */
+#ifdef HAVE_INDEX
+	p->seenvs = -1;
+	teximacroopen(p, "Ix");
+	texiputchars(p, "idx");
+	texiputchars(p, p->indexs[i].name);
+	cp = p->indexs[i].index[p->indexs[i].indexsz];
+	while ('\n' != *cp) {
+		assert('\0' != *cp);
+		texiputchar(p, *cp++);
+	}
+	teximacroclose(p);
+#endif
+	p->indexs[i].indexsz++;
+}
+
+/*
+ * Add an index entry named "tok" of length "sz".
+ * This usually consists of two letters, e.g., "cp" or "vr".
+ * This does nothing if the index exists or is zero-sized.
+ */
+void
+texindex_add(struct texi *p, const char *tok, size_t sz)
+{
+	size_t	 i;
+	char	*cp;
+
+	if (0 == sz)
+		return;
+
+	/* Make sure we don't have a duplicate. */
+	for (i = 0; i < p->indexsz; i++) {
+		if (strlen(p->indexs[i].name) != sz)
+			continue;
+		if (strncmp(p->indexs[i].name, tok, sz))
+			continue;
+		return;
+	}
+
+	/* Reallocate indices. */
+	p->indexs = realloc(p->indexs, 
+		sizeof(struct texidex) * (p->indexsz + 1));
+	if (NULL == p->indexs)
+		texiabort(p, NULL);
+	if (NULL == (cp = malloc(sz + 1)))
+		texiabort(p, NULL);
+	memcpy(cp, tok, sz);
+	cp[sz] = '\0';
+	p->indexs[p->indexsz].name = cp;
+	p->indexs[p->indexsz].index = NULL;
+	p->indexs[p->indexsz].indexsz = 0;
+	p->indexsz++;
+}
+
+/*
  * Unmap all files that we're currently using and free all resources
  * that we've allocated during the parse.
  * The utility should exit(...) after this is called.
@@ -128,7 +242,7 @@ texiexit(struct texi *p)
 	for (i = 0; i < p->dirsz; i++)
 		free(p->dirs[i]);
 	for (i = 0; i < p->indexsz; i++)
-		free(p->indexs[i]);
+		texidex_free(&p->indexs[i]);
 	for (i = 0; i < p->valsz; i++) 
 		texivaluefree(&p->vals[i]);
 
@@ -280,14 +394,13 @@ void
 teximacroclose(struct texi *p)
 {
 
-	if (p->ign)
+	if (p->ign || p->literal|| TEXILIST_TABLE == p->list)
 		return;
 
 	if (0 == --p->outmacro) {
 		fputc('\n', p->outfile);
 		p->outcol = p->seenws = 0;
 	}
-	p->seenvs = 0;
 }
 
 /*
@@ -301,7 +414,7 @@ teximacroopen(struct texi *p, const char *s)
 {
 	int	 rc;
 
-	if (p->ign)
+	if (p->ign || p->literal|| TEXILIST_TABLE == p->list)
 		return;
 
 	if (p->outcol && 0 == p->outmacro) {
@@ -309,6 +422,9 @@ teximacroopen(struct texi *p, const char *s)
 		p->outcol = 0;
 	}
 
+	if (p->seenvs > 0 && 0 == p->outmacro)
+		fputs(".Pp\n", p->outfile);
+
 	if (0 == p->outmacro)
 		fputc('.', p->outfile);
 	else
@@ -320,8 +436,7 @@ teximacroopen(struct texi *p, const char *s)
 	fputc(' ', p->outfile);
 	p->outcol++;
 	p->outmacro++;
-	p->seenws = 0;
-	p->seenvs = 0;
+	p->seenws = p->seenvs = 0;
 }
 
 /*
@@ -338,15 +453,15 @@ teximacro(struct texi *p, const char *s)
 		texierr(p, "\"%s\" in open line scope!?", s);
 	if (p->literal)
 		texierr(p, "\"%s\" in a literal scope!?", s);
-
 	if (p->outcol)
 		fputc('\n', p->outfile);
+	if (p->seenvs > 0)
+		fputs(".Pp\n", p->outfile);
 
 	fputc('.', p->outfile);
 	fputs(s, p->outfile);
 	fputc('\n', p->outfile);
 	p->outcol = p->seenws = 0;
-	p->seenvs = 0;
 }
 
 /*
@@ -356,8 +471,8 @@ void
 texivspace(struct texi *p)
 {
 
-	if (TEXILIST_TABLE != p->list)
-		teximacro(p, "Pp");
+	if (TEXILIST_TABLE != p->list && p->seenvs >= 0)
+		p->seenvs = 1;
 }
 
 /*
@@ -420,7 +535,7 @@ texipunctuate(struct texi *p, size_t *pos)
 	if (end == *pos)
 		return;
 	if (end + 1 == BUFSZ(p) || ' ' == BUF(p)[end] || 
-		'\n' == BUF(p)[end]) {
+		'@' == BUF(p)[end] || '\n' == BUF(p)[end]) {
 		for ( ; start < end; start++) {
 			texiputchar(p, ' ');
 			texiputchar(p, BUF(p)[start]);
@@ -448,6 +563,9 @@ advancenext(struct texi *p, size_t *pos)
 
 	while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos])) {
 		p->seenws = 1;
+		if (0 == p->seenvs && '\n' == BUF(p)[*pos])
+			if (*pos + 1 < BUFSZ(p) && '\n' == BUF(p)[*pos + 1])
+				p->seenvs = 1;
 		advance(p, pos);
 	}
 	return(*pos);
@@ -608,11 +726,13 @@ parseword(struct texi *p, size_t *pos, char extra)
 	 * We don't do this if we're in a literal context (we'll print
 	 * out the newlines themselves) nor in a `TS' table.
 	 */
-	if (p->seenvs && 0 == p->literal && TEXILIST_TABLE != p->list)
-		teximacro(p, "Pp");
+	if (p->seenvs > 0 && 0 == p->literal && TEXILIST_TABLE != p->list) {
+		if (p->outcol > 0)
+			fputc('\n', p->outfile);
+		fputs(".Pp\n", p->outfile);
+		p->outcol = 0;
+	}
 
-	p->seenvs = 0;
-
 	/*
 	 * Some line control: if we (non-macro, non-literal) already
 	 * have more than 72 characters written to the screen, then
@@ -684,7 +804,9 @@ parseword(struct texi *p, size_t *pos, char extra)
 			continue;
 		}
 
-		if (*pos < BUFSZ(p) - 2 && 
+		if ('"' == BUF(p)[*pos]) {
+			texiputchars(p, "\\(dq");
+		} else if (*pos < BUFSZ(p) - 2 && 
 			 '-' == BUF(p)[*pos] && 
 			 '-' == BUF(p)[*pos + 1] && 
 			 '-' == BUF(p)[*pos + 2]) {
@@ -712,19 +834,23 @@ parseword(struct texi *p, size_t *pos, char extra)
 		advance(p, pos);
 	}
 
-	if (*pos + 1 < BUFSZ(p) && 
-		'\n' == BUF(p)[*pos] &&
-		'\n' == BUF(p)[*pos + 1])
-		p->seenvs = 1;
-
 	/* 
 	 * New sentence, new line:if we (non-macro, non-literal) see a
 	 * period at the end of the last printed word, then open a
 	 * newline.
 	 */
-	if (0 == p->literal && 0 == p->outmacro && 
-		*pos < BUFSZ(p) && '.' == BUF(p)[*pos - 1])
-		texiputchar(p, '\n');
+	if (0 == p->literal && 0 == p->outmacro && *pos < BUFSZ(p))
+		switch (BUF(p)[*pos - 1]) {
+		case ('.'):
+		case ('!'):
+		case ('?'):
+			texiputchar(p, '\n');
+			break;
+		default:
+			break;
+		}
+
+	p->seenvs = 0;
 }
 
 /*
@@ -778,10 +904,10 @@ texicmd(const struct texi *p, size_t pos, size_t *end,
 
 	/* Look for it in our indices. */
 	for (i = 0; i < p->indexsz; i++) {
-		toksz = strlen(p->indexs[i]);
+		toksz = strlen(p->indexs[i].name);
 		if (len != 5 + toksz)
 			continue;
-		if (strncmp(&BUF(p)[pos], p->indexs[i], toksz))
+		if (strncmp(&BUF(p)[pos], p->indexs[i].name, toksz))
 			continue;
 		if (0 == strncmp(&BUF(p)[pos + toksz], "index", 5))
 			return(TEXICMD_USER_INDEX);
@@ -946,6 +1072,8 @@ parseeoln(struct texi *p, size_t *pos)
 				texiwarn(p, "unexpected \"{\"");
 			advance(p, pos);
 			continue;
+		case ('\n'):
+			continue;
 		case ('@'):
 			break;
 		default:
@@ -968,6 +1096,18 @@ parseeoln(struct texi *p, size_t *pos)
 		advance(p, pos);
 }
 
+enum texicmd
+peeklinecmd(const struct texi *p, size_t pos)
+{
+	size_t		end;
+
+	while (pos < BUFSZ(p) && isws(BUF(p)[pos]))
+		pos++;
+	if (pos == BUFSZ(p) || '@' != BUF(p)[pos])
+		return(TEXICMD__MAX);
+	return(texicmd(p, pos, &end, NULL));
+}
+
 /*
  * Peek to see if there's a command after subsequent whitespace.
  * If so, return the macro identifier.
@@ -1152,6 +1292,9 @@ parseto(struct texi *p, size_t *pos, const char *endto
 		if (NULL != texitoks[cmd].fp) 
 			(*texitoks[cmd].fp)(p, cmd, pos);
 	}
+
+	if (*pos == BUFSZ(p)) 
+		texiwarn(p, "EOF expecting \"%s\" end\n", endtoken);
 }
 
 /*
@@ -1577,6 +1720,7 @@ teximdocopen(struct texi *p, size_t *pos)
 	t = time(NULL);
 	strftime(date, sizeof(date), "%F", localtime(&t));
 
+	p->seenvs = -1;
 	teximacroopen(p, "Dd");
 	texiputchars(p, date);
 	teximacroclose(p);