===================================================================
RCS file: /cvs/texi2mdoc/util.c,v
retrieving revision 1.14
retrieving revision 1.21
diff -u -p -r1.14 -r1.21
--- texi2mdoc/util.c	2015/02/25 14:37:17	1.14
+++ texi2mdoc/util.c	2015/03/01 13:39:51	1.21
@@ -1,4 +1,4 @@
-/*	$Id: util.c,v 1.14 2015/02/25 14:37:17 kristaps Exp $ */
+/*	$Id: util.c,v 1.21 2015/03/01 13:39:51 kristaps Exp $ */
 /*
  * Copyright (c) 2015 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -20,8 +20,6 @@
 #include <assert.h>
 #include <ctype.h>
 #include <fcntl.h>
-#include <getopt.h>
-#include <libgen.h>
 #include <limits.h>
 #include <stdarg.h>
 #include <stdio.h>
@@ -79,7 +77,9 @@ texiexit(struct texi *p)
 
 	/* Make sure we're newline-terminated. */
 	if (p->outcol)
-		putchar('\n');
+		fputc('\n', p->outfile);
+	if (NULL != p->chapters)
+		teximdocclose(p, 1);
 
 	/* Unmap all files. */
 	while (p->filepos > 0)
@@ -122,12 +122,19 @@ texiabort(struct texi *p, const char *errstring)
 void
 texiwarn(const struct texi *p, const char *fmt, ...)
 {
-	va_list	 ap;
+	va_list	 	 	 ap;
+	const struct texifile	*f;
 
-	fprintf(stderr, "%s:%zu:%zu: warning: ",
-		p->files[p->filepos - 1].name,
-		p->files[p->filepos - 1].line + 1,
-		p->files[p->filepos - 1].col + 1);
+	f = &p->files[p->filepos - 1];
+
+	if (f->insplice)
+		fprintf(stderr, "%s:%zu:%zu (%zuB left in splice): "
+			"warning: ", f->name, f->line + 1, 
+			f->col + 1, f->insplice);
+	else
+		fprintf(stderr, "%s:%zu:%zu: warning: ",
+			f->name, f->line + 1, f->col + 1);
+
 	va_start(ap, fmt);
 	vfprintf(stderr, fmt, ap);
 	va_end(ap);
@@ -141,12 +148,19 @@ texiwarn(const struct texi *p, const char *fmt, ...)
 void
 texierr(struct texi *p, const char *fmt, ...)
 {
-	va_list	 ap;
+	va_list	 	 ap;
+	struct texifile	*f;
 
-	fprintf(stderr, "%s:%zu:%zu: error: ",
-		p->files[p->filepos - 1].name,
-		p->files[p->filepos - 1].line + 1,
-		p->files[p->filepos - 1].col + 1);
+	f = &p->files[p->filepos - 1];
+
+	if (f->insplice) 
+		fprintf(stderr, "%s:%zu:%zu: (%zuB left in splice): "
+			"error: ", f->name, f->line + 1, 
+			f->col + 1, f->insplice);
+	else
+		fprintf(stderr, "%s:%zu:%zu: error: ",
+			f->name, f->line + 1, f->col + 1);
+
 	va_start(ap, fmt);
 	vfprintf(stderr, fmt, ap);
 	va_end(ap);
@@ -166,13 +180,13 @@ texiputchar(struct texi *p, char c)
 	if (p->ign)
 		return;
 	if ('.' == c && 0 == p->outcol)
-		fputs("\\&", stdout);
+		fputs("\\&", p->outfile);
 	if ('\'' == c && 0 == p->outcol)
-		fputs("\\&", stdout);
+		fputs("\\&", p->outfile);
 
-	putchar(c);
+	fputc(c, p->outfile);
 	if ('\\' == c)
-		putchar('e');
+		fputc('e', p->outfile);
 	p->seenvs = 0;
 	if ('\n' == c) {
 		p->outcol = 0;
@@ -193,10 +207,10 @@ texiputchars(struct texi *p, const char *s)
 	if (p->ign)
 		return;
 	if ('.' == *s && 0 == p->outcol)
-		fputs("\\&", stdout);
+		fputs("\\&", p->outfile);
 	if ('\'' == *s && 0 == p->outcol)
-		fputs("\\&", stdout);
-	p->outcol += fputs(s, stdout);
+		fputs("\\&", p->outfile);
+	p->outcol += fputs(s, p->outfile);
 	p->seenvs = 0;
 }
 
@@ -225,7 +239,7 @@ teximacroclose(struct texi *p)
 		return;
 
 	if (0 == --p->outmacro) {
-		putchar('\n');
+		fputc('\n', p->outfile);
 		p->outcol = p->seenws = 0;
 	}
 }
@@ -245,19 +259,19 @@ teximacroopen(struct texi *p, const char *s)
 		return;
 
 	if (p->outcol && 0 == p->outmacro) {
-		putchar('\n');
+		fputc('\n', p->outfile);
 		p->outcol = 0;
 	}
 
 	if (0 == p->outmacro)
-		putchar('.');
+		fputc('.', p->outfile);
 	else
-		putchar(' ');
+		fputc(' ', p->outfile);
 
-	if (EOF != (rc = fputs(s, stdout)))
+	if (EOF != (rc = fputs(s, p->outfile)))
 		p->outcol += rc;
 
-	putchar(' ');
+	fputc(' ', p->outfile);
 	p->outcol++;
 	p->outmacro++;
 	p->seenws = 0;
@@ -279,10 +293,11 @@ teximacro(struct texi *p, const char *s)
 		texierr(p, "\"%s\" in a literal scope!?", s);
 
 	if (p->outcol)
-		putchar('\n');
+		fputc('\n', p->outfile);
 
-	putchar('.');
-	puts(s);
+	fputc('.', p->outfile);
+	fputs(s, p->outfile);
+	fputc('\n', p->outfile);
 	p->outcol = p->seenws = 0;
 }
 
@@ -306,13 +321,22 @@ texivspace(struct texi *p)
 void
 advance(struct texi *p, size_t *pos)
 {
+	struct texifile	*f;
 
-	if ('\n' == BUF(p)[*pos]) {
-		p->files[p->filepos - 1].line++;
-		p->files[p->filepos - 1].col = 0;
-	} else
-		p->files[p->filepos - 1].col++;
+	f = &p->files[p->filepos - 1];
 
+	if (0 == f->insplice) {
+		if ('\n' == BUF(p)[*pos]) {
+			f->line++;
+			f->col = 0;
+		} else
+			f->col++;
+	} else {
+		--f->insplice;
+		if (0 == f->insplice)
+			f->depth = 0;
+	}
+
 	(*pos)++;
 }
 
@@ -419,7 +443,7 @@ advanceto(struct texi *p, size_t *pos, size_t end)
 }
 
 static void
-texiexecmacro(struct texi *p, struct teximacro *m, size_t *pos)
+texiexecmacro(struct texi *p, struct teximacro *m, size_t sv, size_t *pos)
 {
 	size_t		  valsz, realsz, aasz, asz, 
 			   ssz, i, j, k, start, end;
@@ -427,17 +451,31 @@ texiexecmacro(struct texi *p, struct teximacro *m, siz
 	char		**args;
 	const char	 *cp;
 
+	/* Disregard empty macros. */
+	if (0 == (valsz = realsz = strlen(m->value)))
+		return;
+
+	/*
+	 * This is important: it protect us from macros that invoke more
+	 * macros, possibly going on infinitely.
+	 * We use "sv" instead of the current position because we might
+	 * be invoked at the end of the macro (i.e., insplice == 0).
+	 * The "sv" value was initialised at the start of the macro.
+	 */
+	if (sv > 0)
+		if (++p->files[p->filepos].depth > 64)
+			texierr(p, "maximium recursive depth");
+
 	args = argparse(p, pos, &asz, m->argsz);
 	if (asz != m->argsz)
 		texiwarn(p, "invalid macro argument length");
 	aasz = asz < m->argsz ? asz : m->argsz;
 
 	if (0 == aasz) {
-		texisplice(p, m->value, strlen(m->value), pos);
+		texisplice(p, m->value, valsz, *pos);
 		return;
 	}
 
-	valsz = realsz = strlen(m->value);
 	val = strdup(m->value);
 
 	for (i = j = 0; i < realsz; i++) {
@@ -497,7 +535,7 @@ texiexecmacro(struct texi *p, struct teximacro *m, siz
 		i = end;
 	}
 
-	texisplice(p, val, strlen(val), pos);
+	texisplice(p, val, strlen(val), *pos);
 
 	for (i = 0; i < asz; i++)
 		free(args[i]);
@@ -554,7 +592,7 @@ parseword(struct texi *p, size_t *pos, char extra)
  * index after the command name.
  */
 enum texicmd
-texicmd(struct texi *p, size_t pos, size_t *end, struct teximacro **macro)
+texicmd(const struct texi *p, size_t pos, size_t *end, struct teximacro **macro)
 {
 	size_t	 i, len, toksz;
 
@@ -569,7 +607,7 @@ texicmd(struct texi *p, size_t pos, size_t *end, struc
 		return(TEXICMD__MAX);
 
 	/* Alphabetic commands are special. */
-	if ( ! isalpha(BUF(p)[pos])) {
+	if ( ! isalpha((unsigned char)BUF(p)[pos])) {
 		if ((*end = pos + 1) == BUFSZ(p))
 			return(TEXICMD__MAX);
 		for (i = 0; i < TEXICMD__MAX; i++) {
@@ -633,7 +671,7 @@ texicmd(struct texi *p, size_t pos, size_t *end, struc
 int
 parsearg(struct texi *p, size_t *pos, size_t num)
 {
-	size_t		  end;
+	size_t		  end, sv;
 	enum texicmd	  cmd;
 	struct teximacro *macro;
 
@@ -664,10 +702,11 @@ parsearg(struct texi *p, size_t *pos, size_t num)
 			continue;
 		}
 
+		sv = p->files[p->filepos - 1].insplice;
 		cmd = texicmd(p, *pos, &end, &macro);
 		advanceto(p, pos, end);
 		if (NULL != macro)
-			texiexecmacro(p, macro, pos);
+			texiexecmacro(p, macro, sv, pos);
 		if (TEXICMD__MAX == cmd) 
 			continue;
 		if (NULL != texitoks[cmd].fp)
@@ -681,9 +720,9 @@ parsearg(struct texi *p, size_t *pos, size_t num)
  * This will stop in the event of EOF or if we're not at a bracket.
  */
 void
-parsebracket(struct texi *p, size_t *pos)
+parsebracket(struct texi *p, size_t *pos, int dostack)
 {
-	size_t		  end;
+	size_t		  end, sv, stack;
 	enum texicmd	  cmd;
 	struct teximacro *macro;
 
@@ -694,12 +733,25 @@ parsebracket(struct texi *p, size_t *pos)
 		return;
 	advance(p, pos);
 
+	stack = 0;
 	while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
 		switch (BUF(p)[*pos]) {
 		case ('}'):
+			if (stack > 0) {
+				stack--;
+				advance(p, pos);
+				texiputchar(p, '}');
+				continue;
+			}
 			advance(p, pos);
 			return;
 		case ('{'):
+			if (dostack) {
+				stack++;
+				advance(p, pos);
+				texiputchar(p, '{');
+				continue;
+			}
 			if (0 == p->ign)
 				texiwarn(p, "unexpected \"{\"");
 			advance(p, pos);
@@ -711,10 +763,11 @@ parsebracket(struct texi *p, size_t *pos)
 			continue;
 		}
 
+		sv = p->files[p->filepos - 1].insplice;
 		cmd = texicmd(p, *pos, &end, &macro);
 		advanceto(p, pos, end);
 		if (NULL != macro)
-			texiexecmacro(p, macro, pos);
+			texiexecmacro(p, macro, sv, pos);
 		if (TEXICMD__MAX == cmd) 
 			continue;
 		if (NULL != texitoks[cmd].fp)
@@ -730,7 +783,7 @@ parsebracket(struct texi *p, size_t *pos)
 void
 parseeoln(struct texi *p, size_t *pos)
 {
-	size_t		  end;
+	size_t		  end, sv;
 	enum texicmd	  cmd;
 	struct teximacro *macro;
 
@@ -759,10 +812,11 @@ parseeoln(struct texi *p, size_t *pos)
 			continue;
 		}
 
+		sv = p->files[p->filepos - 1].insplice;
 		cmd = texicmd(p, *pos, &end, &macro);
 		advanceto(p, pos, end);
 		if (NULL != macro)
-			texiexecmacro(p, macro, pos);
+			texiexecmacro(p, macro, sv, pos);
 		if (TEXICMD__MAX == cmd) 
 			continue;
 		if (NULL != texitoks[cmd].fp)
@@ -774,13 +828,30 @@ parseeoln(struct texi *p, size_t *pos)
 }
 
 /*
+ * Peek to see if there's a command after subsequent whitespace.
+ * If so, return the macro identifier.
+ * This DOES NOT work with user-defined macros.
+ */
+enum texicmd
+peekcmd(const struct texi *p, size_t pos)
+{
+	size_t		end;
+
+	while (pos < BUFSZ(p) && ismspace(BUF(p)[pos]))
+		pos++;
+	if (pos == BUFSZ(p) || '@' != BUF(p)[pos])
+		return(TEXICMD__MAX);
+	return(texicmd(p, pos, &end, NULL));
+}
+
+/*
  * Parse a single word or command.
  * This will return immediately at the EOF.
  */
 static void
 parsesingle(struct texi *p, size_t *pos)
 {
-	size_t		  end;
+	size_t		  end, sv;
 	enum texicmd	  cmd;
 	struct teximacro *macro;
 
@@ -805,10 +876,11 @@ parsesingle(struct texi *p, size_t *pos)
 		return;
 	}
 
+	sv = p->files[p->filepos - 1].insplice;
 	cmd = texicmd(p, *pos, &end, &macro);
 	advanceto(p, pos, end);
 	if (NULL != macro)
-		texiexecmacro(p, macro, pos);
+		texiexecmacro(p, macro, sv, pos);
 	if (TEXICMD__MAX == cmd) 
 		return;
 	if (NULL != texitoks[cmd].fp)
@@ -833,7 +905,7 @@ parselinearg(struct texi *p, size_t *pos)
 	}
 
 	if (*pos < BUFSZ(p) && '{' == BUF(p)[*pos])
-		parsebracket(p, pos);
+		parsebracket(p, pos, 0);
 	else if (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos])
 		parsesingle(p, pos);
 	else
@@ -855,7 +927,7 @@ parseeof(struct texi *p)
 }
 
 void
-texisplice(struct texi *p, const char *buf, size_t sz, size_t *pos)
+texisplice(struct texi *p, const char *buf, size_t sz, size_t pos)
 {
 	char		*cp;
 	struct texifile	*f;
@@ -871,8 +943,9 @@ texisplice(struct texi *p, const char *buf, size_t sz,
 		f->map = cp;
 	}
 
-	memmove(f->map + *pos + sz, f->map + *pos, f->mapsz - *pos);
-	memcpy(f->map + *pos, buf, sz);
+	f->insplice += sz;
+	memmove(f->map + pos + sz, f->map + pos, f->mapsz - pos);
+	memcpy(f->map + pos, buf, sz);
 	f->mapsz += sz;
 }
 
@@ -884,7 +957,7 @@ texisplice(struct texi *p, const char *buf, size_t sz,
 void
 parseto(struct texi *p, size_t *pos, const char *endtoken)
 {
-	size_t		  end;
+	size_t		  end, sv;
 	enum texicmd	  cmd;
 	size_t		  endtoksz;
 	struct teximacro *macro;
@@ -911,6 +984,7 @@ parseto(struct texi *p, size_t *pos, const char *endto
 			continue;
 		}
 
+		sv = p->files[p->filepos - 1].insplice;
 		cmd = texicmd(p, *pos, &end, &macro);
 		advanceto(p, pos, end);
 		if (TEXICMD_END == cmd) {
@@ -931,7 +1005,7 @@ parseto(struct texi *p, size_t *pos, const char *endto
 			continue;
 		} 
 		if (NULL != macro)
-			texiexecmacro(p, macro, pos);
+			texiexecmacro(p, macro, sv, pos);
 		if (TEXICMD__MAX == cmd) 
 			continue;
 		if (NULL != texitoks[cmd].fp) 
@@ -1188,6 +1262,9 @@ argparse(struct texi *p, size_t *pos, size_t *argsz, s
 	args = NULL;
 	*argsz = 0;
 
+	if (*pos == BUFSZ(p))
+		return(args);
+
 	if ('{' != BUF(p)[*pos] && hint) {
 		/*
 		 * Special case: if we encounter an unbracketed argument
@@ -1214,6 +1291,8 @@ argparse(struct texi *p, size_t *pos, size_t *argsz, s
 	} else if ('{' != BUF(p)[*pos])
 		return(args);
 
+	assert('{' == BUF(p)[*pos]);
+
 	/* Parse til the closing '}', putting into the array. */
 	advance(p, pos);
 	while (*pos < BUFSZ(p)) {
@@ -1270,3 +1349,112 @@ argparse(struct texi *p, size_t *pos, size_t *argsz, s
 	advance(p, pos);
 	return(args);
 }
+
+/*
+ * If we're printing chapters, then do some naviation here and then
+ * close our outfile.
+ * I want to call this the SEE ALSO section, but that's not really what
+ * it is: we'll refer to the "initial" (top) node and the next and
+ * previous chapters.
+ */
+void
+teximdocclose(struct texi *p, int last)
+{
+	char	 buf[PATH_MAX];
+
+	if (NULL == p->chapters || 0 == p->chapnum)
+		return;
+
+	teximacro(p, "Sh INFO NAVIGATION");
+
+	/* Print a reference to the "top" node. */
+	if (p->chapnum > 1) {
+		snprintf(buf, sizeof(buf), "node1 7");
+		teximacroopen(p, "Xr ");
+		texiputchars(p, buf);
+		texiputchars(p, " ,");
+		teximacroclose(p);
+	}
+
+	/* Print a reference to the previous node. */
+	if (p->chapnum > 2) {
+		snprintf(buf, sizeof(buf),
+			"node%zu 7", p->chapnum - 1);
+		teximacroopen(p, "Xr ");
+		texiputchars(p, buf);
+		if ( ! last) 
+			texiputchars(p, " ,");
+		teximacroclose(p);
+	} 
+
+	/* Print a reference to the next node. */
+	if ( ! last) {
+		snprintf(buf, sizeof(buf),
+			"node%zu 7", p->chapnum + 1);
+		teximacroopen(p, "Xr ");
+		texiputchars(p, buf);
+		teximacroclose(p);
+	}
+
+	fclose(p->outfile);
+}
+
+/*
+ * Open a mdoc(7) context.
+ * If we're printing chapters, then open the outfile here, too.
+ * Otherwise just print the mdoc(7) prologue.
+ */
+void
+teximdocopen(struct texi *p, size_t *pos)
+{
+	const char	*cp;
+	time_t		 t;
+	char		 date[32];
+	char	 	 fname[PATH_MAX];
+
+	if (NULL != p->chapters) {
+		snprintf(fname, sizeof(fname), "%s/node%zu.7", 
+			p->chapters, ++p->chapnum);
+		p->outfile = fopen(fname, "w");
+		if (NULL == p->outfile)
+			texiabort(p, fname);
+	}
+
+	/*
+	 * Here we print our standard mdoc(7) prologue.
+	 * We use the title set with @settitle for the `Nd' description
+	 * and the source document filename (the first one as invoked on
+	 * the command line) for the title.
+	 * The date is set to the current date.
+	 */
+	t = time(NULL);
+	strftime(date, sizeof(date), "%F", localtime(&t));
+
+	teximacroopen(p, "Dd");
+	texiputchars(p, date);
+	teximacroclose(p);
+	teximacroopen(p, "Dt");
+	for (cp = p->title; '\0' != *cp; cp++)
+		texiputchar(p, toupper((unsigned int)*cp));
+	texiputchars(p, " 7");
+	teximacroclose(p);
+	teximacro(p, "Os");
+	teximacro(p, "Sh NAME");
+	teximacroopen(p, "Nm");
+	for (cp = p->title; '\0' != *cp; cp++)
+		texiputchar(p, *cp);
+	teximacroclose(p);
+	teximacroopen(p, "Nd");
+	/*
+	 * The subtitle `Nd' can consist of arbitrary macros, so paste
+	 * it and parse to the end of the line.
+	 */
+	if (NULL != p->subtitle) {
+		texisplice(p, p->subtitle, strlen(p->subtitle), *pos);
+		parseeoln(p, pos);
+	} else
+		texiputchars(p, "Unknown description");
+	teximacroclose(p);
+	p->seenvs = 1;
+}
+