===================================================================
RCS file: /cvs/texi2mdoc/main.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -p -r1.1 -r1.2
--- texi2mdoc/main.c	2015/02/16 22:24:43	1.1
+++ texi2mdoc/main.c	2015/02/17 10:27:18	1.2
@@ -1,4 +1,4 @@
-/*	$Id: main.c,v 1.1 2015/02/16 22:24:43 kristaps Exp $ */
+/*	$Id: main.c,v 1.2 2015/02/17 10:27:18 kristaps Exp $ */
 /*
  * Copyright (c) 2015 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -21,6 +21,8 @@
 #include <ctype.h>
 #include <fcntl.h>
 #include <getopt.h>
+#include <libgen.h>
+#include <limits.h>
 #include <stdarg.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -32,8 +34,11 @@
  * different story.
  */
 enum	texicmd {
+	TEXICMD_ACRONYM,
 	TEXICMD_A4PAPER,
 	TEXICMD_ANCHOR,
+	TEXICMD_APPENDIX,
+	TEXICMD_APPENDIXSEC,
 	TEXICMD_AT,
 	TEXICMD_BYE,
 	TEXICMD_CHAPTER,
@@ -41,22 +46,27 @@ enum	texicmd {
 	TEXICMD_CODE,
 	TEXICMD_COMMAND,
 	TEXICMD_COMMENT,
+	TEXICMD_COMMENT_LONG,
 	TEXICMD_CONTENTS,
 	TEXICMD_COPYING,
 	TEXICMD_COPYRIGHT,
 	TEXICMD_DETAILMENU,
 	TEXICMD_DIRCATEGORY,
 	TEXICMD_DIRENTRY,
+	TEXICMD_DOTS,
 	TEXICMD_EMAIL,
 	TEXICMD_EMPH,
 	TEXICMD_END,
+	TEXICMD_ENUMERATE,
 	TEXICMD_EXAMPLE,
 	TEXICMD_FILE,
+	TEXICMD_HEADING,
 	TEXICMD_I,
 	TEXICMD_IFHTML,
 	TEXICMD_IFNOTTEX,
 	TEXICMD_IFTEX,
 	TEXICMD_IMAGE,
+	TEXICMD_INCLUDE,
 	TEXICMD_ITEM,
 	TEXICMD_ITEMIZE,
 	TEXICMD_KBD,
@@ -65,6 +75,7 @@ enum	texicmd {
 	TEXICMD_NODE,
 	TEXICMD_QUOTATION,
 	TEXICMD_PARINDENT,
+	TEXICMD_PRINTINDEX,
 	TEXICMD_REF,
 	TEXICMD_SAMP,
 	TEXICMD_SECTION,
@@ -79,6 +90,7 @@ enum	texicmd {
 	TEXICMD_TITLEPAGE,
 	TEXICMD_TOP,
 	TEXICMD_UNNUMBERED,
+	TEXICMD_UNNUMBEREDSEC,
 	TEXICMD_URL,
 	TEXICMD_VAR,
 	TEXICMD__MAX
@@ -98,6 +110,9 @@ struct	texifile {
 
 struct	texi;
 
+/*
+ * Callback for functions implementing texi commands.
+ */
 typedef	void (*texicmdfp)(struct texi *, 
 	enum texicmd, const char *, size_t, size_t *);
 
@@ -124,24 +139,30 @@ struct	texi {
 	size_t		 outcol; /* column of output */
 	int		 outmacro; /* whether output is in line macro */
 	int		 seenws; /* whitespace has been ignored */
+	char		*dir; /* texi directory */
 };
 
+/* FIXME: don't use this crap. */
 #define	ismpunct(_x) \
 	('.' == (_x) || \
 	 ',' == (_x) || \
 	 ';' == (_x))
+#define	isws(_x) \
+	(' ' == (_x) || '\t' == (_x))
 
 static	void doarg1(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void dobracket(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void dobye(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void docommand(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void doemph(struct texi *, enum texicmd, const char *, size_t, size_t *);
+static	void doenumerate(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void doexample(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void dofile(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void doifnottex(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void doignblock(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void doignbracket(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void doignline(struct texi *, enum texicmd, const char *, size_t, size_t *);
+static	void doinclude(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void doitalic(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void doitem(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void doitemize(struct texi *, enum texicmd, const char *, size_t, size_t *);
@@ -155,8 +176,11 @@ static	void dosubsection(struct texi *, enum texicmd, 
 static	void dosymbol(struct texi *, enum texicmd, const char *, size_t, size_t *);
 
 static	const struct texitok texitoks[TEXICMD__MAX] = {
+	{ doarg1, "acronym", 7 }, /* TEXICMD_ACRONYM */
 	{ doignline, "afourpaper", 10 }, /* TEXICMD_A4PAPER */
 	{ doignbracket, "anchor", 6 }, /* TEXICMD_ANCHOR */
+	{ dosh, "appendix", 8 }, /* TEXICMD_APPENDIX */
+	{ dosh, "appendixsec", 11 }, /* TEXICMD_APPENDIXSEC */
 	{ dosymbol, "@", 1 }, /* TEXICMD_AT */
 	{ dobye, "bye", 3 }, /* TEXICMD_BYE */
 	{ dosh, "chapter", 7 }, /* TEXICMD_CHAPTER */
@@ -164,22 +188,27 @@ static	const struct texitok texitoks[TEXICMD__MAX] = {
 	{ doliteral, "code", 4 }, /* TEXICMD_CODE */
 	{ docommand, "command", 7 }, /* TEXICMD_COMMAND */
 	{ doignline, "c", 1 }, /* TEXICMD_COMMENT */
+	{ doignline, "comment", 7 }, /* TEXICMD_COMMENT_LONG */
 	{ doignline, "contents", 8 }, /* TEXICMD_CONTENTS */
 	{ doignblock, "copying", 7 }, /* TEXICMD_COPYING */
 	{ dosymbol, "copyright", 9 }, /* TEXICMD_COPYRIGHT */
 	{ doignblock, "detailmenu", 10 }, /* TEXICMD_DETAILMENU */
 	{ doignline, "dircategory", 11 }, /* TEXICMD_DIRCATEGORY */
 	{ doignblock, "direntry", 8 }, /* TEXICMD_DIRENTRY */
+	{ dosymbol, "dots", 4 }, /* TEXICMD_DOTS */
 	{ doarg1, "email", 5 }, /* TEXICMD_EMAIL */
 	{ doemph, "emph", 4 }, /* TEXICMD_EMPH */
 	{ NULL, "end", 3 }, /* TEXICMD_END */
+	{ doenumerate, "enumerate", 9 }, /* TEXICMD_ENUMERATE */
 	{ doexample, "example", 7 }, /* TEXICMD_EXAMPLE */
 	{ dofile, "file", 4 }, /* TEXICMD_FILE */
+	{ dosection, "heading", 7 }, /* TEXICMD_HEADING */
 	{ doitalic, "i", 1 }, /* TEXICMD_I */
 	{ doignblock, "ifhtml", 6 }, /* TEXICMD_IFHTML */
 	{ doifnottex, "ifnottex", 8 }, /* TEXICMD_IFNOTTEX */
 	{ doignblock, "iftex", 5 }, /* TEXICMD_IFTEX */
 	{ doignbracket, "image", 5 }, /* TEXICMD_IMAGE */
+	{ doinclude, "include", 7 }, /* TEXICMD_INCLUDE */
 	{ doitem, "item", 4 }, /* TEXICMD_ITEM */
 	{ doitemize, "itemize", 7 }, /* TEXICMD_ITEMIZE */
 	{ doliteral, "kbd", 3 }, /* TEXICMD_KBD */
@@ -187,6 +216,7 @@ static	const struct texitok texitoks[TEXICMD__MAX] = {
 	{ doignblock, "menu", 4 }, /* TEXICMD_MENU */
 	{ doignline, "node", 4 }, /* TEXICMD_NODE */
 	{ doquotation, "quotation", 9 }, /* TEXICMD_QUOTATION */
+	{ doignline, "printindex", 10 }, /* TEXICMD_PRINTINDEX */
 	{ doignline, "paragraphindent", 14 }, /* TEXICMD_PARINDENT */
 	{ dobracket, "ref", 3 }, /* TEXICMD_REF */
 	{ doliteral, "samp", 4 }, /* TEXICMD_SAMP */
@@ -202,10 +232,14 @@ static	const struct texitok texitoks[TEXICMD__MAX] = {
 	{ doignblock, "titlepage", 9 }, /* TEXICMD_TITLEPAGE */
 	{ dotop, "top", 3 }, /* TEXICMD_TOP */
 	{ dosh, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */
+	{ dosection, "unnumberedsec", 13 }, /* TEXICMD_UNNUMBEREDSEC */
 	{ doarg1, "url", 3 }, /* TEXICMD_URL */
 	{ doliteral, "var", 3 }, /* TEXICMD_VAR */
 };
 
+/*
+ * Unmap the top-most file that we're using.
+ */
 static void
 texifilepop(struct texi *p)
 {
@@ -216,16 +250,25 @@ texifilepop(struct texi *p)
 	munmap(f->map, f->mapsz);
 }
 
+/*
+ * Unmap all files that we're currently using.
+ * The utility should exit(...) after this is called.
+ */
 static void
 texiexit(struct texi *p)
 {
 
 	while (p->filepos > 0)
 		texifilepop(p);
+	free(p->dir);
 }
 
+/*
+ * Fatal error: unmap all files and exit.
+ * The "errstring" is passed to perror(3).
+ */
 static void
-texifatal(struct texi *p, const char *errstring)
+texiabort(struct texi *p, const char *errstring)
 {
 
 	perror(errstring);
@@ -242,7 +285,7 @@ texiwarn(const struct texi *p, const char *fmt, ...)
 {
 	va_list	 ap;
 
-	fprintf(stderr, "%s:%zu:%zu: ",
+	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);
@@ -252,6 +295,23 @@ texiwarn(const struct texi *p, const char *fmt, ...)
 	fputc('\n', stderr);
 }
 
+static void
+texierr(struct texi *p, const char *fmt, ...)
+{
+	va_list	 ap;
+
+	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);
+	va_start(ap, fmt);
+	vfprintf(stderr, fmt, ap);
+	va_end(ap);
+	fputc('\n', stderr);
+	texiexit(p);
+	exit(EXIT_FAILURE);
+}
+
 /*
  * Put a single data character.
  * This MUST NOT be a mdoc(7) command: it should be free text that's
@@ -442,7 +502,7 @@ texicmd(struct texi *p, const char *buf, 
 
 	assert('@' == buf[pos]);
 	for (*end = ++pos; *end < sz && ! isspace(buf[*end]); (*end)++)
-		if ('@' == buf[*end] || '{' == buf[*end])
+		if ((*end > pos && '@' == buf[*end]) || '{' == buf[*end])
 			break;
 
 	len = *end - pos;
@@ -556,7 +616,7 @@ parseto(struct texi *p, const char *buf, 
 		cmd = texicmd(p, buf, *pos, sz, &end);
 		advanceto(p, buf, pos, end);
 		if (TEXICMD_END == cmd) {
-			while (*pos < sz && ' ' == buf[*pos])
+			while (*pos < sz && isws(buf[*pos]))
 				advance(p, buf, pos);
 			/* 
 			 * FIXME: skip tabs and also check the full
@@ -577,6 +637,38 @@ parseto(struct texi *p, const char *buf, 
 }
 
 static void
+parsefile(struct texi *p, const char *fname)
+{
+	struct texifile	 *f;
+	int		  fd;
+	struct stat	  st;
+
+	assert(p->filepos < 64);
+	f = &p->files[p->filepos];
+	memset(f, 0, sizeof(struct texifile));
+
+	f->name = fname;
+	if (-1 == (fd = open(fname, O_RDONLY, 0))) {
+		texiabort(p, fname);
+	} else if (-1 == fstat(fd, &st)) {
+		close(fd);
+		texiabort(p, fname);
+	} 
+
+	f->mapsz = st.st_size;
+	f->map = mmap(NULL, f->mapsz,
+		PROT_READ, MAP_SHARED, fd, 0);
+	close(fd);
+
+	if (MAP_FAILED == f->map)
+		texiabort(p, fname);
+
+	p->filepos++;
+	parseeof(p, f->map, f->mapsz);
+	texifilepop(p);
+}
+
+static void
 doignblock(struct texi *p, enum texicmd cmd, 
 	const char *buf, size_t sz, size_t *pos)
 {
@@ -649,6 +741,47 @@ doinline(struct texi *p, const char *buf, 
 }
 
 static void
+doinclude(struct texi *p, enum texicmd cmd, 
+	const char *buf, size_t sz, size_t *pos)
+{
+	char	 fname[PATH_MAX], path[PATH_MAX];
+	size_t	 i;
+	int	 rc;
+
+	while (*pos < sz && ' ' == buf[*pos])
+		advance(p, buf, pos);
+
+	/* Read in the filename. */
+	for (i = 0; *pos < sz && '\n' != buf[*pos]; i++) {
+		if (i == sizeof(fname) - 1)
+			break;
+		fname[i] = buf[*pos];
+		advance(p, buf, pos);
+	}
+
+	if (i == 0)
+		texierr(p, "path too short");
+	else if ('\n' != buf[*pos])
+		texierr(p, "path too long");
+	else if ('/' == fname[0])
+		texierr(p, "no absolute paths");
+	fname[i] = '\0';
+
+	if (strstr(fname, "../") || strstr(fname, "/.."))
+		texierr(p, "insecure path");
+
+	/* Append filename to original name's directory. */
+	rc = snprintf(path, sizeof(path), "%s/%s", p->dir, fname);
+	if (rc < 0) 
+		texierr(p, "couldn't format filename");
+	else if ((size_t)rc >= sizeof(path))
+		texierr(p, "path too long");
+
+	/* Pump through to parser. */
+	parsefile(p, path);
+}
+
+static void
 doitalic(struct texi *p, enum texicmd cmd, 
 	const char *buf, size_t sz, size_t *pos)
 {
@@ -745,6 +878,9 @@ dosymbol(struct texi *p, enum texicmd cmd, 
 	case (TEXICMD_COPYRIGHT):
 		texiputchars(p, "\\(co");
 		break;
+	case (TEXICMD_DOTS):
+		texiputchars(p, "...");
+		break;
 	case (TEXICMD_LATEX):
 		texiputchars(p, "LaTeX");
 		break;
@@ -835,7 +971,7 @@ dosection(struct texi *p, enum texicmd cmd, 
 		advanceeoln(p, buf, sz, pos);
 		return;
 	}
-	while (*pos < sz && ' ' == buf[*pos]) 
+	while (*pos < sz && isws(buf[*pos]) )
 		advance(p, buf, pos);
 	texifputs(p, ".Ss ");
 	while (*pos < sz && '\n' != buf[*pos]) {
@@ -854,7 +990,7 @@ dosh(struct texi *p, enum texicmd cmd, 
 		advanceeoln(p, buf, sz, pos);
 		return;
 	}
-	while (*pos < sz && ' ' == buf[*pos]) 
+	while (*pos < sz && isws(buf[*pos]))
 		advance(p, buf, pos);
 	texifputs(p, ".Sh ");
 	while (*pos < sz && '\n' != buf[*pos]) {
@@ -871,7 +1007,7 @@ dotop(struct texi *p, enum texicmd cmd, 
 
 	p->flags &= ~TEXI_HEADER;
 	advanceeoln(p, buf, sz, pos);
-	teximacro(p, ".Dd $Mdocdate: February 16 2015 $");
+	teximacro(p, ".Dd $Mdocdate: February 17 2015 $");
 	teximacro(p, ".Dt SOMETHING 7");
 	teximacro(p, ".Os");
 	teximacro(p, ".Sh NAME");
@@ -887,7 +1023,7 @@ doitem(struct texi *p, enum texicmd cmd, 
 
 	/* See if we have arguments... */
 	for (end = *pos; end < sz; end++) 
-		if (' ' != buf[end] && '\t' != buf[end])
+		if ( ! isws(buf[end]))
 			break;
 
 	/* If we have arguments, print them too. */
@@ -914,6 +1050,16 @@ dotable(struct texi *p, enum texicmd cmd, 
 }
 
 static void
+doenumerate(struct texi *p, enum texicmd cmd, 
+	const char *buf, size_t sz, size_t *pos)
+{
+
+	teximacro(p, ".Bl -enum");
+	parseto(p, buf, sz, pos, "enumerate");
+	teximacro(p, ".El");
+}
+
+static void
 doitemize(struct texi *p, enum texicmd cmd, 
 	const char *buf, size_t sz, size_t *pos)
 {
@@ -944,46 +1090,12 @@ doignline(struct texi *p, enum texicmd cmd, 
 		advance(p, buf, pos);
 }
 
-static int
-parsefile(struct texi *p, const char *fname)
-{
-	struct texifile	 *f;
-	int		  fd;
-	struct stat	  st;
-
-	assert(p->filepos < 64);
-	f = &p->files[p->filepos];
-	memset(f, 0, sizeof(struct texifile));
-
-	f->name = fname;
-	if (-1 == (fd = open(fname, O_RDONLY, 0))) {
-		texifatal(p, fname);
-	} else if (-1 == fstat(fd, &st)) {
-		close(fd);
-		texifatal(p, fname);
-	} 
-
-	f->mapsz = st.st_size;
-	f->map = mmap(NULL, f->mapsz,
-		PROT_READ, MAP_SHARED, fd, 0);
-	close(fd);
-
-	if (MAP_FAILED == f->map) {
-		texifatal(p, fname);
-		return(0);
-	}
-
-	p->filepos++;
-	parseeof(p, f->map, f->mapsz);
-	texifilepop(p);
-	return(1);
-}
-
 int
 main(int argc, char *argv[])
 {
 	struct texi	 texi;
-	int		 c, rc;
+	int		 c;
+	char		*path, *dir;
 	const char	*progname;
 
 	progname = strrchr(argv[0], '/');
@@ -1002,11 +1114,22 @@ main(int argc, char *argv[])
 	if (0 == (argc -= optind))
 		goto usage;
 
+	if (NULL == (path = strdup(argv[0]))) {
+		perror(NULL);
+		exit(EXIT_FAILURE);
+	} else if (NULL == (dir = dirname(path))) {
+		perror(argv[0]);
+		free(path);
+		exit(EXIT_FAILURE);
+	}
+	free(path);
+
 	memset(&texi, 0, sizeof(struct texi));
 	texi.flags = TEXI_HEADER;
-	rc = parsefile(&texi, argv[0]);
-	return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
-
+	texi.dir = strdup(dir);
+	parsefile(&texi, argv[0]);
+	texiexit(&texi);
+	return(EXIT_FAILURE);
 usage:
 	fprintf(stderr, "usage: %s file\n", progname);
 	return(EXIT_FAILURE);