===================================================================
RCS file: /cvs/texi2mdoc/main.c,v
retrieving revision 1.8
retrieving revision 1.11
diff -u -p -r1.8 -r1.11
--- texi2mdoc/main.c	2015/02/18 12:03:21	1.8
+++ texi2mdoc/main.c	2015/02/18 16:35:49	1.11
@@ -1,4 +1,4 @@
-/*	$Id: main.c,v 1.8 2015/02/18 12:03:21 kristaps Exp $ */
+/*	$Id: main.c,v 1.11 2015/02/18 16:35:49 kristaps Exp $ */
 /*
  * Copyright (c) 2015 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -27,6 +27,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <time.h>
 #include <unistd.h>
 
 /*
@@ -135,6 +136,7 @@ enum	texicmd {
 	TEXICMD_UREF,
 	TEXICMD_URL,
 	TEXICMD_VAR,
+	TEXICMD_VSKIP,
 	TEXICMD_W,
 	TEXICMD_XREF,
 	TEXICMD__MAX
@@ -185,21 +187,25 @@ struct	texi {
 	size_t		  outcol; /* column in output line */
 	char		**dirs; /* texi directories */
 	size_t		  dirsz; /* number of texi directories */
+	char		 *title; /* title of document */
+	char		 *subtitle; /* subtitle of document */
+	/*
+	 * The following control what we output to the screen.
+	 * The complexity is required to accomodate for mdoc(7).
+	 */
 	enum texilist	  list; /* current list (set recursively) */
 	int		  outmacro; /* if >0, output is in line macro */
 	int		  seenws; /* ws has been seen (and ignored) */
+	int		  seenvs; /* newline has been Pp'd */
 	int		  ign; /* if >0, don't print anything */
 	int		  literal; /* if >0, literal context */
 };
 
-/* FIXME: FIND A BETTER WAY. */
-#define	ismpunct(_x) \
-	('.' == (_x) || \
-	 ',' == (_x) || \
-	 ';' == (_x))
 /* Texi disregards spaces and tabs. */
 #define	isws(_x) \
 	(' ' == (_x) || '\t' == (_x))
+#define	ismspace(_x) \
+	(isws((_x)) || '\n' == (_x))
 
 static	void doblock(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void dobracket(struct texi *, enum texicmd, const char *, size_t, size_t *);
@@ -232,6 +238,7 @@ static	void dosection(struct texi *, enum texicmd, con
 static	void dosp(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void dosubsection(struct texi *, enum texicmd, const char *, size_t, size_t *);
 static	void dosymbol(struct texi *, enum texicmd, const char *, size_t, size_t *);
+static	void dotitle(struct texi *, enum texicmd, const char *, size_t, size_t *);
 
 static	const struct texitok texitoks[TEXICMD__MAX] = {
 	{ doignargn, "acronym", 7 }, /* TEXICMD_ACRONYM */
@@ -312,7 +319,7 @@ static	const struct texitok texitoks[TEXICMD__MAX] = {
 	{ doignline, "set", 3 }, /* TEXICMD_SET */
 	{ doignline, "setchapternewpage", 17 }, /* TEXICMD_SETCHAPNEWPAGE */
 	{ doignline, "setfilename", 11 }, /* TEXICMD_SETFILENAME */
-	{ doignline, "settitle", 8 }, /* TEXICMD_SETTITLE */
+	{ dotitle, "settitle", 8 }, /* TEXICMD_SETTITLE */
 	{ dosp, "sp", 2 }, /* TEXICMD_SP */
 	{ dosymbol, " ", 1 }, /* TEXICMD_SPACE */
 	{ doexample, "smallexample", 12 }, /* TEXICMD_SMALLEXAMPLE */
@@ -334,6 +341,7 @@ static	const struct texitok texitoks[TEXICMD__MAX] = {
 	{ dolink, "uref", 4 }, /* TEXICMD_UREF */
 	{ dolink, "url", 3 }, /* TEXICMD_URL */
 	{ doliteral, "var", 3 }, /* TEXICMD_VAR */
+	{ dosp, "vskip", 5 }, /* TEXICMD_VSKIP */
 	{ dobracket, "w", 1 }, /* TEXICMD_W */
 	{ dolink, "xref", 4 }, /* TEXICMD_XREF */
 };
@@ -369,7 +377,10 @@ texiexit(struct texi *p)
 
 	for (i = 0; i < p->dirsz; i++)
 		free(p->dirs[i]);
+
 	free(p->dirs);
+	free(p->subtitle);
+	free(p->title);
 }
 
 /*
@@ -436,6 +447,7 @@ texiputchar(struct texi *p, char c)
 	if (p->ign)
 		return;
 	putchar(c);
+	p->seenvs = 0;
 	if ('\n' == c) {
 		p->outcol = 0;
 		p->seenws = 0;
@@ -508,6 +520,16 @@ teximacro(struct texi *p, const char *s)
 	texiputchar(p, '\n');
 }
 
+static void
+texivspace(struct texi *p)
+{
+
+	if (p->seenvs)
+		return;
+	teximacro(p, "Pp");
+	p->seenvs = 1;
+}
+
 /*
  * Advance by a single byte in the input stream.
  */
@@ -524,6 +546,40 @@ advance(struct texi *p, const char *buf, size_t *pos)
 	(*pos)++;
 }
 
+static void
+texipunctuate(struct texi *p, const char *buf, size_t sz, size_t *pos)
+{
+	size_t	 start, end;
+
+	if (1 != p->outmacro)
+		return;
+
+	for (start = end = *pos; end < sz; end++) {
+		switch (buf[end]) {
+		case (','):
+		case (')'):
+		case ('.'):
+		case ('"'):
+		case (':'):
+		case ('!'):
+		case ('?'):
+			continue;
+		default:
+			break;
+		}
+		break;
+	}
+	if (end == *pos)
+		return;
+	if (end + 1 == sz || ' ' == buf[end] || '\n' == buf[end]) {
+		for ( ; start < end; start++) {
+			texiputchar(p, ' ');
+			texiputchar(p, buf[start]);
+			advance(p, buf, pos);
+		}
+	}
+}
+
 /*
  * Advance to the next non-whitespace word in the input stream.
  * If we're in literal mode, then print all of the whitespace as we're
@@ -534,7 +590,7 @@ advancenext(struct texi *p, const char *buf, size_t sz
 {
 	
 	if (p->literal) {
-		while (*pos < sz && isspace(buf[*pos])) {
+		while (*pos < sz && ismspace(buf[*pos])) {
 			if (*pos && '\n' == buf[*pos] && 
 				'\\' == buf[*pos - 1])
 				texiputchar(p, 'e');
@@ -544,16 +600,15 @@ advancenext(struct texi *p, const char *buf, size_t sz
 		return(*pos);
 	} 
 
-	while (*pos < sz && isspace(buf[*pos])) {
+	while (*pos < sz && ismspace(buf[*pos])) {
 		p->seenws = 1;
 		/* 
 		 * If it looks like we've printed a double-line, then
 		 * output a paragraph.
 		 * FIXME: this is stupid.
 		 */
-		if (*pos && '\n' == buf[*pos] && 
-			'\n' == buf[*pos - 1])
-			teximacro(p, "Pp");
+		if (*pos && '\n' == buf[*pos] && '\n' == buf[*pos - 1])
+			texivspace(p);
 		advance(p, buf, pos);
 	}
 	return(*pos);
@@ -606,7 +661,7 @@ texiword(struct texi *p, const char *buf, 
 
 	p->seenws = 0;
 
-	while (*pos < sz && ! isspace(buf[*pos])) {
+	while (*pos < sz && ! ismspace(buf[*pos])) {
 		switch (buf[*pos]) {
 		case ('@'):
 		case ('}'):
@@ -644,12 +699,15 @@ texicmd(struct texi *p, const char *buf, 
 
 	assert('@' == buf[pos]);
 
-	if (++pos >= sz)
+	if ((*end = pos) == sz)
 		return(TEXICMD__MAX);
+	else if ((*end = ++pos) == sz)
+		return(TEXICMD__MAX);
 
 	/* Alphabetic commands are special. */
 	if ( ! isalpha(buf[pos])) {
-		*end = pos + 1;
+		if ((*end = pos + 1) == sz)
+			return(TEXICMD__MAX);
 		for (i = 0; i < TEXICMD__MAX; i++) {
 			if (1 != texitoks[i].len)
 				continue;
@@ -660,7 +718,7 @@ texicmd(struct texi *p, const char *buf, 
 		return(TEXICMD__MAX);
 	}
 
-	for (*end = pos; *end < sz && ! isspace(buf[*end]); (*end)++)
+	for (*end = pos; *end < sz && ! ismspace(buf[*end]); (*end)++)
 		if ((*end > pos && ('@' == buf[*end] || 
 			  '{' == buf[*end] || '}' == buf[*end])))
 			break;
@@ -692,7 +750,7 @@ parsearg(struct texi *p, const char *buf, 
 	size_t		 end;
 	enum texicmd	 cmd;
 
-	while (*pos < sz && isspace(buf[*pos]))
+	while (*pos < sz && ismspace(buf[*pos]))
 		advance(p, buf, pos);
 	if (*pos == sz || (0 == num && '{' != buf[*pos]))
 		return(0);
@@ -739,7 +797,7 @@ parsebracket(struct texi *p, const char *buf, size_t s
 	size_t		 end;
 	enum texicmd	 cmd;
 
-	while (*pos < sz && isspace(buf[*pos]))
+	while (*pos < sz && ismspace(buf[*pos]))
 		advance(p, buf, pos);
 
 	if (*pos == sz || '{' != buf[*pos])
@@ -977,7 +1035,7 @@ dodeftypevar(struct texi *p, enum texicmd cmd, 
 		return;
 	}
 
-	teximacro(p, "Pp");
+	texivspace(p);
 	if (TEXICMD_DEFTYPEVR == cmd) {
 		parsebracket(p, buf, sz, pos);
 		texiputchars(p, ":\n");
@@ -985,7 +1043,7 @@ dodeftypevar(struct texi *p, enum texicmd cmd, 
 	teximacroopen(p, "Vt");
 	parseeoln(p, buf, sz, pos);
 	teximacroclose(p);
-	teximacro(p, "Pp");
+	texivspace(p);
 	parseto(p, buf, sz, pos, blk);
 }
 
@@ -1014,7 +1072,7 @@ dodeftypefun(struct texi *p, enum texicmd cmd, 
 	switch (cmd) {
 	case (TEXICMD_DEFTYPEFN):
 	case (TEXICMD_DEFTYPEFUN):
-		teximacro(p, "Pp");
+		texivspace(p);
 		break;
 	default:
 		break;
@@ -1033,7 +1091,7 @@ dodeftypefun(struct texi *p, enum texicmd cmd, 
 	teximacroopen(p, "Li");
 	parseeoln(p, buf, sz, pos);
 	teximacroclose(p);
-	teximacro(p, "Pp");
+	texivspace(p);
 	if (NULL != blk)
 		parseto(p, buf, sz, pos, blk);
 }
@@ -1064,13 +1122,7 @@ doinline(struct texi *p, const char *buf, 
 	teximacroopen(p, macro);
 	p->seenws = 0;
 	parsebracket(p, buf, sz, pos);
-	if (*pos < sz - 1 && 
-		 ismpunct(buf[*pos]) && 
-		 isspace(buf[*pos + 1])) {
-		texiputchar(p, ' ');
-		texiputchar(p, buf[*pos]);
-		advance(p, buf, pos);
-	}
+	texipunctuate(p, buf, sz, pos);
 	teximacroclose(p);
 }
 
@@ -1197,6 +1249,7 @@ dodisplay(struct texi *p, enum texicmd cmd, 
 {
 
 	teximacro(p, "Bd -display -offset indent");
+	p->seenvs = 1;
 	advanceeoln(p, buf, sz, pos, 1);
 	parseto(p, buf, sz, pos, "display");
 	teximacro(p, "Ed");
@@ -1228,6 +1281,23 @@ dobye(struct texi *p, enum texicmd cmd, 
 }
 
 static void
+dotitle(struct texi *p, enum texicmd cmd, 
+	const char *buf, size_t sz, size_t *pos)
+{
+	size_t	 start, end;
+
+	while (*pos < sz && isws(buf[*pos]))
+		advance(p, buf, pos);
+	start = end = *pos;
+	while (end < sz && '\n' != buf[end])
+		end++;
+	free(p->subtitle);
+	p->subtitle = malloc(end - start + 1);
+	memcpy(p->subtitle, &buf[start], end - start);
+	p->subtitle[end - start] = '\0';
+}
+
+static void
 dosymbol(struct texi *p, enum texicmd cmd, 
 	const char *buf, size_t sz, size_t *pos)
 {
@@ -1363,14 +1433,7 @@ dolink(struct texi *p, enum texicmd cmd, 
 		c = parsearg(p, buf, sz, pos, 1);
 	p->ign--;
 
-	if (*pos < sz - 1 && 
-		 ismpunct(buf[*pos]) && 
-		 isspace(buf[*pos + 1])) {
-		texiputchar(p, ' ');
-		texiputchar(p, buf[*pos]);
-		advance(p, buf, pos);
-	}
-
+	texipunctuate(p, buf, sz, pos);
 	teximacroclose(p);
 }
 
@@ -1392,11 +1455,11 @@ dosubsection(struct texi *p, enum texicmd cmd, 
 		const char *buf, size_t sz, size_t *pos)
 {
 
-	teximacro(p, "Pp");
+	texivspace(p);
 	teximacroopen(p, "Em");
 	parseeoln(p, buf, sz, pos);
 	teximacroclose(p);
-	teximacro(p, "Pp");
+	texivspace(p);
 }
 
 static void
@@ -1412,6 +1475,7 @@ dosection(struct texi *p, enum texicmd cmd, 
 	teximacroopen(p, "Ss");
 	parseeoln(p, buf, sz, pos);
 	teximacroclose(p);
+	p->seenvs = 1;
 }
 
 static void
@@ -1419,7 +1483,7 @@ dosp(struct texi *p, enum texicmd cmd, 
 	const char *buf, size_t sz, size_t *pos)
 {
 
-	teximacro(p, "Pp");
+	texivspace(p);
 	advanceeoln(p, buf, sz, pos, 1);
 }
 
@@ -1436,21 +1500,40 @@ dochapter(struct texi *p, enum texicmd cmd, 
 	teximacroopen(p, "Sh");
 	parseeoln(p, buf, sz, pos);
 	teximacroclose(p);
+	p->seenvs = 1;
 }
 
 static void
 dotop(struct texi *p, enum texicmd cmd, 
 	const char *buf, size_t sz, size_t *pos)
 {
+	const char	*cp;
+	time_t		 t;
+	char		 date[32];
 
+	t = time(NULL);
+	strftime(date, sizeof(date), "%F", localtime(&t));
+
 	p->ign--;
-	advanceeoln(p, buf, sz, pos, 1);
-	teximacro(p, "Dd $Mdocdate: February 18 2015 $");
-	teximacro(p, "Dt SOMETHING 7");
+	teximacroopen(p, "Dd");
+	texiputchars(p, date);
+	teximacroclose(p);
+	teximacroopen(p, "Dt");
+	for (cp = p->title; '\0' != *cp; cp++)
+		texiputchar(p, toupper(*cp));
+	texiputchars(p, " 7");
+	teximacroclose(p);
 	teximacro(p, "Os");
 	teximacro(p, "Sh NAME");
-	teximacro(p, "Nm Something");
-	teximacro(p, "Nd Something");
+	teximacroopen(p, "Nm");
+	texiputchars(p, p->title);
+	teximacroclose(p);
+	teximacroopen(p, "Nd");
+	texiputchars(p, NULL != p->subtitle ?
+		p->subtitle : "Unknown description");
+	teximacroclose(p);
+	p->seenvs = 1;
+	dochapter(p, cmd, buf, sz, pos);
 }
 
 static void
@@ -1471,15 +1554,16 @@ doitem(struct texi *p, enum texicmd cmd, 
 		teximacro(p, "It");
 		break;
 	default:
-		teximacro(p, "Pp");
+		texivspace(p);
 		break;
 	}
+	p->seenvs = 1;
 
 	parseeoln(p, buf, sz, pos);
 
 	if (TEXILIST_ITEM == p->list)
 		teximacroclose(p);
-	else
+	else if (p->outcol > 0)
 		texiputchar(p, '\n');
 }
 
@@ -1491,6 +1575,7 @@ dotable(struct texi *p, enum texicmd cmd, 
 
 	p->list = TEXILIST_ITEM;
 	teximacro(p, "Bl -tag -width Ds");
+	p->seenvs = 1;
 	parseto(p, buf, sz, pos, "table");
 	teximacro(p, "El");
 	p->list = sv;
@@ -1504,6 +1589,7 @@ doenumerate(struct texi *p, enum texicmd cmd, 
 
 	p->list = TEXILIST_NOITEM;
 	teximacro(p, "Bl -enum");
+	p->seenvs = 1;
 	parseto(p, buf, sz, pos, "enumerate");
 	teximacro(p, "El");
 	p->list = sv;
@@ -1517,6 +1603,7 @@ doitemize(struct texi *p, enum texicmd cmd, 
 
 	p->list = TEXILIST_ITEM;
 	teximacro(p, "Bl -bullet");
+	p->seenvs = 1;
 	parseto(p, buf, sz, pos, "itemize");
 	teximacro(p, "El");
 	p->list = sv;
@@ -1592,7 +1679,7 @@ main(int argc, char *argv[])
 	struct texi	 texi;
 	int		 c;
 	char		*path, *dir;
-	const char	*progname, *Idir;
+	const char	*progname, *Idir, *cp;
 
 	progname = strrchr(argv[0], '/');
 	if (progname == NULL)
@@ -1600,7 +1687,9 @@ main(int argc, char *argv[])
 	else
 		++progname;
 
+	memset(&texi, 0, sizeof(struct texi));
 	Idir = NULL;
+
 	while (-1 != (c = getopt(argc, argv, "I:"))) 
 		switch (c) {
 		case ('I'):
@@ -1624,7 +1713,17 @@ main(int argc, char *argv[])
 	}
 	free(path);
 
-	memset(&texi, 0, sizeof(struct texi));
+	if (NULL != (cp = strrchr(argv[0], '/'))) 
+		texi.title = strdup(cp + 1);
+	else
+		texi.title = strdup(argv[0]);
+
+	if (NULL == texi.title) {
+		perror(NULL);
+		exit(EXIT_FAILURE);
+	} else if (NULL != (path = strchr(texi.title, '.')))
+		*path = '\0';
+
 	texi.ign = 1;
 	texi.dirs = parsedirs(dir, Idir, &texi.dirsz);
 	parsefile(&texi, argv[0]);