===================================================================
RCS file: /cvs/mandoc/html.c,v
retrieving revision 1.31
retrieving revision 1.33
diff -u -p -r1.31 -r1.33
--- mandoc/html.c	2009/09/17 07:41:28	1.31
+++ mandoc/html.c	2009/09/17 13:17:30	1.33
@@ -1,4 +1,4 @@
-/*	$Id: html.c,v 1.31 2009/09/17 07:41:28 kristaps Exp $ */
+/*	$Id: html.c,v 1.33 2009/09/17 13:17:30 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -17,16 +17,22 @@
 #include <sys/queue.h>
 
 #include <assert.h>
+#include <ctype.h>
 #include <err.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 
+#include "chars.h"
 #include "mdoc.h"
 #include "man.h"
 
 #define	DOCTYPE		"-//W3C//DTD HTML 4.01//EN"
 #define	DTD		"http://www.w3.org/TR/html4/strict.dtd"
 
+#define	INDENT		 5
+#define	HALFINDENT	 3
+
 enum	htmltag {
 	TAG_HTML,
 	TAG_HEAD,
@@ -41,6 +47,10 @@ enum	htmltag {
 	TAG_LINK,
 	TAG_BR,
 	TAG_A,
+	TAG_TABLE,
+	TAG_COL,
+	TAG_TR,
+	TAG_TD,
 	TAG_MAX
 };
 
@@ -53,6 +63,9 @@ enum	htmlattr {
 	ATTR_TYPE,
 	ATTR_MEDIA,
 	ATTR_CLASS,
+	ATTR_STYLE,
+	ATTR_WIDTH,
+	ATTR_VALIGN,
 	ATTR_MAX
 };
 
@@ -68,7 +81,7 @@ static	const struct htmldata htmltags[TAG_MAX] = {
 	{"head",	HTML_CLRLINE}, /* TAG_HEAD */
 	{"body",	HTML_CLRLINE}, /* TAG_BODY */
 	{"meta",	HTML_CLRLINE | HTML_NOSTACK}, /* TAG_META */
-	{"title",	HTML_CLRLINE | HTML_NOSTACK}, /* TAG_TITLE */
+	{"title",	HTML_CLRLINE}, /* TAG_TITLE */
 	{"div",		HTML_CLRLINE}, /* TAG_DIV */
 	{"h1",		0}, /* TAG_H1 */
 	{"h2",		0}, /* TAG_H2 */
@@ -77,6 +90,10 @@ static	const struct htmldata htmltags[TAG_MAX] = {
 	{"link",	HTML_CLRLINE | HTML_NOSTACK}, /* TAG_LINK */
 	{"br",		HTML_CLRLINE | HTML_NOSTACK}, /* TAG_LINK */
 	{"a",		0}, /* TAG_A */
+	{"table",	HTML_CLRLINE}, /* TAG_TABLE */
+	{"col",		HTML_CLRLINE | HTML_NOSTACK}, /* TAG_COL */
+	{"tr",		HTML_CLRLINE}, /* TAG_TR */
+	{"td",		HTML_CLRLINE}, /* TAG_TD */
 };
 
 static	const char	 *const htmlattrs[ATTR_MAX] = {
@@ -87,7 +104,10 @@ static	const char	 *const htmlattrs[ATTR_MAX] = {
 	"href",
 	"type",
 	"media",
-	"class"
+	"class",
+	"style",
+	"width",
+	"valign",
 };
 
 struct	htmlpair {
@@ -107,6 +127,7 @@ struct	html {
 #define	HTML_NOSPACE	 (1 << 0)
 #define	HTML_NEWLINE	 (1 << 1)
 	struct tagq	  stack;
+	void		 *symtab;
 };
 
 #define	MDOC_ARGS	  const struct mdoc_meta *m, \
@@ -134,12 +155,23 @@ static	struct tag	 *print_otag(struct html *, enum htm
 static	void		  print_tagq(struct html *, const struct tag *);
 static	void		  print_stagq(struct html *, const struct tag *);
 static	void		  print_ctag(struct html *, enum htmltag);
-static	void		  print_encode(const char *);
+static	void		  print_encode(struct html *, const char *);
+static	void		  print_escape(struct html *, const char **);
 static	void		  print_text(struct html *, const char *);
+static	void		  print_res(struct html *, const char *, int);
+static	void		  print_spec(struct html *, const char *, int);
+
+static	int		  a2width(const char *);
+static	int		  a2offs(const char *);
+
+static	int		  mdoc_list_pre(MDOC_ARGS);
 static	int		  mdoc_root_pre(MDOC_ARGS);
+static	int		  mdoc_hang_pre(MDOC_ARGS);
 
 static	int		  mdoc_ar_pre(MDOC_ARGS);
+static	int		  mdoc_bl_pre(MDOC_ARGS);
 static	int		  mdoc_fl_pre(MDOC_ARGS);
+static	int		  mdoc_it_pre(MDOC_ARGS);
 static	int		  mdoc_nd_pre(MDOC_ARGS);
 static	int		  mdoc_nm_pre(MDOC_ARGS);
 static	int		  mdoc_ns_pre(MDOC_ARGS);
@@ -149,7 +181,13 @@ static	int		  mdoc_pp_pre(MDOC_ARGS);
 static	int		  mdoc_sh_pre(MDOC_ARGS);
 static	int		  mdoc_ss_pre(MDOC_ARGS);
 static	int		  mdoc_xr_pre(MDOC_ARGS);
+static	int		  mdoc_xx_pre(MDOC_ARGS);
 
+#ifdef __linux__
+extern	size_t	  strlcpy(char *, const char *, size_t);
+extern	size_t	  strlcat(char *, const char *, size_t);
+#endif
+
 static	const struct htmlmdoc mdocs[MDOC_MAX] = {
 	{NULL, NULL}, /* Ap */
 	{NULL, NULL}, /* Dd */
@@ -162,9 +200,9 @@ static	const struct htmlmdoc mdocs[MDOC_MAX] = {
 	{NULL, NULL}, /* Dl */
 	{NULL, NULL}, /* Bd */
 	{NULL, NULL}, /* Ed */
-	{NULL, NULL}, /* Bl */
+	{mdoc_bl_pre, NULL}, /* Bl */
 	{NULL, NULL}, /* El */
-	{NULL, NULL}, /* It */
+	{mdoc_it_pre, NULL}, /* It */
 	{NULL, NULL}, /* Ad */ 
 	{NULL, NULL}, /* An */
 	{mdoc_ar_pre, NULL}, /* Ar */
@@ -211,7 +249,7 @@ static	const struct htmlmdoc mdocs[MDOC_MAX] = {
 	{NULL, NULL}, /* Bf */ 
 	{NULL, NULL}, /* Bo */
 	{NULL, NULL}, /* Bq */
-	{NULL, NULL}, /* Bsx */
+	{mdoc_xx_pre, NULL}, /* Bsx */
 	{NULL, NULL}, /* Bx */
 	{NULL, NULL}, /* Db */
 	{NULL, NULL}, /* Dc */
@@ -221,12 +259,12 @@ static	const struct htmlmdoc mdocs[MDOC_MAX] = {
 	{NULL, NULL}, /* Ef */
 	{NULL, NULL}, /* Em */ 
 	{NULL, NULL}, /* Eo */
-	{NULL, NULL}, /* Fx */
+	{mdoc_xx_pre, NULL}, /* Fx */
 	{NULL, NULL}, /* Ms */
 	{NULL, NULL}, /* No */
 	{mdoc_ns_pre, NULL}, /* Ns */
-	{NULL, NULL}, /* Nx */
-	{NULL, NULL}, /* Ox */
+	{mdoc_xx_pre, NULL}, /* Nx */
+	{mdoc_xx_pre, NULL}, /* Ox */
 	{NULL, NULL}, /* Pc */
 	{NULL, NULL}, /* Pf */
 	{NULL, NULL}, /* Po */
@@ -244,7 +282,7 @@ static	const struct htmlmdoc mdocs[MDOC_MAX] = {
 	{NULL, NULL}, /* Sx */
 	{NULL, NULL}, /* Sy */
 	{NULL, NULL}, /* Tn */
-	{NULL, NULL}, /* Ux */
+	{mdoc_xx_pre, NULL}, /* Ux */
 	{NULL, NULL}, /* Xc */
 	{NULL, NULL}, /* Xo */
 	{NULL, NULL}, /* Fo */ 
@@ -267,7 +305,7 @@ static	const struct htmlmdoc mdocs[MDOC_MAX] = {
 	{NULL, NULL}, /* %C */ 
 	{NULL, NULL}, /* Es */ 
 	{NULL, NULL}, /* En */ 
-	{NULL, NULL}, /* Dx */ 
+	{mdoc_xx_pre, NULL}, /* Dx */ 
 	{NULL, NULL}, /* %Q */ 
 	{NULL, NULL}, /* br */
 	{NULL, NULL}, /* sp */ 
@@ -317,6 +355,10 @@ html_alloc(void)
 		return(NULL);
 
 	SLIST_INIT(&h->stack);
+	if (NULL == (h->symtab = chars_init(CHARS_HTML))) {
+		free(h);
+		return(NULL);
+	}
 	return(h);
 }
 
@@ -393,7 +435,7 @@ print_mdoc_head(MDOC_ARGS)
 
 	print_gen_head(h);
 	print_otag(h, TAG_TITLE, 0, NULL);
-	print_encode(m->title);
+	print_encode(h, m->title);
 }
 
 
@@ -471,7 +513,7 @@ print_man_head(MAN_ARGS)
 
 	print_gen_head(h);
 	print_otag(h, TAG_TITLE, 0, NULL);
-	print_encode(m->title);
+	print_encode(h, m->title);
 }
 
 
@@ -485,13 +527,151 @@ print_man_body(MAN_ARGS)
 
 
 static void
-print_encode(const char *p)
+print_spec(struct html *h, const char *p, int len)
 {
+	const char	*rhs;
+	int		 i;
+	size_t		 sz;
 
-	printf("%s", p); /* XXX */
+	rhs = chars_a2ascii(h->symtab, p, (size_t)len, &sz);
+
+	if (NULL == rhs) 
+		return;
+	for (i = 0; i < (int)sz; i++) 
+		putchar(rhs[i]);
 }
 
 
+static void
+print_res(struct html *h, const char *p, int len)
+{
+	const char	*rhs;
+	int		 i;
+	size_t		 sz;
+
+	rhs = chars_a2res(h->symtab, p, (size_t)len, &sz);
+
+	if (NULL == rhs)
+		return;
+	for (i = 0; i < (int)sz; i++) 
+		putchar(rhs[i]);
+}
+
+
+static void
+print_escape(struct html *h, const char **p)
+{
+	int		 j, type;
+	const char	*wp;
+
+	wp = *p;
+	type = 1;
+
+	if (0 == *(++wp)) {
+		*p = wp;
+		return;
+	}
+
+	if ('(' == *wp) {
+		wp++;
+		if (0 == *wp || 0 == *(wp + 1)) {
+			*p = 0 == *wp ? wp : wp + 1;
+			return;
+		}
+
+		print_spec(h, wp, 2);
+		*p = ++wp;
+		return;
+
+	} else if ('*' == *wp) {
+		if (0 == *(++wp)) {
+			*p = wp;
+			return;
+		}
+
+		switch (*wp) {
+		case ('('):
+			wp++;
+			if (0 == *wp || 0 == *(wp + 1)) {
+				*p = 0 == *wp ? wp : wp + 1;
+				return;
+			}
+
+			print_res(h, wp, 2);
+			*p = ++wp;
+			return;
+		case ('['):
+			type = 0;
+			break;
+		default:
+			print_res(h, wp, 1);
+			*p = wp;
+			return;
+		}
+	
+	} else if ('f' == *wp) {
+		if (0 == *(++wp)) {
+			*p = wp;
+			return;
+		}
+
+		switch (*wp) {
+		case ('B'):
+			/* TODO */
+			break;
+		case ('I'):
+			/* TODO */
+			break;
+		case ('P'):
+			/* FALLTHROUGH */
+		case ('R'):
+			/* TODO */
+			break;
+		default:
+			break;
+		}
+
+		*p = wp;
+		return;
+
+	} else if ('[' != *wp) {
+		print_spec(h, wp, 1);
+		*p = wp;
+		return;
+	}
+
+	wp++;
+	for (j = 0; *wp && ']' != *wp; wp++, j++)
+		/* Loop... */ ;
+
+	if (0 == *wp) {
+		*p = wp;
+		return;
+	}
+
+	if (type)
+		print_spec(h, wp - j, j);
+	else
+		print_res(h, wp - j, j);
+
+	*p = wp;
+}
+
+
+static void
+print_encode(struct html *h, const char *p)
+{
+
+	for (; *p; p++) {
+		if ('\\' != *p) {
+			putchar(*p);
+			continue;
+		}
+		print_escape(h, &p);
+	}
+}
+
+
 static struct tag *
 print_otag(struct html *h, enum htmltag tag, 
 		int sz, const struct htmlpair *p)
@@ -515,7 +695,7 @@ print_otag(struct html *h, enum htmltag tag, 
 	for (i = 0; i < sz; i++) {
 		printf(" %s=\"", htmlattrs[p[i].key]);
 		assert(p->val);
-		print_encode(p[i].val);
+		print_encode(h, p[i].val);
 		printf("\"");
 	}
 	printf(">");
@@ -590,7 +770,7 @@ print_text(struct html *h, const char *p)
 	h->flags &= ~HTML_NEWLINE;
 
 	if (p)
-		print_encode(p);
+		print_encode(h, p);
 
 	if (*p && 0 == *(p + 1))
 		switch (*p) {
@@ -639,6 +819,54 @@ print_stagq(struct html *h, const struct tag *suntil)
 }
 
 
+static int
+a2offs(const char *p)
+{
+	int		 len, i;
+
+	if (0 == strcmp(p, "left"))
+		return(0);
+	if (0 == strcmp(p, "indent"))
+		return(INDENT + 1);
+	if (0 == strcmp(p, "indent-two"))
+		return((INDENT + 1) * 2);
+
+	if (0 == (len = (int)strlen(p)))
+		return(0);
+
+	for (i = 0; i < len - 1; i++) 
+		if ( ! isdigit((u_char)p[i]))
+			break;
+
+	if (i == len - 1) 
+		if ('n' == p[len - 1] || 'm' == p[len - 1])
+			return(atoi(p));
+
+	return(len);
+}
+
+
+static int
+a2width(const char *p)
+{
+	int		 i, len;
+
+	if (0 == (len = (int)strlen(p)))
+		return(0);
+	for (i = 0; i < len - 1; i++) 
+		if ( ! isdigit((u_char)p[i]))
+			break;
+
+	if (i == len - 1) 
+		if ('n' == p[len - 1] || 'm' == p[len - 1])
+			return(atoi(p) + 2);
+
+	return(len + 2);
+}
+
+
+
+
 /* ARGSUSED */
 static int
 mdoc_root_pre(MDOC_ARGS)
@@ -659,7 +887,7 @@ mdoc_ss_pre(MDOC_ARGS)
 {
 
 	if (MDOC_BODY == n->type)
-		print_otag(h, TAG_P, 0, NULL);
+		print_otag(h, TAG_DIV, 0, NULL);
 	if (MDOC_HEAD == n->type)
 		print_otag(h, TAG_H2, 0, NULL);
 	return(1);
@@ -755,7 +983,7 @@ mdoc_sh_pre(MDOC_ARGS)
 {
 
 	if (MDOC_BODY == n->type)
-		print_otag(h, TAG_P, 0, NULL);
+		print_otag(h, TAG_DIV, 0, NULL);
 	if (MDOC_HEAD == n->type)
 		print_otag(h, TAG_H1, 0, NULL);
 	return(1);
@@ -798,7 +1026,6 @@ mdoc_ns_pre(MDOC_ARGS)
 	return(1);
 }
 
-
 /* ARGSUSED */
 static int
 mdoc_ar_pre(MDOC_ARGS)
@@ -810,4 +1037,191 @@ mdoc_ar_pre(MDOC_ARGS)
 
 	print_otag(h, TAG_SPAN, 1, &tag);
 	return(1);
+}
+
+/* ARGSUSED */
+static int
+mdoc_xx_pre(MDOC_ARGS)
+{
+	const char	*pp;
+
+	switch (n->tok) {
+	case (MDOC_Bsx):
+		pp = "BSDI BSD/OS";
+		break;
+	case (MDOC_Dx):
+		pp = "DragonFlyBSD";
+		break;
+	case (MDOC_Fx):
+		pp = "FreeBSD";
+		break;
+	case (MDOC_Nx):
+		pp = "NetBSD";
+		break;
+	case (MDOC_Ox):
+		pp = "OpenBSD";
+		break;
+	case (MDOC_Ux):
+		pp = "UNIX";
+		break;
+	default:
+		return(1);
+	}
+
+	print_text(h, pp);
+	return(1);
+}
+
+
+static int
+mdoc_hang_pre(MDOC_ARGS)
+{
+	int			 i, width, offs;
+	struct htmlpair		 tag;
+	char			 buf[BUFSIZ];
+	const struct mdoc_node	*bl;
+
+	if (MDOC_BODY == n->type) {
+		print_otag(h, TAG_DIV, 0, NULL);
+		return(1);
+	}
+
+	bl = n->parent->parent;
+	if (MDOC_BLOCK != n->type) 
+		bl = bl->parent;
+
+	assert(bl->args);
+
+	for (width = i = 0; i < (int)bl->args->argc; i++) 
+		if (MDOC_Width == bl->args->argv[i].arg) {
+			assert(bl->args->argv[i].sz);
+			width = a2width(bl->args->argv[i].value[0]);
+			break;
+		} else if (MDOC_Offset == bl->args->argv[i].arg) {
+			assert(bl->args->argv[i].sz);
+			offs = a2offs(bl->args->argv[i].value[0]);
+			break;
+		}
+
+	if (0 == width)
+		width = 10;
+
+	width *= 10;
+
+	if (MDOC_BLOCK == n->type)
+		snprintf(buf, BUFSIZ - 1, "margin-left: %dpx; "
+				"clear: both;", width);
+	else
+		snprintf(buf, BUFSIZ - 1, "float: left; "
+				"margin-left: -%dpx;", width);
+
+	tag.key = ATTR_STYLE;
+	tag.val = buf;
+
+	print_otag(h, TAG_DIV, 1, &tag);
+
+	return(1);
+}
+
+
+static int
+mdoc_list_pre(MDOC_ARGS)
+{
+
+	/* TODO */
+	return(0);
+}
+
+
+static int
+mdoc_it_pre(MDOC_ARGS)
+{
+	int		 	 i, len;
+	const struct mdoc_node	*bl;
+
+	if (MDOC_BLOCK == n->type)
+		bl = n->parent->parent;
+	else
+		bl = n->parent->parent->parent;
+
+	assert(bl->args);
+	len = (int)bl->args->argc;
+
+	for (i = 0; i < len; i++) 
+		switch (bl->args->argv[i].arg) {
+		case (MDOC_Bullet):
+			/* FALLTHROUGH */
+		case (MDOC_Dash):
+			/* FALLTHROUGH */
+		case (MDOC_Enum):
+			/* FALLTHROUGH */
+		case (MDOC_Hyphen):
+			return(0); /* TODO */
+		case (MDOC_Tag):
+			return(0);
+		case (MDOC_Inset):
+			/* FALLTHROUGH */
+		case (MDOC_Diag):
+			/* FALLTHROUGH */
+		case (MDOC_Item):
+			/* FALLTHROUGH */
+		case (MDOC_Column):
+			/* FALLTHROUGH */
+		case (MDOC_Hang):
+			return(mdoc_hang_pre(m, n, h));
+			/* FALLTHROUGH */
+		case (MDOC_Ohang):
+			return(0); /* TODO */
+		default:
+			abort();
+			/* NOTREACHED */
+		}
+
+	abort();
+	/* NOTREACHED */
+}
+
+
+static int
+mdoc_bl_pre(MDOC_ARGS)
+{
+	int		 i, len;
+
+	if (MDOC_BLOCK != n->type)
+		return(1);
+
+	assert(n->args);
+	len = (int)n->args->argc;
+
+	for (i = 0; i < len; i++) 
+		switch (n->args->argv[i].arg) {
+		case (MDOC_Bullet):
+			/* FALLTHROUGH */
+		case (MDOC_Dash):
+			/* FALLTHROUGH */
+		case (MDOC_Enum):
+			/* FALLTHROUGH */
+		case (MDOC_Hyphen):
+			return(mdoc_list_pre(m, n, h));
+		case (MDOC_Tag):
+			return(0);
+		case (MDOC_Inset):
+			/* FALLTHROUGH */
+		case (MDOC_Diag):
+			/* FALLTHROUGH */
+		case (MDOC_Item):
+			/* FALLTHROUGH */
+		case (MDOC_Column):
+			/* FALLTHROUGH */
+		case (MDOC_Hang):
+			return(1);
+		case (MDOC_Ohang):
+			return(0); /* TODO */
+		default:
+			abort();
+			/* NOTREACHED */
+		}
+
+	abort();
+	/* NOTREACHED */
 }