http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=242599#21

* file.c
(is_html_type): added.
(examineFile, loadGeneralFile, _saveBuffer)
(openGeneralPagerBuffer, reloadBuffer): use is_html_type() instead of strcasecmp().
(loadGeneralFile): set f.guess_tupe
* display.c
(displayBuffer): use is_html_type() instead of strcasecmp().
* buffer.c
(reshapeBuffer): use is_html_type() instead of strcasecmp().
* backend.c
(internal_get): use is_html_type() instead of strcasecmp().

8 files changed, 52 insertions, 25 deletions

diff --git a/ChangeLog b/ChangeLog
index 7391a2f..b9dc043 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,16 @@
+2010-07-18 d+w3m@vdr.jp
+
+	* [w3m-dev 04319] Re: w3m's bugs from bugs.debian.org
+	http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=242599#21
+	* file.c (is_html_type): added.
+	(examineFile, loadGeneralFile, _saveBuffer)
+	(openGeneralPagerBuffer, reloadBuffer): use is_html_type() instead
+	of strcasecmp().
+	(loadGeneralFile): set f.guess_tupe
+	* display.c (displayBuffer): use is_html_type() instead of strcasecmp().
+	* buffer.c (reshapeBuffer): use is_html_type() instead of strcasecmp().
+	* backend.c (internal_get): use is_html_type() instead of strcasecmp().
+
 2010-07-18 Hironori SAKAMOTO <hsaka@mth.biglobe.ne.jp>
 
 	* [w3m-dev 04286] Re: break textform when buffer back
@@ -8951,4 +8964,4 @@ a	* [w3m-dev 03276] compile error on EWS4800
 	* release-0-2-1
 	* import w3m-0.2.1
 
-$Id: ChangeLog,v 1.1007 2010/07/18 13:48:48 htrb Exp $
+$Id: ChangeLog,v 1.1008 2010/07/18 14:10:09 htrb Exp $
diff --git a/backend.c b/backend.c
index 101e67e..c418095 100644
--- a/backend.c
+++ b/backend.c
@@ -1,4 +1,4 @@
-/* $Id: backend.c,v 1.13 2003/09/22 21:02:16 ukai Exp $ */
+/* $Id: backend.c,v 1.14 2010/07/18 14:10:09 htrb Exp $ */
 #include <stdio.h>
 #include <string.h>
 #include <sys/types.h>
@@ -95,7 +95,7 @@ internal_get(char *url, int flag, FormList *request)
     buf = loadGeneralFile(url, NULL, NO_REFERER, 0, request);
     do_download = FALSE;
     if (buf != NULL && buf != NO_BUFFER) {
-	if (!strcasecmp(buf->type, "text/html") && backend_halfdump_buf) {
+	if (is_html_type(buf->type) && backend_halfdump_buf) {
 	    TextLineListItem *p;
 	    Str first, last;
 	    int len = 0;
diff --git a/buffer.c b/buffer.c
index f258dd6..5afc26a 100644
--- a/buffer.c
+++ b/buffer.c
@@ -1,4 +1,4 @@
-/* $Id: buffer.c,v 1.29 2003/09/26 17:59:51 ukai Exp $ */
+/* $Id: buffer.c,v 1.30 2010/07/18 14:10:09 htrb Exp $ */
 #include "fm.h"
 
 #ifdef USE_MOUSE
@@ -558,7 +558,7 @@ reshapeBuffer(Buffer *buf)
     WcOption.auto_detect = WC_OPT_DETECT_OFF;
     UseContentCharset = FALSE;
 #endif
-    if (!strcasecmp(buf->type, "text/html"))
+    if (is_html_type(buf->type))
 	loadHTMLBuffer(&f, buf);
     else
 	loadBuffer(&f, buf);
@@ -590,7 +590,7 @@ reshapeBuffer(Buffer *buf)
 		gotoLine(buf, cur->linenumber);
 	}
 	buf->pos -= buf->currentLine->bpos;
-	if (FoldLine && strcasecmp(buf->type, "text/html"))
+	if (FoldLine && !is_html_type(buf->type))
 	    buf->currentColumn = 0;
 	else
 	    buf->currentColumn = sbuf.currentColumn;
diff --git a/display.c b/display.c
index 386c42f..e00eb0c 100644
--- a/display.c
+++ b/display.c
@@ -1,4 +1,4 @@
-/* $Id: display.c,v 1.70 2007/05/29 12:07:02 inu Exp $ */
+/* $Id: display.c,v 1.71 2010/07/18 14:10:09 htrb Exp $ */
 #include <signal.h>
 #include "fm.h"
 
@@ -380,7 +380,7 @@ displayBuffer(Buffer *buf, int mode)
     if (buf->height == 0)
 	buf->height = LASTLINE + 1;
     if ((buf->width != INIT_BUFFER_WIDTH &&
-	 ((buf->type && !strcmp(buf->type, "text/html")) || FoldLine))
+	 (is_html_type(buf->type) || FoldLine))
 	|| buf->need_reshape) {
 	buf->need_reshape = TRUE;
 	reshapeBuffer(buf);
diff --git a/file.c b/file.c
index 0e559c6..bbdd432 100644
--- a/file.c
+++ b/file.c
@@ -1,4 +1,4 @@
-/* $Id: file.c,v 1.254 2007/05/23 15:06:05 inu Exp $ */
+/* $Id: file.c,v 1.255 2010/07/18 14:10:09 htrb Exp $ */
 #include "fm.h"
 #include <sys/types.h>
 #include "myctype.h"
@@ -272,6 +272,13 @@ is_plain_text_type(char *type)
 	    (is_text_type(type) && !is_dump_text_type(type)));
 }
 
+int
+is_html_type(char *type)
+{
+    return (type && (strcasecmp(type, "text/html") == 0 ||
+		     strcasecmp(type, "application/xhtml+xml") == 0));
+}
+
 static void
 check_compression(char *path, URLFile *uf)
 {
@@ -373,7 +380,7 @@ examineFile(char *path, URLFile *uf)
 	    uf->guess_type = guessContentType(path);
 	    if (uf->guess_type == NULL)
 		uf->guess_type = "text/plain";
-	    if (strcasecmp(uf->guess_type, "text/html") == 0)
+	    if (is_html_type(uf->guess_type))
 		return;
 	    if ((fp = lessopen_stream(path))) {
 		UFclose(uf);
@@ -2055,6 +2062,10 @@ loadGeneralFile(char *path, ParsedURL *volatile current, char *referer,
 	    t = f.guess_type;
     }
 
+    /* XXX: can we use guess_type to give the type to loadHTMLstream
+     *      to support default utf8 encoding for XHTML here? */
+    f.guess_type = t;
+    
   page_loaded:
     if (page) {
 	FILE *src;
@@ -2166,7 +2177,7 @@ loadGeneralFile(char *path, ParsedURL *volatile current, char *referer,
     }
 #endif
 
-    if (!strcasecmp(t, "text/html"))
+    if (is_html_type(t))
 	proc = loadHTMLBuffer;
     else if (is_plain_text_type(t))
 	proc = loadBuffer;
@@ -2230,7 +2241,7 @@ loadGeneralFile(char *path, ParsedURL *volatile current, char *referer,
 	b->real_type = real_type;
 	if (b->currentURL.host == NULL && b->currentURL.file == NULL)
 	    copyParsedURL(&b->currentURL, &pu);
-	if (!strcasecmp(t, "text/html"))
+	if (is_html_type(t))
 	    b->type = "text/html";
 	else if (w3m_backend) {
 	    Str s = Strnew_charp(t);
@@ -6949,6 +6960,8 @@ loadHTMLstream(URLFile *f, Buffer *newBuf, FILE * src, int internal)
     }
     if (content_charset && UseContentCharset)
 	doc_charset = content_charset;
+    else if (f->guess_type && !strcasecmp(f->guess_type, "application/xhtml+xml"))
+	doc_charset = WC_CES_UTF_8;
     meta_charset = 0;
 #endif
 #if	0
@@ -7383,8 +7396,7 @@ _saveBuffer(Buffer *buf, Line *l, FILE * f, int cont)
     wc_ces charset = DisplayCharset ? DisplayCharset : WC_CES_US_ASCII;
 #endif
 
-    if (buf->type && !strcasecmp(buf->type, "text/html"))
-	is_html = TRUE;
+    is_html = is_html_type(buf->type);
 
   pager_next:
     for (; l != NULL; l = l->next) {
@@ -7541,7 +7553,7 @@ openGeneralPagerBuffer(InputStream stream)
 	t = DefaultType;
 	DefaultType = NULL;
     }
-    if (!strcasecmp(t, "text/html")) {
+    if (is_html_type(t)) {
 	buf = loadHTMLBuffer(&uf, t_buf);
 	buf->type = "text/html";
     }
@@ -8351,7 +8363,7 @@ reloadBuffer(Buffer *buf)
 	buf->hmarklist->nmark = 0;
     if (buf->imarklist)
 	buf->imarklist->nmark = 0;
-    if (!strcasecmp(buf->type, "text/html"))
+    if (is_html_type(buf->type))
 	loadHTMLBuffer(&uf, buf);
     else
 	loadBuffer(&uf, buf);
diff --git a/main.c b/main.c
index 361dc20..ceaff34 100644
--- a/main.c
+++ b/main.c
@@ -1,4 +1,4 @@
-/* $Id: main.c,v 1.259 2007/06/04 13:21:10 inu Exp $ */
+/* $Id: main.c,v 1.260 2010/07/18 14:10:09 htrb Exp $ */
 #define MAINPROGRAM
 #include "fm.h"
 #include <signal.h>
@@ -4616,10 +4616,10 @@ DEFUN(vwSrc, SOURCE VIEW, "View HTML source")
 
     buf = newBuffer(INIT_BUFFER_WIDTH);
 
-    if (!strcasecmp(Currentbuf->type, "text/html")) {
+    if (is_html_type(Currentbuf->type)) {
 	buf->type = "text/plain";
 	if (Currentbuf->real_type &&
-	    !strcasecmp(Currentbuf->real_type, "text/html"))
+	    is_html_type(Currentbuf->real_type))
 	    buf->real_type = "text/plain";
 	else
 	    buf->real_type = Currentbuf->real_type;
@@ -4769,8 +4769,8 @@ DEFUN(reload, RELOAD, "Reload buffer")
     repBuffer(Currentbuf, buf);
     if ((buf->type != NULL) && (sbuf.type != NULL) &&
 	((!strcasecmp(buf->type, "text/plain") &&
-	  !strcasecmp(sbuf.type, "text/html")) ||
-	 (!strcasecmp(buf->type, "text/html") &&
+	  is_html_type(sbuf.type)) ||
+	 (is_html_type(buf->type) &&
 	  !strcasecmp(sbuf.type, "text/plain")))) {
 	vwSrc();
 	if (Currentbuf != buf)
@@ -5092,7 +5092,7 @@ DEFUN(dispI, DISPLAY_IMAGE, "Restart loading and drawing of images")
 	return;
     displayImage = TRUE;
     /*
-     * if (!(Currentbuf->type && !strcmp(Currentbuf->type, "text/html")))
+     * if (!(Currentbuf->type && is_html_type(Currentbuf->type)))
      * return;
      */
     Currentbuf->image_flag = IMG_FLAG_AUTO;
@@ -5105,7 +5105,7 @@ DEFUN(stopI, STOP_IMAGE, "Stop loading and drawing of images")
     if (!activeImage)
 	return;
     /*
-     * if (!(Currentbuf->type && !strcmp(Currentbuf->type, "text/html")))
+     * if (!(Currentbuf->type && is_html_type(Currentbuf->type)))
      * return;
      */
     Currentbuf->image_flag = IMG_FLAG_SKIP;
diff --git a/proto.h b/proto.h
index b8f4885..ce7e85c 100644
--- a/proto.h
+++ b/proto.h
@@ -1,4 +1,4 @@
-/* $Id: proto.h,v 1.101 2006/04/07 13:21:12 inu Exp $ */
+/* $Id: proto.h,v 1.102 2010/07/18 14:10:09 htrb Exp $ */
 /* 
  *   This file was automatically generated by version 1.7 of cextract.
  *   Manual editing not recommended.
@@ -165,6 +165,7 @@ extern ParsedURL *schemeToProxy(int scheme);
 extern void examineFile(char *path, URLFile *uf);
 extern char *acceptableEncoding();
 extern int dir_exist(char *path);
+extern int is_html_type(char *type);
 #ifdef USE_M17N
 extern char **get_symbol(wc_ces charset, int *width);
 extern char **set_symbol(int width);
diff --git a/url.c b/url.c
index 4faafce..e9fc59d 100644
--- a/url.c
+++ b/url.c
@@ -1,4 +1,4 @@
-/* $Id: url.c,v 1.95 2007/05/23 15:06:06 inu Exp $ */
+/* $Id: url.c,v 1.96 2010/07/18 14:10:09 htrb Exp $ */
 #include "fm.h"
 #ifndef __MINGW32_VERSION
 #include <sys/types.h>
@@ -101,6 +101,7 @@ static struct table2 DefaultGuess[] = {
     {"html", "text/html"},
     {"htm", "text/html"},
     {"shtml", "text/html"},
+    {"xhtml", "application/xhtml+xml"},
     {"gif", "image/gif"},
     {"jpeg", "image/jpeg"},
     {"jpg", "image/jpeg"},