[w3m-dev 03616] Re: data: URL scheme

* file.c (loadGeneralFile): check SCM_DATA
	(loadImageBuffer): newBuffer()
* html.h (SCM_DATA): added
* indep.c (url_unquote): deleted
	(Str_url_unquote): renamed from Str_form_unquote
			+ is decoded is_form only
* indep.h (url_unquote): deleted
	(Str_url_unquote): added
	(Str_form_unquote): define by Str_url_unquote
* main.c (followA): file_unquote
	(cmd_loadURL): file_unquote
* url.c (DefaultPort): add for data:
	(schemetable): add "data"
	(DefaultFile): SCM_FTPDIR
	(parseURL): scheme copied from current
	(parseURL2): SCM_DATA
		check SCM_FTP, SCM_FTPDIR
	(_parsedURL2Str): add data in scheme_str
		handle SCM_DATA
		SCM_FTPDIR
	(openURL): file_unquote
		handle SCM_DATA
From: Hironori SAKAMOTO <hsaka@mth.biglobe.ne.jp>

7 files changed, 106 insertions, 51 deletions

diff --git a/ChangeLog b/ChangeLog
index 4faf61a..7a15c26 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,29 @@
+2003-01-09  Hironori SAKAMOTO <hsaka@mth.biglobe.ne.jp>
+
+	* [w3m-dev 03616] Re: data: URL scheme
+	* file.c (loadGeneralFile): check SCM_DATA
+		(loadImageBuffer): newBuffer()
+	* html.h (SCM_DATA): added
+	* indep.c (url_unquote): deleted
+		(Str_url_unquote): renamed from Str_form_unquote
+				+ is decoded is_form only
+	* indep.h (url_unquote): deleted
+		(Str_url_unquote): added
+		(Str_form_unquote): define by Str_url_unquote
+	* main.c (followA): file_unquote
+		(cmd_loadURL): file_unquote
+	* url.c (DefaultPort): add for data:
+		(schemetable): add "data"
+		(DefaultFile): SCM_FTPDIR
+		(parseURL): scheme copied from current
+		(parseURL2): SCM_DATA
+			check SCM_FTP, SCM_FTPDIR
+		(_parsedURL2Str): add data in scheme_str
+			handle SCM_DATA
+			SCM_FTPDIR
+		(openURL): file_unquote
+			handle SCM_DATA
+
 2003-01-08  Hironori SAKAMOTO <hsaka@mth.biglobe.ne.jp>
 
 	* [w3m-dev 03611] relative URL
@@ -6192,4 +6218,4 @@ a	* [w3m-dev 03276] compile error on EWS4800
 	* release-0-2-1
 	* import w3m-0.2.1
 
-$Id: ChangeLog,v 1.658 2003/01/07 15:53:42 ukai Exp $
+$Id: ChangeLog,v 1.659 2003/01/08 17:24:11 ukai Exp $
diff --git a/file.c b/file.c
index 876f35e..4345057 100644
--- a/file.c
+++ b/file.c
@@ -1,4 +1,4 @@
-/* $Id: file.c,v 1.175 2003/01/06 15:36:58 ukai Exp $ */
+/* $Id: file.c,v 1.176 2003/01/08 17:24:12 ukai Exp $ */
 #include "fm.h"
 #include <sys/types.h>
 #include "myctype.h"
@@ -1870,6 +1870,9 @@ loadGeneralFile(char *path, ParsedURL *volatile current, char *referer,
 	}
 #endif
     }
+    else if (pu.scheme == SCM_DATA) {
+	t = f.guess_type;
+    }
     else if (searchHeader) {
 	t_buf = newBuffer(INIT_BUFFER_WIDTH);
 	readHeader(&f, t_buf, searchHeader_through, &pu);
@@ -6867,13 +6870,13 @@ loadImageBuffer(URLFile *uf, Buffer *newBuf)
     cache->index = 0;
 
   image_buffer:
+    if (newBuf == NULL)
+	newBuf = newBuffer(INIT_BUFFER_WIDTH);
     cache->loaded |= IMG_FLAG_DONT_REMOVE;
     if (uf->scheme != SCM_LOCAL)
 	newBuf->sourcefile = cache->file;
 
     tmp = Sprintf("<img src=\"%s\"><br><br>", html_quote(image->url));
-    if (newBuf == NULL)
-	newBuf = newBuffer(INIT_BUFFER_WIDTH);
     tmpf = tmpfname(TMPF_SRC, ".html");
     src = fopen(tmpf->ptr, "w");
     newBuf->mailcap_source = tmpf->ptr;
diff --git a/html.h b/html.h
index 2be27a9..6f5ea11 100644
--- a/html.h
+++ b/html.h
@@ -1,4 +1,4 @@
-/* $Id: html.h,v 1.18 2003/01/06 15:36:59 ukai Exp $ */
+/* $Id: html.h,v 1.19 2003/01/08 17:24:12 ukai Exp $ */
 #ifndef _HTML_H
 #define _HTML_H
 #ifdef USE_SSL
@@ -364,9 +364,10 @@ struct environment {
 #define SCM_NNTP_GROUP	8
 #define SCM_NEWS	9
 #define SCM_NEWS_GROUP	10
-#define SCM_MAILTO      11
+#define SCM_DATA	11
+#define SCM_MAILTO      12
 #ifdef USE_SSL
-#define SCM_HTTPS       12
+#define SCM_HTTPS       13
 #endif				/* USE_SSL */
 
 #endif				/* _HTML_H */
diff --git a/indep.c b/indep.c
index dd62c9f..f28d6be 100644
--- a/indep.c
+++ b/indep.c
@@ -1,4 +1,4 @@
-/* $Id: indep.c,v 1.27 2002/12/24 17:20:47 ukai Exp $ */
+/* $Id: indep.c,v 1.28 2003/01/08 17:24:12 ukai Exp $ */
 #include "fm.h"
 #include <stdio.h>
 #include <pwd.h>
@@ -535,34 +535,6 @@ url_quote(char *str)
 }
 
 char *
-url_unquote(char *str)
-{
-    Str tmp = NULL;
-    char *p, *q;
-    int c;
-
-    for (p = str; *p;) {
-	if (*p == '%') {
-	    q = p;
-	    c = url_unquote_char(&q);
-	    if (c >= 0 && c != '\0' && c != '\n' && c != '\r') {
-		if (tmp == NULL)
-		    tmp = Strnew_charp_n(str, (int)(p - str));
-		Strcat_char(tmp, (char)c);
-		p = q;
-		continue;
-	    }
-	}
-	if (tmp)
-	    Strcat_char(tmp, *p);
-	p++;
-    }
-    if (tmp)
-	return tmp->ptr;
-    return str;
-}
-
-char *
 file_quote(char *str)
 {
     Str tmp = NULL;
@@ -661,15 +633,16 @@ Str_form_quote(Str x)
     return x;
 }
 
+
 Str
-Str_form_unquote(Str x)
+Str_url_unquote(Str x, int is_form)
 {
     Str tmp = NULL;
     char *p = x->ptr, *ep = x->ptr + x->length, *q;
     int c;
 
     for (; p < ep;) {
-	if (*p == '+') {
+	if (is_form && *p == '+') {
 	    if (tmp == NULL)
 		tmp = Strnew_charp_n(x->ptr, (int)(p - x->ptr));
 	    Strcat_char(tmp, ' ');
diff --git a/indep.h b/indep.h
index 1ada968..fc068ac 100644
--- a/indep.h
+++ b/indep.h
@@ -1,4 +1,4 @@
-/* $Id: indep.h,v 1.10 2002/11/09 21:55:24 ukai Exp $ */
+/* $Id: indep.h,v 1.11 2003/01/08 17:24:12 ukai Exp $ */
 #ifndef INDEP_H
 #define INDEP_H
 #include "gc.h"
@@ -45,9 +45,9 @@ extern char *html_unquote(char *str);
 extern char *file_quote(char *str);
 extern char *file_unquote(char *str);
 extern char *url_quote(char *str);
-extern char *url_unquote(char *str);
+extern Str Str_url_unquote(Str x, int is_form);
 extern Str Str_form_quote(Str x);
-extern Str Str_form_unquote(Str x);
+#define Str_form_unquote(x) Str_url_unquote((x), TRUE)
 extern char *shell_quote(char *str);
 
 extern char *w3m_auxbin_dir();
diff --git a/main.c b/main.c
index 486e360..86e46db 100644
--- a/main.c
+++ b/main.c
@@ -1,4 +1,4 @@
-/* $Id: main.c,v 1.184 2003/01/06 15:36:59 ukai Exp $ */
+/* $Id: main.c,v 1.185 2003/01/08 17:24:13 ukai Exp $ */
 #define MAINPROGRAM
 #include "fm.h"
 #include <signal.h>
@@ -2864,7 +2864,7 @@ followA(void)
 	    Strtruncate(to, pos - to->ptr);
 #endif
 	fmTerm();
-	system(myExtCommand(Mailer, shell_quote(url_unquote(to->ptr)),
+	system(myExtCommand(Mailer, shell_quote(file_unquote(to->ptr)),
 			    FALSE)->ptr);
 	fmInit();
 	displayBuffer(Currentbuf, B_FORCE_REDRAW);
@@ -3819,7 +3819,7 @@ cmd_loadURL(char *url, ParsedURL *current, char *referer)
 	    Strtruncate(to, pos - to->ptr);
 #endif
 	fmTerm();
-	system(myExtCommand(Mailer, shell_quote(url_unquote(to->ptr)),
+	system(myExtCommand(Mailer, shell_quote(file_unquote(to->ptr)),
 			    FALSE)->ptr);
 	fmInit();
 	displayBuffer(Currentbuf, B_FORCE_REDRAW);
diff --git a/url.c b/url.c
index 7216495..2c4b656 100644
--- a/url.c
+++ b/url.c
@@ -1,4 +1,4 @@
-/* $Id: url.c,v 1.65 2003/01/07 15:53:43 ukai Exp $ */
+/* $Id: url.c,v 1.66 2003/01/08 17:24:13 ukai Exp $ */
 #include "fm.h"
 #include <sys/types.h>
 #include <sys/socket.h>
@@ -56,6 +56,7 @@ static int
     119,			/* nntp group */
     119,			/* news */
     119,			/* news group */
+    0,				/* data - not defined */
     0,				/* mailto - not defined */
 #ifdef USE_SSL
     443,			/* https */
@@ -73,6 +74,7 @@ struct cmdtable schemetable[] = {
     /*  {"nntp", SCM_NNTP_GROUP}, */
     {"news", SCM_NEWS},
     /*  {"news", SCM_NEWS_GROUP}, */
+    {"data", SCM_DATA},
 #ifndef USE_W3MMAILER
     {"mailto", SCM_MAILTO},
 #endif
@@ -226,6 +228,7 @@ DefaultFile(int scheme)
     case SCM_LOCAL:
     case SCM_LOCAL_CGI:
     case SCM_FTP:
+    case SCM_FTPDIR:
 	return allocStr("/", -1);
     }
     return NULL;
@@ -708,9 +711,29 @@ parseURL(char *url, ParsedURL *p_url, ParsedURL *current)
 	 * denotes a filename (therefore the scheme is SCM_LOCAL).
 	 */
 	if (current) {
-	    p_url->scheme = current->scheme;
-	    if (p_url->scheme == SCM_LOCAL_CGI)
-		p_url->scheme = SCM_LOCAL;
+	    switch (current->scheme) {
+		case SCM_LOCAL:
+		case SCM_LOCAL_CGI:
+		    p_url->scheme = SCM_LOCAL;
+		    break;
+		case SCM_FTP:
+		case SCM_FTPDIR:
+		    p_url->scheme = SCM_FTP;
+		    break;
+#ifdef USE_NNTP
+		case SCM_NNTP:
+		case SCM_NNTP_GROUP:
+		    p_url->scheme = SCM_NNTP;
+		    break;
+		case SCM_NEWS:
+		case SCM_NEWS_GROUP:
+		    p_url->scheme = SCM_NEWS;
+		    break;
+#endif
+		default:
+		    p_url->scheme = current->scheme;
+		    break;
+	    }
 	}
 	else
 	    p_url->scheme = SCM_LOCAL;
@@ -970,6 +993,8 @@ parseURL2(char *url, ParsedURL *pu, ParsedURL *current)
     if (pu->scheme == SCM_MAILTO)
 	return;
 #endif
+    if (pu->scheme == SCM_DATA)
+	return;
     if (pu->scheme == SCM_NEWS || pu->scheme == SCM_NEWS_GROUP) {
 	if (pu->file && !strchr(pu->file, '@') &&
 	    (!(p = strchr(pu->file, '/')) || strchr(p + 1, '-') ||
@@ -1000,7 +1025,9 @@ parseURL2(char *url, ParsedURL *pu, ParsedURL *current)
     if (pu->scheme == SCM_LOCAL)
 	pu->file = expandName(pu->file);
 
-    if (current && pu->scheme == current->scheme && pu->host == NULL) {
+    if (current && (pu->scheme == current->scheme ||
+		    (pu->scheme == SCM_FTP && current->scheme == SCM_FTPDIR))
+	&& pu->host == NULL) {
 	/* Copy omitted element from the current URL */
 	pu->user = current->user;
 	pu->pass = current->pass;
@@ -1120,7 +1147,7 @@ _parsedURL2Str(ParsedURL *pu, int pass)
     Str tmp;
     static char *scheme_str[] = {
 	"http", "gopher", "ftp", "ftp", "file", "file", "exec", "nntp", "nntp",
-	"news", "news", "mailto",
+	"news", "news", "data", "mailto",
 #ifdef USE_SSL
 	"https",
 #endif				/* USE_SSL */
@@ -1156,6 +1183,10 @@ _parsedURL2Str(ParsedURL *pu, int pass)
 	return tmp;
     }
 #endif
+    if (pu->scheme == SCM_DATA) {
+	Strcat_charp(tmp, pu->file);
+	return tmp;
+    }
 #ifdef USE_NNTP
     if (pu->scheme != SCM_NEWS && pu->scheme != SCM_NEWS_GROUP)
 #endif				/* USE_NNTP */
@@ -1593,6 +1624,7 @@ openURL(char *url, ParsedURL *pu, ParsedURL *current,
 	}
 	return uf;
     case SCM_FTP:
+    case SCM_FTPDIR:
 	if (pu->file == NULL)
 	    pu->file = allocStr("/", -1);
 	if (non_null(FTP_proxy) &&
@@ -1769,7 +1801,7 @@ openURL(char *url, ParsedURL *pu, ParsedURL *current,
 		return uf;
 	    if (pu->file == NULL)
 		pu->file = "1";
-	    tmp = Strnew_charp(url_unquote(pu->file));
+	    tmp = Strnew_charp(file_unquote(pu->file));
 	    Strcat_char(tmp, '\n');
 	}
 	write(sock, tmp->ptr, tmp->length);
@@ -1787,6 +1819,26 @@ openURL(char *url, ParsedURL *pu, ParsedURL *current,
 	uf.stream = openNewsStream(pu);
 	return uf;
 #endif				/* USE_NNTP */
+    case SCM_DATA:
+	if (pu->file == NULL)
+	    return uf;
+	p = Strnew_charp(pu->file)->ptr;
+	q = strchr(p, ',');
+	if (q == NULL)
+	    return uf;
+	*q++ = '\0';
+	tmp = Strnew_charp(q);
+	q = strrchr(p, ';');
+	if (q != NULL && !strcmp(q, ";base64"))
+	{
+	    *q = '\0';
+	    uf.encoding = ENC_BASE64;
+	}
+	else
+	    tmp = Str_url_unquote(tmp, FALSE);
+	uf.stream = newStrStream(tmp);
+	uf.guess_type = (*p != '\0') ? p : "text/plain";
+	return uf;
     case SCM_UNKNOWN:
     default:
 	return uf;