diff libpurple/util.c @ 18041:678d78b7fa34

propagate from branch 'im.pidgin.pidgin' (head a58972b72c7aa0fa0899c5a6b96e51cd6c427ab4) to branch 'im.pidgin.pidgin.2.1.0' (head 03df10bd904eed59317242a557aed2b8430d9630)

--- a/libpurple/util.c	Mon Jun 04 05:47:51 2007 +0000
+++ b/libpurple/util.c	Mon Jun 04 05:55:13 2007 +0000
@@ -47,6 +47,7 @@
 	} website;
 
 	char *url;
+	int num_times_redirected;
 	gboolean full;
 	char *user_agent;
 	gboolean http11;
@@ -1285,14 +1286,17 @@
 								pt->dest_tag = y; \
 								tags = g_list_prepend(tags, pt); \
 							} \
-							xhtml = g_string_append(xhtml, "<" y); \
-							c += strlen("<" x ); \
-							xhtml = g_string_append(xhtml, innards->str); \
-							xhtml = g_string_append_c(xhtml, '>'); \
+							if(xhtml) { \
+								xhtml = g_string_append(xhtml, "<" y); \
+								xhtml = g_string_append(xhtml, innards->str); \
+								xhtml = g_string_append_c(xhtml, '>'); \
+							} \
 							c = p + 1; \
 						} else { \
-							xhtml = g_string_append(xhtml, "&lt;"); \
-							plain = g_string_append_c(plain, '<'); \
+							if(xhtml) \
+								xhtml = g_string_append(xhtml, "&lt;"); \
+							if(plain) \
+								plain = g_string_append_c(plain, '<'); \
 							c++; \
 						} \
 						g_string_free(innards, TRUE); \
@@ -1301,16 +1305,19 @@
 						if(!g_ascii_strncasecmp(c, "<" x, strlen("<" x)) && \
 								(*(c+strlen("<" x)) == '>' || \
 								 !g_ascii_strncasecmp(c+strlen("<" x), "/>", 2))) { \
-							xhtml = g_string_append(xhtml, "<" y); \
+							if(xhtml) \
+								xhtml = g_string_append(xhtml, "<" y); \
 							c += strlen("<" x); \
 							if(*c != '/') { \
 								struct purple_parse_tag *pt = g_new0(struct purple_parse_tag, 1); \
 								pt->src_tag = x; \
 								pt->dest_tag = y; \
 								tags = g_list_prepend(tags, pt); \
-								xhtml = g_string_append_c(xhtml, '>'); \
+								if(xhtml) \
+									xhtml = g_string_append_c(xhtml, '>'); \
 							} else { \
-								xhtml = g_string_append(xhtml, "/>");\
+								if(xhtml) \
+									xhtml = g_string_append(xhtml, "/>");\
 							} \
 							c = strchr(c, '>') + 1; \
 							continue; \
@@ -1320,11 +1327,18 @@
 purple_markup_html_to_xhtml(const char *html, char **xhtml_out,
 						  char **plain_out)
 {
-	GString *xhtml = g_string_new("");
-	GString *plain = g_string_new("");
+	GString *xhtml = NULL;
+	GString *plain = NULL;
 	GList *tags = NULL, *tag;
 	const char *c = html;
 
+	g_return_if_fail(xhtml_out != NULL || plain_out != NULL);
+
+	if(xhtml_out)
+		xhtml = g_string_new("");
+	if(plain_out)
+		plain = g_string_new("");
+
 	while(c && *c) {
 		if(*c == '<') {
 			if(*(c+1) == '/') { /* closing tag */
@@ -1340,7 +1354,8 @@
 				if(tag) {
 					while(tags) {
 						struct purple_parse_tag *pt = tags->data;
-						g_string_append_printf(xhtml, "</%s>", pt->dest_tag);
+						if(xhtml)
+							g_string_append_printf(xhtml, "</%s>", pt->dest_tag);
 						if(tags == tag)
 							break;
 						tags = g_list_remove(tags, pt);
@@ -1358,8 +1373,10 @@
 					if(*end == '>') {
 						c = end+1;
 					} else {
-						xhtml = g_string_append(xhtml, "&lt;");
-						plain = g_string_append_c(plain, '<');
+						if(xhtml)
+							xhtml = g_string_append(xhtml, "&lt;");
+						if(plain)
+							plain = g_string_append_c(plain, '<');
 						c++;
 					}
 				}
@@ -1388,7 +1405,8 @@
 				ALLOW_TAG("span");
 				ALLOW_TAG("strong");
 				ALLOW_TAG("ul");
-
+				ALLOW_TAG("img");
+				
 				/* we skip <HR> because it's not legal in XHTML-IM.  However,
 				 * we still want to send something sensible, so we put a
 				 * linebreak in its place. <BR> also needs special handling
@@ -1399,8 +1417,9 @@
 							!g_ascii_strncasecmp(c+3, "/>", 2) ||
 							!g_ascii_strncasecmp(c+3, " />", 3))) {
 					c = strchr(c, '>') + 1;
-					xhtml = g_string_append(xhtml, "<br/>");
-					if(*c != '\n')
+					if(xhtml)
+						xhtml = g_string_append(xhtml, "<br/>");
+					if(plain && *c != '\n')
 						plain = g_string_append_c(plain, '\n');
 					continue;
 				}
@@ -1444,7 +1463,8 @@
 					pt->dest_tag = "span";
 					tags = g_list_prepend(tags, pt);
 					c = strchr(c, '>') + 1;
-					xhtml = g_string_append(xhtml, "<span style='font-weight: bold;'>");
+					if(xhtml)
+						xhtml = g_string_append(xhtml, "<span style='font-weight: bold;'>");
 					continue;
 				}
 				if(!g_ascii_strncasecmp(c, "<u>", 3) || !g_ascii_strncasecmp(c, "<underline>", strlen("<underline>"))) {
@@ -1453,7 +1473,8 @@
 					pt->dest_tag = "span";
 					tags = g_list_prepend(tags, pt);
 					c = strchr(c, '>') + 1;
-					xhtml = g_string_append(xhtml, "<span style='text-decoration: underline;'>");
+					if (xhtml)
+						xhtml = g_string_append(xhtml, "<span style='text-decoration: underline;'>");
 					continue;
 				}
 				if(!g_ascii_strncasecmp(c, "<s>", 3) || !g_ascii_strncasecmp(c, "<strike>", strlen("<strike>"))) {
@@ -1462,7 +1483,8 @@
 					pt->dest_tag = "span";
 					tags = g_list_prepend(tags, pt);
 					c = strchr(c, '>') + 1;
-					xhtml = g_string_append(xhtml, "<span style='text-decoration: line-through;'>");
+					if(xhtml)
+						xhtml = g_string_append(xhtml, "<span style='text-decoration: line-through;'>");
 					continue;
 				}
 				if(!g_ascii_strncasecmp(c, "<sub>", 5)) {
@@ -1471,7 +1493,8 @@
 					pt->dest_tag = "span";
 					tags = g_list_prepend(tags, pt);
 					c = strchr(c, '>') + 1;
-					xhtml = g_string_append(xhtml, "<span style='vertical-align:sub;'>");
+					if(xhtml)
+						xhtml = g_string_append(xhtml, "<span style='vertical-align:sub;'>");
 					continue;
 				}
 				if(!g_ascii_strncasecmp(c, "<sup>", 5)) {
@@ -1480,7 +1503,8 @@
 					pt->dest_tag = "span";
 					tags = g_list_prepend(tags, pt);
 					c = strchr(c, '>') + 1;
-					xhtml = g_string_append(xhtml, "<span style='vertical-align:super;'>");
+					if(xhtml)
+						xhtml = g_string_append(xhtml, "<span style='vertical-align:super;'>");
 					continue;
 				}
 				if(!g_ascii_strncasecmp(c, "<font", 5) && (*(c+5) == '>' || *(c+5) == ' ')) {
@@ -1574,7 +1598,10 @@
 					pt->dest_tag = "span";
 					tags = g_list_prepend(tags, pt);
 					if(style->len)
-						g_string_append_printf(xhtml, "<span style='%s'>", g_strstrip(style->str));
+					{
+						if(xhtml)
+							g_string_append_printf(xhtml, "<span style='%s'>", g_strstrip(style->str));
+					}
 					else
 						pt->ignore = TRUE;
 					g_string_free(style, TRUE);
@@ -1594,7 +1621,8 @@
 								color = g_string_append_c(color, *q);
 								q++;
 							}
-							g_string_append_printf(xhtml, "<span style='background: %s;'>", g_strstrip(color->str));
+							if(xhtml)
+								g_string_append_printf(xhtml, "<span style='background: %s;'>", g_strstrip(color->str));
 							g_string_free(color, TRUE);
 							if ((c = strchr(c, '>')) != NULL)
 								c++;
@@ -1615,14 +1643,17 @@
 				if(!g_ascii_strncasecmp(c, "<!--", strlen("<!--"))) {
 					char *p = strstr(c + strlen("<!--"), "-->");
 					if(p) {
-						xhtml = g_string_append(xhtml, "<!--");
+						if(xhtml)
+							xhtml = g_string_append(xhtml, "<!--");
 						c += strlen("<!--");
 						continue;
 					}
 				}
 
-				xhtml = g_string_append(xhtml, "&lt;");
-				plain = g_string_append_c(plain, '<');
+				if(xhtml)
+					xhtml = g_string_append(xhtml, "&lt;");
+				if(plain)
+					plain = g_string_append_c(plain, '<');
 				c++;
 			}
 		} else if(*c == '&') {
@@ -1635,29 +1666,31 @@
 				g_snprintf(buf, sizeof(buf), "%c", *c);
 				pln = buf;
 			}
-			xhtml = g_string_append_len(xhtml, c, len);
-			plain = g_string_append(plain, pln);
+			if(xhtml)
+				xhtml = g_string_append_len(xhtml, c, len);
+			if(plain)
+				plain = g_string_append(plain, pln);
 			c += len;
 		} else {
-			xhtml = g_string_append_c(xhtml, *c);
-			plain = g_string_append_c(plain, *c);
+			if(xhtml)
+				xhtml = g_string_append_c(xhtml, *c);
+			if(plain)
+				plain = g_string_append_c(plain, *c);
 			c++;
 		}
 	}
-	tag = tags;
-	while(tag) {
-		struct purple_parse_tag *pt = tag->data;
-		if(!pt->ignore)
-			g_string_append_printf(xhtml, "</%s>", pt->dest_tag);
-		tag = tag->next;
+	if(xhtml) {
+		for (tag = tags; tag ; tag = tag->next) {
+			struct purple_parse_tag *pt = tag->data;
+			if(!pt->ignore)
+				g_string_append_printf(xhtml, "</%s>", pt->dest_tag);
+		}
 	}
 	g_list_free(tags);
 	if(xhtml_out)
-		*xhtml_out = g_strdup(xhtml->str);
+		*xhtml_out = g_string_free(xhtml, FALSE);
 	if(plain_out)
-		*plain_out = g_strdup(plain->str);
-	g_string_free(xhtml, TRUE);
-	g_string_free(plain, TRUE);
+		*plain_out = g_string_free(plain, FALSE);
 }
 
 /* The following are probably reasonable changes:
@@ -3273,6 +3306,17 @@
 		g_hash_table_destroy(params);
 }
 
+/*
+ * TODO: Should probably add a "gboolean *ret_ishttps" parameter that
+ *       is set to TRUE if this URL is https, otherwise it is set to
+ *       FALSE.  But that change will break the API.
+ *
+ *       This is important for Yahoo! web messenger login.  They now
+ *       force https login, and if you access the web messenger login
+ *       page via http then it redirects you to the https version, but
+ *       purple_util_fetch_url() ignores the "https" and attempts to
+ *       fetch the URL via http again, which gets redirected again.
+ */
 gboolean
 purple_url_parse(const char *url, char **ret_host, int *ret_port,
 			   char **ret_path, char **ret_user, char **ret_passwd)
@@ -3293,12 +3337,16 @@
 
 	g_return_val_if_fail(url != NULL, FALSE);
 
-	if ((turl = strstr(url, "http://")) != NULL ||
-		(turl = strstr(url, "HTTP://")) != NULL)
+	if ((turl = purple_strcasestr(url, "http://")) != NULL)
 	{
 		turl += 7;
 		url = turl;
 	}
+	else if ((turl = purple_strcasestr(url, "https://")) != NULL)
+	{
+		turl += 8;
+		url = turl;
+	}
 
 	/* parse out authentication information if supplied */
 	/* Only care about @ char BEFORE the first / */
@@ -3378,85 +3426,92 @@
 			   PurpleUtilFetchUrlData *gfud)
 {
 	gchar *s;
-
-	if ((s = g_strstr_len(data, data_len, "Location: ")) != NULL)
+	gchar *new_url, *temp_url, *end;
+	gboolean full;
+	int len;
+
+	if ((s = g_strstr_len(data, data_len, "Location: ")) == NULL)
+		/* We're not being redirected */
+		return FALSE;
+
+	s += strlen("Location: ");
+	end = strchr(s, '\r');
+
+	/* Just in case :) */
+	if (end == NULL)
+		end = strchr(s, '\n');
+
+	if (end == NULL)
+		return FALSE;
+
+	len = end - s;
+
+	new_url = g_malloc(len + 1);
+	strncpy(new_url, s, len);
+	new_url[len] = '\0';
+
+	full = gfud->full;
+
+	if (*new_url == '/' || g_strstr_len(new_url, len, "://") == NULL)
 	{
-		gchar *new_url, *temp_url, *end;
-		gboolean full;
-		int len;
-
-		s += strlen("Location: ");
-		end = strchr(s, '\r');
-
-		/* Just in case :) */
-		if (end == NULL)
-			end = strchr(s, '\n');
-
-		if (end == NULL)
-			return FALSE;
-
-		len = end - s;
-
-		new_url = g_malloc(len + 1);
-		strncpy(new_url, s, len);
-		new_url[len] = '\0';
-
-		full = gfud->full;
-
-		if (*new_url == '/' || g_strstr_len(new_url, len, "://") == NULL)
-		{
-			temp_url = new_url;
-
-			new_url = g_strdup_printf("%s:%d%s", gfud->website.address,
-									  gfud->website.port, temp_url);
-
-			g_free(temp_url);
-
-			full = FALSE;
-		}
-
-		purple_debug_info("util", "Redirecting to %s\n", new_url);
-
-		/*
-		 * Try again, with this new location.  This code is somewhat
-		 * ugly, but we need to reuse the gfud because whoever called
-		 * us is holding a reference to it.
-		 */
-		g_free(gfud->url);
-		gfud->url = new_url;
-		gfud->full = full;
-		g_free(gfud->request);
-		gfud->request = NULL;
-
-		purple_input_remove(gfud->inpa);
-		gfud->inpa = 0;
-		close(gfud->fd);
-		gfud->fd = -1;
-		gfud->request_written = 0;
-		gfud->len = 0;
-		gfud->data_len = 0;
-
-		g_free(gfud->website.user);
-		g_free(gfud->website.passwd);
-		g_free(gfud->website.address);
-		g_free(gfud->website.page);
-		purple_url_parse(new_url, &gfud->website.address, &gfud->website.port,
-					   &gfud->website.page, &gfud->website.user, &gfud->website.passwd);
-
-		gfud->connect_data = purple_proxy_connect(NULL, NULL,
-				gfud->website.address, gfud->website.port,
-				url_fetch_connect_cb, gfud);
-
-		if (gfud->connect_data == NULL)
-		{
-			purple_util_fetch_url_error(gfud, _("Unable to connect to %s"),
-					gfud->website.address);
-		}
-
+		temp_url = new_url;
+
+		new_url = g_strdup_printf("%s:%d%s", gfud->website.address,
+								  gfud->website.port, temp_url);
+
+		g_free(temp_url);
+
+		full = FALSE;
+	}
+
+	purple_debug_info("util", "Redirecting to %s\n", new_url);
+
+	gfud->num_times_redirected++;
+	if (gfud->num_times_redirected >= 5)
+	{
+		purple_util_fetch_url_error(gfud,
+				_("Could not open %s: Redirected too many times"),
+				gfud->url);
 		return TRUE;
 	}
 
-	return FALSE;
+	/*
+	 * Try again, with this new location.  This code is somewhat
+	 * ugly, but we need to reuse the gfud because whoever called
+	 * us is holding a reference to it.
+	 */
+	g_free(gfud->url);
+	gfud->url = new_url;
+	gfud->full = full;
+	g_free(gfud->request);
+	gfud->request = NULL;
+
+	purple_input_remove(gfud->inpa);
+	gfud->inpa = 0;
+	close(gfud->fd);
+	gfud->fd = -1;
+	gfud->request_written = 0;
+	gfud->len = 0;
+	gfud->data_len = 0;
+
+	g_free(gfud->website.user);
+	g_free(gfud->website.passwd);
+	g_free(gfud->website.address);
+	g_free(gfud->website.page);
+	purple_url_parse(new_url, &gfud->website.address, &gfud->website.port,
+				   &gfud->website.page, &gfud->website.user, &gfud->website.passwd);
+
+	gfud->connect_data = purple_proxy_connect(NULL, NULL,
+			gfud->website.address, gfud->website.port,
+			url_fetch_connect_cb, gfud);
+
+	if (gfud->connect_data == NULL)
+	{
+		purple_util_fetch_url_error(gfud, _("Unable to connect to %s"),
+				gfud->website.address);
+	}
+
+	return TRUE;
 }
 
 static size_t