Fix parsing error when an empty HTML data is returned. Add HTTP status and other information to status bar.

svn path=/trunk/netsurf/; revision=3140

7 files changed, 233 insertions, 116 deletions

diff --git a/!NetSurf/Resources/en/Messages b/!NetSurf/Resources/en/Messages
index 27b0d3d78..4cb99fa2a 100644
--- a/!NetSurf/Resources/en/Messages
+++ b/!NetSurf/Resources/en/Messages
@@ -450,11 +450,7 @@ Complete:Page complete (%gs)
 Redirecting:Redirecting...
 Processing:Processing document
 Formatting:Formatting document
-FetchObjs:Loading %u %s
-FetchObjs2:Loading %u %s: %s
 Done:Document done
-FetchStyle:Loading %u %s
-FetchStyle2:Loading %u %s: %s
 
 # Fetch warning/error messages - displayed in status bar
 #
@@ -471,6 +467,47 @@ PNGError:PNG library error.
 MNGError:MNG library error.
 BadSprite:Invalid or corrupt Sprite data.
 
+# HTTP status codes
+#
+HTTP0:OK
+HTTP200:OK
+HTTP201:Created
+HTTP202:Accepted
+HTTP203:Non-authoritative information
+HTTP204:No content
+HTTP205:Reset content
+HTTP206:Partial content
+HTTP300:Multiple choices
+HTTP301:Moved permanently
+HTTP302:Found
+HTTP303:See other
+HTTP304:Not modified
+HTTP305:Use proxy
+HTTP307:Temporary redirect
+HTTP400:Bad request
+HTTP401:Unauthorized
+HTTP402:Payment required
+HTTP403:Forbidden
+HTTP404:Not found
+HTTP405:Method not allowed
+HTTP406:Not acceptable
+HTTP407:Proxy authentication required
+HTTP408:Request timeout
+HTTP409:Conflict
+HTTP410:Gone
+HTTP411:Length required
+HTTP412:Precondition failed
+HTTP413:Request entity too large
+HTTP414:Request-URI too long
+HTTP415:Unsupported media type
+HTTP416:Requested range not satisfiable
+HTTP417:Expectation failed
+HTTP500:Internal server error
+HTTP501:Not implemented
+HTTP502:Bad gateway
+HTTP503:Service unavailable
+HTTP504:Gateway timeout
+HTTP505:HTTP version not supported
 
 # User interface
 # ==============
diff --git a/content/content.c b/content/content.c
index 0f23d25cc..d211bb109 100644
--- a/content/content.c
+++ b/content/content.c
@@ -2,7 +2,7 @@
  * This file is part of NetSurf, http://netsurf-browser.org/
  * Licensed under the GNU General Public License,
  *		  http://www.opensource.org/licenses/gpl-license
- * Copyright 2005 James Bursa <bursa@users.sourceforge.net>
+ * Copyright 2005-2007 James Bursa <bursa@users.sourceforge.net>
  */
 
 /** \file
@@ -18,6 +18,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <time.h>
 #include "netsurf/utils/config.h"
 #include "netsurf/content/content.h"
 #include "netsurf/content/fetch.h"
@@ -304,6 +305,7 @@ static const struct handler_entry handler_map[] = {
 #define HANDLER_MAP_COUNT (sizeof(handler_map) / sizeof(handler_map[0]))
 
 
+static void content_update_status(struct content *c);
 static void content_destroy(struct content *c);
 static void content_stop_check(struct content *c);
 
@@ -373,6 +375,7 @@ struct content * content_create(const char *url)
 	c->refresh = 0;
 	c->bitmap = 0;
 	c->fresh = false;
+	c->time = clock();
 	c->size = 0;
 	c->title = 0;
 	c->active = 0;
@@ -380,13 +383,14 @@ struct content * content_create(const char *url)
 	user_sentinel->p1 = user_sentinel->p2 = 0;
 	user_sentinel->next = 0;
 	c->user_list = user_sentinel;
-	content_set_status(c, messages_get("Loading"));
+	c->sub_status[0] = 0;
 	c->locked = false;
 	c->fetch = 0;
 	c->source_data = 0;
 	c->source_size = 0;
 	c->source_allocated = 0;
 	c->total_size = 0;
+	c->http_code = 0;
 	c->no_error_pages = false;
 	c->download = false;
 	c->error_count = 0;
@@ -400,6 +404,8 @@ struct content * content_create(const char *url)
 	c->cache_data->etag = 0;
 	c->cache_data->last_modified = 0;
 
+	content_set_status(c, messages_get("Loading"));
+
 	c->prev = 0;
 	c->next = content_list;
 	if (content_list)
@@ -592,11 +598,38 @@ void content_set_status(struct content *c, const char *status_message, ...)
 	int len;
 
 	va_start(ap, status_message);
-	if ((len = vsnprintf(c->status_message, sizeof (c->status_message),
+	if ((len = vsnprintf(c->sub_status, sizeof (c->sub_status),
 			status_message, ap)) < 0 ||
-			(int)sizeof (c->status_message) <= len)
-		c->status_message[sizeof (c->status_message) - 1] = '\0';
+			(int)sizeof (c->sub_status) <= len)
+		c->sub_status[sizeof (c->sub_status) - 1] = '\0';
 	va_end(ap);
+
+	content_update_status(c);
+}
+
+
+void content_update_status(struct content *c)
+{
+	char token[20];
+	const char *status;
+	clock_t time;
+
+	snprintf(token, sizeof token, "HTTP%li", c->http_code);
+	status = messages_get(token);
+	if (status == token)
+		status = token + 4;
+
+	if (c->status == CONTENT_STATUS_TYPE_UNKNOWN ||
+			c->status == CONTENT_STATUS_LOADING ||
+			c->status == CONTENT_STATUS_READY)
+		time = clock() - c->time;
+	else
+		time = c->time;
+
+	snprintf(c->status_message, sizeof (c->status_message),
+			"%s (%.1fs) %s", status,
+			(float) time / CLOCKS_PER_SEC, c->sub_status);
+	/* LOG(("%s", c->status_message)); */
 }
 
 
@@ -704,7 +737,22 @@ void content_convert(struct content *c, int width, int height)
 			c->status == CONTENT_STATUS_DONE);
 	content_broadcast(c, CONTENT_MSG_READY, msg_data);
 	if (c->status == CONTENT_STATUS_DONE)
-		content_broadcast(c, CONTENT_MSG_DONE, msg_data);
+		content_set_done(c);
+}
+
+
+/**
+ * Put a content in status CONTENT_STATUS_DONE.
+ */
+
+void content_set_done(struct content *c)
+{
+	union content_msg_data msg_data;
+
+	c->status = CONTENT_STATUS_DONE;
+	c->time = clock() - c->time;
+	content_update_status(c);
+	content_broadcast(c, CONTENT_MSG_DONE, msg_data);
 }
 
 
diff --git a/content/content.h b/content/content.h
index 123733ba9..268ba31a1 100644
--- a/content/content.h
+++ b/content/content.h
@@ -2,7 +2,7 @@
  * This file is part of NetSurf, http://netsurf-browser.org/
  * Licensed under the GNU General Public License,
  *                http://www.opensource.org/licenses/gpl-license
- * Copyright 2005 James Bursa <bursa@users.sourceforge.net>
+ * Copyright 2005-2007 James Bursa <bursa@users.sourceforge.net>
  * Copyright 2003 Philip Pemberton <philpem@users.sourceforge.net>
  */
 
@@ -16,6 +16,7 @@
 #define _NETSURF_DESKTOP_CONTENT_H_
 
 #include <stdint.h>
+#include <time.h>
 #include "netsurf/utils/config.h"
 #include "netsurf/content/content_type.h"
 #include "netsurf/css/css.h"
@@ -178,6 +179,9 @@ struct content {
 	 *  shared between users. */
 	bool fresh;
 	struct cache_data *cache_data;	/**< Cache control data */
+	clock_t time;			/**< Creation time, if TYPE_UNKNOWN,
+					  LOADING or READY,
+					  otherwise total time. */
 
 	unsigned int size;		/**< Estimated size of all data
 					  associated with this content, except
@@ -186,7 +190,8 @@ struct content {
 	unsigned int active;		/**< Number of child fetches or
 					  conversions currently in progress. */
 	struct content_user *user_list;	/**< List of users. */
-	char status_message[80];	/**< Text for status bar. */
+	char status_message[120];	/**< Full text for status bar. */
+	char sub_status[80];		/**< Status of content. */
 	/** Content is being processed: data structures may be inconsistent
 	 * and content must not be redrawn or modified. */
 	bool locked;
@@ -196,6 +201,7 @@ struct content {
 	unsigned long source_size;	/**< Amount of data fetched so far. */
 	unsigned long source_allocated;	/**< Amount of space allocated so far. */
 	unsigned long total_size;	/**< Total data size, 0 if unknown. */
+	long http_code;			/**< HTTP status code, 0 if not HTTP. */
 
 	bool no_error_pages;		/**< Used by fetchcache(). */
 	bool download;			/**< Used by fetchcache(). */
@@ -226,6 +232,7 @@ void content_set_status(struct content *c, const char *status_message, ...);
 bool content_process_data(struct content *c, const char *data,
 		unsigned int size);
 void content_convert(struct content *c, int width, int height);
+void content_set_done(struct content *c);
 void content_reformat(struct content *c, int width, int height);
 void content_clean(void);
 void content_reset(struct content *c);
diff --git a/content/fetchcache.c b/content/fetchcache.c
index 220359df0..0a630661b 100644
--- a/content/fetchcache.c
+++ b/content/fetchcache.c
@@ -376,6 +376,7 @@ void fetchcache_callback(fetch_msg msg, void *p, const void *data,
 	switch (msg) {
 		case FETCH_TYPE:
 			c->total_size = size;
+			c->http_code = fetch_http_code(c->fetch);
 			mime_type = fetchcache_parse_type(data, &params);
 			if (!mime_type) {
 				msg_data.error = messages_get("NoMemory");
diff --git a/desktop/browser.c b/desktop/browser.c
index 628fbc9b3..b3fe85127 100644
--- a/desktop/browser.c
+++ b/desktop/browser.c
@@ -20,7 +20,6 @@
 #include <stdint.h>
 #include <stdlib.h>
 #include <string.h>
-#include <time.h>
 #include "curl/curl.h"
 #include "netsurf/utils/config.h"
 #include "netsurf/content/fetch.h"
@@ -286,7 +285,6 @@ void browser_window_go_post(struct browser_window *bw, const char *url,
 
 	browser_window_set_status(bw, messages_get("Loading"));
 	bw->history_add = history_add;
-	bw->time0 = clock();
 	c = fetchcache(url2, browser_window_callback, (intptr_t) bw, 0,
 			width, height, false,
 			post_urlenc, post_multipart, true, download);
@@ -347,6 +345,7 @@ void browser_window_callback(content_msg msg, struct content *c,
 			url[sizeof url - 1] = 0;
 			gui_window_set_url(bw->window, url);
 			bw->refresh_interval = -1;
+			browser_window_set_status(bw, c->status_message);
 		}
 		break;
 
@@ -414,10 +413,7 @@ void browser_window_callback(content_msg msg, struct content *c,
 		assert(bw->current_content == c);
 
 		browser_window_update(bw, false);
-		sprintf(status, messages_get("Complete"),
-				((float) (clock() - bw->time0)) /
-				CLOCKS_PER_SEC);
-		browser_window_set_status(bw, status);
+		browser_window_set_status(bw, c->status_message);
 		browser_window_stop_throbber(bw);
 		history_update(bw->history, c);
 		hotlist_visited(c);
diff --git a/desktop/browser.h b/desktop/browser.h
index c7bbaf8e3..8f0e74e89 100644
--- a/desktop/browser.h
+++ b/desktop/browser.h
@@ -74,8 +74,6 @@ struct browser_window {
 	bool throbbing;
 	/** Add loading_content to the window history when it loads. */
 	bool history_add;
-	/** Start time of fetching loading_content. */
-	clock_t time0;
 
 	/** Fragment identifier for current_content. */
 	char *frag_id;
diff --git a/render/html.c b/render/html.c
index 1e50a77fb..7b7844ae8 100644
--- a/render/html.c
+++ b/render/html.c
@@ -55,6 +55,20 @@ static bool html_object_type_permitted(const content_type type,
 static void html_object_refresh(void *p);
 static void html_destroy_frameset(struct content_html_frames *frameset);
 static void html_destroy_iframe(struct content_html_iframe *iframe);
+static void html_set_status(struct content *c, const char *extra);
+
+static const char empty_document[] =
+	"<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\""
+        "	\"http://www.w3.org/TR/html4/strict.dtd\">"
+	"<html>"
+	"<head>"
+	"<title>Empty document</title>"
+	"</head>"
+	"<body>"
+	"<h1>Empty document</h1>"
+	"<p>The document sent by the server is empty.</p>"
+	"</body>"
+	"</html>";
 
 
 /**
@@ -264,11 +278,15 @@ bool html_convert(struct content *c, int width, int height)
 	union content_msg_data msg_data;
 
 	/* finish parsing */
+	if (c->source_size == 0)
+		htmlParseChunk(c->data.html.parser, empty_document,
+				sizeof empty_document, 0);
 	htmlParseChunk(c->data.html.parser, "", 0, 1);
 	document = c->data.html.parser->myDoc;
 	/*xmlDebugDumpDocument(stderr, c->data.html.parser->myDoc);*/
 	htmlFreeParserCtxt(c->data.html.parser);
 	c->data.html.parser = 0;
+
 	if (!document) {
 		LOG(("Parsing failed"));
 		msg_data.error = messages_get("ParsingFail");
@@ -353,24 +371,92 @@ bool html_convert(struct content *c, int width, int height)
 	xmlFreeDoc(document);
 
 	/* layout the box tree */
-	content_set_status(c, messages_get("Formatting"));
+	html_set_status(c, messages_get("Formatting"));
 	content_broadcast(c, CONTENT_MSG_STATUS, msg_data);
 	LOG(("Layout document"));
 	html_reformat(c, width, height);
 	/*box_dump(c->data.html.layout->children, 0);*/
 
-	if (c->active == 0) {
+	if (c->active == 0)
 		c->status = CONTENT_STATUS_DONE;
-		content_set_status(c, messages_get("Done"));
-	} else {
+	else
 		c->status = CONTENT_STATUS_READY;
-		content_set_status(c, messages_get("FetchObjs"), c->active,
-			messages_get((c->active == 1) ? "obj" : "objs"));
-	}
+	html_set_status(c, "");
+
+	return true;
+}
+
+
+/**
+ * Process elements in <head>.
+ *
+ * \param  c     content structure
+ * \param  head  xml node of head element
+ * \return  true on success, false on memory exhaustion
+ *
+ * The title and base href are extracted if present.
+ */
+
+bool html_head(struct content *c, xmlNode *head)
+{
+	xmlNode *node;
+	xmlChar *s;
+
+	c->title = 0;
+
+	for (node = head->children; node != 0; node = node->next) {
+		if (node->type != XML_ELEMENT_NODE)
+			continue;
+
+		LOG(("Node: %s", node->name));
+		if (!c->title && strcmp(node->name, "title") == 0) {
+			xmlChar *title = xmlNodeGetContent(node);
+			if (!title)
+				return false;
+			char *title2 = squash_whitespace(title);
+			xmlFree(title);
+			if (!title2)
+				return false;
+			c->title = talloc_strdup(c, title2);
+			free(title2);
+			if (!c->title)
+				return false;
 
+		} else if (strcmp(node->name, "base") == 0) {
+			char *href = (char *) xmlGetProp(node,
+					(const xmlChar *) "href");
+			if (href) {
+				char *url;
+				url_func_result res;
+				res = url_normalize(href, &url);
+				if (res == URL_FUNC_OK) {
+					c->data.html.base_url =
+							talloc_strdup(c, url);
+					free(url);
+				}
+				xmlFree(href);
+			}
+			/* don't use the central values to ease freeing later on */
+			if ((s = xmlGetProp(node, (const xmlChar *) "target"))) {
+				if ((!strcasecmp(s, "_blank")) || (!strcasecmp(s, "_top")) ||
+						(!strcasecmp(s, "_parent")) ||
+						(!strcasecmp(s, "_self")) ||
+						('a' <= s[0] && s[0] <= 'z') ||
+						('A' <= s[0] && s[0] <= 'Z')) {  /* [6.16] */
+					c->data.html.base_target = talloc_strdup(c, s);
+					if (!c->data.html.base_target) {
+						xmlFree(s);
+						return false;
+					}
+				}
+				xmlFree(s);
+			}
+		}
+	}
 	return true;
 }
 
+
 /**
  * Search for meta refresh
  *
@@ -483,75 +569,6 @@ bool html_meta_refresh(struct content *c, xmlNode *head)
 	return true;
 }
 
-/**
- * Process elements in <head>.
- *
- * \param  c     content structure
- * \param  head  xml node of head element
- * \return  true on success, false on memory exhaustion
- *
- * The title and base href are extracted if present.
- */
-
-bool html_head(struct content *c, xmlNode *head)
-{
-	xmlNode *node;
-	xmlChar *s;
-
-	c->title = 0;
-
-	for (node = head->children; node != 0; node = node->next) {
-		if (node->type != XML_ELEMENT_NODE)
-			continue;
-
-		LOG(("Node: %s", node->name));
-		if (!c->title && strcmp(node->name, "title") == 0) {
-			xmlChar *title = xmlNodeGetContent(node);
-			if (!title)
-				return false;
-			char *title2 = squash_whitespace(title);
-			xmlFree(title);
-			if (!title2)
-				return false;
-			c->title = talloc_strdup(c, title2);
-			free(title2);
-			if (!c->title)
-				return false;
-
-		} else if (strcmp(node->name, "base") == 0) {
-			char *href = (char *) xmlGetProp(node,
-					(const xmlChar *) "href");
-			if (href) {
-				char *url;
-				url_func_result res;
-				res = url_normalize(href, &url);
-				if (res == URL_FUNC_OK) {
-					c->data.html.base_url =
-							talloc_strdup(c, url);
-					free(url);
-				}
-				xmlFree(href);
-			}
-			/* don't use the central values to ease freeing later on */
-			if ((s = xmlGetProp(node, (const xmlChar *) "target"))) {
-				if ((!strcasecmp(s, "_blank")) || (!strcasecmp(s, "_top")) ||
-						(!strcasecmp(s, "_parent")) ||
-						(!strcasecmp(s, "_self")) ||
-						('a' <= s[0] && s[0] <= 'z') ||
-						('A' <= s[0] && s[0] <= 'Z')) {  /* [6.16] */
-					c->data.html.base_target = talloc_strdup(c, s);
-					if (!c->data.html.base_target) {
-						xmlFree(s);
-						return false;
-					}
-				}
-				xmlFree(s);
-			}
-		}
-	}
-	return true;
-}
-
 
 /**
  * Process inline stylesheets and fetch linked stylesheets.
@@ -762,9 +779,7 @@ bool html_find_stylesheets(struct content *c, xmlNode *head)
 	/* complete the fetches */
 	while (c->active != 0) {
 		if (c->active != last_active) {
-			content_set_status(c, messages_get("FetchStyle"),
-				c->active,
-				messages_get((c->active == 1) ? "styl" : "styls"));
+			html_set_status(c, "");
 			content_broadcast(c, CONTENT_MSG_STATUS, msg_data);
 			last_active = c->active;
 		}
@@ -820,7 +835,7 @@ void html_convert_css_callback(content_msg msg, struct content *css,
 				c->data.html.stylesheet_content[i] = 0;
 				c->active--;
 				content_add_error(c, "NotCSS", 0);
-				content_set_status(c, messages_get("NotCSS"));
+				html_set_status(c, messages_get("NotCSS"));
 				content_broadcast(c, CONTENT_MSG_STATUS, data);
 				content_remove_user(css,
 						html_convert_css_callback,
@@ -854,10 +869,7 @@ void html_convert_css_callback(content_msg msg, struct content *css,
 			break;
 
 		case CONTENT_MSG_STATUS:
-			content_set_status(c, messages_get("FetchStyle2"),
-				c->active,
-				messages_get((c->active == 1) ? "styl" : "styls"),
-				css->status_message);
+			html_set_status(c, css->status_message);
 			content_broadcast(c, CONTENT_MSG_STATUS, data);
 			break;
 
@@ -1063,7 +1075,7 @@ void html_object_callback(content_msg msg, struct content *object,
 			c->data.html.object[i].content = 0;
 			c->active--;
 			content_add_error(c, "?", 0);
-			content_set_status(c, messages_get("BadObject"));
+			html_set_status(c, messages_get("BadObject"));
 			content_broadcast(c, CONTENT_MSG_STATUS, data);
 			content_remove_user(object, html_object_callback,
 					(intptr_t) c, i);
@@ -1094,18 +1106,14 @@ void html_object_callback(content_msg msg, struct content *object,
 			c->data.html.object[i].content = 0;
 			c->active--;
 			content_add_error(c, "?", 0);
-			content_set_status(c, messages_get("ObjError"),
-					data.error);
+			html_set_status(c, data.error);
 			content_broadcast(c, CONTENT_MSG_STATUS, data);
 			html_object_failed(box, c,
 					c->data.html.object[i].background);
 			break;
 
 		case CONTENT_MSG_STATUS:
-			content_set_status(c, messages_get("FetchObjs2"),
-				c->active,
-				messages_get((c->active == 1) ? "obj" : "objs"),
-				object->status_message);
+			html_set_status(c, object->status_message);
 			/* content_broadcast(c, CONTENT_MSG_STATUS, 0); */
 			break;
 
@@ -1202,13 +1210,11 @@ void html_object_callback(content_msg msg, struct content *object,
 			msg == CONTENT_MSG_AUTH)) {
 		/* all objects have arrived */
 		content_reformat(c, c->available_width, c->height);
-		c->status = CONTENT_STATUS_DONE;
-		content_set_status(c, messages_get("Done"));
-		content_broadcast(c, CONTENT_MSG_DONE, data);
+		html_set_status(c, "");
+		content_set_done(c);
 	}
 	if (c->status == CONTENT_STATUS_READY)
-		content_set_status(c, messages_get("FetchObjs"), c->active,
-			messages_get((c->active == 1) ? "obj" : "objs"));
+		html_set_status(c, "");
 }
 
 
@@ -1537,6 +1543,30 @@ void html_destroy_iframe(struct content_html_iframe *iframe) {
 
 
 /**
+ * Set the content status.
+ */
+
+void html_set_status(struct content *c, const char *extra)
+{
+	unsigned int stylesheets = 0, objects = 0;
+	if (c->data.html.object_count == 0)
+		stylesheets = c->data.html.stylesheet_count - c->active;
+	else {
+		stylesheets = c->data.html.stylesheet_count;
+		objects = c->data.html.object_count - c->active;
+	}
+	content_set_status(c, "%u/%u %s %u/%u %s  %s",
+			stylesheets, c->data.html.stylesheet_count,
+			messages_get((c->data.html.stylesheet_count == 1) ?
+					"styl" : "styls"),
+			objects, c->data.html.object_count,
+			messages_get((c->data.html.object_count == 1) ?
+					"obj" : "objs"),
+			extra);
+}
+
+
+/**
  * Handle a window containing a CONTENT_HTML being opened.
  */