Add url_fragment to extract fragment from URL

Optionally allow url_compare to ignore fragments in comparison
Fix handling of url_compare result in a few places
Fix redirects which contain fragments in the Location header

svn path=/trunk/netsurf/; revision=3826

content/content.c

@@ -608,6 +608,7 @@ bool content_set_type(struct content *c, content_type type,
 				return false;
 			}
 			content_remove_user(c, callback, p1, p2);
+			msg_data.new_url = NULL;
 			content_broadcast(clone, CONTENT_MSG_NEWPTR, msg_data);
 			fetchcache_go(clone, referer,
 					callback, p1, p2,

content/content.h

@@ -84,7 +84,7 @@ typedef enum {
 	CONTENT_MSG_STATUS,    /**< new status string */
 	CONTENT_MSG_REFORMAT,  /**< content_reformat done */
 	CONTENT_MSG_REDRAW,    /**< needs redraw (eg. new animation frame) */
-	CONTENT_MSG_NEWPTR,    /**< address of structure has changed */
+	CONTENT_MSG_NEWPTR,    /**< structure has been replaced */
 	CONTENT_MSG_REFRESH,   /**< wants refresh */
 #ifdef WITH_AUTH
 	CONTENT_MSG_AUTH,      /**< authentication required */
@@ -97,7 +97,8 @@ typedef enum {
 /** Extra data for some content_msg messages. */
 union content_msg_data {
 	const char *error;	/**< Error message, for CONTENT_MSG_ERROR. */
-	char *redirect;	/**< Redirect URL, for CONTENT_MSG_REDIRECT. */
+	const char *new_url;	/**< Replacement URL (or NULL if the same 
+				 * as previous), for CONTENT_MSG_NEWPTR. */
 	/** Area of content which needs redrawing, for CONTENT_MSG_REDRAW. */
 	struct {
 		float x, y, width, height;

content/fetchcache.c

@@ -689,6 +689,8 @@ void fetchcache_notmodified(struct content *c, const void *data)
 			}
 
 			content_remove_user(c, callback, p1, p2);
+
+			msg_data.new_url = NULL;
 			callback(CONTENT_MSG_NEWPTR, fb, p1, p2, msg_data);
 
 			/* and catch user up with fallback's state */
@@ -786,25 +788,26 @@ void fetchcache_redirect(struct content *c, const void *data,
 {
 	char *url, *url1;
 	char *referer, *parent_url;
-	long http_code = fetch_http_code(c->fetch);
-	const char *ref = fetch_get_referer(c->fetch);
-	const char *parent = fetch_get_parent_url(c->fetch);
+	long http_code;
+	const char *ref;
+	const char *parent;
 	union content_msg_data msg_data;
 	url_func_result result;
 
 	/* Preconditions */
 	assert(c && data);
 	assert(c->status == CONTENT_STATUS_TYPE_UNKNOWN);
-	/* Ensure a redirect happened */
-	assert(300 <= http_code && http_code <= 399);
-	/* 304 is handled by fetch_notmodified() */
-	assert(http_code != 304);
 
 	/* Extract fetch details */
 	http_code = fetch_http_code(c->fetch);
 	ref = fetch_get_referer(c->fetch);
 	parent = fetch_get_parent_url(c->fetch);
 
+	/* Ensure a redirect happened */
+	assert(300 <= http_code && http_code <= 399);
+	/* 304 is handled by fetch_notmodified() */
+	assert(http_code != 304);
+
 	/* Clone referer and parent url
 	 * originals are destroyed in fetch_abort() */
 	referer = ref ? strdup(ref) : NULL;
@@ -941,6 +944,7 @@ void fetchcache_redirect(struct content *c, const void *data,
 		replacement->redirect_count = c->redirect_count + 1;
 
 		/* Notify user that content has changed */
+		msg_data.new_url = url;
 		callback(CONTENT_MSG_NEWPTR, replacement, p1, p2, msg_data);
 
 		/* Start fetching the replacement content */

desktop/browser.c

@@ -262,7 +262,7 @@ void browser_window_go_post(struct browser_window *bw, const char *url,
 {
 	struct content *c;
 	char *url2;
-	char *hash;
+	char *fragment;
 	url_func_result res;
 	char url_buf[256];
 	int depth = 0;
@@ -296,38 +296,38 @@ void browser_window_go_post(struct browser_window *bw, const char *url,
 	if (!download)
 		gui_window_set_url(bw->window, url2);
 
+	free(bw->frag_id);
+	bw->frag_id = NULL;
+
 	/* find any fragment identifier on end of URL */
-	hash = strchr(url2, '#');
-	if (bw->frag_id) {
-		free(bw->frag_id);
-		bw->frag_id = 0;
-	}
-	if (hash) {
-		char *frag = curl_unescape(hash+1, strlen(hash + 1));
-		if (!frag) {
-			free(url2);
-			warn_user("NoMemory", 0);
-			return;
-		}
+	res = url_fragment(url2, &fragment);
+	if (res == URL_FUNC_NOMEM) {
+		free(url2);
+		warn_user("NoMemory", 0);
+		return;
+	} else if (res == URL_FUNC_OK) {
+		bool same_url = false;
 
-		bw->frag_id = strdup(frag);
-		curl_free(frag);
+		bw->frag_id = fragment;
 
-		if (!bw->frag_id) {
-			free(url2);
-			warn_user("NoMemory", 0);
-			return;
+		/* Compare new URL with existing one (ignoring fragments) */
+		if (bw->current_content && bw->current_content->url) {
+			res = url_compare(bw->current_content->url, url2, 
+					true, &same_url);
+			if (res == URL_FUNC_NOMEM) {
+				free(url2);
+				warn_user("NoMemory", 0);
+				return;
+			} else if (res == URL_FUNC_FAILED) {
+				same_url = false;
+			}
 		}
 
 		/* if we're simply moving to another ID on the same page,
 		 * don't bother to fetch, just update the window.
 		 */
-		if (!post_urlenc && !post_multipart && !strchr(url2, '?') &&
-			bw->current_content && bw->current_content->url &&
-				strncasecmp(bw->current_content->url,
-						url2, hash - url2) == 0 &&
-				strlen(bw->current_content->url) ==
-					(unsigned int)(hash - url2)) {
+		if (same_url && !post_urlenc && !post_multipart && 
+				!strchr(url2, '?')) {
 			free(url2);
 			browser_window_update(bw, false);
 			snprintf(url_buf, sizeof url_buf, "%s#%s",
@@ -532,6 +532,27 @@ void browser_window_callback(content_msg msg, struct content *c,
 
 	case CONTENT_MSG_NEWPTR:
 		bw->loading_content = c;
+		if (data.new_url) {
+			/* Replacement URL too, so check for new fragment */
+			char *fragment;
+			url_func_result res;
+
+			/* Remove any existing fragment */
+			free(bw->frag_id);
+			bw->frag_id = NULL;
+
+			/* Extract new one, if any */
+			res = url_fragment(data.new_url, &fragment);
+			if (res == URL_FUNC_OK) {
+				/* Save for later use */
+				bw->frag_id = fragment;
+			}
+			/* Ignore memory exhaustion here -- it'll simply result
+			 * in the window being scrolled to the top rather than
+			 * to the fragment. That's acceptable, given that it's
+			 * likely that more important things will complain
+			 * about memory shortage. */
+		}
 		break;
 
 #ifdef WITH_AUTH

render/directory.c

@@ -96,8 +96,8 @@ bool directory_convert(struct content *c, int width, int height) {
 
 	res = url_parent(c->url, &up);
 	if (res == URL_FUNC_OK) {
-		res = url_compare(c->url, up, &compare);
-		if (!compare) {
+		res = url_compare(c->url, up, false, &compare);
+		if ((res == URL_FUNC_OK) && !compare) {
 			snprintf(buffer, sizeof(buffer),
 				"<a href=\"..\">[..]</a>\n");
 			htmlParseChunk(c->data.html.parser, buffer,

riscos/menus.c

@@ -1989,8 +1989,10 @@ void ro_gui_menu_prepare_action(wimp_w owner, menu_action action,
 			if (result) {
 				res = url_parent(c->url, &parent);
 				if (res == URL_FUNC_OK) {
-				  	res = url_compare(c->url, parent, &compare);
-				  	result = !compare;
+				  	res = url_compare(c->url, parent, 
+							false, &compare);
+					if (res == URL_FUNC_OK)
+					  	result = !compare;
 				  	free(parent);
        	                	} else {
        	                	  	result = false;

riscos/window.c

@@ -3329,8 +3329,8 @@ bool ro_gui_window_navigate_up(struct gui_window *g, const char *url) {
 
 	res = url_parent(url, &parent);
 	if (res == URL_FUNC_OK) {
-		res = url_compare(url, parent, &compare);
-		if (!compare && (res == URL_FUNC_OK))
+		res = url_compare(url, parent, false, &compare);
+		if ((res == URL_FUNC_OK) && !compare)
 			browser_window_go(g->bw, parent, 0, true);
 		free(parent);
 	}

utils/url.c

@@ -30,6 +30,7 @@
 #include <strings.h>
 #include <sys/types.h>
 #include <regex.h>
+#include "curl/curl.h"
 #include "utils/log.h"
 #include "utils/url.h"
 #include "utils/utils.h"
@@ -718,6 +719,44 @@ url_func_result url_leafname(const char *url, char **result)
 	return status;
 }
 
+/**
+ * Extract fragment from an URL
+ * This will unescape any %xx entities in the fragment
+ *
+ * \param url     an absolute URL
+ * \param result  pointer to pointer to buffer to hold result
+ * \return URL_FUNC_OK on success
+ */
+
+url_func_result url_fragment(const char *url, char **result)
+{
+	url_func_result status;
+	struct url_components components;
+
+	assert(url);
+
+	status = url_get_components(url, &components);
+	if (status == URL_FUNC_OK) {
+		if (!components.fragment) {
+			status = URL_FUNC_FAILED;
+		} else {
+			char *frag = curl_unescape(components.fragment,
+					strlen(components.fragment));
+			if (!frag) {
+				status = URL_FUNC_NOMEM;
+			} else {
+				*result = strdup(frag);
+				if (!(*result))
+					status = URL_FUNC_NOMEM;
+				curl_free(frag);
+			}
+		}
+	}
+
+	url_destroy_components(&components);
+	return status;
+}
+
 /**
  * Attempt to find a nice filename for a URL.
  *
@@ -897,12 +936,14 @@ url_func_result url_escape(const char *unescaped, bool sptoplus,
 /**
  * Compare two absolute, normalized URLs
  *
- * \param url1 URL 1
- * \param url2 URL 2
- * \param result Pointer to location to store result (true if URLs match)
+ * \param url1     URL 1
+ * \param url2     URL 2
+ * \param nofrag   Ignore fragment part in comparison
+ * \param result   Pointer to location to store result (true if URLs match)
  * \return URL_FUNC_OK on success
  */
-url_func_result url_compare(const char *url1, const char *url2, bool *result)
+url_func_result url_compare(const char *url1, const char *url2,
+		bool nofrag, bool *result)
 {
 	url_func_result status;
 	struct url_components c1, c2;
@@ -930,7 +971,7 @@ url_func_result url_compare(const char *url1, const char *url2, bool *result)
 			((c1.path && c2.path) || (!c1.path && !c2.path)) &&
 			((c1.query && c2.query) ||
 					(!c1.query && !c2.query)) &&
-			((c1.fragment && c2.fragment) ||
+			(nofrag || (c1.fragment && c2.fragment) ||
 					(!c1.fragment && !c2.fragment))) {
 
 		if (c1.scheme)
@@ -946,7 +987,7 @@ url_func_result url_compare(const char *url1, const char *url2, bool *result)
 		if (c1.query)
 			res &= strcmp(c1.query, c2.query) == 0;
 
-		if (c1.fragment)
+		if (!nofrag && c1.fragment)
 			res &= strcmp(c1.fragment, c2.fragment) == 0;
 	} else {
 		/* Can't match */

utils/url.h

@@ -54,8 +54,9 @@ url_func_result url_parent(const char *url, char **result);
 url_func_result url_plq(const char *url, char **result);
 url_func_result url_path(const char *url, char **result);
 url_func_result url_leafname(const char *url, char **result);
+url_func_result url_fragment(const char *url, char **result);
 url_func_result url_compare(const char *url1, const char *url2,
-		bool *result);
+		bool nofrag, bool *result);
 
 url_func_result url_get_components(const char *url,
 		struct url_components *result);