Make the high-level cache drop contents of unacceptable type on the floor.

svn path=/trunk/netsurf/; revision=10238

7 files changed, 123 insertions, 39 deletions

diff --git a/content/hlcache.c b/content/hlcache.c
index 05a416b0b..298fe8151 100644
--- a/content/hlcache.c
+++ b/content/hlcache.c
@@ -26,7 +26,9 @@
 
 #include "content/content.h"
 #include "content/hlcache.h"
+#include "utils/http.h"
 #include "utils/log.h"
+#include "utils/messages.h"
 #include "utils/url.h"
 
 typedef struct hlcache_entry hlcache_entry;
@@ -38,6 +40,8 @@ struct hlcache_retrieval_ctx {
 
 	hlcache_handle *handle;		/**< High-level handle for object */
 
+	const content_type *accepted_types;	/**< Accepted types, or NULL */
+
 	hlcache_child_context child;	/**< Child context */	
 };
 
@@ -62,6 +66,8 @@ static hlcache_entry *hlcache_content_list;
 
 static nserror hlcache_llcache_callback(llcache_handle *handle,
 		const llcache_event *event, void *pw);
+static bool hlcache_type_is_acceptable(llcache_handle *llcache, 
+		const content_type *accepted_types);
 static nserror hlcache_find_content(hlcache_retrieval_ctx *ctx);
 static void hlcache_content_callback(struct content *c,
 		content_msg msg, union content_msg_data data, void *pw);
@@ -74,7 +80,8 @@ static void hlcache_content_callback(struct content *c,
 nserror hlcache_handle_retrieve(const char *url, uint32_t flags,
 		const char *referer, llcache_post_data *post,
 		hlcache_handle_callback cb, void *pw,
-		hlcache_child_context *child, hlcache_handle **result)
+		hlcache_child_context *child, 
+		const content_type *accepted_types, hlcache_handle **result)
 {
 	hlcache_retrieval_ctx *ctx;
 	nserror error;
@@ -97,6 +104,8 @@ nserror hlcache_handle_retrieve(const char *url, uint32_t flags,
 		ctx->child.quirks = child->quirks;
 	}
 
+	ctx->accepted_types = accepted_types;
+
 	ctx->handle->cb = cb;
 	ctx->handle->pw = pw;
 
@@ -218,9 +227,26 @@ nserror hlcache_llcache_callback(llcache_handle *handle,
 
 	switch (event->type) {
 	case LLCACHE_EVENT_HAD_HEADERS:
-		error = hlcache_find_content(ctx);
-		if (error != NSERROR_OK)
-			return error;
+		if (hlcache_type_is_acceptable(handle, ctx->accepted_types)) {
+			error = hlcache_find_content(ctx);
+			if (error != NSERROR_OK)
+				return error;
+		} else {
+			/* Unacceptable type: abort fetch and report error */
+			llcache_handle_abort(handle);
+			llcache_handle_release(handle);
+
+			if (ctx->handle->cb != NULL) {
+				hlcache_event hlevent;
+
+				hlevent.type = CONTENT_MSG_ERROR;
+				hlevent.data.error = messages_get("BadType");
+
+				ctx->handle->cb(ctx->handle, &hlevent, 
+						ctx->handle->pw);
+			}
+		}
+
 		/* No longer require retrieval context */
 		free(ctx);
 		break;
@@ -247,6 +273,50 @@ nserror hlcache_llcache_callback(llcache_handle *handle,
 }
 
 /**
+ * Determine if the type of a low-level cache object is acceptable
+ *
+ * \param llcache         Low-level cache object to consider
+ * \param accepted_types  Array of acceptable types, or NULL for any
+ * \return True if the type is acceptable, false otherwise
+ */
+bool hlcache_type_is_acceptable(llcache_handle *llcache, 
+		const content_type *accepted_types)
+{
+	const char *content_type_header;
+	char *mime_type;
+	http_parameter *params;
+	content_type type;
+	nserror error;
+
+	if (accepted_types == NULL)
+		return true;
+
+	content_type_header = 
+			llcache_handle_get_header(llcache, "Content-Type");
+	if (content_type_header == NULL)
+		content_type_header = "text/plain";
+
+	error = http_parse_content_type(content_type_header, &mime_type, 
+			&params);
+	if (error != NSERROR_OK)
+		return false;
+
+	type = content_lookup(mime_type);
+
+	free(mime_type);
+	http_parameter_list_destroy(params);
+
+	while (*accepted_types != CONTENT_UNKNOWN) {
+		if (*accepted_types == type)
+			break;
+
+		accepted_types++;
+	}
+
+	return *accepted_types == type;
+}
+
+/**
  * Find a content for the high-level cache handle
  *
  * \param ctx   High-level cache retrieval context
diff --git a/content/hlcache.h b/content/hlcache.h
index 5ef42a301..1208b15c4 100644
--- a/content/hlcache.h
+++ b/content/hlcache.h
@@ -56,20 +56,23 @@ typedef nserror (*hlcache_handle_callback)(hlcache_handle *handle,
 /**
  * Retrieve a high-level cache handle for an object
  *
- * \param url      URL of the object to retrieve handle for
- * \param flags    Object retrieval flags
- * \param referer  Referring URL, or NULL if none
- * \param post     POST data, or NULL for a GET request
- * \param cb       Callback to handle object events
- * \param pw       Pointer to client-specific data for callback
- * \param child    Child retrieval context, or NULL for top-level content
- * \param result   Pointer to location to recieve cache handle
+ * \param url             URL of the object to retrieve handle for
+ * \param flags           Object retrieval flags
+ * \param referer         Referring URL, or NULL if none
+ * \param post            POST data, or NULL for a GET request
+ * \param cb              Callback to handle object events
+ * \param pw              Pointer to client-specific data for callback
+ * \param child           Child retrieval context, or NULL for top-level content
+ * \param accepted_types  Array of acceptable content types, or NULL for any
+ * \param result          Pointer to location to recieve cache handle
  * \return NSERROR_OK on success, appropriate error otherwise
  *
  * Child contents are keyed on the tuple < URL, quirks >.
  * The quirks field is ignored for child contents whose behaviour is not
  * affected by quirks mode.
  *
+ * The \a accepted_types array must be terminated with CONTENT_UNKNOWN
+ *
  * \todo The above rules should be encoded in the handler_map.
  *
  * \todo Is there any way to sensibly reduce the number of parameters here?
@@ -77,7 +80,8 @@ typedef nserror (*hlcache_handle_callback)(hlcache_handle *handle,
 nserror hlcache_handle_retrieve(const char *url, uint32_t flags,
 		const char *referer, llcache_post_data *post,
 		hlcache_handle_callback cb, void *pw,
-		hlcache_child_context *child, hlcache_handle **result);
+		hlcache_child_context *child, 
+		const content_type *accepted_types, hlcache_handle **result);
 
 /**
  * Release a high-level cache handle
diff --git a/css/css.c b/css/css.c
index ef38d76aa..1bbbfe608 100644
--- a/css/css.c
+++ b/css/css.c
@@ -200,6 +200,7 @@ bool nscss_convert(struct content *c)
  */
 css_error nscss_convert_css_data(struct content_css_data *c)
 {
+	static const content_type accept[] = { CONTENT_CSS, CONTENT_UNKNOWN };
 	const char *referer;
 	uint32_t i = 0;
 	css_error error;
@@ -252,7 +253,8 @@ css_error nscss_convert_css_data(struct content_css_data *c)
 		c->imports[c->import_count].media = media;
 		nerror = hlcache_handle_retrieve(lwc_string_data(uri),
 				0, referer, NULL, nscss_import, c,
-				&child, &c->imports[c->import_count++].c);
+				&child, accept,
+				&c->imports[c->import_count++].c);
 		if (error != NSERROR_OK) {
 			return CSS_NOMEM;
 		}
diff --git a/desktop/browser.c b/desktop/browser.c
index 70d6b8cf4..e1b071fdb 100644
--- a/desktop/browser.c
+++ b/desktop/browser.c
@@ -409,7 +409,8 @@ void browser_window_go_post(struct browser_window *bw, const char *url,
 	error = hlcache_handle_retrieve(url2, 0, referer,
 			fetch_is_post ? &post : NULL,
 			browser_window_callback, bw,
-			parent != NULL ? &child : NULL, &c);
+			parent != NULL ? &child : NULL,
+			NULL, &c);
 	if (error == NSERROR_NO_FETCH_HANDLER) {
 		gui_launch_url(url2);
 		free(url2);
diff --git a/desktop/searchweb.c b/desktop/searchweb.c
index aa993b6a8..89f86ed93 100644
--- a/desktop/searchweb.c
+++ b/desktop/searchweb.c
@@ -206,6 +206,12 @@ char *search_web_get_url(const char *encsearchterm)
 
 void search_web_retrieve_ico(bool localdefault)
 {
+	static const content_type accept[] = {
+#ifdef WITH_BMP
+		CONTENT_ICO,
+#endif
+		CONTENT_UNKNOWN
+	};
 	char *url;
 	nserror error;
 
@@ -230,7 +236,8 @@ void search_web_retrieve_ico(bool localdefault)
 	}
 
 	error = hlcache_handle_retrieve(url, 0, NULL, NULL,
-			search_web_ico_callback, NULL, NULL, &search_ico);
+			search_web_ico_callback, NULL, NULL, accept,
+			&search_ico);
 	if (error != NSERROR_OK)
 		search_ico = NULL;
 
@@ -269,6 +276,8 @@ nserror search_web_ico_callback(hlcache_handle *ico,
 		} else 
 #endif
 		{
+			hlcache_handle_release(ico);
+			search_ico = NULL;
 			search_web_retrieve_ico(true);
 		}
 		break;
diff --git a/render/favicon.c b/render/favicon.c
index 06ff0b94f..d1eaaacdd 100644
--- a/render/favicon.c
+++ b/render/favicon.c
@@ -148,6 +148,18 @@ char *favicon_get_icon_ref(struct content *c, xmlNode *html)
 
 bool favicon_get_icon(struct content *c, xmlNode *html)
 {
+	static const content_type permitted_types[] = {
+#ifdef WITH_BMP
+		CONTENT_ICO,
+#endif
+#if defined(WITH_MNG) || defined(WITH_PNG)
+		CONTENT_PNG,
+#endif
+#ifdef WITH_GIF
+		CONTENT_GIF,
+#endif
+		CONTENT_UNKNOWN
+	};
 	char *url;
 	nserror error;
 
@@ -157,7 +169,7 @@ bool favicon_get_icon(struct content *c, xmlNode *html)
 
 	error = hlcache_handle_retrieve(url, LLCACHE_RETRIEVE_NO_ERROR_PAGES, 
 			content__get_url(c), NULL, favicon_callback, c, NULL, 
-			&c->data.html.favicon);	
+			permitted_types, &c->data.html.favicon);	
 
 	free(url);
 
@@ -171,29 +183,12 @@ bool favicon_get_icon(struct content *c, xmlNode *html)
 nserror favicon_callback(hlcache_handle *icon,
 		const hlcache_event *event, void *pw)
 {
-	static const content_type permitted_types[] = {
-#ifdef WITH_BMP
-		CONTENT_ICO,
-#endif
-#if defined(WITH_MNG) || defined(WITH_PNG)
-		CONTENT_PNG,
-#endif
-#ifdef WITH_GIF
-		CONTENT_GIF,
-#endif
-		CONTENT_UNKNOWN
-	};
 	struct content *c = pw;
-	const content_type *type;
 
 	switch (event->type) {
 	case CONTENT_MSG_LOADING:
 		/* check that the favicon is really a correct image type */
-		for (type = permitted_types; *type != CONTENT_UNKNOWN; type++)
-			if (content_get_type(icon) == *type)
-				break;
-
-		if (*type == CONTENT_UNKNOWN) {
+		if (content_get_type(icon) == CONTENT_UNKNOWN) {
 			union content_msg_data msg_data;
 
 			LOG(("%s is not a favicon", content_get_url(icon)));
diff --git a/render/html.c b/render/html.c
index 2ead62fc3..f50146d95 100644
--- a/render/html.c
+++ b/render/html.c
@@ -783,6 +783,7 @@ bool html_meta_refresh(struct content *c, xmlNode *head)
 
 bool html_find_stylesheets(struct content *c, xmlNode *html)
 {
+	static const content_type accept[] = { CONTENT_CSS, CONTENT_UNKNOWN };
 	xmlNode *node;
 	char *rel, *type, *media, *href, *url, *url2;
 	unsigned int i = STYLESHEET_START;
@@ -819,7 +820,7 @@ bool html_find_stylesheets(struct content *c, xmlNode *html)
 
 	ns_error = hlcache_handle_retrieve(default_stylesheet_url, 0,
 			content__get_url(c), NULL,
-			html_convert_css_callback, c, &child, 
+			html_convert_css_callback, c, &child, accept,
 			&c->data.html.stylesheets[
 					STYLESHEET_BASE].data.external);
 	if (ns_error != NSERROR_OK)
@@ -830,7 +831,7 @@ bool html_find_stylesheets(struct content *c, xmlNode *html)
 	if (c->data.html.quirks == BINDING_QUIRKS_MODE_FULL) {
 		ns_error = hlcache_handle_retrieve(quirks_stylesheet_url, 0,
 				content__get_url(c), NULL,
-				html_convert_css_callback, c, &child,
+				html_convert_css_callback, c, &child, accept,
 				&c->data.html.stylesheets[
 					STYLESHEET_QUIRKS].data.external);
 		if (ns_error != NSERROR_OK)
@@ -842,7 +843,7 @@ bool html_find_stylesheets(struct content *c, xmlNode *html)
 	if (option_block_ads) {
 		ns_error = hlcache_handle_retrieve(adblock_stylesheet_url, 0,
 				content__get_url(c), NULL,
-				html_convert_css_callback, c, &child,
+				html_convert_css_callback, c, &child, accept,
 				&c->data.html.stylesheets[
 					STYLESHEET_ADBLOCK].data.external);
 		if (ns_error != NSERROR_OK)
@@ -951,6 +952,7 @@ bool html_find_stylesheets(struct content *c, xmlNode *html)
 			ns_error = hlcache_handle_retrieve(url2, 0,
 					content__get_url(c), NULL,
 					html_convert_css_callback, c, &child,
+					accept,
 					&c->data.html.stylesheets[i].
 							data.external);
 
@@ -1246,7 +1248,7 @@ bool html_fetch_object(struct content *c, const char *url, struct box *box,
 	}
 
 	error = hlcache_handle_retrieve(url2, 0, content__get_url(c), NULL,
-			html_object_callback, c, &child,
+			html_object_callback, c, &child, permitted_types,
 			&c_fetch);
 
 	/* No longer need normalized url */
@@ -1313,6 +1315,7 @@ bool html_replace_object(struct content *c, unsigned int i, const char *url)
 	/* initialise fetch */
 	error = hlcache_handle_retrieve(url2, 0, content__get_url(c), NULL,
 			html_object_callback, c, &child,
+			c->data.html.object[i].permitted_types,
 			&c_fetch);
 
 	free(url2);