Improve error handling in html content

3 files changed, 117 insertions, 62 deletions

diff --git a/render/html.c b/render/html.c
index 09256a6fa..4e79e085e 100644
--- a/render/html.c
+++ b/render/html.c
@@ -113,13 +113,14 @@ static void html_box_convert_done(html_content *c, bool success)
 	LOG(("Done XML to box (%p)", c));
 
 	/* Clean up and report error if unsuccessful or aborted */
-	if ((success == false) || c->aborted) {
+	if ((success == false) || (c->aborted)) {
+		if (success == false) {
+			msg_data.errorcode = NSERROR_BOX_CONVERT;
+		} else {
+			msg_data.errorcode = NSERROR_STOPPED;
+		}
 		html_destroy_objects(c);
-		if (success == false)
-			msg_data.error = messages_get("NoMemory");
-		else
-			msg_data.error = messages_get("Stopped");
-		content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
+		content_broadcast(&c->base, CONTENT_MSG_ERRORCODE, msg_data);
 		content_set_error(&c->base);
 		return;
 	}
@@ -134,19 +135,22 @@ static void html_box_convert_done(html_content *c, bool success)
 
 	exc = dom_document_get_document_element(c->document, (void *) &html);
 	if ((exc != DOM_NO_ERR) || (html == NULL)) {
+		/** @todo should this call html_destroy_objects(c);
+		 * like the other error paths 
+		 */
 		LOG(("error retrieving html element from dom"));
-		msg_data.error = messages_get("ParsingFail");
-		content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
+		msg_data.errorcode = NSERROR_DOM;
+		content_broadcast(&c->base, CONTENT_MSG_ERRORCODE, msg_data);
 		content_set_error(&c->base);
 		return;
 	}
 
 	/* extract image maps - can't do this sensibly in xml_to_box */
-	if (imagemap_extract(c) == false) {
+	msg_data.errorcode = imagemap_extract(c);
+	if (msg_data.errorcode != NSERROR_OK) {
 		LOG(("imagemap extraction failed"));
 		html_destroy_objects(c);
-		msg_data.error = messages_get("NoMemory");
-		content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
+		content_broadcast(&c->base, CONTENT_MSG_ERRORCODE, msg_data);
 		content_set_error(&c->base);
 		dom_node_unref(html);
 		return;
@@ -159,8 +163,9 @@ static void html_box_convert_done(html_content *c, bool success)
 
 	content_set_ready(&c->base);
 
-	if (c->base.active == 0)
+	if (c->base.active == 0) {
 		content_set_done(&c->base);
+	}
 
 	html_set_status(c, "");
 	dom_node_unref(html);
@@ -413,9 +418,62 @@ html_create(const content_handler *handler,
 	return NSERROR_OK;
 }
 
+static nserror
+parse_chunk_to_nserror(dom_hubbub_error error)
+{
+	switch (error) {
 
+	/* HUBBUB_REPROCESS is not handled here because it can
+	 * never occur outside the hubbub treebuilder
+	 */
 
-static bool
+	case DOM_HUBBUB_OK: 
+		/* parsed ok */
+		return NSERROR_OK;
+
+	case (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_PAUSED): 
+		/* hubbub input paused */
+		return NSERROR_OK;
+
+	case DOM_HUBBUB_NOMEM:
+		/* out of memory error from DOM */
+		return NSERROR_NOMEM;
+
+	case (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_ENCODINGCHANGE):
+		/* encoding changed */
+		return NSERROR_ENCODING_CHANGE;
+
+	case (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_NOMEM):
+		/* out of memory error from parser */
+		return NSERROR_NOMEM;
+
+	case (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_BADPARM):
+		return NSERROR_BAD_PARAMETER;
+
+	case (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_INVALID):
+		return NSERROR_INVALID;
+
+	case (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_FILENOTFOUND):
+		return NSERROR_NOT_FOUND;
+
+	case (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_NEEDDATA):
+		return NSERROR_NEED_DATA;
+
+	case (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_BADENCODING):
+		return NSERROR_BAD_ENCODING;
+
+	case (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_UNKNOWN):
+		/* currently only generated by the libdom hubbub binding */
+	default:
+		/* unknown error */
+		/** @todo better error handling and reporting */
+		return NSERROR_UNKNOWN;
+	}
+	return NSERROR_UNKNOWN;
+}
+
+
+static nserror
 html_process_encoding_change(struct content *c, 
 			     const char *data, 
 			     unsigned int size)
@@ -425,34 +483,30 @@ html_process_encoding_change(struct content *c,
 	const char *encoding;
 	const char *source_data;
 	unsigned long source_size;
-	union content_msg_data msg_data;
 
 	/* Retrieve new encoding */
 	encoding = dom_hubbub_parser_get_encoding(html->parser, 
 						  &html->encoding_source);
-
 	if (encoding == NULL) {
-		msg_data.error = messages_get("NoMemory");
-		content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
-		return false;
+		return NSERROR_NOMEM;
 	}
 
-	if (html->encoding != NULL)
+	if (html->encoding != NULL) {
 		free(html->encoding);
+	}
 
 	html->encoding = strdup(encoding);
 	if (html->encoding == NULL) {
-		msg_data.error = messages_get("NoMemory");
-		content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
-		return false;
+		return NSERROR_NOMEM;
 	}
 
 	/* Destroy binding */
 	dom_hubbub_parser_destroy(html->parser);
 	html->parser = NULL;
 
-	if (html->document != NULL)
+	if (html->document != NULL) {
 		dom_node_unref(html->document);
+	}
 
 	/* Create new binding, using the new encoding */
 	html->parser = dom_hubbub_parser_create(html->encoding, 
@@ -468,9 +522,7 @@ html_process_encoding_change(struct content *c,
 		free(html->encoding);
 		html->encoding = strdup("Windows-1252");
 		if (html->encoding == NULL) {
-			msg_data.error = messages_get("NoMemory");
-			content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
-			return false;
+			return NSERROR_NOMEM;
 		}
 
 		html->parser = dom_hubbub_parser_create(html->encoding, 
@@ -486,10 +538,7 @@ html_process_encoding_change(struct content *c,
 			 * parser errors back instead of everything being
 			 * OOM 
 			 */
-
-			msg_data.error = messages_get("NoMemory");
-			content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
-			return false;
+			return NSERROR_NOMEM;
 		}
 
 	}
@@ -498,21 +547,16 @@ html_process_encoding_change(struct content *c,
 
 	/* Reprocess all the data.  This is safe because
 	 * the encoding is now specified at parser start which means
-	 * it cannot be changed again. */
-	error = dom_hubbub_parser_parse_chunk(html->parser, (const uint8_t *)source_data, source_size);
-
-	if ((error == DOM_HUBBUB_OK) || 
-	    (error == (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_PAUSED))) {
-		return true;
-	}
-
-	msg_data.error = messages_get("NoMemory");
-	content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
-
-	return false;
+	 * it cannot be changed again. 
+	 */
+	error = dom_hubbub_parser_parse_chunk(html->parser, 
+					      (const uint8_t *)source_data, 
+					      source_size);
 
+	return parse_chunk_to_nserror(error);
 }
 
+
 /**
  * Process data for CONTENT_HTML.
  */
@@ -524,21 +568,27 @@ html_process_data(struct content *c, const char *data, unsigned int size)
 	dom_hubbub_error error;
 	union content_msg_data msg_data;
 
-	error = dom_hubbub_parser_parse_chunk(html->parser, (const uint8_t *) data, size);
+	msg_data.errorcode = NSERROR_OK; /* assume its all going to be ok */
 
-	if ((error == DOM_HUBBUB_OK) || 
-	    (error == (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_PAUSED))) {
-		return true;
-	} else if (error == (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_ENCODINGCHANGE)) {
-		return html_process_encoding_change(c, data, size);
-	} 
+	error = dom_hubbub_parser_parse_chunk(html->parser, 
+					      (const uint8_t *) data, 
+					      size);
 
-	/** @todo better error handling and reporting */
-	msg_data.error = messages_get("NoMemory");
-	content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
 
-	return false;
-	
+	msg_data.errorcode = parse_chunk_to_nserror(error);
+
+	/* deal with encoding change */
+	if (msg_data.errorcode == NSERROR_ENCODING_CHANGE) {
+		 msg_data.errorcode = html_process_encoding_change(c, data, size);
+	}
+
+	/* broadcast the error if necessary */
+	if (msg_data.errorcode != NSERROR_OK) {
+		content_broadcast(c, CONTENT_MSG_ERRORCODE, msg_data);
+		return false;
+	}
+
+	return true;	
 }
 
 
diff --git a/render/imagemap.c b/render/imagemap.c
index bf90443c4..c41b86d0b 100644
--- a/render/imagemap.c
+++ b/render/imagemap.c
@@ -251,23 +251,25 @@ void imagemap_dump(html_content *c)
  * \param map_str A dom_string which is "map"
  * \return false on memory exhaustion, true otherwise
  */
-bool
+nserror
 imagemap_extract(html_content *c)
 {
 	dom_nodelist *nlist;
 	dom_exception exc;
 	unsigned long mapnr;
 	uint32_t maybe_maps;
-	
+	nserror ret = NSERROR_OK;
+
 	exc = dom_document_get_elements_by_tag_name(c->document, 
 						    corestring_dom_map, 
 						    &nlist);
 	if (exc != DOM_NO_ERR) {
-		return false;
+		return NSERROR_DOM;
 	}
 	
 	exc = dom_nodelist_get_length(nlist, &maybe_maps);
 	if (exc != DOM_NO_ERR) {
+		ret = NSERROR_DOM;
 		goto out_nlist;
 	}
 	
@@ -276,6 +278,7 @@ imagemap_extract(html_content *c)
 		dom_string *name;
 		exc = dom_nodelist_item(nlist, mapnr, &node);
 		if (exc != DOM_NO_ERR) {
+			ret = NSERROR_DOM;
 			goto out_nlist;
 		}
 		
@@ -283,6 +286,7 @@ imagemap_extract(html_content *c)
 						&name);
 		if (exc != DOM_NO_ERR) {
 			dom_node_unref(node);
+			ret = NSERROR_DOM;
 			goto out_nlist;
 		}
 		
@@ -292,6 +296,7 @@ imagemap_extract(html_content *c)
 							&name);
 			if (exc != DOM_NO_ERR) {
 				dom_node_unref(node);
+				ret = NSERROR_DOM;
 				goto out_nlist;
 			}
 		}
@@ -301,6 +306,7 @@ imagemap_extract(html_content *c)
 			if (imagemap_extract_map(node, c, &entry) == false) {
 				dom_string_unref(name);
 				dom_node_unref(node);
+				ret = NSERROR_NOMEM; /** @todo check this */
 				goto out_nlist;
 			}
 			
@@ -313,6 +319,7 @@ imagemap_extract(html_content *c)
 			    (imagemap_add(c, name, entry) == false)) {
 				dom_string_unref(name);
 				dom_node_unref(node);
+				ret = NSERROR_NOMEM; /** @todo check this */
 				goto out_nlist;
 			}
 		}
@@ -321,14 +328,12 @@ imagemap_extract(html_content *c)
 		dom_node_unref(node);
 	}
 	
-	dom_nodelist_unref(nlist);
-	
-	return true;
 	
 out_nlist:
 	
 	dom_nodelist_unref(nlist);
-	return false;
+
+	return ret;
 }
 
 /**
diff --git a/render/imagemap.h b/render/imagemap.h
index edbfea08f..d450cda86 100644
--- a/render/imagemap.h
+++ b/render/imagemap.h
@@ -28,7 +28,7 @@ struct hlcache_handle;
 
 void imagemap_destroy(struct html_content *c);
 void imagemap_dump(struct html_content *c);
-bool imagemap_extract(struct html_content *c);
+nserror imagemap_extract(struct html_content *c);
 
 nsurl *imagemap_get(struct html_content *c, const char *key,
 		unsigned long x, unsigned long y,