1 files changed, 108 insertions, 405 deletions

diff --git a/render/html.c b/render/html.c
index 996bfc3a5..8c11f35b9 100644
--- a/render/html.c
+++ b/render/html.c
@@ -28,8 +28,6 @@
 #include <strings.h>
 #include <stdlib.h>
 
-#include <dom/dom.h>
-
 #include "utils/config.h"
 #include "content/content_protected.h"
 #include "content/fetch.h"
@@ -38,7 +36,6 @@
 #include "desktop/options.h"
 #include "desktop/selection.h"
 #include "desktop/scrollbar.h"
-#include "javascript/js.h"
 #include "image/bitmap.h"
 #include "render/box.h"
 #include "render/font.h"
@@ -86,15 +83,15 @@ static dom_string *html_dom_string_head;
 static dom_string *html_dom_string_rel;
 dom_string *html_dom_string_href;
 static dom_string *html_dom_string_hreflang;
-static dom_string *html_dom_string_type;
+dom_string *html_dom_string_type;
 static dom_string *html_dom_string_media;
 static dom_string *html_dom_string_sizes;
 static dom_string *html_dom_string_title;
 static dom_string *html_dom_string_base;
 static dom_string *html_dom_string_link;
 static dom_string *html_dom_string_script;
-static dom_string *html_dom_string_text_javascript;
-static dom_string *html_dom_string_src;
+dom_string *html_dom_string_text_javascript;
+dom_string *html_dom_string_src;
 dom_string *html_dom_string_target;
 static dom_string *html_dom_string__parent;
 static dom_string *html_dom_string__self;
@@ -117,107 +114,6 @@ dom_string *html_dom_string_circle;
 dom_string *html_dom_string_poly;
 dom_string *html_dom_string_polygon;
 
-typedef bool (script_handler_t)(struct jscontext *jscontext, const char *data, size_t size) ;
-
-
-static script_handler_t *select_script_handler(content_type ctype)
-{
-	if (ctype == CONTENT_JS) {
-		return js_exec;
-	}
-	return NULL;
-}
-
-
-/* attempt to progress script execution
- *
- * execute scripts using algorithm found in:
- * http://www.whatwg.org/specs/web-apps/current-work/multipage/scripting-1.html#the-script-element
- *
- */
-static bool html_scripts_exec(html_content *c)
-{
-	unsigned int i;
-	struct html_script *s;
-	script_handler_t *script_handler;
-
-	if (c->jscontext == NULL)
-		return false;
-
-	for (i = 0, s = c->scripts; i != c->scripts_count; i++, s++) {
-		if (s->already_started) {
-			continue;
-		}
-
-		assert((s->type == HTML_SCRIPT_EXTERNAL) ||
-		       (s->type == HTML_SCRIPT_INTERNAL));
-
-		if (s->type == HTML_SCRIPT_EXTERNAL) {
-			/* ensure script content is present */
-			if (s->data.external == NULL)
-				continue;
-
-			/* ensure script content fetch status is not an error */
-			if (content_get_status(s->data.external) == CONTENT_STATUS_ERROR)
-				continue;
-
-			/* ensure script handler for content type */
-			script_handler = select_script_handler(content_get_type(s->data.external));
-			if (script_handler == NULL)
-				continue; /* unsupported type */
-
-			if (content_get_status(s->data.external) == CONTENT_STATUS_DONE) {
-				/* external script is now available */
-				const char *data;
-				unsigned long size;
-				data = content_get_source_data(s->data.external, &size );
-				script_handler(c->jscontext, data, size);
-
-				s->already_started = true;
-
-			} else {
-				/* script not yet available */
-
-				/* check if deferable or asynchronous */
-				if (!s->defer && !s->async) {
-					break;
-				}
-			}
-		}
-	}
-
-	return true;
-}
-
-/* create new html script entry */
-static struct html_script *
-html_process_new_script(html_content *c, enum html_script_type type)
-{
-	struct html_script *nscript;
-	/* add space for new script entry */
-	nscript = realloc(c->scripts,
-			  sizeof(struct html_script) * (c->scripts_count + 1));
-	if (nscript == NULL) {
-		return NULL;
-	}
-
-	c->scripts = nscript;
-
-	/* increment script entry count */
-	nscript = &c->scripts[c->scripts_count];
-	c->scripts_count++;
-
-	nscript->already_started = false;
-	nscript->parser_inserted = false;
-	nscript->force_async = true;
-	nscript->ready_exec = false;
-	nscript->async = false;
-	nscript->defer = false;
-
-	nscript->type = type;
-
-	return nscript;
-}
 
 static void html_destroy_objects(html_content *html)
 {
@@ -293,8 +189,8 @@ static void html_box_convert_done(html_content *c, bool success)
 	/*imagemap_dump(c);*/
 
 	/* Destroy the parser binding */
-	binding_destroy_tree(c->parser_binding);
-	c->parser_binding = NULL;
+	dom_hubbub_parser_destroy(c->parser);
+	c->parser = NULL;
 
 	content_set_ready(&c->base);
 
@@ -309,7 +205,7 @@ static void html_box_convert_done(html_content *c, bool success)
  *
  * \param c  Content to convert
  */
-static void html_finish_conversion(html_content *c)
+void html_finish_conversion(html_content *c)
 {
 	union content_msg_data msg_data;
 	dom_exception exc; /* returned by libdom functions */
@@ -400,217 +296,17 @@ static void html_finish_conversion(html_content *c)
 }
 
 
-/**
- * Callback for fetchcache() for linked stylesheets.
- */
-
-static nserror
-html_convert_script_callback(hlcache_handle *script,
-			  const hlcache_event *event,
-			  void *pw)
-{
-	html_content *parent = pw;
-	unsigned int i;
-	struct html_script *s;
-
-	/* Find script */
-	for (i = 0, s = parent->scripts; i != parent->scripts_count; i++, s++) {
-		if (s->type == HTML_SCRIPT_EXTERNAL &&
-		    s->data.external == script)
-			break;
-	}
-
-	assert(i != parent->scripts_count);
-
-	switch (event->type) {
-	case CONTENT_MSG_LOADING:
-		break;
-
-	case CONTENT_MSG_READY:
-		break;
-
-	case CONTENT_MSG_DONE:
-		LOG(("script %d done '%s'", i,
-				nsurl_access(hlcache_handle_get_url(script))));
-		parent->base.active--;
-		LOG(("%d fetches active", parent->base.active));
-
-		/* script finished loading so try and continue execution */
-		html_scripts_exec(parent);
-		break;
-
-	case CONTENT_MSG_ERROR:
-		LOG(("script %s failed: %s",
-				nsurl_access(hlcache_handle_get_url(script)),
-				event->data.error));
-		hlcache_handle_release(script);
-		s->data.external = NULL;
-		parent->base.active--;
-		LOG(("%d fetches active", parent->base.active));
-		content_add_error(&parent->base, "?", 0);
-
-		/* script failed loading so try and continue execution */
-		html_scripts_exec(parent);
-
-		break;
-
-	case CONTENT_MSG_STATUS:
-		html_set_status(parent, content_get_status_message(script));
-		content_broadcast(&parent->base, CONTENT_MSG_STATUS,
-				event->data);
-		break;
-
-	default:
-		assert(0);
-	}
-
-	if (parent->base.active == 0)
-		html_finish_conversion(parent);
-
-	return NSERROR_OK;
-}
-
-/** process script node
- *
- *
- */
-static dom_hubbub_error
-html_process_script(void *ctx, dom_node *node)
-{
-	html_content *c = (html_content *)ctx;
-	dom_exception exc; /* returned by libdom functions */
-	dom_string *src, *script, *mimetype;
-	struct html_script *nscript;
-	union content_msg_data msg_data;
-
-	/* ensure javascript context is available */
-	if (c->jscontext == NULL) {
-		union content_msg_data msg_data;
-
-		msg_data.jscontext = &c->jscontext;
-		content_broadcast(&c->base, CONTENT_MSG_GETCTX, msg_data);
-		LOG(("javascript context %p ", c->jscontext));
-		if (c->jscontext == NULL) {
-			/* no context and it could not be created, abort */
-			return DOM_HUBBUB_OK;
-		}
-	}
-
-	LOG(("content %p parser %p node %p",c,c->parser_binding, node));
-
-	exc = dom_element_get_attribute(node, html_dom_string_type, &mimetype);
-	if (exc != DOM_NO_ERR || mimetype == NULL) {
-		mimetype = dom_string_ref(html_dom_string_text_javascript);
-	}
-
-	exc = dom_element_get_attribute(node, html_dom_string_src, &src);
-	if (exc != DOM_NO_ERR || src == NULL) {
-		struct lwc_string_s *lwcmimetype;
-		script_handler_t *script_handler;
-
-		/* does not appear to be a src so script is inline content */
-		exc = dom_node_get_text_content(node, &script);
-		if ((exc != DOM_NO_ERR) || (script == NULL)) {
-			dom_string_unref(mimetype);
-			return DOM_HUBBUB_OK; /* no contents, skip */
-		}
-
-		nscript = html_process_new_script(c, HTML_STYLESHEET_INTERNAL);
-		if (nscript == NULL) {
-			dom_string_unref(mimetype);
-			dom_string_unref(script);
-			goto html_process_script_no_memory;
-		}
-
-		nscript->data.internal = script;
-		nscript->mimetype = mimetype;
-		nscript->already_started = true;
-
-		/* charset (encoding) */
-
-		/* ensure script handler for content type */
-		dom_string_intern(mimetype, &lwcmimetype);
-		script_handler = select_script_handler(content_factory_type_from_mime_type(lwcmimetype));
-		lwc_string_unref(lwcmimetype);
-
-		if (script_handler != NULL) {
-			script_handler(c->jscontext,
-				       dom_string_data(script),
-				       dom_string_byte_length(script));
-		}
-
-
-	} else {
-		/* script with a src tag */
-		nserror ns_error;
-		nsurl *joined;
-		hlcache_child_context child;
-
-
-		nscript = html_process_new_script(c, HTML_STYLESHEET_EXTERNAL);
-		if (nscript == NULL) {
-			dom_string_unref(src);
-			dom_string_unref(mimetype);
-			goto html_process_script_no_memory;
-		}
-
-		/* charset (encoding) */
-
-		ns_error = nsurl_join(c->base_url, dom_string_data(src), &joined);
-		dom_string_unref(src);
-		if (ns_error != NSERROR_OK) {
-			dom_string_unref(mimetype);
-			goto html_process_script_no_memory;
-		}
-
-		nscript->mimetype = mimetype; /* keep reference to mimetype */
-
-		LOG(("script %i '%s'", c->scripts_count, nsurl_access(joined)));
-
-		child.charset = c->encoding;
-		child.quirks = c->base.quirks;
-
-		ns_error = hlcache_handle_retrieve(joined,
-						   0,
-						   content_get_url(&c->base),
-						   NULL,
-						   html_convert_script_callback,
-						   c,
-						   &child,
-						   CONTENT_SCRIPT,
-						   &nscript->data.external);
-
-		nsurl_unref(joined);
-
-		if (ns_error != NSERROR_OK) {
-			goto html_process_script_no_memory;
-		}
-
-		c->base.active++; /* ensure base content knows the fetch is active */
-		LOG(("%d fetches active", c->base.active));
-
-	}
-	html_scripts_exec(c);
-
-	return DOM_HUBBUB_OK;
-
-html_process_script_no_memory:
-	msg_data.error = messages_get("NoMemory");
-	content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
-	return DOM_HUBBUB_NOMEM;
-}
 
 static nserror
 html_create_html_data(html_content *c, const http_parameter *params)
 {
 	lwc_string *charset;
 	union content_msg_data msg_data;
-	binding_error error;
 	nserror nerror;
 
-	c->parser_binding = NULL;
+	c->parser = NULL;
 	c->document = NULL;
-	c->quirks = BINDING_QUIRKS_MODE_NONE;
+	c->quirks = DOM_DOCUMENT_QUIRKS_MODE_NONE;
 	c->encoding = NULL;
 	c->base_url = nsurl_ref(content_get_url(&c->base));
 	c->base_target = NULL;
@@ -637,8 +333,10 @@ html_create_html_data(html_content *c, const http_parameter *params)
 	c->jscontext = NULL;
 
 	if (lwc_intern_string("*", SLEN("*"), &c->universal) != lwc_error_ok) {
-		error = BINDING_NOMEM;
-		goto error;
+		msg_data.error = messages_get("NoMemory");
+		content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
+
+		return NSERROR_NOMEM;
 	}
 
 	selection_prepare(&c->sel, (struct content *)c, true);
@@ -650,60 +348,56 @@ html_create_html_data(html_content *c, const http_parameter *params)
 		lwc_string_unref(charset);
 
 		if (c->encoding == NULL) {
-			error = BINDING_NOMEM;
-			goto error;
+			lwc_string_unref(c->universal);
+			c->universal = NULL;
+
+			msg_data.error = messages_get("NoMemory");
+			content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
+
+			return NSERROR_NOMEM;
+
 		}
-		c->encoding_source = ENCODING_SOURCE_HEADER;
+		c->encoding_source = DOM_HUBBUB_ENCODING_SOURCE_HEADER;
 	}
 
 	/* Create the parser binding */
-	error = binding_create_tree(&c->parser_binding,
-				    c->encoding,
-				    nsoption_bool(enable_javascript),
-				    html_process_script,
-				    c);
-	if (error == BINDING_BADENCODING && c->encoding != NULL) {
+	c->parser = dom_hubbub_parser_create(c->encoding, 
+					     true, 
+					     nsoption_bool(enable_javascript), 
+					     NULL, 
+					     html_process_script, 
+					     c);
+	if ((c->parser == NULL) && (c->encoding != NULL)) {
 		/* Ok, we don't support the declared encoding. Bailing out
 		 * isn't exactly user-friendly, so fall back to autodetect */
 		talloc_free(c->encoding);
 		c->encoding = NULL;
 
-		error = binding_create_tree(&c->parser_binding,
-					    c->encoding,
-					    nsoption_bool(enable_javascript),
-					    html_process_script,
-					    c);
+		c->parser = dom_hubbub_parser_create(c->encoding, 
+						     true, 
+						     nsoption_bool(enable_javascript), 
+						     NULL, 
+						     html_process_script, 
+						     c);
 
-	}
-
-	if (error != BINDING_OK)
-		goto error;
 
-	return NSERROR_OK;
-
-error:
-	if (error == BINDING_BADENCODING) {
-		LOG(("Bad encoding: %s", c->encoding ? c->encoding : ""));
-		msg_data.error = messages_get("ParsingFail");
-		nerror = NSERROR_BAD_ENCODING;
-	} else {
-		msg_data.error = messages_get("NoMemory");
-		nerror = NSERROR_NOMEM;
 	}
 
-	content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
+	if (c->parser == NULL) {
+		nsurl_unref(c->base_url);
+		c->base_url = NULL;
 
-	if (c->universal != NULL) {
 		lwc_string_unref(c->universal);
 		c->universal = NULL;
-	}
 
-	if (c->base_url != NULL) {
-		nsurl_unref(c->base_url);
-		c->base_url = NULL;
+		msg_data.error = messages_get("NoMemory");
+		content_broadcast(&c->base, CONTENT_MSG_ERROR, msg_data);
+
+		return NSERROR_NOMEM;
 	}
 
-	return nerror;
+	return NSERROR_OK;
+
 }
 
 /**
@@ -757,14 +451,16 @@ static bool
 html_process_data(struct content *c, const char *data, unsigned int size)
 {
 	html_content *html = (html_content *) c;
-	binding_error err;
+	dom_hubbub_error error;
 	const char *encoding;
+	const char *source_data;
+	unsigned long source_size;
 
-	err = binding_parse_chunk(html->parser_binding,
-			(const uint8_t *) data, size);
-	if (err == BINDING_ENCODINGCHANGE) {
+	error = dom_hubbub_parser_parse_chunk(html->parser, (const uint8_t *) data, size);
+
+	if (error == (DOM_HUBBUB_HUBBUB_ERR | HUBBUB_ENCODINGCHANGE)) {
 		goto encoding_change;
-	} else if (err != BINDING_OK) {
+	} else if (error != DOM_HUBBUB_OK) {
 		union content_msg_data msg_data;
 
 		msg_data.error = messages_get("NoMemory");
@@ -778,9 +474,8 @@ html_process_data(struct content *c, const char *data, unsigned int size)
 encoding_change:
 
 	/* Retrieve new encoding */
-	encoding = binding_get_encoding(
-			html->parser_binding,
-			&html->encoding_source);
+	encoding = dom_hubbub_parser_get_encoding(html->parser, 
+						  &html->encoding_source);
 
 	if (html->encoding != NULL)
 		talloc_free(html->encoding);
@@ -795,16 +490,17 @@ encoding_change:
 	}
 
 	/* Destroy binding */
-	binding_destroy_tree(html->parser_binding);
-	html->parser_binding = NULL;
+	dom_hubbub_parser_destroy(html->parser);
+	html->parser = NULL;
 
 	/* Create new binding, using the new encoding */
-	err = binding_create_tree(&html->parser_binding,
-				  html->encoding,
-				  nsoption_bool(enable_javascript),
-				  html_process_script,
-				  html);
-	if (err == BINDING_BADENCODING) {
+	html->parser = dom_hubbub_parser_create(html->encoding, 
+				true, 
+				nsoption_bool(enable_javascript), 
+				NULL, 
+				html_process_script, 
+				html);
+	if (html->parser == NULL) {
 		/* Ok, we don't support the declared encoding. Bailing out
 		 * isn't exactly user-friendly, so fall back to Windows-1252 */
 		talloc_free(html->encoding);
@@ -817,37 +513,35 @@ encoding_change:
 			return false;
 		}
 
-		err = binding_create_tree(&html->parser_binding,
-					  html->encoding,
-					  nsoption_bool(enable_javascript),
-					  html_process_script,
-					  html);
-	}
+		html->parser = dom_hubbub_parser_create(html->encoding, 
+							true, 
+							nsoption_bool(enable_javascript), 
+							NULL, 
+							html_process_script, 
+							html);
 
-	if (err != BINDING_OK) {
-		union content_msg_data msg_data;
+		if (html->parser == NULL) {
+			union content_msg_data msg_data;
+
+			/** @todo add a message callback function and pass the
+			 * parser errors back instead of everything being
+			 * OOM 
+			 */
 
-		if (err == BINDING_BADENCODING) {
-			LOG(("Bad encoding: %s", html->encoding
-					? html->encoding : ""));
-			msg_data.error = messages_get("ParsingFail");
-		} else
 			msg_data.error = messages_get("NoMemory");
-		content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
-		return false;
-	}
+			content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
+			return false;
+		}
 
-	{
-		const char *source_data;
-		unsigned long source_size;
+	}
 
-		source_data = content__get_source_data(c, &source_size);
+	source_data = content__get_source_data(c, &source_size);
 
-		/* Recurse to reprocess all the data.  This is safe because
-		 * the encoding is now specified at parser start which means
-		 * it cannot be changed again. */
-		return html_process_data(c, source_data, source_size);
-	}
+	/* Recurse to reprocess all the data.  This is safe because
+	 * the encoding is now specified at parser start which means
+	 * it cannot be changed again. */
+	return html_process_data(c, source_data, source_size);
+	
 }
 
 
@@ -2156,7 +1850,7 @@ static bool html_find_stylesheets(html_content *c, dom_node *html)
 	c->base.active++;
 	LOG(("%d fetches active", c->base.active));
 
-	if (c->quirks == BINDING_QUIRKS_MODE_FULL) {
+	if (c->quirks == DOM_DOCUMENT_QUIRKS_MODE_FULL) {
 		ns_error = hlcache_handle_retrieve(html_quirks_stylesheet_url,
 				0, content_get_url(&c->base), NULL,
 				html_convert_css_callback, c, &child,
@@ -2224,7 +1918,7 @@ html_find_stylesheets_no_memory:
 static bool html_convert(struct content *c)
 {
 	html_content *htmlc = (html_content *) c;
-	binding_error err;
+	dom_hubbub_error err;
 	dom_node *html, *head;
 	union content_msg_data msg_data;
 	unsigned long size;
@@ -2235,18 +1929,18 @@ static bool html_convert(struct content *c)
 	/* finish parsing */
 	content__get_source_data(c, &size);
 
-	err = binding_parse_completed(htmlc->parser_binding);
-	if (err != BINDING_OK) {
+	err = dom_hubbub_parser_completed(htmlc->parser);
+	if (err != DOM_HUBBUB_OK) {
 		union content_msg_data msg_data;
 
+		/** @todo Improve precessing of errors */
 		msg_data.error = messages_get("NoMemory");
 		content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
 
 		return false;
 	}
 
-	htmlc->document = binding_get_document(htmlc->parser_binding,
-					&htmlc->quirks);
+	htmlc->document = dom_hubbub_parser_get_document(htmlc->parser);
 
 	if (htmlc->document == NULL) {
 		LOG(("Parsing failed"));
@@ -2255,10 +1949,18 @@ static bool html_convert(struct content *c)
 		return false;
 	}
 
+	exc = dom_document_get_quirks_mode(htmlc->document, &htmlc->quirks);
+	if (exc != DOM_NO_ERR) {
+		LOG(("error retrieving quirks"));
+	} 
+
+	LOG(("quirks set to %d", htmlc->quirks));
+
+
 	if (htmlc->encoding == NULL) {
-		const char *encoding = binding_get_encoding(
-				htmlc->parser_binding,
-				&htmlc->encoding_source);
+		const char *encoding;
+		encoding = dom_hubbub_parser_get_encoding(htmlc->parser,
+					&htmlc->encoding_source);
 
 		htmlc->encoding = talloc_strdup(c, encoding);
 		if (htmlc->encoding == NULL) {
@@ -2667,13 +2369,14 @@ static void html_destroy(struct content *c)
 	if (html->base_url)
 		nsurl_unref(html->base_url);
 
-	if (html->parser_binding != NULL) {
-		binding_destroy_tree(html->parser_binding);
-		html->parser_binding = NULL;
+	if (html->parser != NULL) {
+		dom_hubbub_parser_destroy(html->parser);
+		html->parser = NULL;
 	}
 
-	if (html->document != NULL)
-		binding_destroy_document(html->document);
+	if (html->document != NULL) {
+		dom_node_unref(html->document);
+	}
 
 	/* Free base target */
 	if (html->base_target != NULL) {
@@ -3244,7 +2947,7 @@ const char *html_get_encoding(hlcache_handle *h)
  * \param h  Content to retrieve charset from
  * \return Pointer to charset, or NULL
  */
-binding_encoding_source html_get_encoding_source(hlcache_handle *h)
+dom_hubbub_encoding_source html_get_encoding_source(hlcache_handle *h)
 {
 	html_content *c = (html_content *) hlcache_handle_get_content(h);