[js] switch to libxml++ . Begining.

meson.build

@@ -88,6 +88,7 @@ conf_data.set('CONFIG_LIBEVENT', get_option('libevent'))
 conf_data.set('CONFIG_X', get_option('x'))
 conf_data.set('CONFIG_LIBDOM', get_option('libdom'))
 conf_data.set('CONFIG_HTMLCXX', get_option('htmlcxx'))
+conf_data.set('CONFIG_XML', get_option('xml'))
 
 #CONFIG_BOOKMARKS=true
 #CONFIG_XBEL_BOOKMARKS=true
@@ -301,6 +302,14 @@ if conf_data.get('CONFIG_HTMLCXX')
     deps += htmlcxxdeps
 endif
 
+if conf_data.get('CONFIG_XML')
+    xml2deps = dependency('libxml2')
+    deps += xml2deps
+    xmlplusplusdeps = dependency('libxml++-5.0')
+    deps += xmlplusplusdeps
+endif
+
+
 if conf_data.get('CONFIG_SCRIPTING_PYTHON')
     python3deps = dependency('python3-embed')
     deps += python3deps

meson_options.txt

@@ -62,3 +62,4 @@ option('libevent', type: 'boolean', value: false, description: 'compile with lib
 option('x', type: 'boolean', value: false, description: 'use the X Window System')
 option('libdom', type: 'boolean', value: false, description: 'libdom')
 option('htmlcxx', type: 'boolean', value: false, description: 'htmlcxx')
+option('xml', type: 'boolean', value: false, description: 'libxml++')

src/ecmascript/spidermonkey/document.c

@@ -49,8 +49,9 @@
 #include "viewer/text/link.h"
 #include "viewer/text/vs.h"
 
-#include <htmlcxx/html/ParserDom.h>
+#include <libxml/tree.h>
-using namespace htmlcxx;
+#include <libxml/HTMLparser.h>
+#include <libxml++/libxml++.h>
 
 #include <iostream>
 
@@ -698,12 +699,14 @@ document_parse(struct document *document)
 
 	add_bytes_to_string(&str, f->data, f->length);
 
-	HTML::ParserDom parser;
+	 // Parse HTML and create a DOM tree
-	tree<HTML::Node> *dom = new tree<HTML::Node>;
+	xmlDoc* doc = htmlReadDoc((xmlChar*)str.source, NULL, NULL, HTML_PARSE_RECOVER | HTML_PARSE_NOERROR | HTML_PARSE_NOWARNING);
-	*dom = parser.parseTree(str.source);
+	// Encapsulate raw libxml document in a libxml++ wrapper
+	xmlNode* r = xmlDocGetRootElement(doc);
+	xmlpp::Element* root = new xmlpp::Element(r);
 	done_string(&str);
 
-	return (void *)dom;
+	return (void *)root;
 }
 
 
@@ -731,9 +734,7 @@ document_getElementById(JSContext *ctx, unsigned int argc, JS::Value *vp)
 		return true;
 	}
 
-	tree<HTML::Node> *dom = document->dom;
+	xmlpp::Element* root = (xmlpp::Element *)document->dom;
-	tree<HTML::Node>::iterator it = dom->begin();
-	tree<HTML::Node>::iterator end = dom->end();
 
 	struct string idstr;
 
@@ -741,21 +742,22 @@ document_getElementById(JSContext *ctx, unsigned int argc, JS::Value *vp)
 	jshandle_value_to_char_string(&idstr, ctx, &args[0]);
 	std::string id = idstr.source;
 
-	JSObject *elem = nullptr;
+	std::string xpath = "//*[@id=\"";
-
+	xpath += id;
-	for (; it != end; ++it) {
+	xpath += "\"]";
-		if (it->isTag()) {
-			it->parseAttributes();
-			if (it->attribute("id").first && it->attribute("id").second == id) {
-				tree<HTML::Node> *node = new tree<HTML::Node>;
-				*node = *it;
-				elem = getElement(ctx, node);
-				break;
-			}
-		}
-	}
 
 	done_string(&idstr);
+
+	auto elements = root->find(xpath);
+
+	if (elements.size() == 0) {
+		args.rval().setNull();
+		return true;
+	}
+
+	auto node = elements[0];
+	JSObject *elem = getElement(ctx, node);
+
 	if (elem) {
 		args.rval().setObject(*elem);
 	} else {

src/ecmascript/spidermonkey/element.c

@@ -46,12 +46,12 @@
 #include "viewer/text/link.h"
 #include "viewer/text/vs.h"
 
+#include <libxml++/libxml++.h>
+
 #include <iostream>
 #include <algorithm>
 #include <string>
 
-using namespace htmlcxx;
-
 static bool element_get_property_childElementCount(JSContext *ctx, unsigned int argc, JS::Value *vp);
 static bool element_get_property_className(JSContext *ctx, unsigned int argc, JS::Value *vp);
 static bool element_set_property_className(JSContext *ctx, unsigned int argc, JS::Value *vp);
@@ -84,16 +84,16 @@ JSClass element_class = {
 };
 
 JSPropertySpec element_props[] = {
-	JS_PSG("childElementCount",	element_get_property_childElementCount, JSPROP_ENUMERATE),
+//	JS_PSG("childElementCount",	element_get_property_childElementCount, JSPROP_ENUMERATE),
-	JS_PSGS("className",	element_get_property_className, element_set_property_className, JSPROP_ENUMERATE),
+//	JS_PSGS("className",	element_get_property_className, element_set_property_className, JSPROP_ENUMERATE),
-	JS_PSGS("dir",	element_get_property_dir, element_set_property_dir, JSPROP_ENUMERATE),
+//	JS_PSGS("dir",	element_get_property_dir, element_set_property_dir, JSPROP_ENUMERATE),
 	JS_PSGS("id",	element_get_property_id, element_set_property_id, JSPROP_ENUMERATE),
-	JS_PSGS("innerHTML",	element_get_property_innerHtml, element_set_property_innerHtml, JSPROP_ENUMERATE),
+//	JS_PSGS("innerHTML",	element_get_property_innerHtml, element_set_property_innerHtml, JSPROP_ENUMERATE),
-	JS_PSGS("lang",	element_get_property_lang, element_set_property_lang, JSPROP_ENUMERATE),
+//	JS_PSGS("lang",	element_get_property_lang, element_set_property_lang, JSPROP_ENUMERATE),
-	JS_PSGS("outerHTML",	element_get_property_outerHtml, element_set_property_outerHtml, JSPROP_ENUMERATE),
+//	JS_PSGS("outerHTML",	element_get_property_outerHtml, element_set_property_outerHtml, JSPROP_ENUMERATE),
-	JS_PSG("tagName",	element_get_property_tagName, JSPROP_ENUMERATE),
+//	JS_PSG("tagName",	element_get_property_tagName, JSPROP_ENUMERATE),
-	JS_PSGS("textContent",	element_get_property_textContent, element_set_property_textContent, JSPROP_ENUMERATE),
+//	JS_PSGS("textContent",	element_get_property_textContent, element_set_property_textContent, JSPROP_ENUMERATE),
-	JS_PSGS("title",	element_get_property_title, element_set_property_title, JSPROP_ENUMERATE),
+//	JS_PSGS("title",	element_get_property_title, element_set_property_title, JSPROP_ENUMERATE),
 	JS_PS_END
 };
 
@@ -268,17 +268,14 @@ element_get_property_id(JSContext *ctx, unsigned int argc, JS::Value *vp)
 		return false;
 	}
 
-	tree<HTML::Node> *el = JS_GetPrivate(hobj);
+	xmlpp::Element *el = JS_GetPrivate(hobj);
 
 	if (!el) {
 		args.rval().setNull();
 		return true;
 	}
 
-	tree<HTML::Node>::iterator it = el->begin();
+	std::string v = el->get_attribute_value("id");
-	it->parseAttributes();
-	std::string v = it->attribute("id").second;
-
 	args.rval().setString(JS_NewStringCopyZ(ctx, v.c_str()));
 
 	return true;
@@ -877,8 +874,8 @@ static bool element_hasAttribute(JSContext *ctx, unsigned int argc, JS::Value *r
 static bool element_hasAttributes(JSContext *ctx, unsigned int argc, JS::Value *rval);
 
 const spidermonkeyFunctionSpec element_funcs[] = {
-	{ "hasAttribute",		element_hasAttribute,	1 },
+//	{ "hasAttribute",		element_hasAttribute,	1 },
-	{ "hasAttributes",		element_hasAttributes,	0 },
+//	{ "hasAttributes",		element_hasAttributes,	0 },
 	{ NULL }
 };