parse5 1.4.0 → 1.5.1

package/lib/{tree_construction → common}/doctype.js

@@ -81,6 +81,16 @@ var VALID_DOCTYPE_NAME = 'html',
         'html'
     ];
 
+
+//Utils
+function enquoteDoctypeId(id) {
+    var quote = id.indexOf('"') !== -1 ? '\'' : '"';
+
+    return quote + id + quote;
+}
+
+
+//API
 exports.isQuirks = function (name, publicId, systemId) {
     if (name !== VALID_DOCTYPE_NAME)
         return true;
@@ -107,3 +117,18 @@ exports.isQuirks = function (name, publicId, systemId) {
 
     return false;
 };
+
+exports.serializeContent = function (name, publicId, systemId) {
+    var str = '!DOCTYPE ' + name;
+
+    if (publicId !== null)
+        str += ' PUBLIC ' + enquoteDoctypeId(publicId);
+
+    else if (systemId !== null)
+        str += ' SYSTEM';
+
+    if (systemId !== null)
+        str += ' ' + enquoteDoctypeId(systemId);
+
+    return str;
+};

package/lib/serialization/serializer.js

@@ -1,194 +1,178 @@
-'use strict';
-
-var DefaultTreeAdapter = require('../tree_adapters/default'),
-    Utils = require('../common/utils'),
-    HTML = require('../common/html');
-
-//Aliases
-var $ = HTML.TAG_NAMES,
-    NS = HTML.NAMESPACES;
-
-//Default serializer options
-var DEFAULT_OPTIONS = {
-    encodeHtmlEntities: true
-};
-
-//Escaping regexes
-var AMP_REGEX = /&/g,
-    NBSP_REGEX = /\u00a0/g,
-    DOUBLE_QUOTE_REGEX = /"/g,
-    LT_REGEX = /</g,
-    GT_REGEX = />/g;
-
-//Escape string
-function escapeString(str, attrMode) {
-    str = str
-        .replace(AMP_REGEX, '&amp;')
-        .replace(NBSP_REGEX, '&nbsp;');
-
-    if (attrMode)
-        str = str.replace(DOUBLE_QUOTE_REGEX, '&quot;');
-
-    else {
-        str = str
-            .replace(LT_REGEX, '&lt;')
-            .replace(GT_REGEX, '&gt;');
-    }
-
-    return str;
-}
-
-
-//Enquote doctype ID
-function enquoteDoctypeId(id) {
-    var quote = id.indexOf('"') !== -1 ? '\'' : '"';
-
-    return quote + id + quote;
-}
-
-
-//Serializer
-var Serializer = module.exports = function (treeAdapter, options) {
-    this.treeAdapter = treeAdapter || DefaultTreeAdapter;
-    this.options = Utils.mergeOptions(DEFAULT_OPTIONS, options);
-};
-
-
-//API
-Serializer.prototype.serialize = function (node) {
-    this.html = '';
-    this._serializeChildNodes(node);
-
-    return this.html;
-};
-
-
-//Internals
-Serializer.prototype._serializeChildNodes = function (parentNode) {
-    var childNodes = this.treeAdapter.getChildNodes(parentNode);
-
-    if (childNodes) {
-        for (var i = 0, cnLength = childNodes.length; i < cnLength; i++) {
-            var currentNode = childNodes[i];
-
-            if (this.treeAdapter.isElementNode(currentNode))
-                this._serializeElement(currentNode);
-
-            else if (this.treeAdapter.isTextNode(currentNode))
-                this._serializeTextNode(currentNode);
-
-            else if (this.treeAdapter.isCommentNode(currentNode))
-                this._serializeCommentNode(currentNode);
-
-            else if (this.treeAdapter.isDocumentTypeNode(currentNode))
-                this._serializeDocumentTypeNode(currentNode);
-        }
-    }
-};
-
-Serializer.prototype._serializeElement = function (node) {
-    var tn = this.treeAdapter.getTagName(node),
-        ns = this.treeAdapter.getNamespaceURI(node),
-        qualifiedTn = (ns === NS.HTML || ns === NS.SVG || ns === NS.MATHML) ? tn : (ns + ':' + tn);
-
-    this.html += '<' + qualifiedTn;
-    this._serializeAttributes(node);
-    this.html += '>';
-
-    if (tn !== $.AREA && tn !== $.BASE && tn !== $.BASEFONT && tn !== $.BGSOUND && tn !== $.BR && tn !== $.BR &&
-        tn !== $.COL && tn !== $.EMBED && tn !== $.FRAME && tn !== $.HR && tn !== $.IMG && tn !== $.INPUT &&
-        tn !== $.KEYGEN && tn !== $.LINK && tn !== $.MENUITEM && tn !== $.META && tn !== $.PARAM && tn !== $.SOURCE &&
-        tn !== $.TRACK && tn !== $.WBR) {
-
-        if (tn === $.PRE || tn === $.TEXTAREA || tn === $.LISTING) {
-            var firstChild = this.treeAdapter.getFirstChild(node);
-
-            if (firstChild && this.treeAdapter.isTextNode(firstChild)) {
-                var content = this.treeAdapter.getTextNodeContent(firstChild);
-
-                if (content[0] === '\n')
-                    this.html += '\n';
-            }
-        }
-
-        var childNodesHolder = tn === $.TEMPLATE && ns === NS.HTML ?
-                               this.treeAdapter.getChildNodes(node)[0] :
-                               node;
-
-        this._serializeChildNodes(childNodesHolder);
-        this.html += '</' + qualifiedTn + '>';
-    }
-};
-
-Serializer.prototype._serializeAttributes = function (node) {
-    var attrs = this.treeAdapter.getAttrList(node);
-
-    for (var i = 0, attrsLength = attrs.length; i < attrsLength; i++) {
-        var attr = attrs[i],
-            value = this.options.encodeHtmlEntities ? escapeString(attr.value, true) : attr.value;
-
-        this.html += ' ';
-
-        if (!attr.namespace)
-            this.html += attr.name;
-
-        else if (attr.namespace === NS.XML)
-            this.html += 'xml:' + attr.name;
-
-        else if (attr.namespace === NS.XMLNS) {
-            if (attr.name !== 'xmlns')
-                this.html += 'xmlns:';
-
-            this.html += attr.name;
-        }
-
-        else if (attr.namespace === NS.XLINK)
-            this.html += 'xlink:' + attr.name;
-
-        else
-            this.html += attr.namespace + ':' + attr.name;
-
-        this.html += '="' + value + '"';
-    }
-};
-
-Serializer.prototype._serializeTextNode = function (node) {
-    var content = this.treeAdapter.getTextNodeContent(node),
-        parent = this.treeAdapter.getParentNode(node),
-        parentTn = void 0;
-
-    if (parent && this.treeAdapter.isElementNode(parent))
-        parentTn = this.treeAdapter.getTagName(parent);
-
-    if (parentTn === $.STYLE || parentTn === $.SCRIPT || parentTn === $.XMP || parentTn === $.IFRAME ||
-        parentTn === $.NOEMBED || parentTn === $.NOFRAMES || parentTn === $.PLAINTEXT || parentTn === $.NOSCRIPT) {
-        this.html += content;
-    }
-
-    else
-        this.html += this.options.encodeHtmlEntities ? escapeString(content, false) : content;
-};
-
-Serializer.prototype._serializeCommentNode = function (node) {
-    this.html += '<!--' + this.treeAdapter.getCommentNodeContent(node) + '-->';
-};
-
-Serializer.prototype._serializeDocumentTypeNode = function (node) {
-    var name = this.treeAdapter.getDocumentTypeNodeName(node),
-        publicId = this.treeAdapter.getDocumentTypeNodePublicId(node),
-        systemId = this.treeAdapter.getDocumentTypeNodeSystemId(node);
-
-    this.html += '<!DOCTYPE ' + name;
-
-    if (publicId !== null)
-        this.html += ' PUBLIC ' + enquoteDoctypeId(publicId);
-
-    else if (systemId !== null)
-        this.html += ' SYSTEM';
-
-    if (systemId !== null)
-        this.html += ' ' + enquoteDoctypeId(systemId);
-
-
-    this.html += '>';
-};
+'use strict';
+
+var DefaultTreeAdapter = require('../tree_adapters/default'),
+    Doctype = require('../common/doctype'),
+    Utils = require('../common/utils'),
+    HTML = require('../common/html');
+
+//Aliases
+var $ = HTML.TAG_NAMES,
+    NS = HTML.NAMESPACES;
+
+//Default serializer options
+var DEFAULT_OPTIONS = {
+    encodeHtmlEntities: true
+};
+
+//Escaping regexes
+var AMP_REGEX = /&/g,
+    NBSP_REGEX = /\u00a0/g,
+    DOUBLE_QUOTE_REGEX = /"/g,
+    LT_REGEX = /</g,
+    GT_REGEX = />/g;
+
+//Escape string
+function escapeString(str, attrMode) {
+    str = str
+        .replace(AMP_REGEX, '&amp;')
+        .replace(NBSP_REGEX, '&nbsp;');
+
+    if (attrMode)
+        str = str.replace(DOUBLE_QUOTE_REGEX, '&quot;');
+
+    else {
+        str = str
+            .replace(LT_REGEX, '&lt;')
+            .replace(GT_REGEX, '&gt;');
+    }
+
+    return str;
+}
+
+
+//Enquote doctype ID
+
+
+
+//Serializer
+var Serializer = module.exports = function (treeAdapter, options) {
+    this.treeAdapter = treeAdapter || DefaultTreeAdapter;
+    this.options = Utils.mergeOptions(DEFAULT_OPTIONS, options);
+};
+
+
+//API
+Serializer.prototype.serialize = function (node) {
+    this.html = '';
+    this._serializeChildNodes(node);
+
+    return this.html;
+};
+
+
+//Internals
+Serializer.prototype._serializeChildNodes = function (parentNode) {
+    var childNodes = this.treeAdapter.getChildNodes(parentNode);
+
+    if (childNodes) {
+        for (var i = 0, cnLength = childNodes.length; i < cnLength; i++) {
+            var currentNode = childNodes[i];
+
+            if (this.treeAdapter.isElementNode(currentNode))
+                this._serializeElement(currentNode);
+
+            else if (this.treeAdapter.isTextNode(currentNode))
+                this._serializeTextNode(currentNode);
+
+            else if (this.treeAdapter.isCommentNode(currentNode))
+                this._serializeCommentNode(currentNode);
+
+            else if (this.treeAdapter.isDocumentTypeNode(currentNode))
+                this._serializeDocumentTypeNode(currentNode);
+        }
+    }
+};
+
+Serializer.prototype._serializeElement = function (node) {
+    var tn = this.treeAdapter.getTagName(node),
+        ns = this.treeAdapter.getNamespaceURI(node);
+
+    this.html += '<' + tn;
+    this._serializeAttributes(node);
+    this.html += '>';
+
+    if (tn !== $.AREA && tn !== $.BASE && tn !== $.BASEFONT && tn !== $.BGSOUND && tn !== $.BR && tn !== $.BR &&
+        tn !== $.COL && tn !== $.EMBED && tn !== $.FRAME && tn !== $.HR && tn !== $.IMG && tn !== $.INPUT &&
+        tn !== $.KEYGEN && tn !== $.LINK && tn !== $.MENUITEM && tn !== $.META && tn !== $.PARAM && tn !== $.SOURCE &&
+        tn !== $.TRACK && tn !== $.WBR) {
+
+        if (tn === $.PRE || tn === $.TEXTAREA || tn === $.LISTING) {
+            var firstChild = this.treeAdapter.getFirstChild(node);
+
+            if (firstChild && this.treeAdapter.isTextNode(firstChild)) {
+                var content = this.treeAdapter.getTextNodeContent(firstChild);
+
+                if (content[0] === '\n')
+                    this.html += '\n';
+            }
+        }
+
+        var childNodesHolder = tn === $.TEMPLATE && ns === NS.HTML ?
+                               this.treeAdapter.getChildNodes(node)[0] :
+                               node;
+
+        this._serializeChildNodes(childNodesHolder);
+        this.html += '</' + tn + '>';
+    }
+};
+
+Serializer.prototype._serializeAttributes = function (node) {
+    var attrs = this.treeAdapter.getAttrList(node);
+
+    for (var i = 0, attrsLength = attrs.length; i < attrsLength; i++) {
+        var attr = attrs[i],
+            value = this.options.encodeHtmlEntities ? escapeString(attr.value, true) : attr.value;
+
+        this.html += ' ';
+
+        if (!attr.namespace)
+            this.html += attr.name;
+
+        else if (attr.namespace === NS.XML)
+            this.html += 'xml:' + attr.name;
+
+        else if (attr.namespace === NS.XMLNS) {
+            if (attr.name !== 'xmlns')
+                this.html += 'xmlns:';
+
+            this.html += attr.name;
+        }
+
+        else if (attr.namespace === NS.XLINK)
+            this.html += 'xlink:' + attr.name;
+
+        else
+            this.html += attr.namespace + ':' + attr.name;
+
+        this.html += '="' + value + '"';
+    }
+};
+
+Serializer.prototype._serializeTextNode = function (node) {
+    var content = this.treeAdapter.getTextNodeContent(node),
+        parent = this.treeAdapter.getParentNode(node),
+        parentTn = void 0;
+
+    if (parent && this.treeAdapter.isElementNode(parent))
+        parentTn = this.treeAdapter.getTagName(parent);
+
+    if (parentTn === $.STYLE || parentTn === $.SCRIPT || parentTn === $.XMP || parentTn === $.IFRAME ||
+        parentTn === $.NOEMBED || parentTn === $.NOFRAMES || parentTn === $.PLAINTEXT || parentTn === $.NOSCRIPT) {
+        this.html += content;
+    }
+
+    else
+        this.html += this.options.encodeHtmlEntities ? escapeString(content, false) : content;
+};
+
+Serializer.prototype._serializeCommentNode = function (node) {
+    this.html += '<!--' + this.treeAdapter.getCommentNodeContent(node) + '-->';
+};
+
+Serializer.prototype._serializeDocumentTypeNode = function (node) {
+    var name = this.treeAdapter.getDocumentTypeNodeName(node),
+        publicId = this.treeAdapter.getDocumentTypeNodePublicId(node),
+        systemId = this.treeAdapter.getDocumentTypeNodeSystemId(node);
+
+    this.html += '<' + Doctype.serializeContent(name, publicId, systemId) + '>';
+};

package/lib/simple_api/simple_api_parser.js

@@ -1,107 +1,107 @@
-'use strict';
-
-var Tokenizer = require('../tokenization/tokenizer'),
-    TokenizerProxy = require('./tokenizer_proxy'),
-    Utils = require('../common/utils');
-
-//Default options
-var DEFAULT_OPTIONS = {
-    decodeHtmlEntities: true,
-    locationInfo: false
-};
-
-//Skipping handler
-function skip() {
-    //NOTE: do nothing =)
-}
-
-//SimpleApiParser
-var SimpleApiParser = module.exports = function (handlers, options) {
-    this.options = Utils.mergeOptions(DEFAULT_OPTIONS, options);
-    this.handlers = {
-        doctype: this._wrapHandler(handlers.doctype),
-        startTag: this._wrapHandler(handlers.startTag),
-        endTag: this._wrapHandler(handlers.endTag),
-        text: this._wrapHandler(handlers.text),
-        comment: this._wrapHandler(handlers.comment)
-    };
-};
-
-SimpleApiParser.prototype._wrapHandler = function (handler) {
-    var parser = this;
-
-    handler = handler || skip;
-
-    if (this.options.locationInfo) {
-        return function () {
-            var args = Array.prototype.slice.call(arguments);
-            args.push(parser.currentTokenLocation);
-            handler.apply(handler, args);
-        };
-    }
-
-    return handler;
-};
-
-//API
-SimpleApiParser.prototype.parse = function (html) {
-    var token = null;
-
-    this._reset(html);
-
-    do {
-        token = this.tokenizerProxy.getNextToken();
-
-        if (token.type === Tokenizer.CHARACTER_TOKEN ||
-            token.type === Tokenizer.WHITESPACE_CHARACTER_TOKEN ||
-            token.type === Tokenizer.NULL_CHARACTER_TOKEN) {
-
-            if (this.options.locationInfo) {
-                if (this.pendingText === null)
-                    this.currentTokenLocation = token.location;
-
-                else
-                    this.currentTokenLocation.end = token.location.end;
-            }
-
-            this.pendingText = (this.pendingText || '') + token.chars;
-        }
-
-        else {
-            this._emitPendingText();
-            this._handleToken(token);
-        }
-    } while (token.type !== Tokenizer.EOF_TOKEN);
-};
-
-//Internals
-SimpleApiParser.prototype._handleToken = function (token) {
-    if (this.options.locationInfo)
-        this.currentTokenLocation = token.location;
-
-    if (token.type === Tokenizer.START_TAG_TOKEN)
-        this.handlers.startTag(token.tagName, token.attrs, token.selfClosing);
-
-    else if (token.type === Tokenizer.END_TAG_TOKEN)
-        this.handlers.endTag(token.tagName);
-
-    else if (token.type === Tokenizer.COMMENT_TOKEN)
-        this.handlers.comment(token.data);
-
-    else if (token.type === Tokenizer.DOCTYPE_TOKEN)
-        this.handlers.doctype(token.name, token.publicId, token.systemId);
-
-};
-
-SimpleApiParser.prototype._reset = function (html) {
-    this.tokenizerProxy = new TokenizerProxy(html, this.options);
-    this.pendingText = null;
-    this.currentTokenLocation = null;
-};
-
-SimpleApiParser.prototype._emitPendingText = function () {
-    if (this.pendingText !== null) {
-        this.handlers.text(this.pendingText);
-        this.pendingText = null;
-    }
-};
+'use strict';
+
+var Tokenizer = require('../tokenization/tokenizer'),
+    TokenizerProxy = require('./tokenizer_proxy'),
+    Utils = require('../common/utils');
+
+//Default options
+var DEFAULT_OPTIONS = {
+    decodeHtmlEntities: true,
+    locationInfo: false
+};
+
+//Skipping handler
+function skip() {
+    //NOTE: do nothing =)
+}
+
+//SimpleApiParser
+var SimpleApiParser = module.exports = function (handlers, options) {
+    this.options = Utils.mergeOptions(DEFAULT_OPTIONS, options);
+    this.handlers = {
+        doctype: this._wrapHandler(handlers.doctype),
+        startTag: this._wrapHandler(handlers.startTag),
+        endTag: this._wrapHandler(handlers.endTag),
+        text: this._wrapHandler(handlers.text),
+        comment: this._wrapHandler(handlers.comment)
+    };
+};
+
+SimpleApiParser.prototype._wrapHandler = function (handler) {
+    var parser = this;
+
+    handler = handler || skip;
+
+    if (this.options.locationInfo) {
+        return function () {
+            var args = Array.prototype.slice.call(arguments);
+            args.push(parser.currentTokenLocation);
+            handler.apply(handler, args);
+        };
+    }
+
+    return handler;
+};
+
+//API
+SimpleApiParser.prototype.parse = function (html) {
+    var token = null;
+
+    this._reset(html);
+
+    do {
+        token = this.tokenizerProxy.getNextToken();
+
+        if (token.type === Tokenizer.CHARACTER_TOKEN ||
+            token.type === Tokenizer.WHITESPACE_CHARACTER_TOKEN ||
+            token.type === Tokenizer.NULL_CHARACTER_TOKEN) {
+
+            if (this.options.locationInfo) {
+                if (this.pendingText === null)
+                    this.currentTokenLocation = token.location;
+
+                else
+                    this.currentTokenLocation.end = token.location.end;
+            }
+
+            this.pendingText = (this.pendingText || '') + token.chars;
+        }
+
+        else {
+            this._emitPendingText();
+            this._handleToken(token);
+        }
+    } while (token.type !== Tokenizer.EOF_TOKEN);
+};
+
+//Internals
+SimpleApiParser.prototype._handleToken = function (token) {
+    if (this.options.locationInfo)
+        this.currentTokenLocation = token.location;
+
+    if (token.type === Tokenizer.START_TAG_TOKEN)
+        this.handlers.startTag(token.tagName, token.attrs, token.selfClosing);
+
+    else if (token.type === Tokenizer.END_TAG_TOKEN)
+        this.handlers.endTag(token.tagName);
+
+    else if (token.type === Tokenizer.COMMENT_TOKEN)
+        this.handlers.comment(token.data);
+
+    else if (token.type === Tokenizer.DOCTYPE_TOKEN)
+        this.handlers.doctype(token.name, token.publicId, token.systemId);
+
+};
+
+SimpleApiParser.prototype._reset = function (html) {
+    this.tokenizerProxy = new TokenizerProxy(html, this.options);
+    this.pendingText = null;
+    this.currentTokenLocation = null;
+};
+
+SimpleApiParser.prototype._emitPendingText = function () {
+    if (this.pendingText !== null) {
+        this.handlers.text(this.pendingText);
+        this.pendingText = null;
+    }
+};