parse5 3.0.3 → 5.1.0

package/lib/sax/index.js

@@ -1,118 +0,0 @@
-'use strict';
-
-var TransformStream = require('stream').Transform,
-    DevNullStream = require('./dev_null_stream'),
-    inherits = require('util').inherits,
-    Tokenizer = require('../tokenizer'),
-    LocationInfoTokenizerMixin = require('../extensions/location_info/tokenizer_mixin'),
-    ParserFeedbackSimulator = require('./parser_feedback_simulator'),
-    mergeOptions = require('../utils/merge_options');
-
-var DEFAULT_OPTIONS = {
-    locationInfo: false
-};
-
-var SAXParser = module.exports = function (options) {
-    TransformStream.call(this);
-
-    this.options = mergeOptions(DEFAULT_OPTIONS, options);
-
-    this.tokenizer = new Tokenizer(options);
-
-    if (this.options.locationInfo)
-        new LocationInfoTokenizerMixin(this.tokenizer);
-
-    this.parserFeedbackSimulator = new ParserFeedbackSimulator(this.tokenizer);
-
-    this.pendingText = null;
-    this.currentTokenLocation = void 0;
-
-    this.lastChunkWritten = false;
-    this.stopped = false;
-
-    // NOTE: always pipe stream to the /dev/null stream to avoid
-    // `highWaterMark` hit even if we don't have consumers.
-    // (see: https://github.com/inikulin/parse5/issues/97#issuecomment-171940774)
-    this.pipe(new DevNullStream());
-};
-
-inherits(SAXParser, TransformStream);
-
-//TransformStream implementation
-SAXParser.prototype._transform = function (chunk, encoding, callback) {
-    if (!this.stopped) {
-        this.tokenizer.write(chunk.toString('utf8'), this.lastChunkWritten);
-        this._runParsingLoop();
-    }
-
-    this.push(chunk);
-
-    callback();
-};
-
-SAXParser.prototype._flush = function (callback) {
-    callback();
-};
-
-SAXParser.prototype.end = function (chunk, encoding, callback) {
-    this.lastChunkWritten = true;
-    TransformStream.prototype.end.call(this, chunk, encoding, callback);
-};
-
-SAXParser.prototype.stop = function () {
-    this.stopped = true;
-};
-
-//Internals
-SAXParser.prototype._runParsingLoop = function () {
-    do {
-        var token = this.parserFeedbackSimulator.getNextToken();
-
-        if (token.type === Tokenizer.HIBERNATION_TOKEN)
-            break;
-
-        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.endOffset = token.location.endOffset;
-            }
-
-            this.pendingText = (this.pendingText || '') + token.chars;
-        }
-
-        else {
-            this._emitPendingText();
-            this._handleToken(token);
-        }
-    } while (!this.stopped && token.type !== Tokenizer.EOF_TOKEN);
-};
-
-SAXParser.prototype._handleToken = function (token) {
-    if (this.options.locationInfo)
-        this.currentTokenLocation = token.location;
-
-    if (token.type === Tokenizer.START_TAG_TOKEN)
-        this.emit('startTag', token.tagName, token.attrs, token.selfClosing, this.currentTokenLocation);
-
-    else if (token.type === Tokenizer.END_TAG_TOKEN)
-        this.emit('endTag', token.tagName, this.currentTokenLocation);
-
-    else if (token.type === Tokenizer.COMMENT_TOKEN)
-        this.emit('comment', token.data, this.currentTokenLocation);
-
-    else if (token.type === Tokenizer.DOCTYPE_TOKEN)
-        this.emit('doctype', token.name, token.publicId, token.systemId, this.currentTokenLocation);
-};
-
-SAXParser.prototype._emitPendingText = function () {
-    if (this.pendingText !== null) {
-        this.emit('text', this.pendingText, this.currentTokenLocation);
-        this.pendingText = null;
-    }
-};

package/lib/sax/parser_feedback_simulator.js

@@ -1,153 +0,0 @@
-'use strict';
-
-var Tokenizer = require('../tokenizer'),
-    foreignContent = require('../common/foreign_content'),
-    UNICODE = require('../common/unicode'),
-    HTML = require('../common/html');
-
-
-//Aliases
-var $ = HTML.TAG_NAMES,
-    NS = HTML.NAMESPACES;
-
-
-//ParserFeedbackSimulator
-//Simulates adjustment of the Tokenizer which performed by standard parser during tree construction.
-var ParserFeedbackSimulator = module.exports = function (tokenizer) {
-    this.tokenizer = tokenizer;
-
-    this.namespaceStack = [];
-    this.namespaceStackTop = -1;
-    this._enterNamespace(NS.HTML);
-};
-
-ParserFeedbackSimulator.prototype.getNextToken = function () {
-    var token = this.tokenizer.getNextToken();
-
-    if (token.type === Tokenizer.START_TAG_TOKEN)
-        this._handleStartTagToken(token);
-
-    else if (token.type === Tokenizer.END_TAG_TOKEN)
-        this._handleEndTagToken(token);
-
-    else if (token.type === Tokenizer.NULL_CHARACTER_TOKEN && this.inForeignContent) {
-        token.type = Tokenizer.CHARACTER_TOKEN;
-        token.chars = UNICODE.REPLACEMENT_CHARACTER;
-    }
-
-    else if (this.skipNextNewLine) {
-        if (token.type !== Tokenizer.HIBERNATION_TOKEN)
-            this.skipNextNewLine = false;
-
-        if (token.type === Tokenizer.WHITESPACE_CHARACTER_TOKEN && token.chars[0] === '\n') {
-            if (token.chars.length === 1)
-                return this.getNextToken();
-
-            token.chars = token.chars.substr(1);
-        }
-    }
-
-    return token;
-};
-
-//Namespace stack mutations
-ParserFeedbackSimulator.prototype._enterNamespace = function (namespace) {
-    this.namespaceStackTop++;
-    this.namespaceStack.push(namespace);
-
-    this.inForeignContent = namespace !== NS.HTML;
-    this.currentNamespace = namespace;
-    this.tokenizer.allowCDATA = this.inForeignContent;
-};
-
-ParserFeedbackSimulator.prototype._leaveCurrentNamespace = function () {
-    this.namespaceStackTop--;
-    this.namespaceStack.pop();
-
-    this.currentNamespace = this.namespaceStack[this.namespaceStackTop];
-    this.inForeignContent = this.currentNamespace !== NS.HTML;
-    this.tokenizer.allowCDATA = this.inForeignContent;
-};
-
-//Token handlers
-ParserFeedbackSimulator.prototype._ensureTokenizerMode = function (tn) {
-    if (tn === $.TEXTAREA || tn === $.TITLE)
-        this.tokenizer.state = Tokenizer.MODE.RCDATA;
-
-    else if (tn === $.PLAINTEXT)
-        this.tokenizer.state = Tokenizer.MODE.PLAINTEXT;
-
-    else if (tn === $.SCRIPT)
-        this.tokenizer.state = Tokenizer.MODE.SCRIPT_DATA;
-
-    else if (tn === $.STYLE || tn === $.IFRAME || tn === $.XMP ||
-             tn === $.NOEMBED || tn === $.NOFRAMES || tn === $.NOSCRIPT)
-        this.tokenizer.state = Tokenizer.MODE.RAWTEXT;
-};
-
-ParserFeedbackSimulator.prototype._handleStartTagToken = function (token) {
-    var tn = token.tagName;
-
-    if (tn === $.SVG)
-        this._enterNamespace(NS.SVG);
-
-    else if (tn === $.MATH)
-        this._enterNamespace(NS.MATHML);
-
-    if (this.inForeignContent) {
-        if (foreignContent.causesExit(token)) {
-            this._leaveCurrentNamespace();
-            return;
-        }
-
-        var currentNs = this.currentNamespace;
-
-        if (currentNs === NS.MATHML)
-            foreignContent.adjustTokenMathMLAttrs(token);
-
-        else if (currentNs === NS.SVG) {
-            foreignContent.adjustTokenSVGTagName(token);
-            foreignContent.adjustTokenSVGAttrs(token);
-        }
-
-        foreignContent.adjustTokenXMLAttrs(token);
-
-        tn = token.tagName;
-
-        if (!token.selfClosing && foreignContent.isIntegrationPoint(tn, currentNs, token.attrs))
-            this._enterNamespace(NS.HTML);
-    }
-
-    else {
-        if (tn === $.PRE || tn === $.TEXTAREA || tn === $.LISTING)
-            this.skipNextNewLine = true;
-
-        else if (tn === $.IMAGE)
-            token.tagName = $.IMG;
-
-        this._ensureTokenizerMode(tn);
-    }
-};
-
-ParserFeedbackSimulator.prototype._handleEndTagToken = function (token) {
-    var tn = token.tagName;
-
-    if (!this.inForeignContent) {
-        var previousNs = this.namespaceStack[this.namespaceStackTop - 1];
-
-        if (previousNs === NS.SVG && foreignContent.SVG_TAG_NAMES_ADJUSTMENT_MAP[tn])
-            tn = foreignContent.SVG_TAG_NAMES_ADJUSTMENT_MAP[tn];
-
-        //NOTE: check for exit from integration point
-        if (foreignContent.isIntegrationPoint(tn, previousNs, token.attrs))
-            this._leaveCurrentNamespace();
-    }
-
-    else if (tn === $.SVG && this.currentNamespace === NS.SVG ||
-             tn === $.MATH && this.currentNamespace === NS.MATHML)
-        this._leaveCurrentNamespace();
-
-    // NOTE: adjust end tag name as well for consistency
-    if (this.currentNamespace === NS.SVG)
-        foreignContent.adjustTokenSVGTagName(token);
-};

package/lib/serializer/serializer_stream.js

@@ -1,28 +0,0 @@
-'use strict';
-
-var ReadableStream = require('stream').Readable,
-    inherits = require('util').inherits,
-    Serializer = require('./index');
-
-var SerializerStream = module.exports = function (node, options) {
-    ReadableStream.call(this);
-
-    this.serializer = new Serializer(node, options);
-
-    Object.defineProperty(this.serializer, 'html', {
-        //NOTE: To make `+=` concat operator work properly we define
-        //getter which always returns empty string
-        get: function () {
-            return '';
-        },
-        set: this.push.bind(this)
-    });
-};
-
-inherits(SerializerStream, ReadableStream);
-
-//Readable stream implementation
-SerializerStream.prototype._read = function () {
-    this.serializer.serialize();
-    this.push(null);
-};

package/lib/tree_adapters/default.js

@@ -1,209 +0,0 @@
-'use strict';
-
-var DOCUMENT_MODE = require('../common/html').DOCUMENT_MODE;
-
-//Node construction
-exports.createDocument = function () {
-    return {
-        nodeName: '#document',
-        mode: DOCUMENT_MODE.NO_QUIRKS,
-        childNodes: []
-    };
-};
-
-exports.createDocumentFragment = function () {
-    return {
-        nodeName: '#document-fragment',
-        childNodes: []
-    };
-};
-
-exports.createElement = function (tagName, namespaceURI, attrs) {
-    return {
-        nodeName: tagName,
-        tagName: tagName,
-        attrs: attrs,
-        namespaceURI: namespaceURI,
-        childNodes: [],
-        parentNode: null
-    };
-};
-
-exports.createCommentNode = function (data) {
-    return {
-        nodeName: '#comment',
-        data: data,
-        parentNode: null
-    };
-};
-
-var createTextNode = function (value) {
-    return {
-        nodeName: '#text',
-        value: value,
-        parentNode: null
-    };
-};
-
-
-//Tree mutation
-var appendChild = exports.appendChild = function (parentNode, newNode) {
-    parentNode.childNodes.push(newNode);
-    newNode.parentNode = parentNode;
-};
-
-var insertBefore = exports.insertBefore = function (parentNode, newNode, referenceNode) {
-    var insertionIdx = parentNode.childNodes.indexOf(referenceNode);
-
-    parentNode.childNodes.splice(insertionIdx, 0, newNode);
-    newNode.parentNode = parentNode;
-};
-
-exports.setTemplateContent = function (templateElement, contentElement) {
-    templateElement.content = contentElement;
-};
-
-exports.getTemplateContent = function (templateElement) {
-    return templateElement.content;
-};
-
-exports.setDocumentType = function (document, name, publicId, systemId) {
-    var doctypeNode = null;
-
-    for (var i = 0; i < document.childNodes.length; i++) {
-        if (document.childNodes[i].nodeName === '#documentType') {
-            doctypeNode = document.childNodes[i];
-            break;
-        }
-    }
-
-    if (doctypeNode) {
-        doctypeNode.name = name;
-        doctypeNode.publicId = publicId;
-        doctypeNode.systemId = systemId;
-    }
-
-    else {
-        appendChild(document, {
-            nodeName: '#documentType',
-            name: name,
-            publicId: publicId,
-            systemId: systemId
-        });
-    }
-};
-
-exports.setDocumentMode = function (document, mode) {
-    document.mode = mode;
-};
-
-exports.getDocumentMode = function (document) {
-    return document.mode;
-};
-
-exports.detachNode = function (node) {
-    if (node.parentNode) {
-        var idx = node.parentNode.childNodes.indexOf(node);
-
-        node.parentNode.childNodes.splice(idx, 1);
-        node.parentNode = null;
-    }
-};
-
-exports.insertText = function (parentNode, text) {
-    if (parentNode.childNodes.length) {
-        var prevNode = parentNode.childNodes[parentNode.childNodes.length - 1];
-
-        if (prevNode.nodeName === '#text') {
-            prevNode.value += text;
-            return;
-        }
-    }
-
-    appendChild(parentNode, createTextNode(text));
-};
-
-exports.insertTextBefore = function (parentNode, text, referenceNode) {
-    var prevNode = parentNode.childNodes[parentNode.childNodes.indexOf(referenceNode) - 1];
-
-    if (prevNode && prevNode.nodeName === '#text')
-        prevNode.value += text;
-    else
-        insertBefore(parentNode, createTextNode(text), referenceNode);
-};
-
-exports.adoptAttributes = function (recipient, attrs) {
-    var recipientAttrsMap = [];
-
-    for (var i = 0; i < recipient.attrs.length; i++)
-        recipientAttrsMap.push(recipient.attrs[i].name);
-
-    for (var j = 0; j < attrs.length; j++) {
-        if (recipientAttrsMap.indexOf(attrs[j].name) === -1)
-            recipient.attrs.push(attrs[j]);
-    }
-};
-
-
-//Tree traversing
-exports.getFirstChild = function (node) {
-    return node.childNodes[0];
-};
-
-exports.getChildNodes = function (node) {
-    return node.childNodes;
-};
-
-exports.getParentNode = function (node) {
-    return node.parentNode;
-};
-
-exports.getAttrList = function (element) {
-    return element.attrs;
-};
-
-//Node data
-exports.getTagName = function (element) {
-    return element.tagName;
-};
-
-exports.getNamespaceURI = function (element) {
-    return element.namespaceURI;
-};
-
-exports.getTextNodeContent = function (textNode) {
-    return textNode.value;
-};
-
-exports.getCommentNodeContent = function (commentNode) {
-    return commentNode.data;
-};
-
-exports.getDocumentTypeNodeName = function (doctypeNode) {
-    return doctypeNode.name;
-};
-
-exports.getDocumentTypeNodePublicId = function (doctypeNode) {
-    return doctypeNode.publicId;
-};
-
-exports.getDocumentTypeNodeSystemId = function (doctypeNode) {
-    return doctypeNode.systemId;
-};
-
-//Node types
-exports.isTextNode = function (node) {
-    return node.nodeName === '#text';
-};
-
-exports.isCommentNode = function (node) {
-    return node.nodeName === '#comment';
-};
-
-exports.isDocumentTypeNode = function (node) {
-    return node.nodeName === '#documentType';
-};
-
-exports.isElementNode = function (node) {
-    return !!node.tagName;
-};