parse5 2.1.5 → 2.2.3

package/lib/sax/dev_null_stream.js

@@ -1,14 +1,14 @@
-'use strict';
-
-var WritableStream = require('stream').Writable,
-    util = require('util');
-
-var DevNullStream = module.exports = function () {
-    WritableStream.call(this);
-};
-
-util.inherits(DevNullStream, WritableStream);
-
-DevNullStream.prototype._write = function (chunk, encoding, cb) {
-    cb();
-};
+'use strict';
+
+var WritableStream = require('stream').Writable,
+    util = require('util');
+
+var DevNullStream = module.exports = function () {
+    WritableStream.call(this);
+};
+
+util.inherits(DevNullStream, WritableStream);
+
+DevNullStream.prototype._write = function (chunk, encoding, cb) {
+    cb();
+};

package/lib/sax/index.js

@@ -124,13 +124,11 @@ SAXParser.prototype.stop = function () {
 //Internals
 SAXParser.prototype._runParsingLoop = function () {
     do {
-        var token = this.tokenizer.getNextToken();
+        var token = this.parserFeedbackSimulator.getNextToken();
 
         if (token.type === Tokenizer.HIBERNATION_TOKEN)
             break;
 
-        this.parserFeedbackSimulator.adjustTokenizer(token);
-
         if (token.type === Tokenizer.CHARACTER_TOKEN ||
             token.type === Tokenizer.WHITESPACE_CHARACTER_TOKEN ||
             token.type === Tokenizer.NULL_CHARACTER_TOKEN) {
@@ -140,7 +138,7 @@ SAXParser.prototype._runParsingLoop = function () {
                     this.currentTokenLocation = token.location;
 
                 else
-                    this.currentTokenLocation.end = token.location.end;
+                    this.currentTokenLocation.endOffset = token.location.endOffset;
             }
 
             this.pendingText = (this.pendingText || '') + token.chars;
@@ -165,7 +163,7 @@ SAXParser.prototype._handleToken = function (token) {
          * @instance
          * @type {Function}
          * @param {String} name - Tag name.
-         * @param {String} attributes - List of attributes in the `{ key: String, value: String }` form.
+         * @param {Array} attrs - List of attributes in the `{ name: String, value: String, prefix?: String }` form.
          * @param {Boolean} selfClosing - Indicates if the tag is self-closing.
          * @param {StartTagLocationInfo} [location] - Start tag source code location info.
          * Available if location info is enabled in {@link SAXParserOptions}.

package/lib/sax/parser_feedback_simulator.js

@@ -1,116 +1,153 @@
-'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.currentNamespace = null;
-    this.inForeignContent = false;
-};
-
-//API
-ParserFeedbackSimulator.prototype.adjustTokenizer = function (token) {
-    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;
-    }
-
-    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);
-
-    else if (this.inForeignContent) {
-        if (foreignContent.causesExit(token))
-            this._leaveCurrentNamespace();
-
-        else if (foreignContent.isMathMLTextIntegrationPoint(tn, this.currentNamespace) ||
-                 foreignContent.isHtmlIntegrationPoint(tn, this.currentNamespace, token.attrs))
-            this._enterNamespace(NS.HTML);
-    }
-
-    else
-        this._ensureTokenizerMode(tn);
-};
-
-ParserFeedbackSimulator.prototype._handleEndTagToken = function (token) {
-    var tn = token.tagName;
-
-    if (!this.inForeignContent) {
-        var previousNs = this.namespaceStack[this.namespaceStackTop - 1];
-
-        //NOTE: check for exit from integration point
-        if (foreignContent.isMathMLTextIntegrationPoint(tn, previousNs) ||
-            foreignContent.isHtmlIntegrationPoint(tn, previousNs, token.attrs))
-            this._leaveCurrentNamespace();
-
-        else if (tn === $.SCRIPT)
-            this.tokenizer.state = Tokenizer.MODE.DATA;
-    }
-
-    else if (tn === $.SVG && this.currentNamespace === NS.SVG ||
-             tn === $.MATH && this.currentNamespace === NS.MATHML)
-        this._leaveCurrentNamespace();
-};
+'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);
+};