parse5 2.2.2 → 3.0.2

package/lib/index.js

@@ -1,108 +1,44 @@
-'use strict';
-
-var Parser = require('./parser'),
-    Serializer = require('./serializer');
-
-/** @namespace parse5 */
-
-/**
- * Parses an HTML string.
- * @function parse
- * @memberof parse5
- * @instance
- * @param {string} html - Input HTML string.
- * @param {ParserOptions} [options] - Parsing options.
- * @returns {ASTNode<Document>} document
- * @example
- * var parse5 = require('parse5');
- *
- * var document = parse5.parse('<!DOCTYPE html><html><head></head><body>Hi there!</body></html>');
- */
-exports.parse = function parse(html, options) {
-    var parser = new Parser(options);
-
-    return parser.parse(html);
-};
-
-/**
- * Parses an HTML fragment.
- * @function parseFragment
- * @memberof parse5
- * @instance
- * @param {ASTNode} [fragmentContext] - Parsing context element. If specified, given fragment
- * will be parsed as if it was set to the context element's `innerHTML` property.
- * @param {string} html - Input HTML fragment string.
- * @param {ParserOptions} [options] - Parsing options.
- * @returns {ASTNode<DocumentFragment>} documentFragment
- * @example
- * var parse5 = require('parse5');
- *
- * var documentFragment = parse5.parseFragment('<table></table>');
- *
- * // Parses the html fragment in the context of the parsed <table> element.
- * var trFragment = parser.parseFragment(documentFragment.childNodes[0], '<tr><td>Shake it, baby</td></tr>');
- */
-exports.parseFragment = function parseFragment(fragmentContext, html, options) {
-    if (typeof fragmentContext === 'string') {
-        options = html;
-        html = fragmentContext;
-        fragmentContext = null;
-    }
-
-    var parser = new Parser(options);
-
-    return parser.parseFragment(html, fragmentContext);
-};
-
-/**
- * Serializes an AST node to an HTML string.
- * @function serialize
- * @memberof parse5
- * @instance
- * @param {ASTNode} node - Node to serialize.
- * @param {SerializerOptions} [options] - Serialization options.
- * @returns {String} html
- * @example
- * var parse5 = require('parse5');
- *
- * var document = parse5.parse('<!DOCTYPE html><html><head></head><body>Hi there!</body></html>');
- *
- * // Serializes a document.
- * var html = parse5.serialize(document);
- *
- * // Serializes the <body> element content.
- * var bodyInnerHtml = parse5.serialize(document.childNodes[0].childNodes[1]);
- */
-exports.serialize = function (node, options) {
-    var serializer = new Serializer(node, options);
-
-    return serializer.serialize();
-};
-
-/**
- * Provides built-in tree adapters that can be used for parsing and serialization.
- * @var treeAdapters
- * @memberof parse5
- * @instance
- * @property {TreeAdapter} default - Default tree format for parse5.
- * @property {TreeAdapter} htmlparser2 - Quite popular [htmlparser2](https://github.com/fb55/htmlparser2) tree format
- * (e.g. used by [cheerio](https://github.com/MatthewMueller/cheerio) and [jsdom](https://github.com/tmpvar/jsdom)).
- * @example
- * var parse5 = require('parse5');
- *
- * // Uses the default tree adapter for parsing.
- * var document = parse5.parse('<div></div>', { treeAdapter: parse5.treeAdapters.default });
- *
- * // Uses the htmlparser2 tree adapter with the SerializerStream.
- * var serializer = new parse5.SerializerStream(node, { treeAdapter: parse5.treeAdapters.htmlparser2 });
- */
-exports.treeAdapters = {
-    default: require('./tree_adapters/default'),
-    htmlparser2: require('./tree_adapters/htmlparser2')
-};
-
-
-// Streaming
-exports.ParserStream = require('./parser/stream');
-exports.SerializerStream = require('./serializer/stream');
-exports.SAXParser = require('./sax');
+'use strict';
+
+var Parser = require('./parser'),
+    Serializer = require('./serializer');
+
+
+// Shorthands
+exports.parse = function parse(html, options) {
+    var parser = new Parser(options);
+
+    return parser.parse(html);
+};
+
+exports.parseFragment = function parseFragment(fragmentContext, html, options) {
+    if (typeof fragmentContext === 'string') {
+        options = html;
+        html = fragmentContext;
+        fragmentContext = null;
+    }
+
+    var parser = new Parser(options);
+
+    return parser.parseFragment(html, fragmentContext);
+};
+
+exports.serialize = function (node, options) {
+    var serializer = new Serializer(node, options);
+
+    return serializer.serialize();
+};
+
+
+// Tree adapters
+exports.treeAdapters = {
+    default: require('./tree_adapters/default'),
+    htmlparser2: require('./tree_adapters/htmlparser2')
+};
+
+
+// Streaming
+exports.ParserStream = require('./parser/parser_stream');
+exports.PlainTextConversionStream = require('./parser/plain_text_conversion_stream');
+exports.SerializerStream = require('./serializer/serializer_stream');
+exports.SAXParser = require('./sax');

package/lib/location_info/parser_mixin.js

@@ -8,53 +8,6 @@ var OpenElementStack = require('../parser/open_element_stack'),
 //Aliases
 var $ = HTML.TAG_NAMES;
 
-
-function setEndLocation(element, closingToken, treeAdapter) {
-    var loc = element.__location;
-
-    if (!loc)
-        return;
-
-    /**
-     * @typedef {Object} ElementLocationInfo
-     * @extends StartTagLocationInfo
-     *
-     * @property {StartTagLocationInfo} startTag - Element's start tag location info.
-     * @property {LocationInfo} endTag - Element's end tag location info.
-     */
-    if (!loc.startTag) {
-        loc.startTag = {
-            line: loc.line,
-            col: loc.col,
-            startOffset: loc.startOffset,
-            endOffset: loc.endOffset
-        };
-        if (loc.attrs)
-            loc.startTag.attrs = loc.attrs;
-    }
-
-    if (closingToken.location) {
-        var ctLocation = closingToken.location,
-            tn = treeAdapter.getTagName(element),
-        // NOTE: For cases like <p> <p> </p> - First 'p' closes without a closing tag and
-        // for cases like <td> <p> </td> - 'p' closes without a closing tag
-            isClosingEndTag = closingToken.type === Tokenizer.END_TAG_TOKEN &&
-                              tn === closingToken.tagName;
-
-        if (isClosingEndTag) {
-            loc.endTag = {
-                line: ctLocation.line,
-                col: ctLocation.col,
-                startOffset: ctLocation.startOffset,
-                endOffset: ctLocation.endOffset
-            };
-        }
-
-        loc.endOffset = ctLocation.startOffset;
-    }
-}
-
-
 exports.assign = function (parser) {
     //NOTE: obtain Parser proto this way to avoid module circular references
     var parserProto = Object.getPrototypeOf(parser),
@@ -63,6 +16,50 @@ exports.assign = function (parser) {
         lastFosterParentingLocation = null,
         currentToken = null;
 
+    function setEndLocation(element, closingToken) {
+        var loc = element.__location;
+
+        if (!loc)
+            return;
+
+        if (!loc.startTag) {
+            loc.startTag = {
+                line: loc.line,
+                col: loc.col,
+                startOffset: loc.startOffset,
+                endOffset: loc.endOffset
+            };
+
+            if (loc.attrs)
+                loc.startTag.attrs = loc.attrs;
+        }
+
+        if (closingToken.location) {
+            var ctLocation = closingToken.location,
+                tn = treeAdapter.getTagName(element),
+            // NOTE: For cases like <p> <p> </p> - First 'p' closes without a closing tag and
+            // for cases like <td> <p> </td> - 'p' closes without a closing tag
+                isClosingEndTag = closingToken.type === Tokenizer.END_TAG_TOKEN &&
+                                  tn === closingToken.tagName;
+
+            if (isClosingEndTag) {
+                loc.endTag = {
+                    line: ctLocation.line,
+                    col: ctLocation.col,
+                    startOffset: ctLocation.startOffset,
+                    endOffset: ctLocation.endOffset
+                };
+            }
+
+            if (isClosingEndTag)
+                loc.endOffset = ctLocation.endOffset;
+            else
+                loc.endOffset = ctLocation.startOffset;
+        }
+
+        else if (closingToken.type === Tokenizer.EOF_TOKEN)
+            loc.endOffset = parser.tokenizer.preprocessor.sourcePos;
+    }
 
     //NOTE: patch _bootstrap method
     parser._bootstrap = function (document, fragmentContext) {
@@ -74,23 +71,32 @@ exports.assign = function (parser) {
 
         //OpenElementStack
         parser.openElements.pop = function () {
-            setEndLocation(this.current, currentToken, treeAdapter);
+            setEndLocation(this.current, currentToken);
             OpenElementStack.prototype.pop.call(this);
         };
 
         parser.openElements.popAllUpToHtmlElement = function () {
             for (var i = this.stackTop; i > 0; i--)
-                setEndLocation(this.items[i], currentToken, treeAdapter);
+                setEndLocation(this.items[i], currentToken);
 
             OpenElementStack.prototype.popAllUpToHtmlElement.call(this);
         };
 
         parser.openElements.remove = function (element) {
-            setEndLocation(element, currentToken, treeAdapter);
+            setEndLocation(element, currentToken);
             OpenElementStack.prototype.remove.call(this, element);
         };
     };
 
+    parser._runParsingLoop = function (scriptHandler) {
+        parserProto._runParsingLoop.call(this, scriptHandler);
+
+        // NOTE: generate location info for elements
+        // that remains on open element stack
+        for (var i = parser.openElements.stackTop; i >= 0; i--)
+            setEndLocation(parser.openElements.items[i], currentToken);
+    };
+
 
     //Token processing
     parser._processTokenInForeignContent = function (token) {
@@ -111,7 +117,7 @@ exports.assign = function (parser) {
                 var element = this.openElements.items[i];
 
                 if (this.treeAdapter.getTagName(element) === token.tagName) {
-                    setEndLocation(element, token, treeAdapter);
+                    setEndLocation(element, token);
                     break;
                 }
             }

package/lib/location_info/tokenizer_mixin.js

@@ -13,20 +13,11 @@ exports.assign = function (tokenizer) {
         tokenCol = -1,
         tokenLine = 1,
         isEol = false,
-        lineStartPosStack = [0],
         lineStartPos = 0,
         col = -1,
         line = 1;
 
     function attachLocationInfo(token) {
-        /**
-         * @typedef {Object} LocationInfo
-         *
-         * @property {Number} line - One-based line index
-         * @property {Number} col - One-based column index
-         * @property {Number} startOffset - Zero-based first character index
-         * @property {Number} endOffset - Zero-based last character index
-         */
         token.location = {
             line: tokenLine,
             col: tokenCol,
@@ -43,7 +34,6 @@ exports.assign = function (tokenizer) {
         if (isEol) {
             isEol = false;
             line++;
-            lineStartPosStack.push(this.preprocessor.sourcePos);
             lineStartPos = this.preprocessor.sourcePos;
         }
 
@@ -59,11 +49,6 @@ exports.assign = function (tokenizer) {
         tokenizerProto._unconsume.call(this);
         isEol = false;
 
-        while (lineStartPos > this.preprocessor.sourcePos && lineStartPosStack.length > 1) {
-            lineStartPos = lineStartPosStack.pop();
-            line--;
-        }
-
         col = this.preprocessor.sourcePos - lineStartPos + 1;
     };
 
@@ -117,14 +102,8 @@ exports.assign = function (tokenizer) {
         this.currentAttrLocation.endOffset = this.preprocessor.sourcePos;
 
         if (!this.currentToken.location.attrs)
-            this.currentToken.location.attrs = {};
-
-        /**
-         * @typedef {Object} StartTagLocationInfo
-         * @extends LocationInfo
-         *
-         * @property {Dictionary<String, LocationInfo>} attrs - Start tag attributes' location info.
-         */
+            this.currentToken.location.attrs = Object.create(null);
+
         this.currentToken.location.attrs[this.currentAttr.name] = this.currentAttrLocation;
     };