parse5 2.2.3 → 3.0.3

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/parser/formatting_element_list.js

@@ -1,167 +1,167 @@
-'use strict';
-
-//Const
-var NOAH_ARK_CAPACITY = 3;
-
-//List of formatting elements
-var FormattingElementList = module.exports = function (treeAdapter) {
-    this.length = 0;
-    this.entries = [];
-    this.treeAdapter = treeAdapter;
-    this.bookmark = null;
-};
-
-//Entry types
-FormattingElementList.MARKER_ENTRY = 'MARKER_ENTRY';
-FormattingElementList.ELEMENT_ENTRY = 'ELEMENT_ENTRY';
-
-//Noah Ark's condition
-//OPTIMIZATION: at first we try to find possible candidates for exclusion using
-//lightweight heuristics without thorough attributes check.
-FormattingElementList.prototype._getNoahArkConditionCandidates = function (newElement) {
-    var candidates = [];
-
-    if (this.length >= NOAH_ARK_CAPACITY) {
-        var neAttrsLength = this.treeAdapter.getAttrList(newElement).length,
-            neTagName = this.treeAdapter.getTagName(newElement),
-            neNamespaceURI = this.treeAdapter.getNamespaceURI(newElement);
-
-        for (var i = this.length - 1; i >= 0; i--) {
-            var entry = this.entries[i];
-
-            if (entry.type === FormattingElementList.MARKER_ENTRY)
-                break;
-
-            var element = entry.element,
-                elementAttrs = this.treeAdapter.getAttrList(element),
-                isCandidate = this.treeAdapter.getTagName(element) === neTagName &&
-                              this.treeAdapter.getNamespaceURI(element) === neNamespaceURI &&
-                              elementAttrs.length === neAttrsLength;
-
-            if (isCandidate)
-                candidates.push({idx: i, attrs: elementAttrs});
-        }
-    }
-
-    return candidates.length < NOAH_ARK_CAPACITY ? [] : candidates;
-};
-
-FormattingElementList.prototype._ensureNoahArkCondition = function (newElement) {
-    var candidates = this._getNoahArkConditionCandidates(newElement),
-        cLength = candidates.length;
-
-    if (cLength) {
-        var neAttrs = this.treeAdapter.getAttrList(newElement),
-            neAttrsLength = neAttrs.length,
-            neAttrsMap = {};
-
-        //NOTE: build attrs map for the new element so we can perform fast lookups
-        for (var i = 0; i < neAttrsLength; i++) {
-            var neAttr = neAttrs[i];
-
-            neAttrsMap[neAttr.name] = neAttr.value;
-        }
-
-        for (i = 0; i < neAttrsLength; i++) {
-            for (var j = 0; j < cLength; j++) {
-                var cAttr = candidates[j].attrs[i];
-
-                if (neAttrsMap[cAttr.name] !== cAttr.value) {
-                    candidates.splice(j, 1);
-                    cLength--;
-                }
-
-                if (candidates.length < NOAH_ARK_CAPACITY)
-                    return;
-            }
-        }
-
-        //NOTE: remove bottommost candidates until Noah's Ark condition will not be met
-        for (i = cLength - 1; i >= NOAH_ARK_CAPACITY - 1; i--) {
-            this.entries.splice(candidates[i].idx, 1);
-            this.length--;
-        }
-    }
-};
-
-//Mutations
-FormattingElementList.prototype.insertMarker = function () {
-    this.entries.push({type: FormattingElementList.MARKER_ENTRY});
-    this.length++;
-};
-
-FormattingElementList.prototype.pushElement = function (element, token) {
-    this._ensureNoahArkCondition(element);
-
-    this.entries.push({
-        type: FormattingElementList.ELEMENT_ENTRY,
-        element: element,
-        token: token
-    });
-
-    this.length++;
-};
-
-FormattingElementList.prototype.insertElementAfterBookmark = function (element, token) {
-    var bookmarkIdx = this.length - 1;
-
-    for (; bookmarkIdx >= 0; bookmarkIdx--) {
-        if (this.entries[bookmarkIdx] === this.bookmark)
-            break;
-    }
-
-    this.entries.splice(bookmarkIdx + 1, 0, {
-        type: FormattingElementList.ELEMENT_ENTRY,
-        element: element,
-        token: token
-    });
-
-    this.length++;
-};
-
-FormattingElementList.prototype.removeEntry = function (entry) {
-    for (var i = this.length - 1; i >= 0; i--) {
-        if (this.entries[i] === entry) {
-            this.entries.splice(i, 1);
-            this.length--;
-            break;
-        }
-    }
-};
-
-FormattingElementList.prototype.clearToLastMarker = function () {
-    while (this.length) {
-        var entry = this.entries.pop();
-
-        this.length--;
-
-        if (entry.type === FormattingElementList.MARKER_ENTRY)
-            break;
-    }
-};
-
-//Search
-FormattingElementList.prototype.getElementEntryInScopeWithTagName = function (tagName) {
-    for (var i = this.length - 1; i >= 0; i--) {
-        var entry = this.entries[i];
-
-        if (entry.type === FormattingElementList.MARKER_ENTRY)
-            return null;
-
-        if (this.treeAdapter.getTagName(entry.element) === tagName)
-            return entry;
-    }
-
-    return null;
-};
-
-FormattingElementList.prototype.getElementEntry = function (element) {
-    for (var i = this.length - 1; i >= 0; i--) {
-        var entry = this.entries[i];
-
-        if (entry.type === FormattingElementList.ELEMENT_ENTRY && entry.element === element)
-            return entry;
-    }
-
-    return null;
-};
+'use strict';
+
+//Const
+var NOAH_ARK_CAPACITY = 3;
+
+//List of formatting elements
+var FormattingElementList = module.exports = function (treeAdapter) {
+    this.length = 0;
+    this.entries = [];
+    this.treeAdapter = treeAdapter;
+    this.bookmark = null;
+};
+
+//Entry types
+FormattingElementList.MARKER_ENTRY = 'MARKER_ENTRY';
+FormattingElementList.ELEMENT_ENTRY = 'ELEMENT_ENTRY';
+
+//Noah Ark's condition
+//OPTIMIZATION: at first we try to find possible candidates for exclusion using
+//lightweight heuristics without thorough attributes check.
+FormattingElementList.prototype._getNoahArkConditionCandidates = function (newElement) {
+    var candidates = [];
+
+    if (this.length >= NOAH_ARK_CAPACITY) {
+        var neAttrsLength = this.treeAdapter.getAttrList(newElement).length,
+            neTagName = this.treeAdapter.getTagName(newElement),
+            neNamespaceURI = this.treeAdapter.getNamespaceURI(newElement);
+
+        for (var i = this.length - 1; i >= 0; i--) {
+            var entry = this.entries[i];
+
+            if (entry.type === FormattingElementList.MARKER_ENTRY)
+                break;
+
+            var element = entry.element,
+                elementAttrs = this.treeAdapter.getAttrList(element),
+                isCandidate = this.treeAdapter.getTagName(element) === neTagName &&
+                              this.treeAdapter.getNamespaceURI(element) === neNamespaceURI &&
+                              elementAttrs.length === neAttrsLength;
+
+            if (isCandidate)
+                candidates.push({idx: i, attrs: elementAttrs});
+        }
+    }
+
+    return candidates.length < NOAH_ARK_CAPACITY ? [] : candidates;
+};
+
+FormattingElementList.prototype._ensureNoahArkCondition = function (newElement) {
+    var candidates = this._getNoahArkConditionCandidates(newElement),
+        cLength = candidates.length;
+
+    if (cLength) {
+        var neAttrs = this.treeAdapter.getAttrList(newElement),
+            neAttrsLength = neAttrs.length,
+            neAttrsMap = Object.create(null);
+
+        //NOTE: build attrs map for the new element so we can perform fast lookups
+        for (var i = 0; i < neAttrsLength; i++) {
+            var neAttr = neAttrs[i];
+
+            neAttrsMap[neAttr.name] = neAttr.value;
+        }
+
+        for (i = 0; i < neAttrsLength; i++) {
+            for (var j = 0; j < cLength; j++) {
+                var cAttr = candidates[j].attrs[i];
+
+                if (neAttrsMap[cAttr.name] !== cAttr.value) {
+                    candidates.splice(j, 1);
+                    cLength--;
+                }
+
+                if (candidates.length < NOAH_ARK_CAPACITY)
+                    return;
+            }
+        }
+
+        //NOTE: remove bottommost candidates until Noah's Ark condition will not be met
+        for (i = cLength - 1; i >= NOAH_ARK_CAPACITY - 1; i--) {
+            this.entries.splice(candidates[i].idx, 1);
+            this.length--;
+        }
+    }
+};
+
+//Mutations
+FormattingElementList.prototype.insertMarker = function () {
+    this.entries.push({type: FormattingElementList.MARKER_ENTRY});
+    this.length++;
+};
+
+FormattingElementList.prototype.pushElement = function (element, token) {
+    this._ensureNoahArkCondition(element);
+
+    this.entries.push({
+        type: FormattingElementList.ELEMENT_ENTRY,
+        element: element,
+        token: token
+    });
+
+    this.length++;
+};
+
+FormattingElementList.prototype.insertElementAfterBookmark = function (element, token) {
+    var bookmarkIdx = this.length - 1;
+
+    for (; bookmarkIdx >= 0; bookmarkIdx--) {
+        if (this.entries[bookmarkIdx] === this.bookmark)
+            break;
+    }
+
+    this.entries.splice(bookmarkIdx + 1, 0, {
+        type: FormattingElementList.ELEMENT_ENTRY,
+        element: element,
+        token: token
+    });
+
+    this.length++;
+};
+
+FormattingElementList.prototype.removeEntry = function (entry) {
+    for (var i = this.length - 1; i >= 0; i--) {
+        if (this.entries[i] === entry) {
+            this.entries.splice(i, 1);
+            this.length--;
+            break;
+        }
+    }
+};
+
+FormattingElementList.prototype.clearToLastMarker = function () {
+    while (this.length) {
+        var entry = this.entries.pop();
+
+        this.length--;
+
+        if (entry.type === FormattingElementList.MARKER_ENTRY)
+            break;
+    }
+};
+
+//Search
+FormattingElementList.prototype.getElementEntryInScopeWithTagName = function (tagName) {
+    for (var i = this.length - 1; i >= 0; i--) {
+        var entry = this.entries[i];
+
+        if (entry.type === FormattingElementList.MARKER_ENTRY)
+            return null;
+
+        if (this.treeAdapter.getTagName(entry.element) === tagName)
+            return entry;
+    }
+
+    return null;
+};
+
+FormattingElementList.prototype.getElementEntry = function (element) {
+    for (var i = this.length - 1; i >= 0; i--) {
+        var entry = this.entries[i];
+
+        if (entry.type === FormattingElementList.ELEMENT_ENTRY && entry.element === element)
+            return entry;
+    }
+
+    return null;
+};