parse5 3.0.1 → 3.0.2

package/lib/location_info/parser_mixin.js

@@ -8,49 +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;
-
-    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;
-    }
-}
-
-
 exports.assign = function (parser) {
     //NOTE: obtain Parser proto this way to avoid module circular references
     var parserProto = Object.getPrototypeOf(parser),
@@ -59,6 +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) {
@@ -70,23 +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) {
@@ -107,7 +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/parser/index.js

@@ -319,7 +319,7 @@ Parser.prototype.parse = function (html) {
 
     this._bootstrap(document, null);
     this.tokenizer.write(html, true);
-    this._runParsingLoop(null, null);
+    this._runParsingLoop(null);
 
     return document;
 };
@@ -345,7 +345,7 @@ Parser.prototype.parseFragment = function (html, fragmentContext) {
     this._resetInsertionMode();
     this._findFormInFragmentContext();
     this.tokenizer.write(html, true);
-    this._runParsingLoop(null, null);
+    this._runParsingLoop(null);
 
     var rootElement = this.treeAdapter.getFirstChild(documentMock),
         fragment = this.treeAdapter.createDocumentFragment();
@@ -386,7 +386,7 @@ Parser.prototype._bootstrap = function (document, fragmentContext) {
 };
 
 //Parsing loop
-Parser.prototype._runParsingLoop = function (writeCallback, scriptHandler) {
+Parser.prototype._runParsingLoop  = function (scriptHandler) {
     while (!this.stopped) {
         this._setupTokenizerCDATAMode();
 
@@ -411,6 +411,10 @@ Parser.prototype._runParsingLoop = function (writeCallback, scriptHandler) {
         if (scriptHandler && this.pendingScript)
             break;
     }
+};
+
+Parser.prototype.runParsingLoopForCurrentChunk = function (writeCallback, scriptHandler) {
+    this._runParsingLoop(scriptHandler);
 
     if (scriptHandler && this.pendingScript) {
         var script = this.pendingScript;

package/lib/parser/parser_stream.js

@@ -40,7 +40,7 @@ ParserStream.prototype.end = function (chunk, encoding, callback) {
 
 //Scriptable parser implementation
 ParserStream.prototype._runParsingLoop = function () {
-    this.parser._runParsingLoop(this.writeCallback, this._scriptHandler);
+    this.parser.runParsingLoopForCurrentChunk(this.writeCallback, this._scriptHandler);
 };
 
 ParserStream.prototype._resume = function () {

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/package.json

@@ -1,63 +1,63 @@
-{
-  "name": "parse5",
-  "description": "HTML parsing/serialization toolset for Node.js. WHATWG HTML Living Standard (aka HTML5)-compliant.",
-  "version": "3.0.1",
-  "author": "Ivan Nikulin <ifaaan@gmail.com> (https://github.com/inikulin)",
-  "contributors": "https://github.com/inikulin/parse5/graphs/contributors",
-  "homepage": "https://github.com/inikulin/parse5",
-  "devDependencies": {
-    "del": "^2.0.2",
-    "gulp": "^3.9.0",
-    "gulp-benchmark": "^1.1.1",
-    "gulp-download": "0.0.1",
-    "gulp-eslint": "^3.0.1",
-    "gulp-install": "^0.6.0",
-    "gulp-mocha": "^2.1.3",
-    "gulp-rename": "^1.2.2",
-    "gulp-typedoc": "^2.0.0",
-    "gulp-typescript": "^3.1.2",
-    "publish-please": "^2.2.0",
-    "through2": "^2.0.0",
-    "typedoc": "^0.5.1",
-    "typescript": "^2.0.6"
-  },
-  "keywords": [
-    "html",
-    "parser",
-    "html5",
-    "WHATWG",
-    "specification",
-    "fast",
-    "html parser",
-    "html5 parser",
-    "htmlparser",
-    "parse5",
-    "serializer",
-    "html serializer",
-    "htmlserializer",
-    "sax",
-    "simple api",
-    "parse",
-    "tokenize",
-    "serialize",
-    "tokenizer"
-  ],
-  "license": "MIT",
-  "main": "./lib/index.js",
-  "types": "./lib/index.d.ts",
-  "repository": {
-    "type": "git",
-    "url": "git://github.com/inikulin/parse5.git"
-  },
-  "scripts": {
-    "test": "gulp test",
-    "publish-please": "publish-please",
-    "prepublish": "publish-please guard"
-  },
-  "files": [
-    "lib"
-  ],
-  "dependencies": {
-    "@types/node": "^6.0.46"
-  }
-}
+{
+  "name": "parse5",
+  "description": "HTML parsing/serialization toolset for Node.js. WHATWG HTML Living Standard (aka HTML5)-compliant.",
+  "version": "3.0.2",
+  "author": "Ivan Nikulin <ifaaan@gmail.com> (https://github.com/inikulin)",
+  "contributors": "https://github.com/inikulin/parse5/graphs/contributors",
+  "homepage": "https://github.com/inikulin/parse5",
+  "devDependencies": {
+    "del": "^2.0.2",
+    "gulp": "^3.9.0",
+    "gulp-benchmark": "^1.1.1",
+    "gulp-download": "0.0.1",
+    "gulp-eslint": "^3.0.1",
+    "gulp-install": "^0.6.0",
+    "gulp-mocha": "^2.1.3",
+    "gulp-rename": "^1.2.2",
+    "gulp-typedoc": "^2.0.0",
+    "gulp-typescript": "^3.1.2",
+    "publish-please": "^2.2.0",
+    "through2": "^2.0.0",
+    "typedoc": "^0.5.1",
+    "typescript": "^2.0.6"
+  },
+  "keywords": [
+    "html",
+    "parser",
+    "html5",
+    "WHATWG",
+    "specification",
+    "fast",
+    "html parser",
+    "html5 parser",
+    "htmlparser",
+    "parse5",
+    "serializer",
+    "html serializer",
+    "htmlserializer",
+    "sax",
+    "simple api",
+    "parse",
+    "tokenize",
+    "serialize",
+    "tokenizer"
+  ],
+  "license": "MIT",
+  "main": "./lib/index.js",
+  "types": "./lib/index.d.ts",
+  "repository": {
+    "type": "git",
+    "url": "git://github.com/inikulin/parse5.git"
+  },
+  "scripts": {
+    "test": "gulp test",
+    "publish-please": "publish-please",
+    "prepublish": "publish-please guard"
+  },
+  "files": [
+    "lib"
+  ],
+  "dependencies": {
+    "@types/node": "^6.0.46"
+  }
+}