parse5 3.0.3 → 5.1.0

package/lib/extensions/error-reporting/parser-mixin.js

@@ -0,0 +1,52 @@
+'use strict';
+
+const ErrorReportingMixinBase = require('./mixin-base');
+const ErrorReportingTokenizerMixin = require('./tokenizer-mixin');
+const LocationInfoTokenizerMixin = require('../location-info/tokenizer-mixin');
+const Mixin = require('../../utils/mixin');
+
+class ErrorReportingParserMixin extends ErrorReportingMixinBase {
+    constructor(parser, opts) {
+        super(parser, opts);
+
+        this.opts = opts;
+        this.ctLoc = null;
+        this.locBeforeToken = false;
+    }
+
+    _setErrorLocation(err) {
+        if (this.ctLoc) {
+            err.startLine = this.ctLoc.startLine;
+            err.startCol = this.ctLoc.startCol;
+            err.startOffset = this.ctLoc.startOffset;
+
+            err.endLine = this.locBeforeToken ? this.ctLoc.startLine : this.ctLoc.endLine;
+            err.endCol = this.locBeforeToken ? this.ctLoc.startCol : this.ctLoc.endCol;
+            err.endOffset = this.locBeforeToken ? this.ctLoc.startOffset : this.ctLoc.endOffset;
+        }
+    }
+
+    _getOverriddenMethods(mxn, orig) {
+        return {
+            _bootstrap(document, fragmentContext) {
+                orig._bootstrap.call(this, document, fragmentContext);
+
+                Mixin.install(this.tokenizer, ErrorReportingTokenizerMixin, mxn.opts);
+                Mixin.install(this.tokenizer, LocationInfoTokenizerMixin);
+            },
+
+            _processInputToken(token) {
+                mxn.ctLoc = token.location;
+
+                orig._processInputToken.call(this, token);
+            },
+
+            _err(code, options) {
+                mxn.locBeforeToken = options && options.beforeToken;
+                mxn._reportError(code);
+            }
+        };
+    }
+}
+
+module.exports = ErrorReportingParserMixin;

package/lib/extensions/error-reporting/preprocessor-mixin.js

@@ -0,0 +1,24 @@
+'use strict';
+
+const ErrorReportingMixinBase = require('./mixin-base');
+const PositionTrackingPreprocessorMixin = require('../position-tracking/preprocessor-mixin');
+const Mixin = require('../../utils/mixin');
+
+class ErrorReportingPreprocessorMixin extends ErrorReportingMixinBase {
+    constructor(preprocessor, opts) {
+        super(preprocessor, opts);
+
+        this.posTracker = Mixin.install(preprocessor, PositionTrackingPreprocessorMixin);
+        this.lastErrOffset = -1;
+    }
+
+    _reportError(code) {
+        //NOTE: avoid reporting error twice on advance/retreat
+        if (this.lastErrOffset !== this.posTracker.offset) {
+            this.lastErrOffset = this.posTracker.offset;
+            super._reportError(code);
+        }
+    }
+}
+
+module.exports = ErrorReportingPreprocessorMixin;

package/lib/extensions/error-reporting/tokenizer-mixin.js

@@ -0,0 +1,17 @@
+'use strict';
+
+const ErrorReportingMixinBase = require('./mixin-base');
+const ErrorReportingPreprocessorMixin = require('./preprocessor-mixin');
+const Mixin = require('../../utils/mixin');
+
+class ErrorReportingTokenizerMixin extends ErrorReportingMixinBase {
+    constructor(tokenizer, opts) {
+        super(tokenizer, opts);
+
+        const preprocessorMixin = Mixin.install(tokenizer.preprocessor, ErrorReportingPreprocessorMixin, opts);
+
+        this.posTracker = preprocessorMixin.posTracker;
+    }
+}
+
+module.exports = ErrorReportingTokenizerMixin;

package/lib/extensions/location-info/open-element-stack-mixin.js

@@ -0,0 +1,35 @@
+'use strict';
+
+const Mixin = require('../../utils/mixin');
+
+class LocationInfoOpenElementStackMixin extends Mixin {
+    constructor(stack, opts) {
+        super(stack);
+
+        this.onItemPop = opts.onItemPop;
+    }
+
+    _getOverriddenMethods(mxn, orig) {
+        return {
+            pop() {
+                mxn.onItemPop(this.current);
+                orig.pop.call(this);
+            },
+
+            popAllUpToHtmlElement() {
+                for (let i = this.stackTop; i > 0; i--) {
+                    mxn.onItemPop(this.items[i]);
+                }
+
+                orig.popAllUpToHtmlElement.call(this);
+            },
+
+            remove(element) {
+                mxn.onItemPop(this.current);
+                orig.remove.call(this, element);
+            }
+        };
+    }
+}
+
+module.exports = LocationInfoOpenElementStackMixin;

package/lib/extensions/location-info/parser-mixin.js

@@ -0,0 +1,222 @@
+'use strict';
+
+const Mixin = require('../../utils/mixin');
+const Tokenizer = require('../../tokenizer');
+const LocationInfoTokenizerMixin = require('./tokenizer-mixin');
+const LocationInfoOpenElementStackMixin = require('./open-element-stack-mixin');
+const HTML = require('../../common/html');
+
+//Aliases
+const $ = HTML.TAG_NAMES;
+
+class LocationInfoParserMixin extends Mixin {
+    constructor(parser) {
+        super(parser);
+
+        this.parser = parser;
+        this.treeAdapter = this.parser.treeAdapter;
+        this.posTracker = null;
+        this.lastStartTagToken = null;
+        this.lastFosterParentingLocation = null;
+        this.currentToken = null;
+    }
+
+    _setStartLocation(element) {
+        let loc = null;
+
+        if (this.lastStartTagToken) {
+            loc = Object.assign({}, this.lastStartTagToken.location);
+            loc.startTag = this.lastStartTagToken.location;
+        }
+
+        this.treeAdapter.setNodeSourceCodeLocation(element, loc);
+    }
+
+    _setEndLocation(element, closingToken) {
+        const loc = this.treeAdapter.getNodeSourceCodeLocation(element);
+
+        if (loc) {
+            if (closingToken.location) {
+                const ctLoc = closingToken.location;
+                const tn = this.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.
+                const isClosingEndTag = closingToken.type === Tokenizer.END_TAG_TOKEN && tn === closingToken.tagName;
+
+                if (isClosingEndTag) {
+                    loc.endTag = Object.assign({}, ctLoc);
+                    loc.endLine = ctLoc.endLine;
+                    loc.endCol = ctLoc.endCol;
+                    loc.endOffset = ctLoc.endOffset;
+                } else {
+                    loc.endLine = ctLoc.startLine;
+                    loc.endCol = ctLoc.startCol;
+                    loc.endOffset = ctLoc.startOffset;
+                }
+            }
+        }
+    }
+
+    _getOverriddenMethods(mxn, orig) {
+        return {
+            _bootstrap(document, fragmentContext) {
+                orig._bootstrap.call(this, document, fragmentContext);
+
+                mxn.lastStartTagToken = null;
+                mxn.lastFosterParentingLocation = null;
+                mxn.currentToken = null;
+
+                const tokenizerMixin = Mixin.install(this.tokenizer, LocationInfoTokenizerMixin);
+
+                mxn.posTracker = tokenizerMixin.posTracker;
+
+                Mixin.install(this.openElements, LocationInfoOpenElementStackMixin, {
+                    onItemPop: function(element) {
+                        mxn._setEndLocation(element, mxn.currentToken);
+                    }
+                });
+            },
+
+            _runParsingLoop(scriptHandler) {
+                orig._runParsingLoop.call(this, scriptHandler);
+
+                // NOTE: generate location info for elements
+                // that remains on open element stack
+                for (let i = this.openElements.stackTop; i >= 0; i--) {
+                    mxn._setEndLocation(this.openElements.items[i], mxn.currentToken);
+                }
+            },
+
+            //Token processing
+            _processTokenInForeignContent(token) {
+                mxn.currentToken = token;
+                orig._processTokenInForeignContent.call(this, token);
+            },
+
+            _processToken(token) {
+                mxn.currentToken = token;
+                orig._processToken.call(this, token);
+
+                //NOTE: <body> and <html> are never popped from the stack, so we need to updated
+                //their end location explicitly.
+                const requireExplicitUpdate =
+                    token.type === Tokenizer.END_TAG_TOKEN &&
+                    (token.tagName === $.HTML || (token.tagName === $.BODY && this.openElements.hasInScope($.BODY)));
+
+                if (requireExplicitUpdate) {
+                    for (let i = this.openElements.stackTop; i >= 0; i--) {
+                        const element = this.openElements.items[i];
+
+                        if (this.treeAdapter.getTagName(element) === token.tagName) {
+                            mxn._setEndLocation(element, token);
+                            break;
+                        }
+                    }
+                }
+            },
+
+            //Doctype
+            _setDocumentType(token) {
+                orig._setDocumentType.call(this, token);
+
+                const documentChildren = this.treeAdapter.getChildNodes(this.document);
+                const cnLength = documentChildren.length;
+
+                for (let i = 0; i < cnLength; i++) {
+                    const node = documentChildren[i];
+
+                    if (this.treeAdapter.isDocumentTypeNode(node)) {
+                        this.treeAdapter.setNodeSourceCodeLocation(node, token.location);
+                        break;
+                    }
+                }
+            },
+
+            //Elements
+            _attachElementToTree(element) {
+                //NOTE: _attachElementToTree is called from _appendElement, _insertElement and _insertTemplate methods.
+                //So we will use token location stored in this methods for the element.
+                mxn._setStartLocation(element);
+                mxn.lastStartTagToken = null;
+                orig._attachElementToTree.call(this, element);
+            },
+
+            _appendElement(token, namespaceURI) {
+                mxn.lastStartTagToken = token;
+                orig._appendElement.call(this, token, namespaceURI);
+            },
+
+            _insertElement(token, namespaceURI) {
+                mxn.lastStartTagToken = token;
+                orig._insertElement.call(this, token, namespaceURI);
+            },
+
+            _insertTemplate(token) {
+                mxn.lastStartTagToken = token;
+                orig._insertTemplate.call(this, token);
+
+                const tmplContent = this.treeAdapter.getTemplateContent(this.openElements.current);
+
+                this.treeAdapter.setNodeSourceCodeLocation(tmplContent, null);
+            },
+
+            _insertFakeRootElement() {
+                orig._insertFakeRootElement.call(this);
+                this.treeAdapter.setNodeSourceCodeLocation(this.openElements.current, null);
+            },
+
+            //Comments
+            _appendCommentNode(token, parent) {
+                orig._appendCommentNode.call(this, token, parent);
+
+                const children = this.treeAdapter.getChildNodes(parent);
+                const commentNode = children[children.length - 1];
+
+                this.treeAdapter.setNodeSourceCodeLocation(commentNode, token.location);
+            },
+
+            //Text
+            _findFosterParentingLocation() {
+                //NOTE: store last foster parenting location, so we will be able to find inserted text
+                //in case of foster parenting
+                mxn.lastFosterParentingLocation = orig._findFosterParentingLocation.call(this);
+
+                return mxn.lastFosterParentingLocation;
+            },
+
+            _insertCharacters(token) {
+                orig._insertCharacters.call(this, token);
+
+                const hasFosterParent = this._shouldFosterParentOnInsertion();
+
+                const parent =
+                    (hasFosterParent && mxn.lastFosterParentingLocation.parent) ||
+                    this.openElements.currentTmplContent ||
+                    this.openElements.current;
+
+                const siblings = this.treeAdapter.getChildNodes(parent);
+
+                const textNodeIdx =
+                    hasFosterParent && mxn.lastFosterParentingLocation.beforeElement
+                        ? siblings.indexOf(mxn.lastFosterParentingLocation.beforeElement) - 1
+                        : siblings.length - 1;
+
+                const textNode = siblings[textNodeIdx];
+
+                //NOTE: if we have location assigned by another token, then just update end position
+                const tnLoc = this.treeAdapter.getNodeSourceCodeLocation(textNode);
+
+                if (tnLoc) {
+                    tnLoc.endLine = token.location.endLine;
+                    tnLoc.endCol = token.location.endCol;
+                    tnLoc.endOffset = token.location.endOffset;
+                } else {
+                    this.treeAdapter.setNodeSourceCodeLocation(textNode, token.location);
+                }
+            }
+        };
+    }
+}
+
+module.exports = LocationInfoParserMixin;

package/lib/extensions/location-info/tokenizer-mixin.js

@@ -0,0 +1,146 @@
+'use strict';
+
+const Mixin = require('../../utils/mixin');
+const Tokenizer = require('../../tokenizer');
+const PositionTrackingPreprocessorMixin = require('../position-tracking/preprocessor-mixin');
+
+class LocationInfoTokenizerMixin extends Mixin {
+    constructor(tokenizer) {
+        super(tokenizer);
+
+        this.tokenizer = tokenizer;
+        this.posTracker = Mixin.install(tokenizer.preprocessor, PositionTrackingPreprocessorMixin);
+        this.currentAttrLocation = null;
+        this.ctLoc = null;
+    }
+
+    _getCurrentLocation() {
+        return {
+            startLine: this.posTracker.line,
+            startCol: this.posTracker.col,
+            startOffset: this.posTracker.offset,
+            endLine: -1,
+            endCol: -1,
+            endOffset: -1
+        };
+    }
+
+    _attachCurrentAttrLocationInfo() {
+        this.currentAttrLocation.endLine = this.posTracker.line;
+        this.currentAttrLocation.endCol = this.posTracker.col;
+        this.currentAttrLocation.endOffset = this.posTracker.offset;
+
+        const currentToken = this.tokenizer.currentToken;
+        const currentAttr = this.tokenizer.currentAttr;
+
+        if (!currentToken.location.attrs) {
+            currentToken.location.attrs = Object.create(null);
+        }
+
+        currentToken.location.attrs[currentAttr.name] = this.currentAttrLocation;
+    }
+
+    _getOverriddenMethods(mxn, orig) {
+        const methods = {
+            _createStartTagToken() {
+                orig._createStartTagToken.call(this);
+                this.currentToken.location = mxn.ctLoc;
+            },
+
+            _createEndTagToken() {
+                orig._createEndTagToken.call(this);
+                this.currentToken.location = mxn.ctLoc;
+            },
+
+            _createCommentToken() {
+                orig._createCommentToken.call(this);
+                this.currentToken.location = mxn.ctLoc;
+            },
+
+            _createDoctypeToken(initialName) {
+                orig._createDoctypeToken.call(this, initialName);
+                this.currentToken.location = mxn.ctLoc;
+            },
+
+            _createCharacterToken(type, ch) {
+                orig._createCharacterToken.call(this, type, ch);
+                this.currentCharacterToken.location = mxn.ctLoc;
+            },
+
+            _createEOFToken() {
+                orig._createEOFToken.call(this);
+                this.currentToken.location = mxn._getCurrentLocation();
+            },
+
+            _createAttr(attrNameFirstCh) {
+                orig._createAttr.call(this, attrNameFirstCh);
+                mxn.currentAttrLocation = mxn._getCurrentLocation();
+            },
+
+            _leaveAttrName(toState) {
+                orig._leaveAttrName.call(this, toState);
+                mxn._attachCurrentAttrLocationInfo();
+            },
+
+            _leaveAttrValue(toState) {
+                orig._leaveAttrValue.call(this, toState);
+                mxn._attachCurrentAttrLocationInfo();
+            },
+
+            _emitCurrentToken() {
+                const ctLoc = this.currentToken.location;
+
+                //NOTE: if we have pending character token make it's end location equal to the
+                //current token's start location.
+                if (this.currentCharacterToken) {
+                    this.currentCharacterToken.location.endLine = ctLoc.startLine;
+                    this.currentCharacterToken.location.endCol = ctLoc.startCol;
+                    this.currentCharacterToken.location.endOffset = ctLoc.startOffset;
+                }
+
+                if (this.currentToken.type === Tokenizer.EOF_TOKEN) {
+                    ctLoc.endLine = ctLoc.startLine;
+                    ctLoc.endCol = ctLoc.startCol;
+                    ctLoc.endOffset = ctLoc.startOffset;
+                } else {
+                    ctLoc.endLine = mxn.posTracker.line;
+                    ctLoc.endCol = mxn.posTracker.col + 1;
+                    ctLoc.endOffset = mxn.posTracker.offset + 1;
+                }
+
+                orig._emitCurrentToken.call(this);
+            },
+
+            _emitCurrentCharacterToken() {
+                const ctLoc = this.currentCharacterToken && this.currentCharacterToken.location;
+
+                //NOTE: if we have character token and it's location wasn't set in the _emitCurrentToken(),
+                //then set it's location at the current preprocessor position.
+                //We don't need to increment preprocessor position, since character token
+                //emission is always forced by the start of the next character token here.
+                //So, we already have advanced position.
+                if (ctLoc && ctLoc.endOffset === -1) {
+                    ctLoc.endLine = mxn.posTracker.line;
+                    ctLoc.endCol = mxn.posTracker.col;
+                    ctLoc.endOffset = mxn.posTracker.offset;
+                }
+
+                orig._emitCurrentCharacterToken.call(this);
+            }
+        };
+
+        //NOTE: patch initial states for each mode to obtain token start position
+        Object.keys(Tokenizer.MODE).forEach(modeName => {
+            const state = Tokenizer.MODE[modeName];
+
+            methods[state] = function(cp) {
+                mxn.ctLoc = mxn._getCurrentLocation();
+                orig[state].call(this, cp);
+            };
+        });
+
+        return methods;
+    }
+}
+
+module.exports = LocationInfoTokenizerMixin;

package/lib/extensions/position-tracking/preprocessor-mixin.js

@@ -0,0 +1,64 @@
+'use strict';
+
+const Mixin = require('../../utils/mixin');
+
+class PositionTrackingPreprocessorMixin extends Mixin {
+    constructor(preprocessor) {
+        super(preprocessor);
+
+        this.preprocessor = preprocessor;
+        this.isEol = false;
+        this.lineStartPos = 0;
+        this.droppedBufferSize = 0;
+
+        this.offset = 0;
+        this.col = 0;
+        this.line = 1;
+    }
+
+    _getOverriddenMethods(mxn, orig) {
+        return {
+            advance() {
+                const pos = this.pos + 1;
+                const ch = this.html[pos];
+
+                //NOTE: LF should be in the last column of the line
+                if (mxn.isEol) {
+                    mxn.isEol = false;
+                    mxn.line++;
+                    mxn.lineStartPos = pos;
+                }
+
+                if (ch === '\n' || (ch === '\r' && this.html[pos + 1] !== '\n')) {
+                    mxn.isEol = true;
+                }
+
+                mxn.col = pos - mxn.lineStartPos + 1;
+                mxn.offset = mxn.droppedBufferSize + pos;
+
+                return orig.advance.call(this);
+            },
+
+            retreat() {
+                orig.retreat.call(this);
+
+                mxn.isEol = false;
+                mxn.col = this.pos - mxn.lineStartPos + 1;
+            },
+
+            dropParsedChunk() {
+                const prevPos = this.pos;
+
+                orig.dropParsedChunk.call(this);
+
+                const reduction = prevPos - this.pos;
+
+                mxn.lineStartPos -= reduction;
+                mxn.droppedBufferSize += reduction;
+                mxn.offset = mxn.droppedBufferSize + this.pos;
+            }
+        };
+    }
+}
+
+module.exports = PositionTrackingPreprocessorMixin;

package/lib/index.js

@@ -1,12 +1,11 @@
 'use strict';
 
-var Parser = require('./parser'),
-    Serializer = require('./serializer');
-
+const Parser = require('./parser');
+const Serializer = require('./serializer');
 
 // Shorthands
 exports.parse = function parse(html, options) {
-    var parser = new Parser(options);
+    const parser = new Parser(options);
 
     return parser.parse(html);
 };
@@ -18,27 +17,13 @@ exports.parseFragment = function parseFragment(fragmentContext, html, options) {
         fragmentContext = null;
     }
 
-    var parser = new Parser(options);
+    const parser = new Parser(options);
 
     return parser.parseFragment(html, fragmentContext);
 };
 
-exports.serialize = function (node, options) {
-    var serializer = new Serializer(node, options);
+exports.serialize = function(node, options) {
+    const 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');