Implemented XMLSerializer in PrivateScript.

Source/core/xml/XMLSerializer.js

Index: Source/core/xml/XMLSerializer.js
diff --git a/Source/core/xml/XMLSerializer.js b/Source/core/xml/XMLSerializer.js
new file mode 100644
index 0000000000000000000000000000000000000000..5792432ac804644f0203c17232d1373b5a518bd0
--- /dev/null
+++ b/Source/core/xml/XMLSerializer.js
@@ -0,0 +1,374 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+'use strict';
+
+installClass('XMLSerializer', function(XMLSerializerPrototype) {
+
+    var _EMPTY = '';
+
+    var _XML_NAMESPACE_URI = 'http://www.w3.org/XML/1998/namespace';
+    var _XLINK_NAMESPACE_URI = 'http://www.w3.org/1999/xlink';
+    var _XMLNS_NAMESPACE_URI = 'http://www.w3.org/2000/xmlns';
+    var _HTML_NAMESPACE_URI = 'http://www.w3.org/1999/xhtml';
+
+    var _SERIALIZATION_TYPE_AS_OWNER_DOCUMENT = 0;
+    var _SERIALIZATION_TYPE_FORCED_XML = 1;
+
+    function escapePCDATA(text) {
+        return text.replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
+    }
+
+    function escapeHTMLPCDATA(text) {
+        return escapePCDATA(text).replace(/\u00A0/g, '&nbsp;');
+    }
+
+    function escapeAttributeValue(text) {
+        return text.replace(/&/g, '&amp;').replace(/\u0022/g, '&quot;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
+    }

[Jens Widell, 2014/09/03 10:29:33]

I think \u0022 can be written as ". What's the reason not to?

[tasak, 2014/09/03 11:40:47]

Done.

+
+    function escapeHTMLAttributeValue(text) {
+        return text.replace(/&/g, '&').replace(/\u0022/g, '"').replace(/\u00A0/g, ' ');
+    }
+
+    function validateURI(uri) {
+        return uri ? uri.replace(/\/$/g, '') : '';
+    }
+
+    function cloneNamespace(obj) {
+        var clone = {};
+        for (var i in obj)
+            clone[i] = obj[i];
+        return clone;
+    }
+
+    function isTemplateElement(node) {
+        return isHTMLElement(node) && node.tagName.toLowerCase() == 'template';
+    }
+
+    function generatePrefix(namespaces) {
+        var nsRegex = new RegExp('^ns([0-9]+)$', 'g');
+        var maxValue = 0;
+        for (var key in namespaces) {
+            var match = nsRegex.exec(key);
+            if (match) {
+                var value = parseInt(match[1]);
+                if (maxValue < value)
+                    maxValue = value;
+            }
+        }
+        return 'ns' + (maxValue + 1);
+    }
+
+    function qualifiedTagName(node) {
+        if (isHTMLElement(node))
+            return node.tagName.toLowerCase();
+        return node.tagName;
+    }
+
+    function attributeIsInSerializedNamespace(attr) {
+        var _SERIALIZED_NAMESPACES = {
+            _XML_NAMESPACE_URI:true, _XLINK_NAMESPACE_URI:true,
+            _XMLNS_NAMESPACE_URI:true
+        };
+        return _SERIALIZED_NAMESPACES[validateURI(attr.namespaceURI)];
+    }
+
+    function appendNamespace(prefix, namespaceURI, namespaces) {
+        if (!namespaceURI)
+            return '';
+
+        var result = '';
+        var lookupKey = _EMPTY;
+        if (prefix)
+            lookupKey = prefix;
+        if (namespaces[lookupKey] != namespaceURI) {
+            namespaces[lookupKey] = namespaceURI;
+            result += ' xmlns';
+            if (prefix)
+                result += ':' + prefix;
+            result += '=\"' + namespaceURI + '"';

[Jens Widell, 2014/09/03 10:29:33]

No need to escape " in either string here.

[tasak, 2014/09/03 11:40:46]

Done.

+        }
+        return result;
+    }
+
+    function shouldAddNamespaceAttribute(attr, node) {
+        if (!attr.namespaceURI)
+            return false;
+        if (!attr.prefix)
+            return true;
+        if (!node.hasAttribute('xmlns:' + attr.prefix))
+            return true;
+        return false;
+    }
+
+    function shouldAddNamespaceElement(node, namespaces) {
+        var prefix = node.prefix;
+        if (!node.prefix) {
+            if (node.hasAttribute('xmlns')) {
+                namespaces[_EMPTY] = node.namespaceURI;
+                return false;
+            }
+            return true;
+        }
+        if (!node.hasAttribute('xmlns:' + node.prefix))
+            return true;
+        return false;
+    }
+
+    function isHTMLElement(node) {
+        if (node.nodeType != Node.ELEMENT_NODE)
+            return false;
+
+        for (var runner = node; runner; runner = runner.parentNode) {
+            if (runner.nodeType == Node.ELEMENT_NODE && runner.namespaceURI != null)
+                return validateURI(runner.namespaceURI) == _HTML_NAMESPACE_URI;
+        }
+        return false;
+    }
+
+    function isEndTagOmissible(node) {
+        if (node.nodeType != Node.ELEMENT_NODE)
+            return false;
+        // This tables comes from HTMLElement::ieForbidsInsertHTML().
+        var _END_TAG_OMISSIBLE_TAGS = {
+            'area':true, 'base':true, 'basefont':true, 'br':true, 'col':true, 'embed':true,
+            'frame':true, 'hr':true, 'image':true, 'img':true, 'input':true, 'link':true,
+            'meta':true, 'param':true, 'source':true, 'wbr':true
+        };
+        return _END_TAG_OMISSIBLE_TAGS[node.tagName.toLowerCase()];
+    }
+
+    XMLSerializerPrototype.serializeAsHTMLDocument_ = function(node) {
+        if (this.serializationType_ == _SERIALIZATION_TYPE_FORCED_XML)
+            return false;
+        return node.ownerDocument && (node.ownerDocument instanceof HTMLDocument);
+    }
+
+    XMLSerializerPrototype.serializeAttribute_ = function(node, attr, namespaces) {
+        var documentIsHTML = this.serializeAsHTMLDocument_(node);
+        var result = '';
+
+        var attrPrefix = attr.prefix;
+        var attrName = attr.localName;
+        if (documentIsHTML && !attributeIsInSerializedNamespace(attr)) {
+            result += ' ' + attr.localName;
+        } else {
+            var namespaceURI = validateURI(attr.namespaceURI);
+            if (namespaceURI == _XMLNS_NAMESPACE_URI) {
+                if (!attr.prefix && attr.localName != 'xmlns')
+                    attrPrefix = 'xmlns';
+                if (namespaces) {
+                    var lookupKey = _EMPTY;
+                    if (attr.prefix)
+                        lookupKey = attr.localName;
+                    namespaces[lookupKey] = attr.value;
+                }
+            } else if (namespaceURI == _XML_NAMESPACE_URI) {
+                if (!attr.prefix)
+                    attrPrefix = 'xml';
+            } else {
+                if (namespaceURI == _XLINK_NAMESPACE_URI) {
+                    if (!attr.prefix)
+                        attrPrefix = 'xlink';
+                }
+                if (namespaces && shouldAddNamespaceAttribute(attr, node)) {
+                    if (!attrPrefix) {
+                        for (var i in namespaces) {
+                            if (namespaces[i] == namespaceURI && i) {
+                                attrPrefix = i;
+                                break;
+                            }
+                        }
+                        if (!attrPrefix)
+                            attrPrefix = generatePrefix(namespaces);
+                    }
+                    result += appendNamespace(attrPrefix, namespaceURI, namespaces);
+                }
+            }
+            if (attrPrefix) {
+                result += ' ' + attrPrefix + ':' + attrName;
+            } else {
+                result += ' ' + attrName;
+            }
+        }
+        if (attr.value) {
+            var attrValue;
+            if (documentIsHTML) {
+                attrValue = escapeHTMLAttributeValue(String(attr.value));
+            } else {
+                attrValue = escapeAttributeValue(String(attr.value));
+            }
+            result += '="' + attrValue + '"';
+        }
+        return result;
+    }
+
+    XMLSerializerPrototype.serializeAttributes_ = function(node, namespaces) {
+        var attributes = node.attributes;
+        if (!attributes)
+            return '';
+
+        var sink = '';
+        var attrs = [];
+        for (var index = 0; index < attributes.length; ++index)
+           attrs.push(attributes[index]);
+
+        var serializer = this;
+        attrs.sort(function(a, b) {
+            return a.name <= b.name ? -1 : 0;

[Jens Widell, 2014/09/03 10:29:33]

This is an unusual comparison function:

If a < b, say "a is less than b"
If a == b, say "a is less than b"
If a > b, say "they are equal"

AFAICT, the result in V8 is that it leaves the array in the order it was to
begin with.

[tasak, 2014/09/03 11:40:46]

I see. I found that we don't need to sort attributes.

+        }).forEach(function(attr) {
+            if (attr.name.indexOf('_moz') == 0)
+                return;
+            sink += serializer.serializeAttribute_(node, attr, namespaces);
+        });
+        return sink;
+    }
+
+    XMLSerializerPrototype.serializeOpenTag_ = function(node, namespaces) {
+        var sink = '<' + qualifiedTagName(node);
+        if (!this.serializeAsHTMLDocument_(node) && namespaces && shouldAddNamespaceElement(node, namespaces))
+            sink += appendNamespace(node.prefix, node.namespaceURI, namespaces);
+        return sink;
+    }
+
+    XMLSerializerPrototype.serializeCloseTag_ = function(node, namespaces) {
+        var sink = '';
+        if (this.shouldSelfClose_(node)) {
+            if (isHTMLElement(node))
+                sink += ' ';
+            sink += '/';
+        }
+        sink += '>';
+        return sink;
+    }
+
+    XMLSerializerPrototype.serializeElement_ = function(node, namespaces) {
+        var sink = this.serializeOpenTag_(node, namespaces);
+
+        sink += this.serializeAttributes_(node, namespaces);
+        sink += this.serializeCloseTag_(node, namespaces);
+        return sink;
+    }
+
+    XMLSerializerPrototype.serializeText_ = function(node) {
+        if (!node.nodeValue)
+            return '';
+
+        if (this.serializeAsHTMLDocument_(node)) {
+            if (node.parentNode && isHTMLElement(node.parentNode)) {
+                var tagName = node.parentNode.tagName.toLowerCase();
+                if (tagName == 'script' || tagName == 'style' || tagName == 'xmp')
+                    return node.nodeValue;
+            }
+            return escapeHTMLPCDATA(node.nodeValue);
+        }
+        return escapePCDATA(node.nodeValue);
+    }
+
+    XMLSerializerPrototype.serializeComment_ = function(node) {
+        return '<!--' + node.nodeValue + '-->';
+    }
+
+    XMLSerializerPrototype.serializeDocument_ = function(node) {
+        return this.documentXMLDeclaration_;
+    }
+
+    XMLSerializerPrototype.serializeDocumentType_ = function(node) {
+        var s = '<!DOCTYPE ' + node.name;
+        if (node.publicId) {
+            s += ' PUBLIC "' + node.publicId + '"';
+            if (node.systemId)

[Jens Widell, 2014/09/03 10:29:33]

There can also be just a system id, in which case the output should be <!DOCTYPE
name SYSTEM "url">.

[tasak, 2014/09/03 11:40:47]

Done.

+                s += ' "' + node.systemId + '"';
+        }
+        s += '>';
+        return s;
+    }
+
+    XMLSerializerPrototype.serializeProcessingInstruction_ = function(node) {
+        var s = '<?' + node.target;
+        if (node.nodeValue)
+            s += ' ' + node.nodeValue;
+        s += '?>';
+        return s;
+    }
+
+    XMLSerializerPrototype.serializeCDataSection_ = function(node) {
+        return '<![CDATA[' + node.nodeValue + ']]>';
+    }
+
+    XMLSerializerPrototype.serializeStartTag = function(node, namespaces) {
+        if (node.nodeType == Node.TEXT_NODE) {
+            return this.serializeText_(node);
+        } else if (node.nodeType == Node.COMMENT_NODE) {
+            return this.serializeComment_(node);
+        } else if (node.nodeType == Node.DOCUMENT_NODE) {
+            return this.serializeDocument_(node);
+        } else if (node.nodeType == Node.DOCUMENT_TYPE_NODE) {
+            return this.serializeDocumentType_(node);
+        } else if (node.nodeType == Node.DOCUMENT_FRAGMENT_NODE) {
+            return '';
+        } else if (node.nodeType == Node.PROCESSING_INSTRUCTION_NODE) {
+            return this.serializeProcessingInstruction_(node);
+        } else if (node.nodeType == Node.ELEMENT_NODE) {
+            return this.serializeElement_(node, namespaces);
+        } else if (node.nodeType == Node.CDATA_SECTION_NODE) {
+            return this.serializeCDataSection_(node);
+        } else {
+            return '';
+        }
+    }
+
+    XMLSerializerPrototype.shouldSelfClose_ = function(node) {
+        if (this.serializeAsHTMLDocument_(node))
+            return false;
+        if (node.hasChildNodes())
+            return false;
+        if (isTemplateElement(node) && node.content && node.content.hasChildNodes())
+            return false;
+        if (isHTMLElement(node) && !isEndTagOmissible(node))
+            return false;
+        return true;
+    }
+
+    XMLSerializerPrototype.serializeEndTag_ = function(node) {
+        if (this.shouldSelfClose_(node) || (!node.hasChildNodes() && isEndTagOmissible(node)))
+            return '';
+        return '</' + qualifiedTagName(node) + '>';
+    }
+
+    XMLSerializerPrototype.serializeNodesWithNamespaces_ = function(node, namespaces) {
+        var sink = '';
+        var localNamespaces = cloneNamespace(namespaces);
+        sink += this.serializeStartTag(node, localNamespaces);
+
+        if (!(this.serializeAsHTMLDocument_(node) && isEndTagOmissible(node))) {
+            var current = node;
+            if (isTemplateElement(node))
+                current = node.content;
+            current = current.firstChild;
+            while (current) {
+                sink += this.serializeNodesWithNamespaces_(current, localNamespaces);
+                current = current.nextSibling;
+            }
+        }
+        if (node.nodeType == Node.ELEMENT_NODE)
+            sink += this.serializeEndTag_(node);
+        return sink;
+    }
+
+    XMLSerializerPrototype.serializeNodes_ = function(node) {
+        var namespaces = {};
+        if (!this.serializeAsHTMLDocument_(node))
+            namespaces['xml'] = _XML_NAMESPACE_URI;

[Jens Widell, 2014/09/03 10:29:34]

Could add 'xmlns' as well, I guess. It too is implicitly declared (and IIRC
illegal to declare explicitly.)

[tasak, 2014/09/03 11:40:47]

Done.

+
+        return this.serializeNodesWithNamespaces_(node, namespaces);
+    }
+
+    XMLSerializerPrototype.serializeToStringInternal = function(root, xmlDeclaration) {
+        this.serializationType_ = _SERIALIZATION_TYPE_FORCED_XML;
+        this.documentXMLDeclaration_ = xmlDeclaration ? xmlDeclaration : '';
+        return this.serializeNodes_(root);
+    }
+});