Xml/XPath/XPathDocument.cs

//------------------------------------------------------------------------------
// <copyright file="XPathDocument.cs" company="Microsoft">
//     Copyright (c) Microsoft Corporation.  All rights reserved.
// </copyright>
// <owner current="true" primary="true">Microsoft</owner>
//------------------------------------------------------------------------------

using System;
using System.IO;
using System.Xml;
using System.Xml.Schema;
using System.Collections.Generic;
using MS.Internal.Xml.Cache;
using System.Diagnostics;
using System.Text;
using System.Runtime.Versioning;

namespace System.Xml.XPath {

    /// <summary>
    /// XDocument follows the XPath/XQuery data model.  All nodes in the tree reference the document,
    /// and the document references the root node of the tree.  All namespaces are stored out-of-line,
    /// in an Element --> In-Scope-Namespaces map.
    /// </summary>
    public class XPathDocument : IXPathNavigable {
        private XPathNode[] pageText, pageRoot, pageXmlNmsp;
        private int idxText, idxRoot, idxXmlNmsp;
        private XmlNameTable nameTable;
        private bool hasLineInfo;
        private Dictionary<XPathNodeRef, XPathNodeRef> mapNmsp;
        private Dictionary<string, XPathNodeRef> idValueMap;

        /// <summary>
        /// Flags that control Load behavior.
        /// </summary>
        internal enum LoadFlags {
            None = 0,
            AtomizeNames = 1,       // Do not assume that names passed to XPathDocumentBuilder have been pre-atomized, and atomize them
            Fragment = 2,           // Create a document with no document node
        }


        //-----------------------------------------------
        // Creation Methods
        //-----------------------------------------------

        /// <summary>
        /// Create a new empty document.
        /// </summary>
        internal XPathDocument() {
            this.nameTable = new NameTable();
        }

        /// <summary>
        /// Create a new empty document.  All names should be atomized using "nameTable".
        /// </summary>
        internal XPathDocument(XmlNameTable nameTable) {
            if (nameTable == null)
                throw new ArgumentNullException("nameTable");

            this.nameTable = nameTable;
        }

        /// <summary>
        /// Create a new document and load the content from the reader.
        /// </summary>
        public XPathDocument(XmlReader reader) : this(reader, XmlSpace.Default) {
        }

        /// <summary>
        /// Create a new document from "reader", with whitespace handling controlled according to "space".
        /// </summary>
        public XPathDocument(XmlReader reader, XmlSpace space) {
            if (reader == null)
                throw new ArgumentNullException("reader");

            LoadFromReader(reader, space);
        }

        /// <summary>
        /// Create a new document and load the content from the text reader.
        /// </summary>
        public XPathDocument(TextReader textReader) {
            XmlTextReaderImpl reader = SetupReader(new XmlTextReaderImpl(string.Empty, textReader));

            try {
                LoadFromReader(reader, XmlSpace.Default);
            }
            finally {
                reader.Close();
            }
        }

        /// <summary>
        /// Create a new document and load the content from the stream.
        /// </summary>
        public XPathDocument(Stream stream) {
            XmlTextReaderImpl reader = SetupReader(new XmlTextReaderImpl(string.Empty, stream));

            try {
                LoadFromReader(reader, XmlSpace.Default);
            }
            finally {
                reader.Close();
            }
        }

        /// <summary>
        /// Create a new document and load the content from the Uri.
        /// </summary>
        [ResourceConsumption(ResourceScope.Machine)]
        [ResourceExposure(ResourceScope.Machine)]
        public XPathDocument(string uri) : this(uri, XmlSpace.Default) {
        }

        /// <summary>
        /// Create a new document and load the content from the Uri, with whitespace handling controlled according to "space".
        /// </summary>
        [ResourceConsumption(ResourceScope.Machine)]
        [ResourceExposure(ResourceScope.Machine)]
        public XPathDocument(string uri, XmlSpace space) {
            XmlTextReaderImpl reader = SetupReader(new XmlTextReaderImpl(uri));

            try {
                LoadFromReader(reader, space);
            }
            finally {
                reader.Close();
            }
        }

        /// <summary>
        /// Create a writer that can be used to create nodes in this document.  The root node will be assigned "baseUri", and flags
        /// can be passed to indicate that names should be atomized by the builder and/or a fragment should be created.
        /// </summary>
        internal XmlRawWriter LoadFromWriter(LoadFlags flags, string baseUri) {
            return new XPathDocumentBuilder(this, null, baseUri, flags);
        }

        /// <summary>
        /// Create a writer that can be used to create nodes in this document.  The root node will be assigned "baseUri", and flags
        /// can be passed to indicate that names should be atomized by the builder and/or a fragment should be created.
        /// </summary>
        internal void LoadFromReader(XmlReader reader, XmlSpace space) {
            XPathDocumentBuilder builder;
            IXmlLineInfo lineInfo;
            string xmlnsUri;
            bool topLevelReader;
            int initialDepth;

            if (reader == null)
                throw new ArgumentNullException("reader");

            // Determine line number provider
            lineInfo = reader as IXmlLineInfo;
            if (lineInfo == null || !lineInfo.HasLineInfo())
                lineInfo = null;
            this.hasLineInfo = (lineInfo != null);

            this.nameTable = reader.NameTable;
            builder = new XPathDocumentBuilder(this, lineInfo, reader.BaseURI, LoadFlags.None);

            try {
                // Determine whether reader is in initial state
                topLevelReader = (reader.ReadState == ReadState.Initial);
                initialDepth = reader.Depth;

                // Get atomized xmlns uri
                Debug.Assert((object) this.nameTable.Get(string.Empty) == (object) string.Empty, "NameTable must contain atomized string.Empty");
                xmlnsUri = this.nameTable.Get(XmlReservedNs.NsXmlNs);

                // Read past Initial state; if there are no more events then load is complete
                if (topLevelReader && !reader.Read())
                    return;

                // Read all events
                do {
                    // If reader began in intermediate state, return when all siblings have been read
                    if (!topLevelReader && reader.Depth < initialDepth)
                        return;

                    switch (reader.NodeType) {
                        case XmlNodeType.Element: {
                            bool isEmptyElement = reader.IsEmptyElement;

                            builder.WriteStartElement(reader.Prefix, reader.LocalName, reader.NamespaceURI, reader.BaseURI);

                            // Add attribute and namespace nodes to element
                            while (reader.MoveToNextAttribute()) {
                                string namespaceUri = reader.NamespaceURI;

                                if ((object) namespaceUri == (object) xmlnsUri) {
                                    if (reader.Prefix.Length == 0) {
                                        // Default namespace declaration "xmlns"
                                        Debug.Assert(reader.LocalName == "xmlns");
                                        builder.WriteNamespaceDeclaration(string.Empty, reader.Value);
                                    }
                                    else {
                                        Debug.Assert(reader.Prefix == "xmlns");
                                        builder.WriteNamespaceDeclaration(reader.LocalName, reader.Value);
                                    }
                                }
                                else {
                                    builder.WriteStartAttribute(reader.Prefix, reader.LocalName, namespaceUri);
                                    builder.WriteString(reader.Value, TextBlockType.Text);
                                    builder.WriteEndAttribute();
                                }
                            }

                            if (isEmptyElement)
                                builder.WriteEndElement(true);
                            break;
                        }

                        case XmlNodeType.EndElement:
                            builder.WriteEndElement(false);
                            break;

                        case XmlNodeType.Text:
                        case XmlNodeType.CDATA:
                            builder.WriteString(reader.Value, TextBlockType.Text);
                            break;

                        case XmlNodeType.SignificantWhitespace:
                            if (reader.XmlSpace == XmlSpace.Preserve)
                                builder.WriteString(reader.Value, TextBlockType.SignificantWhitespace);
                            else
                                // Significant whitespace without xml:space="preserve" is not significant in XPath/XQuery data model
                                goto case XmlNodeType.Whitespace;
                            break;

                        case XmlNodeType.Whitespace:
                            // We intentionally ignore the reader.XmlSpace property here and blindly trust
                            //   the reported node type. If the reported information is not in sync
                            //   (in this case if the reader.XmlSpace == Preserve) then we make the choice
                            //   to trust the reported node type. Since we have no control over the input reader
                            //   we can't even assert here.

                            // Always filter top-level whitespace
                            if (space == XmlSpace.Preserve && (!topLevelReader || reader.Depth != 0))
                                builder.WriteString(reader.Value, TextBlockType.Whitespace);
                            break;

                        case XmlNodeType.Comment:
                            builder.WriteComment(reader.Value);
                            break;

                        case XmlNodeType.ProcessingInstruction:
                            builder.WriteProcessingInstruction(reader.LocalName, reader.Value, reader.BaseURI);
                            break;

                        case XmlNodeType.EntityReference:
                            reader.ResolveEntity();
                            break;

                        case XmlNodeType.DocumentType:
                            // Create ID tables
                            IDtdInfo info = reader.DtdInfo;
                            if (info != null)
                                builder.CreateIdTables(info);
                            break;

                        case XmlNodeType.EndEntity:
                        case XmlNodeType.None:
                        case XmlNodeType.XmlDeclaration:
                            break;
                    }
                }
                while (reader.Read());
            }
            finally {
                builder.Close();
            }
        }

        /// <summary>
        /// Create a navigator positioned on the root node of the document.
        /// </summary>
        public XPathNavigator CreateNavigator() {
            return new XPathDocumentNavigator(this.pageRoot, this.idxRoot, null, 0);
        }


        //-----------------------------------------------
        // Document Properties
        //-----------------------------------------------

        /// <summary>
        /// Return the name table used to atomize all name parts (local name, namespace uri, prefix).
        /// </summary>
        internal XmlNameTable NameTable {
            get { return this.nameTable; }
        }

        /// <summary>
        /// Return true if line number information is recorded in the cache.
        /// </summary>
        internal bool HasLineInfo {
            get { return this.hasLineInfo; }
        }

        /// <summary>
        /// Return the singleton collapsed text node associated with the document.  One physical text node
        /// represents each logical text node in the document that is the only content-typed child of its
        /// element parent.
        /// </summary>
        internal int GetCollapsedTextNode(out XPathNode[] pageText) {
            pageText = this.pageText;
            return this.idxText;
        }

        /// <summary>
        /// Set the page and index where the singleton collapsed text node is stored.
        /// </summary>
        internal void SetCollapsedTextNode(XPathNode[] pageText, int idxText) {
            this.pageText = pageText;
            this.idxText = idxText;
        }

        /// <summary>
        /// Return the root node of the document.  This may not be a node of type XPathNodeType.Root if this
        /// is a document fragment.
        /// </summary>
        internal int GetRootNode(out XPathNode[] pageRoot) {
            pageRoot = this.pageRoot;
            return this.idxRoot;
        }

        /// <summary>
        /// Set the page and index where the root node is stored.
        /// </summary>
        internal void SetRootNode(XPathNode[] pageRoot, int idxRoot) {
            this.pageRoot = pageRoot;
            this.idxRoot = idxRoot;
        }

        /// <summary>
        /// Every document has an implicit xmlns:xml namespace node.
        /// </summary>
        internal int GetXmlNamespaceNode(out XPathNode[] pageXmlNmsp) {
            pageXmlNmsp = this.pageXmlNmsp;
            return this.idxXmlNmsp;
        }

        /// <summary>
        /// Set the page and index where the implicit xmlns:xml node is stored.
        /// </summary>
        internal void SetXmlNamespaceNode(XPathNode[] pageXmlNmsp, int idxXmlNmsp) {
            this.pageXmlNmsp = pageXmlNmsp;
            this.idxXmlNmsp = idxXmlNmsp;
        }

        /// <summary>
        /// Associate a namespace node with an element.
        /// </summary>
        internal void AddNamespace(XPathNode[] pageElem, int idxElem, XPathNode[] pageNmsp, int idxNmsp) {
            Debug.Assert(pageElem[idxElem].NodeType == XPathNodeType.Element && pageNmsp[idxNmsp].NodeType == XPathNodeType.Namespace);

            if (this.mapNmsp == null)
                this.mapNmsp = new Dictionary<XPathNodeRef, XPathNodeRef>();

            this.mapNmsp.Add(new XPathNodeRef(pageElem, idxElem), new XPathNodeRef(pageNmsp, idxNmsp));
        }

        /// <summary>
        /// Lookup the namespace nodes associated with an element.
        /// </summary>
        internal int LookupNamespaces(XPathNode[] pageElem, int idxElem, out XPathNode[] pageNmsp) {
            XPathNodeRef nodeRef = new XPathNodeRef(pageElem, idxElem);
            Debug.Assert(pageElem[idxElem].NodeType == XPathNodeType.Element);

            // Check whether this element has any local namespaces
            if (this.mapNmsp == null || !this.mapNmsp.ContainsKey(nodeRef)) {
                pageNmsp = null;
                return 0;
            }

            // Yes, so return the page and index of the first local namespace node
            nodeRef = this.mapNmsp[nodeRef];

            pageNmsp = nodeRef.Page;
            return nodeRef.Index;
        }

        /// <summary>
        /// Add an element indexed by ID value.
        /// </summary>
        internal void AddIdElement(string id, XPathNode[] pageElem, int idxElem) {
            if (this.idValueMap == null)
                this.idValueMap = new Dictionary<string, XPathNodeRef>();

            if (!this.idValueMap.ContainsKey(id))
                this.idValueMap.Add(id, new XPathNodeRef(pageElem, idxElem));
        }

        /// <summary>
        /// Lookup the element node associated with the specified ID value.
        /// </summary>
        internal int LookupIdElement(string id, out XPathNode[] pageElem) {
            XPathNodeRef nodeRef;

            if (this.idValueMap == null || !this.idValueMap.ContainsKey(id)) {
                pageElem = null;
                return 0;
            }

            // Extract page and index from XPathNodeRef
            nodeRef = this.idValueMap[id];
            pageElem = nodeRef.Page;
            return nodeRef.Index;
        }


        //-----------------------------------------------
        // Helper Methods
        //-----------------------------------------------

        /// <summary>
        /// Set properties on the reader so that it is backwards-compatible with V1.
        /// </summary>
        private XmlTextReaderImpl SetupReader(XmlTextReaderImpl reader) {
            reader.EntityHandling = EntityHandling.ExpandEntities;
            reader.XmlValidatingReaderCompatibilityMode = true;
            return reader;
        }
    }
}