Code:
/ 4.0 / 4.0 / DEVDIV_TFS / Dev10 / Releases / RTMRel / ndp / fx / src / Services / Web / System / Web / Services / Discovery / LinkGrep.cs / 1305376 / LinkGrep.cs
//------------------------------------------------------------------------------ //// Copyright (c) Microsoft Corporation. All rights reserved. // //----------------------------------------------------------------------------- namespace System.Web.Services.Discovery { using System; using System.IO; using System.Text.RegularExpressions; using System.Web.Services.Protocols; using System.Globalization; internal class LinkGrep { private LinkGrep() {} private static string ReadEntireStream(TextReader input) { char[] buffer = new char[4096]; int bufferpos = 0; int delta; // for (;;) { delta = input.Read(buffer, bufferpos, buffer.Length - bufferpos); if (delta == 0) break; bufferpos += delta; if (bufferpos == buffer.Length) { char[] newbuf = new char[buffer.Length * 2]; System.Array.Copy(buffer, 0, newbuf, 0, buffer.Length); buffer = newbuf; } } return new string(buffer, 0, bufferpos); } internal static string SearchForLink(Stream stream) { string text = null; text = ReadEntireStream(new StreamReader(stream)); int textpos = 0; Match match; if ((match = doctypeDirectiveRegex.Match(text, textpos)).Success) { textpos += match.Length; } bool oneMatch; for (;;) { // Reset match flag oneMatch = false; // 1: scan for text up to the next tag. // First case: check for whitespace going all the way to the next tag if ((match = whitespaceRegex.Match(text, textpos)).Success) { oneMatch = true; } // Second case: there may be some nonwhitespace; scan it else if ((match = textRegex.Match(text, textpos)).Success) { oneMatch = true; } // we might be done now textpos += match.Length; if (textpos == text.Length) break; // 2: handle constructs that start with < // First, check to see if it's a tag if ((match = tagRegex.Match(text, textpos)).Success) { oneMatch = true; string tag = match.Groups["tagname"].Value; if (String.Compare(tag, "link", StringComparison.OrdinalIgnoreCase) == 0) { CaptureCollection attrnames = match.Groups["attrname"].Captures; CaptureCollection attrvalues = match.Groups["attrval"].Captures; int count = attrnames.Count; bool rightType = false; bool rightRel = false; string href = null; for (int i = 0; i < count; i++) { string attrName = attrnames[i].ToString(); string attrValue = attrvalues[i].ToString(); if (String.Compare(attrName, "type", StringComparison.OrdinalIgnoreCase) == 0 && ContentType.MatchesBase(attrValue, ContentType.TextXml)) { rightType = true; } else if (String.Compare(attrName, "rel", StringComparison.OrdinalIgnoreCase) == 0 && String.Compare(attrValue, "alternate", StringComparison.OrdinalIgnoreCase) == 0) { rightRel = true; } else if (String.Compare(attrName, "href", StringComparison.OrdinalIgnoreCase) == 0) { href = attrValue; } if (rightType && rightRel && href != null) { // Got a link to a disco file! return href; } } } else if (tag == "body") { // If body begins, get out, since link tags should only be defined in the head break; } } // check to see if it's an end tag else if ((match = endtagRegex.Match(text, textpos)).Success) { oneMatch = true; } // check to see if it's a comment else if ((match = commentRegex.Match(text, textpos)).Success) { oneMatch = true; } // we might be done now textpos += match.Length; if (textpos == text.Length) break; // If we couldn't get one single match, it means that it's probably not HTML, so bail if (!oneMatch) { break; } } return null; } private readonly static Regex tagRegex = new Regex ( "\\G<" + // leading < "(?[\\w:.-]+(?=:)|):?" + // optional prefix: "(? [\\w.-]+)" + // tagname "(?:\\s+" + // zero or more attributes "(? [\\w:.-]+(?=:)|):?" + // optional attrprefix: "(? [\\w.-]+)" + // attrname "\\s*=\\s*" + // required equals "(?:" + // quoted value "\"(? [^\"]*)\"" + // double quoted attrval "|'(? [^\']*)'" + // single quoted attrval "|(? [a-zA-Z0-9\\-._:]+)" + // attrval with no quotes (SGML-approved chars) ")" + // end quoted value ")*" + // end attribute "\\s*(? /)?>" // optional trailing /, and trailing > ); private readonly static Regex doctypeDirectiveRegex = new Regex ( @"\G", RegexOptions.IgnoreCase | RegexOptions.Multiline | RegexOptions.IgnorePatternWhitespace ); private readonly static Regex endtagRegex = new Regex ( "\\G" + // leading "(? [\\w:-]+(?=:)|):?" + // optional prefix: "(? [\\w-]+)\\s*>" // tagname ); private readonly static Regex commentRegex = new Regex ( "\\G
- DbProviderFactory.cs
- SmtpTransport.cs
- HttpClientCertificate.cs
- HtmlInputSubmit.cs
- AttributeInfo.cs
- TextDecorationLocationValidation.cs
- MaskedTextBox.cs
- DbProviderFactoriesConfigurationHandler.cs
- LineUtil.cs
- AsynchronousChannel.cs
- PeekCompletedEventArgs.cs
- FileEnumerator.cs
- Mouse.cs
- GradientStop.cs
- SystemKeyConverter.cs
- MgmtResManager.cs
- CodeArgumentReferenceExpression.cs
- XmlElementList.cs
- XmlNamespaceManager.cs
- RuntimeWrappedException.cs
- HttpCacheVary.cs
- XmlChildNodes.cs
- CodeSnippetCompileUnit.cs
- Rss20ItemFormatter.cs
- C14NUtil.cs
- CacheForPrimitiveTypes.cs
- TextFormatterImp.cs
- ObjectStateEntry.cs
- DefinitionBase.cs
- NumberAction.cs
- SpecularMaterial.cs
- GeneratedView.cs
- PerformanceCounterPermission.cs
- BaseTemplateBuildProvider.cs
- XPathAncestorQuery.cs
- xmlfixedPageInfo.cs
- PhysicalAddress.cs
- XmlHierarchicalEnumerable.cs
- SqlConnectionFactory.cs
- ConfigurationStrings.cs
- arclist.cs
- NaturalLanguageHyphenator.cs
- BindingRestrictions.cs
- ValueType.cs
- OptimizedTemplateContentHelper.cs
- LeftCellWrapper.cs
- TypeConverterHelper.cs
- ObjectPersistData.cs
- Track.cs
- StreamInfo.cs
- SafeProcessHandle.cs
- UdpDiscoveryEndpoint.cs
- TemplateInstanceAttribute.cs
- OleDbRowUpdatingEvent.cs
- SimpleMailWebEventProvider.cs
- PeerName.cs
- ColumnClickEvent.cs
- RemoteWebConfigurationHostServer.cs
- BookmarkCallbackWrapper.cs
- DataGridViewComboBoxEditingControl.cs
- DoubleAnimationBase.cs
- COM2ComponentEditor.cs
- NotConverter.cs
- StateWorkerRequest.cs
- FontWeights.cs
- DataGridTableStyleMappingNameEditor.cs
- NewExpression.cs
- UiaCoreTypesApi.cs
- Binding.cs
- MsmqInputMessage.cs
- WeakReferenceEnumerator.cs
- DesignerSerializerAttribute.cs
- Monitor.cs
- Atom10ItemFormatter.cs
- Rights.cs
- ReachIDocumentPaginatorSerializerAsync.cs
- DictionaryBase.cs
- EncryptedData.cs
- controlskin.cs
- FileCodeGroup.cs
- DoubleLinkListEnumerator.cs
- BitmapImage.cs
- TagMapCollection.cs
- ServiceSettingsResponseInfo.cs
- GPPOINTF.cs
- ToolboxItem.cs
- PtsHelper.cs
- SiteMapHierarchicalDataSourceView.cs
- CompilerError.cs
- DataObject.cs
- ScriptRegistrationManager.cs
- WebReferenceOptions.cs
- ToolStripEditorManager.cs
- ISAPIRuntime.cs
- PointCollection.cs
- ProfileGroupSettings.cs
- XmlDataSourceView.cs
- AdRotator.cs
- NetworkStream.cs
- SoapCodeExporter.cs