mcs/class/referencesource/System.Xml/System/Xml/Core/CharEntityEncoderFallback.cs

   1 //------------------------------------------------------------------------------
   2 // <copyright file="CharEntitiesEncodingFallback.cs" company="Microsoft">
   3 //     Copyright (c) Microsoft Corporation.  All rights reserved.
   4 // </copyright>
   5 // <owner current="true" primary="true">Microsoft</owner>
   6 //------------------------------------------------------------------------------
   7
   8 using System.Text;
   9 using System.Diagnostics;
  10 using System.Globalization;
  11
  12 namespace System.Xml {
  13
  14 //
  15 // CharEntityEncoderFallback
  16 //
  17
  18     internal class CharEntityEncoderFallback : EncoderFallback {
  19         private CharEntityEncoderFallbackBuffer fallbackBuffer;
  20
  21         private int[]   textContentMarks;
  22         private int     endMarkPos;
  23         private int     curMarkPos;
  24         private int     startOffset;
  25
  26         internal CharEntityEncoderFallback() {
  27         }
  28
  29         public override EncoderFallbackBuffer CreateFallbackBuffer() {
  30             if ( fallbackBuffer == null ) {
  31                 fallbackBuffer = new CharEntityEncoderFallbackBuffer( this );
  32             }
  33             return fallbackBuffer;
  34         }
  35
  36         public override int MaxCharCount {
  37             get {
  38                 return 12;
  39             }
  40         }
  41
  42         internal int StartOffset {
  43             get {
  44                 return startOffset;
  45             }
  46             set {
  47                 startOffset = value;
  48             }
  49         }
  50
  51         internal void Reset( int[] textContentMarks, int endMarkPos ) {
  52             this.textContentMarks = textContentMarks;
  53             this.endMarkPos = endMarkPos;
  54             curMarkPos = 0;
  55         }
  56
  57         internal bool CanReplaceAt( int index ) {
  58             int mPos = curMarkPos;
  59             int charPos = startOffset + index;
  60             while ( mPos < endMarkPos && charPos >= textContentMarks[mPos+1] ) {
  61                 mPos++;
  62             }
  63             curMarkPos = mPos;
  64
  65             return (mPos & 1) != 0;
  66         }
  67     }
  68
  69 //
  70 // CharEntityFallbackBuffer
  71 //
  72     internal class CharEntityEncoderFallbackBuffer : EncoderFallbackBuffer {
  73         private CharEntityEncoderFallback parent;
  74
  75         private string  charEntity = string.Empty;
  76         private int     charEntityIndex = -1;
  77
  78         internal CharEntityEncoderFallbackBuffer( CharEntityEncoderFallback parent ) {
  79             this.parent = parent;
  80         }
  81
  82         public override bool Fallback( char charUnknown, int index ) {
  83             // If we are already in fallback, throw, it's probably at the suspect character in charEntity
  84             if ( charEntityIndex >= 0 ) {
  85                 (new EncoderExceptionFallback()).CreateFallbackBuffer().Fallback( charUnknown, index );
  86             }
  87
  88             // find out if we can replace the character with entity
  89             if ( parent.CanReplaceAt( index ) ) {
  90                 // Create the replacement character entity
  91                 charEntity = string.Format( CultureInfo.InvariantCulture, "&#x{0:X};", new object[] { (int)charUnknown } );
  92                 charEntityIndex = 0;
  93                 return true;
  94             }
  95             else {
  96                 EncoderFallbackBuffer errorFallbackBuffer = ( new EncoderExceptionFallback() ).CreateFallbackBuffer();
  97                 errorFallbackBuffer.Fallback( charUnknown, index );
  98                 return false;
  99             }
 100         }
 101
 102         public override bool Fallback( char charUnknownHigh, char charUnknownLow, int index ) {
 103             // check input surrogate pair
 104             if ( !char.IsSurrogatePair( charUnknownHigh, charUnknownLow ) ) {
 105                 throw XmlConvert.CreateInvalidSurrogatePairException( charUnknownHigh, charUnknownLow );
 106             }
 107
 108             // If we are already in fallback, throw, it's probably at the suspect character in charEntity
 109             if ( charEntityIndex >= 0 ) {
 110                 (new EncoderExceptionFallback()).CreateFallbackBuffer().Fallback( charUnknownHigh, charUnknownLow, index );
 111             }
 112
 113             if ( parent.CanReplaceAt( index ) ) {
 114                 // Create the replacement character entity
 115                 charEntity = string.Format( CultureInfo.InvariantCulture, "&#x{0:X};", new object[] { SurrogateCharToUtf32( charUnknownHigh, charUnknownLow ) } );
 116                 charEntityIndex = 0;
 117                 return true;
 118             }
 119             else {
 120                 EncoderFallbackBuffer errorFallbackBuffer = ( new EncoderExceptionFallback() ).CreateFallbackBuffer();
 121                 errorFallbackBuffer.Fallback( charUnknownHigh, charUnknownLow, index );
 122                 return false;
 123             }
 124         }
 125
 126         public override char GetNextChar() {
 127             // Bug fix: 35637. The protocol using GetNextChar() and MovePrevious() called by Encoder is not well documented.
 128             // Here we have to to signal to Encoder that the previous read was last character. Only AFTER we can
 129             // mark ourself as done (-1). Otherwise MovePrevious() can still be called, but -1 is already incorrectly set
 130             // and return false from MovePrevious(). Then Encoder ----ing the rest of the bytes.
 131             if (charEntityIndex == charEntity.Length)
 132             {
 133                 charEntityIndex = -1;
 134             }
 135             if ( charEntityIndex == -1 ) {
 136                 return (char)0;
 137             }
 138             else {
 139                 Debug.Assert( charEntityIndex < charEntity.Length );
 140                 char ch = charEntity[charEntityIndex++];
 141                 return ch;
 142             }
 143         }
 144
 145         public override bool MovePrevious() {
 146             if ( charEntityIndex == -1 ) {
 147                 return false;
 148             }
 149             else {
 150                 // Could be == length if just read the last character
 151                 Debug.Assert(charEntityIndex <= charEntity.Length);
 152                 if (charEntityIndex > 0)
 153                 {
 154                     charEntityIndex--;
 155                     return true;
 156                 }
 157                 else {
 158                     return false;
 159                 }
 160             }
 161         }
 162
 163
 164         public override int Remaining {
 165             get {
 166                 if ( charEntityIndex == -1 ) {
 167                     return 0;
 168                 }
 169                 else {
 170                     return charEntity.Length - charEntityIndex;
 171                 }
 172             }
 173         }
 174
 175         public override void Reset() {
 176             charEntityIndex = -1;
 177         }
 178
 179         private int SurrogateCharToUtf32(char highSurrogate, char lowSurrogate) {
 180             return XmlCharType.CombineSurrogateChar(lowSurrogate, highSurrogate);
 181         }
 182     }
 183 }