src/mscorlib/src/System/Text/EncodingTable.Unix.cs

   1 // Licensed to the .NET Foundation under one or more agreements.
   2 // The .NET Foundation licenses this file to you under the MIT license.
   3 // See the LICENSE file in the project root for more information.
   4
   5 using System.Collections.Generic;
   6 using System.Diagnostics;
   7 using System.Globalization;
   8 using System.Text;
   9
  10 namespace System.Text
  11 {
  12     internal static class EncodingTable
  13     {
  14         // Return a list of all EncodingInfo objects describing all of our encodings
  15         internal static EncodingInfo[] GetEncodings()
  16         {
  17             EncodingInfo[] arrayEncodingInfo = new EncodingInfo[s_encodingDataTableItems.Length];
  18
  19             for (int i = 0; i < s_encodingDataTableItems.Length; i++)
  20             {
  21                 CodePageDataItem dataItem = s_encodingDataTableItems[i];
  22
  23                 arrayEncodingInfo[i] = new EncodingInfo(dataItem.CodePage, dataItem.WebName,
  24                     SR.GetResourceString(dataItem.DisplayNameResourceKey));
  25             }
  26
  27             return arrayEncodingInfo;
  28         }
  29
  30         internal static int GetCodePageFromName(string name)
  31         {
  32             if (name == null)
  33             {
  34                 throw new ArgumentNullException(nameof(name));
  35             }
  36
  37             ushort codePage;
  38             if (s_encodingDataTable.TryGetValue(name, out codePage))
  39             {
  40                 return codePage;
  41             }
  42
  43             // The encoding name is not valid.
  44             throw new ArgumentException(
  45                 string.Format(
  46                     CultureInfo.CurrentCulture,
  47                     SR.Argument_EncodingNotSupported, name), nameof(name));
  48         }
  49
  50         internal static CodePageDataItem GetCodePageDataItem(int codepage)
  51         {
  52             CodePageDataItem item;
  53
  54             switch (codepage)
  55             {
  56                 case 1200:
  57                     item = s_encodingDataTableItems[0];
  58                     break;
  59                 case 1201:
  60                     item = s_encodingDataTableItems[1];
  61                     break;
  62                 case 12000:
  63                     item = s_encodingDataTableItems[2];
  64                     break;
  65                 case 12001:
  66                     item = s_encodingDataTableItems[3];
  67                     break;
  68                 case 20127:
  69                     item = s_encodingDataTableItems[4];
  70                     break;
  71                 case 28591:
  72                     item = s_encodingDataTableItems[5];
  73                     break;
  74                 case 65000:
  75                     item = s_encodingDataTableItems[6];
  76                     break;
  77                 case 65001:
  78                     item = s_encodingDataTableItems[7];
  79                     break;
  80                 default:
  81                     item = null;
  82                     break;
  83             }
  84
  85             Debug.Assert(item == null || item.CodePage == codepage, "item.CodePage needs to equal the specified codepage");
  86             return item;
  87         }
  88
  89         // PAL ends here.
  90
  91 #if DEBUG
  92         static EncodingTable()
  93         {
  94             Debug.Assert(
  95                 s_encodingDataTable.Count == EncodingTableCapacity,
  96                 string.Format(CultureInfo.InvariantCulture,
  97                     "EncodingTable s_encodingDataTable's initial capacity (EncodingTableCapacity) is incorrect.{0}Expected (s_encodingDataTable.Count): {1}, Actual (EncodingTableCapacity): {2}",
  98                     Environment.NewLine,
  99                     s_encodingDataTable.Count,
 100                     EncodingTableCapacity));
 101         }
 102 #endif
 103
 104         // NOTE: the following two lists were taken from ~\src\classlibnative\nls\encodingdata.cpp
 105         // and should be kept in sync with those lists
 106
 107         private const int EncodingTableCapacity = 42;
 108         private readonly static Dictionary<string, ushort> s_encodingDataTable =
 109             new Dictionary<string, ushort>(EncodingTableCapacity, StringComparer.OrdinalIgnoreCase)
 110         {
 111             { "ANSI_X3.4-1968", 20127 },
 112             { "ANSI_X3.4-1986", 20127 },
 113             { "ascii", 20127 },
 114             { "cp367", 20127 },
 115             { "cp819", 28591 },
 116             { "csASCII", 20127 },
 117             { "csISOLatin1", 28591 },
 118             { "csUnicode11UTF7", 65000 },
 119             { "IBM367", 20127 },
 120             { "ibm819", 28591 },
 121             { "ISO-10646-UCS-2", 1200 },
 122             { "iso-8859-1", 28591 },
 123             { "iso-ir-100", 28591 },
 124             { "iso-ir-6", 20127 },
 125             { "ISO646-US", 20127 },
 126             { "iso8859-1", 28591 },
 127             { "ISO_646.irv:1991", 20127 },
 128             { "iso_8859-1", 28591 },
 129             { "iso_8859-1:1987", 28591 },
 130             { "l1", 28591 },
 131             { "latin1", 28591 },
 132             { "ucs-2", 1200 },
 133             { "unicode", 1200},
 134             { "unicode-1-1-utf-7", 65000 },
 135             { "unicode-1-1-utf-8", 65001 },
 136             { "unicode-2-0-utf-7", 65000 },
 137             { "unicode-2-0-utf-8", 65001 },
 138             // People get confused about the FFFE here.  We can't change this because it'd break existing apps.
 139             // This has been this way for a long time, including in Mlang.
 140             // Big Endian, BOM seems backwards, think of the BOM in little endian order.
 141             { "unicodeFFFE", 1201},
 142             { "us", 20127 },
 143             { "us-ascii", 20127 },
 144             { "utf-16", 1200 },
 145             { "UTF-16BE", 1201},
 146             { "UTF-16LE", 1200},
 147             { "utf-32", 12000 },
 148             { "UTF-32BE", 12001 },
 149             { "UTF-32LE", 12000 },
 150             { "utf-7", 65000 },
 151             { "utf-8", 65001 },
 152             { "x-unicode-1-1-utf-7", 65000 },
 153             { "x-unicode-1-1-utf-8", 65001 },
 154             { "x-unicode-2-0-utf-7", 65000 },
 155             { "x-unicode-2-0-utf-8", 65001 },
 156         };
 157
 158         // redeclaring these constants here for readability below
 159         private const uint MIMECONTF_MAILNEWS = Encoding.MIMECONTF_MAILNEWS;
 160         private const uint MIMECONTF_BROWSER = Encoding.MIMECONTF_BROWSER;
 161         private const uint MIMECONTF_SAVABLE_MAILNEWS = Encoding.MIMECONTF_SAVABLE_MAILNEWS;
 162         private const uint MIMECONTF_SAVABLE_BROWSER = Encoding.MIMECONTF_SAVABLE_BROWSER;
 163
 164         // keep this array sorted by code page, so the order is consistent for GetEncodings()
 165         // Remember to update GetCodePageDataItem() if this list is updated
 166         private readonly static CodePageDataItem[] s_encodingDataTableItems = new[]
 167         {
 168             new CodePageDataItem(1200, 1200, "utf-16", MIMECONTF_SAVABLE_BROWSER), // "Unicode"
 169             new CodePageDataItem(1201, 1200, "utf-16BE", 0), // Big Endian, old FFFE BOM seems backwards, think of the BOM in little endian order.
 170             new CodePageDataItem(12000, 1200, "utf-32", 0), // "Unicode (UTF-32)"
 171             new CodePageDataItem(12001, 1200, "utf-32BE", 0), // "Unicode (UTF-32 Big Endian)"
 172             new CodePageDataItem(20127, 1252, "us-ascii", MIMECONTF_MAILNEWS | MIMECONTF_SAVABLE_MAILNEWS), // "US-ASCII"
 173             new CodePageDataItem(28591, 1252, "iso-8859-1", MIMECONTF_MAILNEWS | MIMECONTF_BROWSER | MIMECONTF_SAVABLE_MAILNEWS | MIMECONTF_SAVABLE_BROWSER), // "Western European (ISO)"
 174             new CodePageDataItem(65000, 1200, "utf-7", MIMECONTF_MAILNEWS | MIMECONTF_SAVABLE_MAILNEWS), // "Unicode (UTF-7)"
 175             new CodePageDataItem(65001, 1200, "utf-8", MIMECONTF_MAILNEWS | MIMECONTF_BROWSER | MIMECONTF_SAVABLE_MAILNEWS | MIMECONTF_SAVABLE_BROWSER), // "Unicode (UTF-8)"
 176         };
 177     }
 178 }