mcs/class/corlib/Mono.Globalization.Unicode/create-collation-element-table.cs

   1 //
   2 // create-collation-element-table.cs :
   3 //      supports creation of CollationElementTable.cs
   4 //
   5 // Author:
   6 //      Atsushi Enomoto  <atsushi@ximian.com>
   7 //
   8 // Copyright (C) 2005 Novell, Inc (http://www.novell.com)
   9 //
  10 // Permission is hereby granted, free of charge, to any person obtaining
  11 // a copy of this software and associated documentation files (the
  12 // "Software"), to deal in the Software without restriction, including
  13 // without limitation the rights to use, copy, modify, merge, publish,
  14 // distribute, sublicense, and/or sell copies of the Software, and to
  15 // permit persons to whom the Software is furnished to do so, subject to
  16 // the following conditions:
  17 //
  18 // The above copyright notice and this permission notice shall be
  19 // included in all copies or substantial portions of the Software.
  20 //
  21 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  22 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  23 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  24 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
  25 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
  26 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
  27 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  28 //
  29
  30 using System;
  31 using System.Globalization;
  32 using System.IO;
  33
  34 namespace Mono.Globalization.Unicode
  35 {
  36         internal class CollationElementTableCodeGenerator
  37         {
  38                 private int lineCount = 0;
  39
  40                 int keyCount = 1; // 0 indicates that there is no matching entry
  41                 int [] collElem = new int [CollationElementTableUtil.Indexer.TotalCount];
  42                 SortKeyValue [] keyValues = new SortKeyValue [32768];
  43
  44                 public static void Main ()
  45                 {
  46                         new CollationElementTableCodeGenerator ().Run ();
  47                 }
  48
  49                 public void Run ()
  50                 {
  51                         try {
  52                                 Parse ();
  53                                 Serialize ();
  54                         } catch (Exception) {
  55                                 Console.Error.WriteLine ("Internal error at line " + lineCount);
  56                                 throw;
  57                         }
  58                         /*
  59                         int lastQ = -1;
  60                         for (int x = 1; x < CollationElementTableUtil.Indexer.TotalCount; x++) {
  61                                 int i = CollationElementTableUtil.Indexer.GetIndexForCodePoint (x);
  62                                 int q = keyValues [collElem [i]].Quarternary;
  63 //                              if (i != q && q != 0)
  64 //                                      Console.Error.WriteLine ("differs : {0} -> {1}", i, q);
  65                                 if (q != 0) {
  66                                         if (lastQ >= q)
  67                                                 Console.Error.WriteLine ("latter was smaller at {2} : {0} / {1}", lastQ, q, i);
  68                                         lastQ = q;
  69                                 }
  70                         }
  71                         */
  72                 }
  73
  74                 private void Serialize ()
  75                 {
  76                         Console.WriteLine ("static readonly int [] collElem = new int [] {");
  77                         DumpArray (collElem, CollationElementTableUtil.Indexer.TotalCount, true);
  78                         Console.WriteLine ("};");
  79                         Console.WriteLine ("static readonly SortKeyValue [] keyValues = new SortKeyValue [] {");
  80                         for (int i = 0; i < keyCount; i++) {
  81                                 SortKeyValue s = keyValues [i];
  82                                 Console.WriteLine ("    new SortKeyValue ({0}, 0x{1:X04}, 0x{2:X04}, 0x{3:X04}, 0x{4:X04}),",
  83                                         s.Alt ? "true" : "false", s.Primary, s.Secondary,
  84                                         s.Thirtiary, s.Quarternary);
  85                         }
  86                         Console.WriteLine ("};");
  87                 }
  88
  89                 private void DumpArray (int [] array, int count, bool getCP)
  90                 {
  91                         if (array.Length < count)
  92                                 throw new ArgumentOutOfRangeException ("count");
  93                         for (int i = 0; i < count; i++) {
  94                                 if (array [i] == 0)
  95                                         Console.Write ("0, ");
  96                                 else
  97                                         Console.Write ("0x{0:X}, ", array [i]);
  98                                 if (i % 16 == 15) {
  99                                         int l = getCP ? CollationElementTableUtil.Indexer.ToCodePoint (i) : i;
 100                                         Console.WriteLine ("// {0:X04}-{1:X04}", l - 15, l);
 101                                 }
 102                         }
 103                 }
 104
 105                 private void Parse ()
 106                 {
 107                         ushort [] v = new ushort [4];
 108
 109                         TextReader reader = Console.In;
 110                         while (reader.Peek () != -1) {
 111                                 string line = reader.ReadLine ();
 112                                 lineCount++;
 113                                 if (line.StartsWith ("@"))
 114                                         continue; // @version, @variable etc.
 115                                 int idx = line.IndexOf ('#');
 116                                 if (idx >= 0)
 117                                         line = line.Substring (0, idx);
 118                                 if (line.Length == 0)
 119                                         continue;
 120
 121                                 int cp = int.Parse (line.Substring (0, 5), NumberStyles.HexNumber);
 122                                 int collElemIdx = CollationElementTableUtil.Indexer.ToIndex (cp);
 123                                 if (collElemIdx < 0) {
 124                                         Console.Error.WriteLine ("WARNING: handle character {0:x} in collation element table.", cp);
 125                                         continue;
 126                                 }
 127
 128                                 line = line.Substring (line.IndexOf (';') + 1).Trim ();
 129                                 // count entries in a line
 130                                 int entryPerLine = 0;
 131                                 for (int e = 0; (e = line.IndexOf ('[', e) + 1) > 0;)
 132                                         entryPerLine++;
 133
 134                                 int start = 0;
 135                                 for (int e = 0; e < entryPerLine; e++) {
 136                                         start = line.IndexOf ('[', start) + 1;
 137                                         string s = line.Substring (start, line.IndexOf (']', start) - start);
 138
 139                                         bool alt = false;
 140                                         if (s [0] == '*')
 141                                                 alt = true;
 142                                         string [] vslist = s.Substring (1).Split ('.');
 143                                         bool skip = false;
 144                                         for (int i = 0; i < 4; i++) {
 145                                                 if (vslist [i].Length > 4)
 146                                                         skip = true;
 147                                                 else
 148                                                         v [i] = ushort.Parse (vslist [i], NumberStyles.HexNumber);
 149                                         }
 150                                         if (skip) {
 151 //                                              Console.Error.WriteLine ("WARNING: skipped entry {0:X}", cp);
 152                                                 continue;
 153                                         }
 154                                         idx = keyCount;
 155                                         if (entryPerLine == 1) {
 156                                                 // idx = 0 means "no matching entry", so here we start from 1
 157                                                 for (idx = 1; idx < keyCount; idx++) {
 158                                                         SortKeyValue k = keyValues [idx];
 159                                                         if (k.Alt == alt &&
 160                                                                 k.Primary == v [0] &&
 161                                                                 k.Secondary == v [1] &&
 162                                                                 k.Thirtiary == v [2] &&
 163                                                                 k.Quarternary == v [3])
 164                                                                 break;
 165                                                 }
 166                                         }
 167                                         if (idx == keyCount)
 168                                                 AddEntry (alt, v);
 169                                 }
 170                                 if (entryPerLine == 1)
 171                                         collElem [collElemIdx] = idx;
 172                                 else
 173                                         collElem [collElemIdx] =
 174                                                 (short) (keyCount - entryPerLine)
 175                                                 + (short) (entryPerLine << 16);
 176                         }
 177                         reader.Close ();
 178                 }
 179
 180                 private void AddEntry (bool alt, ushort [] v)
 181                 {
 182                         if (keyCount == keyValues.Length) {
 183                                 SortKeyValue [] tmp = new SortKeyValue [keyCount * 2];
 184                                 Array.Copy (keyValues, tmp, keyCount);
 185                                 keyValues = tmp;
 186                         }
 187                         keyValues [keyCount] =
 188                                 new SortKeyValue (alt,
 189                                 v [0], (byte) v [1], (byte) v [2], v [3]);
 190                         keyCount++;
 191                 }
 192         }
 193 }
 194