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

   1 //
   2 //
   3 // CollationbElementTableCodeGenerator.cs
   4 //
   5 // Author:
   6 //      Atsushi Enomoto  <atsushi@ximian.com>
   7 //
   8 // Copyright 2005 Novell, Inc
   9 //
  10 // It supports creation of CollationElementTable.cs
  11 //
  12
  13 using System;
  14 using System.Globalization;
  15 using System.IO;
  16
  17 namespace Mono.Globalization.Unicode
  18 {
  19         internal class CollationElementTableCodeGenerator
  20         {
  21                 private int lineCount = 0;
  22
  23                 int keyCount = 1; // 0 indicates that there is no matching entry
  24                 int [] collElem = new int [CollationElementTableUtil.Indexer.TotalCount];
  25                 SortKeyValue [] keyValues = new SortKeyValue [32768];
  26
  27                 public static void Main ()
  28                 {
  29                         new CollationElementTableCodeGenerator ().Run ();
  30                 }
  31
  32                 public void Run ()
  33                 {
  34                         try {
  35                                 Parse ();
  36                                 Serialize ();
  37                         } catch (Exception) {
  38                                 Console.Error.WriteLine ("Internal error at line " + lineCount);
  39                                 throw;
  40                         }
  41                         /*
  42                         int lastQ = -1;
  43                         for (int x = 1; x < CollationElementTableUtil.Indexer.TotalCount; x++) {
  44                                 int i = CollationElementTableUtil.Indexer.GetIndexForCodePoint (x);
  45                                 int q = keyValues [collElem [i]].Quarternary;
  46 //                              if (i != q && q != 0)
  47 //                                      Console.Error.WriteLine ("differs : {0} -> {1}", i, q);
  48                                 if (q != 0) {
  49                                         if (lastQ >= q)
  50                                                 Console.Error.WriteLine ("latter was smaller at {2} : {0} / {1}", lastQ, q, i);
  51                                         lastQ = q;
  52                                 }
  53                         }
  54                         */
  55                 }
  56
  57                 private void Serialize ()
  58                 {
  59                         Console.WriteLine ("static readonly int [] collElem = new int [] {");
  60                         DumpArray (collElem, CollationElementTableUtil.Indexer.TotalCount, true);
  61                         Console.WriteLine ("};");
  62                         Console.WriteLine ("static readonly SortKeyValue [] keyValues = new SortKeyValue [] {");
  63                         for (int i = 0; i < keyCount; i++) {
  64                                 SortKeyValue s = keyValues [i];
  65                                 Console.WriteLine ("    new SortKeyValue ({0}, 0x{1:X04}, 0x{2:X04}, 0x{3:X04}, 0x{4:X04}),",
  66                                         s.Alt ? "true" : "false", s.Primary, s.Secondary,
  67                                         s.Thirtiary, s.Quarternary);
  68                         }
  69                         Console.WriteLine ("};");
  70                 }
  71
  72                 private void DumpArray (int [] array, int count, bool getCP)
  73                 {
  74                         if (array.Length < count)
  75                                 throw new ArgumentOutOfRangeException ("count");
  76                         for (int i = 0; i < count; i++) {
  77                                 if (array [i] == 0)
  78                                         Console.Write ("0, ");
  79                                 else
  80                                         Console.Write ("0x{0:X}, ", array [i]);
  81                                 if (i % 16 == 15) {
  82                                         int l = getCP ? CollationElementTableUtil.Indexer.ToCodePoint (i) : i;
  83                                         Console.WriteLine ("// {0:X04}-{1:X04}", l - 15, l);
  84                                 }
  85                         }
  86                 }
  87
  88                 private void Parse ()
  89                 {
  90                         ushort [] v = new ushort [4];
  91
  92                         TextReader reader = Console.In;
  93                         while (reader.Peek () != -1) {
  94                                 string line = reader.ReadLine ();
  95                                 lineCount++;
  96                                 if (line.StartsWith ("@"))
  97                                         continue; // @version, @variable etc.
  98                                 int idx = line.IndexOf ('#');
  99                                 if (idx >= 0)
 100                                         line = line.Substring (0, idx);
 101                                 if (line.Length == 0)
 102                                         continue;
 103
 104                                 int cp = int.Parse (line.Substring (0, 5), NumberStyles.HexNumber);
 105                                 int collElemIdx = CollationElementTableUtil.Indexer.ToIndex (cp);
 106                                 if (collElemIdx < 0) {
 107                                         Console.Error.WriteLine ("WARNING: handle character {0:x} in collation element table.", cp);
 108                                         continue;
 109                                 }
 110
 111                                 line = line.Substring (line.IndexOf (';') + 1).Trim ();
 112                                 // count entries in a line
 113                                 int entryPerLine = 0;
 114                                 for (int e = 0; (e = line.IndexOf ('[', e) + 1) > 0;)
 115                                         entryPerLine++;
 116
 117                                 int start = 0;
 118                                 for (int e = 0; e < entryPerLine; e++) {
 119                                         start = line.IndexOf ('[', start) + 1;
 120                                         string s = line.Substring (start, line.IndexOf (']', start) - start);
 121
 122                                         bool alt = false;
 123                                         if (s [0] == '*')
 124                                                 alt = true;
 125                                         string [] vslist = s.Substring (1).Split ('.');
 126                                         bool skip = false;
 127                                         for (int i = 0; i < 4; i++) {
 128                                                 if (vslist [i].Length > 4)
 129                                                         skip = true;
 130                                                 else
 131                                                         v [i] = ushort.Parse (vslist [i], NumberStyles.HexNumber);
 132                                         }
 133                                         if (skip) {
 134 //                                              Console.Error.WriteLine ("WARNING: skipped entry {0:X}", cp);
 135                                                 continue;
 136                                         }
 137                                         idx = keyCount;
 138                                         if (entryPerLine == 1) {
 139                                                 // idx = 0 means "no matching entry", so here we start from 1
 140                                                 for (idx = 1; idx < keyCount; idx++) {
 141                                                         SortKeyValue k = keyValues [idx];
 142                                                         if (k.Alt == alt &&
 143                                                                 k.Primary == v [0] &&
 144                                                                 k.Secondary == v [1] &&
 145                                                                 k.Thirtiary == v [2] &&
 146                                                                 k.Quarternary == v [3])
 147                                                                 break;
 148                                                 }
 149                                         }
 150                                         if (idx == keyCount)
 151                                                 AddEntry (alt, v);
 152                                 }
 153                                 if (entryPerLine == 1)
 154                                         collElem [collElemIdx] = idx;
 155                                 else
 156                                         collElem [collElemIdx] =
 157                                                 (short) (keyCount - entryPerLine)
 158                                                 + (short) (entryPerLine << 16);
 159                         }
 160                         reader.Close ();
 161                 }
 162
 163                 private void AddEntry (bool alt, ushort [] v)
 164                 {
 165                         if (keyCount == keyValues.Length) {
 166                                 SortKeyValue [] tmp = new SortKeyValue [keyCount * 2];
 167                                 Array.Copy (keyValues, tmp, keyCount);
 168                                 keyValues = tmp;
 169                         }
 170                         keyValues [keyCount] =
 171                                 new SortKeyValue (alt,
 172                                 v [0], (byte) v [1], (byte) v [2], v [3]);
 173                         keyCount++;
 174                 }
 175         }
 176 }
 177