//
+// create-mscompat-collation-table.cs : generates Windows-like sortkey tables.
+//
+// Author:
+// Atsushi Enomoto <atsushi@ximian.com>
+//
+// Copyright (C) 2005 Novell, Inc (http://www.novell.com)
+//
+// Permission is hereby granted, free of charge, to any person obtaining
+// a copy of this software and associated documentation files (the
+// "Software"), to deal in the Software without restriction, including
+// without limitation the rights to use, copy, modify, merge, publish,
+// distribute, sublicense, and/or sell copies of the Software, and to
+// permit persons to whom the Software is furnished to do so, subject to
+// the following conditions:
+//
+// The above copyright notice and this permission notice shall be
+// included in all copies or substantial portions of the Software.
+//
+// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+// EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+// MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+// NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
+// LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+// OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+// WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+//
+
//
// There are two kind of sort keys : which are computed and which are laid out
// as an indexed array. Computed sort keys are:
// - Surrogate
// - PrivateUse
//
-// Also, for composite characters it should prepare different index table.
-//
// Though it is possible to "compute" level 3 weights, they are still dumped
// to an array to avoid execution cost.
//
-
-//
-// * sortkey getter signature
-//
-// int GetSortKey (string s, int index, SortKeyBuffer buf)
-// Stores sort key for corresponding character element into buf and
-// returns the length of the consumed _source_ character element in s.
-//
-// * character length to consume
-//
-// If there are characters whose primary weight is 0, they are consumed
-// and considered as a part of the character element.
-//
#define Binary
using System;
const int DecompositionCompat = 0x11;
const int DecompositionCanonical = 0x12;
- TextWriter Result = Console.Out;
+ TextWriter CSResult = Console.Out;
+ TextWriter CResult = TextWriter.Null;
byte [] fillIndex = new byte [256]; // by category
CharMapEntry [] map = new CharMapEntry [char.MaxValue + 1];
string [] diacritics = new string [] {
// LATIN, CYRILLIC etc.
- "UPTURN", "DOUBLE-STRUCK",
+ "VERTICAL LINE ABOVE", "UPTURN", "DOUBLE-STRUCK",
+ "ABKHASIAN",
"MIDDLE HOOK", "WITH VERTICAL LINE ABOVE;", "WITH TONOS",
"WITH ACUTE ACCENT;", "WITH GRAVE ACCENT;",
"WITH ACUTE;", "WITH GRAVE;",
"WITH CIRCUMFLEX ACCENT;", "WITH CIRCUMFLEX;",
"WITH DIALYTIKA;",
"WITH DIAERESIS;", "WITH CARON;", "WITH BREVE;",
- "DIALYTIKA TONOS", "DIALYTIKA AND TONOS", "WITH MACRON;", "WITH TILDE;", "WITH RING ABOVE;",
+ "DIALYTIKA TONOS", "DIALYTIKA AND TONOS",
+ "ABKHASIAN CHE WITH DESCENDER",
+ "WITH MACRON;", "WITH TILDE;", "WITH RING ABOVE;",
"WITH OGONEK;", "WITH CEDILLA;",
//
" DOUBLE ACUTE;", " ACUTE AND DOT ABOVE;",
" BREVE AND TILDE",
" CEDILLA AND BREVE",
" OGONEK AND MACRON",
- //
- "WITH OVERLINE",
+ // 0x40
+ "WITH OVERLINE", "DOUBLE VERTICAL LINE ABOVE",
"WITH HOOK;", "LEFT HOOK;", " WITH HOOK ABOVE;",
" DOUBLE GRAVE",
" INVERTED BREVE",
" LINE BELOW;", " CIRCUMFLEX AND HOOK ABOVE",
" PALATAL HOOK",
" DOT BELOW;",
- " RETROFLEX;", "DIAERESIS BELOW",
- " RING BELOW",
+ " RETROFLEX;", "DIAERESIS BELOW", "RETROFLEX HOOK",
+ " RING BELOW", "LOW VERTICAL LINE",
//
" CIRCUMFLEX BELOW", "HORN AND ACUTE",
" BREVE BELOW;", " HORN AND GRAVE",
+ " LOW MACRON",
" TILDE BELOW",
" TOPBAR",
" DOT BELOW AND DOT ABOVE",
};
byte [] diacriticWeights = new byte [] {
// LATIN.
- 3, 3, 5, 5, 5,
+ 3, 3, 3, 5, 5, 5, 5,
0xE, 0xF,
0xE, 0xF,
//
0x10, 0x11, 0x12, 0x12, 0x13, 0x13, 0x14, 0x15, 0x16,
- 0x16, 0x17, 0x19, 0x1A, 0x1B, 0x1C,
+ 0x16, 0x17, 0x17, 0x19, 0x1A, 0x1B, 0x1C,
//
0x1D, 0x1D, 0x1E, 0x1E, 0x1E, 0x1F, 0x1F, 0x1F,
0x20, 0x21, 0x22, 0x22, 0x23, 0x24,
0x25, 0x25, 0x25, 0x26, 0x28, 0x28, 0x28,
0x29, 0x2A, 0x2B, 0x2C, 0x2F, 0x30,
//
- 0x40, 0x43, 0x43, 0x43, 0x44, 0x46, 0x47, 0x48,
- 0x52, 0x55, 0x55, 0x57, 0x58, 0x59, 0x59, 0x5A,
+ 0x40, 0x41, 0x43, 0x43, 0x43, 0x44, 0x46, 0x47, 0x48,
+ 0x52, 0x55, 0x55, 0x57, 0x58, 0x59, 0x59, 0x59,
+ 0x5A, 0x5A,
//
- 0x60, 0x60, 0x61, 0x61, 0x63, 0x68, 0x68,
+ 0x60, 0x60, 0x61, 0x61, 0x62, 0x63, 0x68, 0x68,
0x69, 0x69, 0x6A, 0x6D, 0x6E,
0x87, 0x95, 0xAA,
// CIRCLED, PARENTHESIZED and so on.
ModifyParsedValues ();
GenerateCore ();
Console.Error.WriteLine ("generation done.");
+ CResult = new StreamWriter ("collation-tables.h", false);
Serialize ();
+ CResult.Close ();
Console.Error.WriteLine ("serialization done.");
/*
StreamWriter sw = new StreamWriter ("agelog.txt");
source, typeof (ushort), i);
}
+ void WriteByte (byte value)
+ {
+
+ }
+
void Serialize ()
{
// Tailorings
byte [] level1 = new byte [map.Length];
byte [] level2 = new byte [map.Length];
byte [] level3 = new byte [map.Length];
- ushort [] widthCompat = new ushort [map.Length];
+// widthCompat is now removed from the mapping table.
+// If it turned out that it is still required, grep this source and uncomment
+// widthCompat related lines. FIXME: remove those lines in the future.
+// ushort [] widthCompat = new ushort [map.Length];
for (int i = 0; i < map.Length; i++) {
categories [i] = map [i].Category;
level1 [i] = map [i].Level1;
level2 [i] = map [i].Level2;
level3 [i] = ComputeLevel3Weight ((char) i);
+/*
// For Japanese Half-width characters, don't
// map widthCompat. It is IgnoreKanaType that
// handles those width differences.
widthCompat [i] = (ushort) decompValues [decompIndex [i]];
break;
}
+*/
}
// compress
level1 = CompressArray (level1, UUtil.Level1);
level2 = CompressArray (level2, UUtil.Level2);
level3 = CompressArray (level3, UUtil.Level3);
- widthCompat = (ushort []) CodePointIndexer.CompressArray (
- widthCompat, typeof (ushort), UUtil.WidthCompat);
+// widthCompat = (ushort []) CodePointIndexer.CompressArray (
+// widthCompat, typeof (ushort), UUtil.WidthCompat);
cjkCHS = CompressArray (cjkCHS, UUtil.CjkCHS);
cjkCHT = CompressArray (cjkCHT,UUtil.Cjk);
cjkJA = CompressArray (cjkJA, UUtil.Cjk);
cjkKOlv2 = CompressArray (cjkKOlv2, UUtil.Cjk);
// Ignorables
- Result.WriteLine ("internal static readonly byte [] ignorableFlags = new byte [] {");
+ CResult.WriteLine ("static const guint8 collation_table_ignorableFlags [] = {");
+ CSResult.WriteLine ("static readonly byte [] ignorableFlagsArr = new byte [] {");
#if Binary
MemoryStream ms = new MemoryStream ();
BinaryWriter binary = new BinaryWriter (ms);
+ binary.Write (UUtil.ResourceVersion);
binary.Write (ignorableFlags.Length);
#endif
for (int i = 0; i < ignorableFlags.Length; i++) {
byte value = ignorableFlags [i];
if (value < 10)
- Result.Write ("{0},", value);
+ CSResult.Write ("{0},", value);
else
- Result.Write ("0x{0:X02},", value);
+ CSResult.Write ("0x{0:X02},", value);
+ CResult.Write ("{0},", value);
#if Binary
binary.Write (value);
#endif
- if ((i & 0xF) == 0xF)
- Result.WriteLine ("// {0:X04}",
+ if ((i & 0xF) == 0xF) {
+ CSResult.WriteLine ("// {0:X04}",
UUtil.Ignorable.ToCodePoint (i - 0xF));
+ CResult.WriteLine ();
+ }
}
- Result.WriteLine ("};");
- Result.WriteLine ();
+ CResult.WriteLine ("0};");
+ CSResult.WriteLine ("};");
+ CSResult.WriteLine ();
// Primary category
- Result.WriteLine ("internal static readonly byte [] categories = new byte [] {");
+ CResult.WriteLine ("static const guint8 collation_table_category [] = {");
+ CSResult.WriteLine ("static readonly byte [] categoriesArr = new byte [] {");
#if Binary
binary.Write (categories.Length);
#endif
for (int i = 0; i < categories.Length; i++) {
byte value = categories [i];
if (value < 10)
- Result.Write ("{0},", value);
+ CSResult.Write ("{0},", value);
else
- Result.Write ("0x{0:X02},", value);
+ CSResult.Write ("0x{0:X02},", value);
+ CResult.Write ("{0},", value);
#if Binary
binary.Write (value);
#endif
- if ((i & 0xF) == 0xF)
- Result.WriteLine ("// {0:X04}",
+ if ((i & 0xF) == 0xF) {
+ CSResult.WriteLine ("// {0:X04}",
UUtil.Category.ToCodePoint (i - 0xF));
+ CResult.WriteLine ();
+ }
}
- Result.WriteLine ("};");
- Result.WriteLine ();
+ CResult.WriteLine ("};");
+ CSResult.WriteLine ("};");
+ CSResult.WriteLine ();
// Primary weight value
- Result.WriteLine ("internal static readonly byte [] level1 = new byte [] {");
+ CResult.WriteLine ("static const guint8 collation_table_level1 [] = {");
+ CSResult.WriteLine ("static readonly byte [] level1Arr = new byte [] {");
#if Binary
binary.Write (level1.Length);
#endif
for (int i = 0; i < level1.Length; i++) {
byte value = level1 [i];
if (value < 10)
- Result.Write ("{0},", value);
+ CSResult.Write ("{0},", value);
else
- Result.Write ("0x{0:X02},", value);
+ CSResult.Write ("0x{0:X02},", value);
+ CResult.Write ("{0},", value);
#if Binary
binary.Write (value);
#endif
- if ((i & 0xF) == 0xF)
- Result.WriteLine ("// {0:X04}",
+ if ((i & 0xF) == 0xF) {
+ CSResult.WriteLine ("// {0:X04}",
UUtil.Level1.ToCodePoint (i - 0xF));
+ CResult.WriteLine ();
+ }
}
- Result.WriteLine ("};");
- Result.WriteLine ();
+ CResult.WriteLine ("0};");
+ CSResult.WriteLine ("};");
+ CSResult.WriteLine ();
// Secondary weight
- Result.WriteLine ("internal static readonly byte [] level2 = new byte [] {");
+ CResult.WriteLine ("static const guint8 collation_table_level2 [] = {");
+ CSResult.WriteLine ("static readonly byte [] level2Arr = new byte [] {");
#if Binary
binary.Write (level2.Length);
#endif
for (int i = 0; i < level2.Length; i++) {
byte value = level2 [i];
if (value < 10)
- Result.Write ("{0},", value);
+ CSResult.Write ("{0},", value);
else
- Result.Write ("0x{0:X02},", value);
+ CSResult.Write ("0x{0:X02},", value);
+ CResult.Write ("{0},", value);
#if Binary
binary.Write (value);
#endif
- if ((i & 0xF) == 0xF)
- Result.WriteLine ("// {0:X04}",
+ if ((i & 0xF) == 0xF) {
+ CSResult.WriteLine ("// {0:X04}",
UUtil.Level2.ToCodePoint (i - 0xF));
+ CResult.WriteLine ();
+ }
}
- Result.WriteLine ("};");
- Result.WriteLine ();
+ CResult.WriteLine ("0};");
+ CSResult.WriteLine ("};");
+ CSResult.WriteLine ();
// Thirtiary weight
- Result.WriteLine ("internal static readonly byte [] level3 = new byte [] {");
+ CResult.WriteLine ("static const guint8 collation_table_level3 [] = {");
+ CSResult.WriteLine ("static readonly byte [] level3Arr = new byte [] {");
#if Binary
binary.Write (level3.Length);
#endif
for (int i = 0; i < level3.Length; i++) {
byte value = level3 [i];
if (value < 10)
- Result.Write ("{0},", value);
+ CSResult.Write ("{0},", value);
else
- Result.Write ("0x{0:X02},", value);
+ CSResult.Write ("0x{0:X02},", value);
+ CResult.Write ("{0},", value);
#if Binary
binary.Write (value);
#endif
- if ((i & 0xF) == 0xF)
- Result.WriteLine ("// {0:X04}",
+ if ((i & 0xF) == 0xF) {
+ CSResult.WriteLine ("// {0:X04}",
UUtil.Level3.ToCodePoint (i - 0xF));
+ CResult.WriteLine ();
+ }
}
- Result.WriteLine ("};");
- Result.WriteLine ();
+ CResult.WriteLine ("0};");
+ CSResult.WriteLine ("};");
+ CSResult.WriteLine ();
+/*
// Width insensitivity mappings
// (for now it is more lightweight than dumping the
// entire NFKD table).
- Result.WriteLine ("internal static readonly ushort [] widthCompat = new ushort [] {");
+ CResult.WriteLine ("static const guint16* widthCompat [] = {");
+ CSResult.WriteLine ("static readonly ushort [] widthCompatArr = new ushort [] {");
#if Binary
binary.Write (widthCompat.Length);
#endif
for (int i = 0; i < widthCompat.Length; i++) {
ushort value = widthCompat [i];
if (value < 10)
- Result.Write ("{0},", value);
+ CSResult.Write ("{0},", value);
else
- Result.Write ("0x{0:X02},", value);
+ CSResult.Write ("0x{0:X02},", value);
+ CResult.Write ("{0},", value);
#if Binary
binary.Write (value);
#endif
- if ((i & 0xF) == 0xF)
- Result.WriteLine ("// {0:X04}",
+ if ((i & 0xF) == 0xF) {
+ CSResult.WriteLine ("// {0:X04}",
UUtil.WidthCompat.ToCodePoint (i - 0xF));
+ CResult.WriteLine ();
+ }
}
- Result.WriteLine ("};");
- Result.WriteLine ();
+ CResult.WriteLine ("0};");
+ CSResult.WriteLine ("};");
+ CSResult.WriteLine ();
+*/
+
#if Binary
- using (FileStream fs = File.Create ("../collation.core.bin")) {
+ using (FileStream fs = File.Create ("../resources/collation.core.bin")) {
byte [] array = ms.ToArray ();
fs.Write (array, 0, array.Length);
}
SerializeCJK ("cjkKOlv2", cjkKOlv2, 0x9FB0);
}
- void SerializeCJK (string name, ushort [] cjk, int max)
+ void SerializeCJK (string name, ushort [] cjk, int max_unused)
{
- int offset = 0;//char.MaxValue - cjk.Length;
- Result.WriteLine ("static ushort [] {0} = new ushort [] {{", name);
+// CResult.WriteLine ("static const int collation_table_collation_cjk_{0}_size [] = {1};", name, cjk.Length);
+ CSResult.WriteLine ("const int {0}ArrLength = {1};", name, cjk.Length);
+
+ int len = cjk.Length;
+ CResult.WriteLine ("static const guint8 collation_table_collation_cjk_{0} [] = {{", name);
+ CSResult.WriteLine ("static byte [] {0}Arr = new byte [] {{", name);
+ // the actual length is *2
+ for (int i = 0; i < 4; i++, len /= 256) {
+ CResult.Write ("{0},", len & 0xFF);
+ CSResult.Write ("0x{0:X04},", len & 0xFF);
+ }
+ CResult.WriteLine ();
+ CSResult.WriteLine ();
#if Binary
MemoryStream ms = new MemoryStream ();
BinaryWriter binary = new BinaryWriter (ms);
- binary.Write (cjk.Length);
+ binary.Write (UUtil.ResourceVersion);
+ binary.Write (cjk.Length); // the actual size is *2.
#endif
+ // category
for (int i = 0; i < cjk.Length; i++) {
- if (i + offset == max)
- break;
- ushort value = cjk [i];
+// if (i == max)
+// break;
+ byte value = (byte) (cjk [i] >> 8);
+ if (value < 10)
+ CSResult.Write ("{0},", value);
+ else
+ CSResult.Write ("0x{0:X02},", value);
+ CResult.Write ("{0},", value);
+#if Binary
+ binary.Write (value);
+#endif
+ if ((i & 0xF) == 0xF) {
+ CSResult.WriteLine ("// {0:X04}", i - 0xF);
+ CResult.WriteLine ();
+ }
+ }
+
+ // level 1
+ for (int i = 0; i < cjk.Length; i++) {
+// if (i == max)
+// break;
+ byte value = (byte) (cjk [i] & 0xFF);
if (value < 10)
- Result.Write ("{0},", value);
+ CSResult.Write ("{0},", value);
else
- Result.Write ("0x{0:X04},", value);
+ CSResult.Write ("0x{0:X02},", value);
+ CResult.Write ("{0},", value);
#if Binary
binary.Write (value);
#endif
- if ((i & 0xF) == 0xF)
- Result.WriteLine ("// {0:X04}", i - 0xF + offset);
+ if ((i & 0xF) == 0xF) {
+ CSResult.WriteLine ("// {0:X04}", i - 0xF);
+ CResult.WriteLine ();
+ }
}
- Result.WriteLine ("};");
- Result.WriteLine ();
+
+ CResult.WriteLine ("0};");
+ CSResult.WriteLine ("};");
+ CSResult.WriteLine ();
#if Binary
- using (FileStream fs = File.Create (String.Format ("../collation.{0}.bin", name))) {
+ using (FileStream fs = File.Create (String.Format ("../resources/collation.{0}.bin", name))) {
byte [] array = ms.ToArray ();
fs.Write (array, 0, array.Length);
}
void SerializeCJK (string name, byte [] cjk, int max)
{
- int offset = 0;//char.MaxValue - cjk.Length;
- Result.WriteLine ("static byte [] {0} = new byte [] {{", name);
+ CResult.WriteLine ("static const guint8 collation_table_collation_cjk_{0} [] = {{", name);
+ CSResult.WriteLine ("static byte [] {0}Arr = new byte [] {{", name);
#if Binary
MemoryStream ms = new MemoryStream ();
BinaryWriter binary = new BinaryWriter (ms);
+ binary.Write (UUtil.ResourceVersion);
#endif
for (int i = 0; i < cjk.Length; i++) {
- if (i + offset == max)
+ if (i == max)
break;
byte value = cjk [i];
if (value < 10)
- Result.Write ("{0},", value);
+ CSResult.Write ("{0},", value);
else
- Result.Write ("0x{0:X02},", value);
+ CSResult.Write ("0x{0:X02},", value);
+ CResult.Write ("{0},", value);
#if Binary
binary.Write (value);
#endif
- if ((i & 0xF) == 0xF)
- Result.WriteLine ("// {0:X04}", i - 0xF + offset);
+ if ((i & 0xF) == 0xF) {
+ CSResult.WriteLine ("// {0:X04}", i - 0xF);
+ CResult.WriteLine ();
+ }
}
- Result.WriteLine ("};");
- Result.WriteLine ();
+ CResult.WriteLine ("0};");
+ CSResult.WriteLine ("};");
+ CSResult.WriteLine ();
#if Binary
- using (FileStream fs = File.Create (String.Format ("../collation.{0}.bin", name))) {
+ using (FileStream fs = File.Create (String.Format ("../resources/collation.{0}.bin", name))) {
byte [] array = ms.ToArray ();
fs.Write (array, 0, array.Length);
}
{
Hashtable indexes = new Hashtable ();
Hashtable counts = new Hashtable ();
- Result.WriteLine ("static char [] tailorings = new char [] {");
+ CResult.WriteLine ("static const guint16 collation_table_tailoring [] = {");
+ CSResult.WriteLine ("static char [] tailoringArr = new char [] {");
int count = 0;
#if Binary
MemoryStream ms = new MemoryStream ();
BinaryWriter binary = new BinaryWriter (ms);
+ // Here we don't need to output resource version.
+ // This is cached.
#endif
foreach (Tailoring t in tailorings) {
if (t.Alias != 0)
continue;
- Result.Write ("/*{0}*/", t.LCID);
+ CResult.Write ("/*{0}*/", t.LCID);
+ CSResult.Write ("/*{0}*/", t.LCID);
indexes.Add (t.LCID, count);
char [] values = t.ItemToCharArray ();
counts.Add (t.LCID, values.Length);
foreach (char c in values) {
- Result.Write ("'\\x{0:X}', ", (int) c);
- if (++count % 16 == 0)
- Result.WriteLine (" // {0:X04}", count - 16);
+ CSResult.Write ("'\\x{0:X}', ", (int) c);
+ CResult.Write ("{0},", (int) c);
+ if (++count % 16 == 0) {
+ CSResult.WriteLine (" // {0:X04}", count - 16);
+ CResult.WriteLine ();
+ }
#if Binary
binary.Write ((ushort) c);
#endif
}
}
- Result.WriteLine ("};");
+ CResult.WriteLine ("0};");
+ CSResult.WriteLine ("};");
- Result.WriteLine ("static TailoringInfo [] tailoringInfos = new TailoringInfo [] {");
+ CResult.WriteLine ("static const guint32 collation_table_tailoring_infos [] = {");
+ CResult.WriteLine ("{0}, /*count*/", tailorings.Count);
+ CSResult.WriteLine ("static TailoringInfo [] tailoringInfos = new TailoringInfo [] {");
#if Binary
byte [] rawdata = ms.ToArray ();
ms = new MemoryStream ();
binary = new BinaryWriter (ms);
+ binary.Write (UUtil.ResourceVersion);
binary.Write (tailorings.Count);
#endif
foreach (Tailoring t in tailorings) {
foreach (Tailoring t2 in tailorings)
if (t2.LCID == t.LCID)
french = t2.FrenchSort;
- Result.WriteLine ("new TailoringInfo ({0}, 0x{1:X}, {2}, {3}), ", t.LCID, idx, cnt, french ? "true" : "false");
+ CSResult.WriteLine ("new TailoringInfo ({0}, 0x{1:X}, {2}, {3}), ", t.LCID, idx, cnt, french ? "true" : "false");
+ CResult.WriteLine ("{0},{1},{2},{3},", t.LCID, idx, cnt, french ? 1 : 0);
#if Binary
binary.Write (t.LCID);
binary.Write (idx);
binary.Write (french);
#endif
}
- Result.WriteLine ("};");
+ CResult.WriteLine ("0};");
+ CSResult.WriteLine ("};");
#if Binary
binary.Write ((byte) 0xFF);
binary.Write ((byte) 0xFF);
binary.Write (rawdata, 0, rawdata.Length);
- using (FileStream fs = File.Create ("../collation.tailoring.bin")) {
+ using (FileStream fs = File.Create ("../resources/collation.tailoring.bin")) {
byte [] array = ms.ToArray ();
fs.Write (array, 0, array.Length);
}
{
StringBuilder sb = new StringBuilder ();
for (int i = 0; i < s.Length; i++) {
- if (s.StartsWith ("\\u")) {
- sb.Append ((char) int.Parse (
- s.Substring (2, 4), NumberStyles.HexNumber),
+ if (i + 5 < s.Length &&
+ s [i] == '\\' && s [i + 1] == 'u') {
+ sb.Append (
+ (char) int.Parse (
+ s.Substring (i + 2, 4),
+ NumberStyles.HexNumber),
1);
i += 5;
}
- else
- sb.Append (s [i]);
+ else
+ sb.Append (s [i]);
}
return sb.ToString ();
}
target = 'B';
else if (s.Substring (offset).StartsWith ("OPEN O"))
target = 'C';
+ else if (s.Substring (offset).StartsWith ("ETH"))
+ target = 'D';
else if (s.Substring (offset).StartsWith ("SCHWA"))
target = 'E';
- else if (s.Substring (offset).StartsWith ("ENG"))
- target = 'N';
else if (s.Substring (offset).StartsWith ("OI;")) // 01A2,01A3
target = 'O';
else if (s.Substring (offset).StartsWith ("YR;")) // 01A2,01A3
target = 'S';
else if (s.Substring (offset).StartsWith ("ESH"))
target = 'S';
+ else if (s.Substring (offset).StartsWith ("OUNCE"))
+ target = 'Z';
// For remaining IPA chars, direct mapping is
// much faster.
switch (cp) {
+ case 0x0166: case 0x0167:
+ // Though they are 'T', they have different weight
+ target = char.MinValue; break;
case 0x0299: target = 'B'; break;
case 0x029A: target = 'E'; break;
case 0x029B: target = 'G'; break;
// diacritical weights by character name
if (diacritics.Length != diacriticWeights.Length)
throw new Exception (String.Format ("Should not happen. weights are {0} while labels are {1}", diacriticWeights.Length, diacritics.Length));
- for (int d = 0; d < diacritics.Length; d++) {
+ for (int d = diacritics.Length - 1; d >= 0; d--) {
if (s.IndexOf (diacritics [d]) > 0) {
diacritical [cp] += diacriticWeights [d];
if (s.IndexOf ("COMBINING") >= 0)
diacritical [cp] -= (byte) 2;
- continue;
+ break;
}
// also process "COMBINING blah" here
// For now it is limited to cp < 0x0370
void ModifyUnidata ()
{
+ ArrayList decompValues = new ArrayList (this.decompValues);
+
+ // Hebrew uppercase letters.
+ foreach (int i in new int []
+ {0x05DB, 0x05DE, 0x05E0, 0x05E4, 0x05E6})
+ isUppercase [i] = true;
+
+
// Modify some decomposition equivalence
for (int i = 0xFE31; i <= 0xFE34; i++) {
decompType [i] = 0;
decompValues [decompIndex [0x3298]] = 0x52DE;
// LAMESPEC: custom remapping (which is not bugs but not fine, non-standard compliant things)
- decompIndex [0xFA0C] = decompIndex [0xF929]; // borrow U+F929 room (being empty)
- decompValues [decompIndex [0xFA0C]] = 0x5140;
+ decompIndex [0xFA0C] = decompValues.Count;
+ decompValues.Add ((int) 0x5140);
decompLength [0xFA0C] = 1;
decompIndex [0xF929] = decompLength [0xF929] = 0;
decompValues [decompIndex [0xF92C]] = 0x90DE;
+
+ decompIndex [0x2125] = decompValues.Count;
+ decompValues.Add ((int) 0x005A);
+ decompLength [0x2125] = 1;
+ decompType [0x2125] = DecompositionFont;
+
+ this.decompValues = decompValues.ToArray (typeof (int)) as int [];
}
void ModifyParsedValues ()
{
+ // Sometimes STROKE don't work fine
+ diacritical [0xD8] = diacritical [0xF8] = 0x21;
+ diacritical [0x141] = diacritical [0x142] = 0x1F;
+ // FIXME: why?
+ diacritical [0xAA] = diacritical [0xBA] = 3;
+ diacritical [0xD0] = diacritical [0xF0] = 0x68;
+ diacritical [0x131] = 3;
+ diacritical [0x138] = 3;
+ // TOPBAR does not work as an identifier for the weight
+ diacritical [0x182] = diacritical [0x183] = 0x68; // B
+ diacritical [0x18B] = diacritical [0x18C] = 0x1E; // D
+ // TONE TWO
+ diacritical [0x1A7] = diacritical [0x1A8] = 0x87;
+ // TONE SIX
+ diacritical [0x184] = diacritical [0x185] = 0x87;
+ // OPEN E
+ diacritical [0x190] = diacritical [0x25B] = 0x7B;
+ // There are many letters w/ diacritical weight 0x7B
+ diacritical [0x0192] = diacritical [0x0194] =
+ diacritical [0x0195] = diacritical [0x0196] =
+ diacritical [0x019C] = diacritical [0x019E] =
+ diacritical [0x01A6] = diacritical [0x01B1] =
+ diacritical [0x01B2] = diacritical [0x01BF] = 0x7B;
+ // ... as well as 0x7C
+ diacritical [0x01A2] = diacritical [0x01A3] = 0x7C;
+
+ // <font> NFKD characters seem to have diacritical
+ // weight as 3,4,5... but the order does not look
+ // by codepoint and I have no idea how they are sorted.
+ diacritical [0x210E] = 3;
+ diacritical [0x210F] = 0x68;
+ diacritical [0x2110] = 4;
+ diacritical [0x2111] = 5;
+ diacritical [0x2112] = 4;
+ diacritical [0x2113] = 4;
+ diacritical [0x211B] = 4;
+ diacritical [0x211C] = 5;
+
// some cyrillic diacritical weight. They seem to be
// based on old character names, so it's quicker to
// set them directly here.
+ // FIXME: they are by mostly unknown reason
diacritical [0x0496] = diacritical [0x0497] = 7;
diacritical [0x0498] = diacritical [0x0499] = 0x1A;
diacritical [0x049A] = diacritical [0x049B] = 0x17;
diacritical [0x04A0] = diacritical [0x04A1] = 0xA;
diacritical [0x04A2] = diacritical [0x04A3] = 7;
diacritical [0x04A4] = diacritical [0x04A5] = 8;
+ diacritical [0x04AA] = diacritical [0x04AB] = 0x1A; // ES CEDILLA?
+ diacritical [0x04AC] = diacritical [0x04AD] = 7; // RIGHT DESCENDER? but U+4B2
+ diacritical [0x04AE] = diacritical [0x04AF] = 0xB; // STRAIGHT U?
+ diacritical [0x04B2] = diacritical [0x04B3] = 0x17; // RIGHT DESCENDER? but U+4AC
+ diacritical [0x04B4] = diacritical [0x04B5] = 3;
+ diacritical [0x04B6] = 8;
+ diacritical [0x04B7] = 7;
+ diacritical [0x04B8] = diacritical [0x04B9] = 9;
+ diacritical [0x04BA] = diacritical [0x04BB] = 9;
// number, secondary weights
byte weight = 0x38;
if (Char.IsNumber ((char) cp))
diacritical [cp] = weight;
+ // Gurmukhi special letters' diacritical weight
+ for (int i = 0x0A50; i < 0x0A60; i++)
+ diacritical [i] = 4;
+ // Oriya special letters' diacritical weight
+ for (int i = 0x0B5C; i < 0x0B60; i++)
+ diacritical [i] = 6;
+
// Update name part of named characters
for (int i = 0; i < sortableCharNames.Count; i++) {
DictionaryEntry de =
fillIndex [6] = 0xA0;
// vowels
for (int i = 0x64B; i <= 0x650; i++)
- AddArabicCharMap ((char) i);
+ AddArabicCharMap ((char) i, 6, 1, 0);
// sukun
AddCharMapGroup ('\u0652', 6, 1, 0);
// shadda
if (!IsIgnorable (i))
AddCharMap ((char) i, 0x1, 1);
+
// FIXME: needs more love here (it should eliminate
// all the hacky code above).
for (int i = 0x0300; i < 0x0370; i++)
if (!IsIgnorable (i) && diacritical [i] != 0
- /* especiall here*/ && !map [i].Defined)
+ && !map [i].Defined)
map [i] = new CharMapEntry (
0x1, 0x1, diacritical [i]);
fillIndex [0x1] = 0xEC;
for (int i = 0x20DD; i <= 0x20E1; i++)
AddCharMap ((char) i, 0x1, 1);
- fillIndex [0x1] = 0x7;
+ fillIndex [0x1] = 0x4;
+ AddCharMap ('\u0CD5', 0x1, 1);
+ AddCharMap ('\u0CD6', 0x1, 1);
+ AddCharMap ('\u093C', 0x1, 1);
for (int i = 0x302A; i <= 0x302D; i++)
AddCharMap ((char) i, 0x1, 1);
+ AddCharMap ('\u0C55', 0x1, 1);
+ AddCharMap ('\u0C56', 0x1, 1);
+
fillIndex [0x1] = 0x50; // I wonder how they are sorted
for (int i = 0x02D4; i <= 0x02D7; i++)
AddCharMap ((char) i, 0x1, 1);
AddCharMapGroup ((char) i, 0xE, 1, 0);
}
- // Greek and Coptic
- fillIndex [0xF] = 02;
- for (int i = 0x0380; i < 0x0390; i++)
+ // IPA extensions
+ // FIXME: this results in not equivalent values to
+ // Windows, but is safer for comparison.
+ char [] ipaArray = new char [0x300 - 0x250 + 0x20];
+ for (int i = 0x40; i < 0x60; i++)
if (Char.IsLetter ((char) i))
- AddLetterMap ((char) i, 0xF, 1);
- fillIndex [0xF] = 02;
- for (int i = 0x0391; i < 0x03CF; i++)
+ ipaArray [i - 0x40] = (char) (i);
+ for (int i = 0x250; i < 0x300; i++)
if (Char.IsLetter ((char) i))
- AddLetterMap ((char) i, 0xF, 1);
+ ipaArray [i - 0x250 + 0x20] = (char) i;
+ Array.Sort (ipaArray, UCAComparer.Instance);
+ int targetASCII = 0;
+ byte latinDiacritical = 0x7B;
+ foreach (char c in ipaArray) {
+ if (c <= 'Z') {
+ targetASCII = c;
+ latinDiacritical = 0x7B;
+ }
+ else
+ map [(int) c] = new CharMapEntry (
+ 0xE,
+ map [targetASCII].Level1,
+ latinDiacritical++);
+ }
+
+ // Greek and Coptic
+
+ // FIXME: this is (mysterious and) incomplete.
+ for (int i = 0x0380; i < 0x0400; i++)
+ if (diacritical [i] == 0 &&
+ decompLength [i] == 1 &&
+ decompType [i] == DecompositionCompat)
+ diacritical [i] = 3;
+
+ fillIndex [0xF] = 2;
+ for (int i = 0x0391; i < 0x03AA; i++)
+ if (i != 0x03A2)
+ AddCharMap ((char) i, 0xF, 1,
+ diacritical [i]);
+ fillIndex [0xF] = 2;
+ for (int i = 0x03B1; i < 0x03CA; i++)
+ if (i != 0x03C2)
+ AddCharMap ((char) i, 0xF, 1,
+ diacritical [i]);
+ // Final Sigma
+ map [0x03C2] = new CharMapEntry (0xF,
+ map [0x03C3].Level1, map [0x03C3].Level2);
+
fillIndex [0xF] = 0x40;
- for (int i = 0x03D0; i < 0x0400; i++)
- if (Char.IsLetter ((char) i))
- AddLetterMap ((char) i, 0xF, 1);
+ for (int i = 0x03DA; i < 0x03F0; i++)
+ AddCharMap ((char) i, 0xF,
+ (byte) (i % 2 == 0 ? 0 : 2),
+ diacritical [i]);
+
+ // NFKD
+ for (int i = 0x0386; i <= 0x0400; i++)
+ FillLetterNFKD (i, true, true);
// Cyrillic.
// Cyrillic letters are sorted like Latin letters i.e.
}
}
+ // NFKD
+ for (int i = 0x0401; i <= 0x045F; i++)
+ FillLetterNFKD (i, false, false);
+
for (int i = 0; i < cymap_src.Length; i++) {
char c = cymap_src [i];
fillIndex [0x10] = map [c].Level1;
// -Letters
fillIndex [0x12] = 0x2;
for (int i = 0x05D0; i < 0x05FF; i++)
- if (Char.IsLetter ((char) i))
- AddLetterMap ((char) i, 0x12, 1);
+ if (Char.IsLetter ((char) i)) {
+ if (isUppercase [i]) {
+ fillIndex [0x12]--;
+ AddLetterMap ((char) i, 0x12, 2);
+ }
+ else
+ AddLetterMap ((char) i, 0x12, 1);
+ }
// -Accents
fillIndex [0x1] = 0x3;
for (int i = 0x0591; i <= 0x05C2; i++) {
case 0x0649: formDiacritical = 5; break;
case 0x064A: formDiacritical = 7; break;
}
- AddLetterMapCore ((char) i, 0x13, 1, formDiacritical, false);
+// AddLetterMapCore ((char) i, 0x13, 1, formDiacritical, false);
+ AddArabicCharMap ((char) i, 0x13, 1, formDiacritical);
}
for (int i = 0x0670; i < 0x0673; i++)
map [i] = new CharMapEntry (0x13, 0xB, (byte) (0xC + i - 0x670));
AddLetterMap ((char) i, 0x1, 1);
continue;
}
- AddLetterMap ((char) i, 0x18, 1);
+ AddLetterMapCore ((char) i, 0x18, 1, 0, true);
}
// Tamil
// Here Windows mapping is not straightforward. It is
// not based on computation but seems manual sorting.
AddCharMapGroup ('+', 0x8, 1, 0); // plus
- AddCharMapGroup ('\u2212', 0x8, 1, 0); // minus
- AddCharMapGroup ('\u229D', 0x8, 1, 0); // minus
- AddCharMapGroup ('\u2297', 0x8, 1, 0); // mul
- AddCharMapGroup ('\u2044', 0x8, 1, 0); // div
- AddCharMapGroup ('\u2215', 0x8, 1, 0); // div
- AddCharMapGroup ('\u2217', 0x8, 1, 0); // mul
- AddCharMapGroup ('\u2218', 0x8, 1, 0); // ring
- AddCharMapGroup ('\u2219', 0x8, 1, 0); // bullet
- AddCharMapGroup ('\u2213', 0x8, 1, 0); // minus-or-plus
- AddCharMapGroup ('\u003C', 0x8, 1, 0); // <
- AddCharMapGroup ('\u227A', 0x8, 1, 0); // precedes relation
- AddCharMapGroup ('\u22B0', 0x8, 1, 0); // precedes under relation
+ AddCharMapGroup ('\u2212', 0x8, 1); // minus
+ AddCharMapGroup ('\u229D', 0x8, 1); // minus
+ AddCharMapGroup ('\u2297', 0x8, 1); // mul
+ AddCharMapGroup ('\u2044', 0x8, 1); // div
+ AddCharMapGroup ('\u2215', 0x8, 0); // div
+ AddCharMapGroup ('\u2298', 0x8, 1); // div slash
+ AddCharMapGroup ('\u2217', 0x8, 0); // mul
+ AddCharMapGroup ('\u229B', 0x8, 1); // asterisk oper
+ AddCharMapGroup ('\u2218', 0x8, 0); // ring
+ AddCharMapGroup ('\u229A', 0x8, 1); // ring
+ AddCharMapGroup ('\u2219', 0x8, 0); // bullet
+ AddCharMapGroup ('\u2299', 0x8, 1); // dot oper
+ AddCharMapGroup ('\u2213', 0x8, 1); // minus-or-plus
+ AddCharMapGroup ('\u003C', 0x8, 1); // <
+ AddCharMapGroup ('\u227A', 0x8, 1); // precedes relation
+ AddCharMapGroup ('\u22B0', 0x8, 1); // precedes under relation
for (int cp = 0; cp < 0x2300; cp++) {
if (cp == 0xAC) // SPECIAL CASE: skip
// Char.GetUnicodeCategory ((char) cp) ==
// UnicodeCategory.MathSymbol)
Char.IsSymbol ((char) cp))
- AddCharMapGroup ((char) cp, 0x8, 1, diacritical [cp]);
+ AddCharMapGroup ((char) cp, 0x8, 1);
// SPECIAL CASES: no idea why Windows sorts as such
switch (cp) {
case 0x3E:
AddCharMap ('\u22B1', 0x8, 1, 0);
break;
case 0xB1:
- AddCharMapGroup ('\u00AB', 0x8, 1, 0);
- AddCharMapGroup ('\u226A', 0x8, 1, 0);
- AddCharMapGroup ('\u00BB', 0x8, 1, 0);
- AddCharMapGroup ('\u226B', 0x8, 1, 0);
+ AddCharMapGroup ('\u00AB', 0x8, 1);
+ AddCharMapGroup ('\u226A', 0x8, 1);
+ AddCharMapGroup ('\u00BB', 0x8, 1);
+ AddCharMapGroup ('\u226B', 0x8, 1);
break;
case 0xF7:
AddCharMap ('\u01C0', 0x8, 1, 0);
mod = diacritical [i];
break;
case 0x13: // Arabic
+ if (i == 0x0621)
+ break; // 0
+ if (diacritical [i] == 0 && decompLength [i] != 0)
+ diacritical [i] = map [decompValues [decompIndex [i]]].Level2;
if (diacritical [i] == 0 && i >= 0xFE8D)
mod = 0x8; // default for arabic
break;
#endregion
}
+ TextInfo ti = CultureInfo.InvariantCulture.TextInfo;
+
+ private void FillLetterNFKD (int i, bool checkUpper, bool greekRemap)
+ {
+ if (map [i].Defined)
+ return;
+ int up = (int) ti.ToUpper ((char) i);
+ if (checkUpper && map [up].Category == 0xF) {
+ if (i == up)
+ return;
+ FillLetterNFKD (up, checkUpper, greekRemap);
+ map [i] = new CharMapEntry (0xF,
+ map [up].Level1,
+ map [up].Level2);
+ } else {
+ int idx = decompIndex [i];
+ if (idx == 0)
+ return;
+ int primary = decompValues [decompIndex [i]];
+ FillLetterNFKD (primary, checkUpper, greekRemap);
+
+ int lv2 = map [primary].Level2;
+ byte off = 0;
+ for (int l = 1; l < decompLength [i]; l++) {
+ int tmp = decompValues [idx + l];
+ if (map [tmp].Category != 1)
+ return;
+ if (greekRemap && map [tmp].Level2 == 0xC)
+ off += 3;
+ else
+ off += map [tmp].Level2;
+ }
+ if (off > 0) {
+ if (lv2 == 0)
+ lv2 += 2;
+ lv2 += off;
+ }
+ // ... but override if the value already exists.
+ if (diacritical [i] != 0)
+ lv2 = diacritical [i];
+ map [i] = new CharMapEntry (
+ map [primary].Category,
+ map [primary].Level1,
+ (byte) lv2);
+ }
+ }
+
private void IncrementSequentialIndex (ref byte hangulCat)
{
fillIndex [hangulCat]++;
DecompositionWide,
DecompositionNarrow,
};
+ private void AddCharMapGroup (char c, byte category, byte updateCount)
+ {
+ AddCharMapGroup (c, category, updateCount, 0, true);
+ }
+
private void AddCharMapGroup (char c, byte category, byte updateCount, byte level2)
{
AddCharMapGroup (c, category, updateCount, level2, false);
}
}
- private void AddArabicCharMap (char c)
+ private void AddArabicCharMap (char c, byte category, byte updateCount, byte level2)
{
- byte category = 6;
- byte updateCount = 1;
- byte level2 = 0;
-
// itself
AddCharMap (c, category, 0, level2);
// Arabic
if ('\u2135' <= c && c <= '\u2138')
return 4;
- byte [] arabicTmp = new byte [] {0x18, 0, 0x8, 0x10};
- if ('\uFEB5' <= c && c < '\uFEED' ||
- '\uFEF1' <= c && c < '\uFEF5')
- return arabicTmp [c % 4];
- if ('\uFE80' <= c && c < '\uFF00') {
+ // I believe that Windows has a bug on setting level 3
+ // weight here. NFKD results in different values.
+ if ('\uFE80' < c && c < '\uFF00') {
// 2(Isolated)/8(Final)/0x18(Medial)
switch (decompType [(int) c]) {
case DecompositionIsolated:
- return 2;
+ return 0; // 2;
case DecompositionFinal:
return 8;
case DecompositionMedial:
return 0x18;
+ case DecompositionInitial:
+ return 0x10;
}
}
+ // I have no idea why those symbols have level 3 weight
+ if (c == '\u2104' || c == '\u212B')
+ return 0x18;
+ if ('\u211E' <= c && c <= '\u212B')
+ return 0x10;
+
// actually I dunno the reason why they have weights.
switch (c) {
case '\u01BC':
byte ret = 0;
switch (c) {
case '\u03C2':
- case '\u2104':
case '\u212B':
ret = 8;
break;