public void RegexStateMachine()
{
// Initialize unicode-related classes to avoid having their cost in the time computation
var sw = Stopwatch.StartNew();
var provider = new UnicodeCharSetProvider();
provider.GetClassSet(CharSetClass.Digit);
provider.GetClassSet(CharSetClass.Dot);
provider.GetClassSet(CharSetClass.Space);
provider.GetClassSet(CharSetClass.Word);
UnicodeRanges.FromUnicodeName("InCombining_Diacritical_Marks");
sw.Stop();
this.output.WriteLine("Unicode init ms: " + sw.ElapsedMilliseconds);
sw = Stopwatch.StartNew();
RegexLexer.CreateStateMachine(out var stateMachine, out var startStateId);
var lapBuild = sw.ElapsedTicks;
var sm = stateMachine.Compile();
var lapCompile = sw.ElapsedTicks;
var startState = new Id <DfaState <LetterId> >(startStateId);
sm(ref startState, ' ');
var lapJit = sw.ElapsedTicks;
sm(ref startState, ' ');
sw.Stop();
this.output.WriteLine("Build ticks: " + lapBuild);
this.output.WriteLine("Compile ticks: " + (lapCompile - lapBuild));
this.output.WriteLine("JIT ticks: " + (lapJit - lapCompile));
this.output.WriteLine("Exec ticks: " + (sw.ElapsedTicks - lapJit));
this.output.WriteLine("Total ms: " + sw.ElapsedMilliseconds);
this.output.WriteLine(stateMachine.ToReadableString());
}
public void CharacterInSet(string ch, bool inSet, string unicodeName)
{
Codepoint parsedChar;
if (ch.StartsWith("\\"))
{
Assert.True(((RangeSetHandle.Static)RegexMatchSet.ParseEscape(ch)).TryGetSingle(out parsedChar));
}
else
{
parsedChar = ch.Single();
}
Assert.Equal(inSet, UnicodeRanges.FromUnicodeName(unicodeName).Contains(parsedChar));
}
public static RangeSetHandle ParseEscape(string escape)
{
var match = rxEscape.Match(escape);
if (!match.Success)
{
throw new ArgumentException("Escape is invalid", "escape");
}
if (match.Groups["name"].Success)
{
return(new RangeSetHandle.Static(UnicodeRanges.FromUnicodeName(match.Groups["name"].Value), match.Groups["c"].Value == "P"));
}
if (match.Groups["hex"].Success)
{
return(new RangeSetHandle.Static(Codepoint.Parse(match.Groups["hex"].Value)));
}
var c = match.Groups["c"].Value[0];
switch (c)
{
case '0':
return(new RangeSetHandle.Static('\0'));
case 'r':
return(new RangeSetHandle.Static('\r'));
case 'n':
return(new RangeSetHandle.Static('\n'));
case 't':
return(new RangeSetHandle.Static('\t'));
case 'a':
return(new RangeSetHandle.Static('\x07'));
case 'e':
return(new RangeSetHandle.Static('\x1B'));
case 'f':
return(new RangeSetHandle.Static('\x0C'));
case 'v':
return(new RangeSetHandle.Static('\x0B'));
case 'd':
return(new RangeSetHandle.Class(CharSetClass.Digit, false));
case 'D':
return(new RangeSetHandle.Class(CharSetClass.Digit, true));
case 'w':
return(new RangeSetHandle.Class(CharSetClass.Word, false));
case 'W':
return(new RangeSetHandle.Class(CharSetClass.Word, true));
case 's':
return(new RangeSetHandle.Class(CharSetClass.Space, false));
case 'S':
return(new RangeSetHandle.Class(CharSetClass.Space, true));
default:
if (char.IsLetterOrDigit(c))
{
throw new ArgumentOutOfRangeException(nameof(escape), "Invalid escape character " + c);
}
return(new RangeSetHandle.Static(c));
}
}
public static RegexMatchSet FromUnicode(string name, bool negate = false)
{
return(new RegexMatchSet($@"\{(negate ? 'P' : 'p')}{{{name}}}", new RangeSetHandle.Static(UnicodeRanges.FromUnicodeName(name), negate)));
}
private bool CheckUnicodeRanges(Validator v, OTFont fontOwner)
{
bool bRet = true;
bool bOk = true;
if (version == 0)
{
if (ulUnicodeRange1 != 0)
{
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "Range[0] was undefined. All bits must be 0.");
bOk = false;
}
if (ulUnicodeRange2 != 0)
{
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "Range[1] was undefined. All bits must be 0.");
bOk = false;
}
if (ulUnicodeRange3 != 0)
{
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "Range[2] was undefined. All bits must be 0.");
bOk = false;
}
if (ulUnicodeRange4 != 0)
{
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "Range[3] was undefined. All bits must be 0.");
bOk = false;
}
if (bOk)
{
v.Pass(P.OS_2_P_UnicodeRanges, m_tag);
}
else
{
bRet = false;
}
return bRet;
}
// count the number of entries in each unicode range in the cmap subtable
UnicodeRanges ur = new UnicodeRanges();
for (uint c = 0; c < 0xffff; c++)
{
// check if c is mapped to a glyph
uint iGlyph = fontOwner.FastMapUnicodeToGlyphID((char)c);
if (iGlyph != 0)
{
UnicodeRanges.Range r = ur.GetRange(c);
if (r != null)
{
r.count++;
}
}
}
uint nCharsInRange;
// bit 0
uint BASIC_LATIN_LOW = 0x0020;
nCharsInRange = ur.GetRange(BASIC_LATIN_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000001, nCharsInRange, "Basic Latin");
// bit 1
uint LATIN_1_SUPPLEMENT_LOW = 0x00A0;
nCharsInRange = ur.GetRange(LATIN_1_SUPPLEMENT_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000002, nCharsInRange, "Latin-1 Supplement");
// bit 2
uint LATIN_EXTENDED_A_LOW = 0x0100;
nCharsInRange = ur.GetRange(LATIN_EXTENDED_A_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000004, nCharsInRange, "Latin Extended-A");
// bit 3
uint LATIN_EXTENDED_B_LOW = 0x0180;
nCharsInRange = ur.GetRange(LATIN_EXTENDED_B_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000008, nCharsInRange, "Latin Extended-B");
// bit 4
uint IPA_EXTENSIONS_LOW = 0x0250;
nCharsInRange = ur.GetRange(IPA_EXTENSIONS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000010, nCharsInRange, "IPA Extensions");
// bit 5
uint SPACING_MODIFIER_LETTERS_LOW = 0x02B0;
nCharsInRange = ur.GetRange(SPACING_MODIFIER_LETTERS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000020, nCharsInRange, "Spacing Modifier Letters");
// bit 6
uint COMBINING_DIACRITICAL_MARKS_LOW = 0x0300;
nCharsInRange = ur.GetRange(COMBINING_DIACRITICAL_MARKS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000040, nCharsInRange, "Combining Diacritical Marks");
// bit 7
uint GREEK_LOW = 0x0370;
nCharsInRange = ur.GetRange(GREEK_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000080, nCharsInRange, "Greek");
// bit 8 reserved
if ((ulUnicodeRange1 & 0x00000100) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #8");
}
// bit 9
uint CYRILLIC_LOW = 0x0400;
uint CYRILLIC_SUPPLEMENTARY_LOW = 0x0500;
nCharsInRange = ur.GetRange(CYRILLIC_LOW).count
+ ur.GetRange(CYRILLIC_SUPPLEMENTARY_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange1, 0x00000200, nCharsInRange, "Cyrillic, Cyrillic Supplementary");
// bit 10
uint ARMENIAN_LOW = 0x0530;
nCharsInRange = ur.GetRange(ARMENIAN_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000400, nCharsInRange, "Armenian");
// bit 11
uint HEBREW_LOW = 0x0590;
nCharsInRange = ur.GetRange(HEBREW_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000800, nCharsInRange, "Hebrew");
if (version > 1)
{
// bit 12 reserved
if ((ulUnicodeRange1 & 0x00001000) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #12");
}
}
// bit 13
uint ARABIC_LOW = 0x0600;
nCharsInRange = ur.GetRange(ARABIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00002000, nCharsInRange, "Arabic");
if (version > 1)
{
// bit 14 reserved
if ((ulUnicodeRange1 & 0x00004000) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #14");
}
}
// bit 15
uint DEVANAGARI_LOW = 0x0900;
nCharsInRange = ur.GetRange(DEVANAGARI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00008000, nCharsInRange, "Devanagari");
// bit 16
uint BENGALI_LOW = 0x0980;
nCharsInRange = ur.GetRange(BENGALI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00010000, nCharsInRange, "Bengali");
// bit 17
uint GURMUKHI_LOW = 0x0A00;
nCharsInRange = ur.GetRange(GURMUKHI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00020000, nCharsInRange, "Gurmukhi");
// bit 18
uint GUJARATI_LOW = 0x0A80;
nCharsInRange = ur.GetRange(GUJARATI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00040000, nCharsInRange, "Gujarati");
// bit 19
uint ORIYA_LOW = 0x0B00;
nCharsInRange = ur.GetRange(ORIYA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00080000, nCharsInRange, "Oriya");
// bit 20
uint TAMIL_LOW = 0x0B80;
nCharsInRange = ur.GetRange(TAMIL_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00100000, nCharsInRange, "Tamil");
// bit 21
uint TELUGU_LOW = 0x0C00;
nCharsInRange = ur.GetRange(TELUGU_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00200000, nCharsInRange, "Telugu");
// bit 22
uint KANNADA_LOW = 0x0C80;
nCharsInRange = ur.GetRange(KANNADA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00400000, nCharsInRange, "Kannada");
// bit 23
uint MALAYALAM_LOW = 0x0D00;
nCharsInRange = ur.GetRange(MALAYALAM_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00800000, nCharsInRange, "Malayalam");
// bit 24
uint THAI_LOW = 0x0E00;
nCharsInRange = ur.GetRange(THAI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x01000000, nCharsInRange, "Thai");
// bit 25
uint LAO_LOW = 0x0E80;
nCharsInRange = ur.GetRange(LAO_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x02000000, nCharsInRange, "Lao");
// bit 26
uint GEORGIAN_LOW = 0x10A0;
nCharsInRange = ur.GetRange(GEORGIAN_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x04000000, nCharsInRange, "Georgian");
if (version > 1)
{
// bit 27 reserved
if ((ulUnicodeRange1 & 0x08000000) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #27");
}
}
// bit 28
uint HANGUL_JAMO_LOW = 0x1100;
nCharsInRange = ur.GetRange(HANGUL_JAMO_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x10000000, nCharsInRange, "Hangul Jamo");
// bit 29
uint LATIN_EXTENDED_ADDITIONAL_LOW = 0x1E00;
nCharsInRange = ur.GetRange(LATIN_EXTENDED_ADDITIONAL_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x20000000, nCharsInRange, "Latin Extended Additional");
// bit 30
uint GREEK_EXTENDED_LOW = 0x1F00;
nCharsInRange = ur.GetRange(GREEK_EXTENDED_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x40000000, nCharsInRange, "Greek Extended");
// bit 31
uint GENERAL_PUNCTUATION_LOW = 0x2000;
nCharsInRange = ur.GetRange(GENERAL_PUNCTUATION_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x80000000, nCharsInRange, "General Punctuation");
// bit 32
uint SUPER_SUB_SCRIPTS_LOW = 0x2070;
nCharsInRange = ur.GetRange(SUPER_SUB_SCRIPTS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000001, nCharsInRange, "Superscripts and Subscripts");
// bit 33
uint CURRENCY_SYMBOLS_LOW = 0x20A0;
nCharsInRange = ur.GetRange(CURRENCY_SYMBOLS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000002, nCharsInRange, "Currency Symbols");
// bit 34
uint SYMBOL_COMBINING_MARKS_LOW = 0x20D0;
nCharsInRange = ur.GetRange(SYMBOL_COMBINING_MARKS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000004, nCharsInRange, "Combining Diacritical marks for symbols");
// bit 35
uint LETTERLIKE_SYMBOLS_LOW = 0x2100;
nCharsInRange = ur.GetRange(LETTERLIKE_SYMBOLS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000008, nCharsInRange, "Letterlike Symbols");
// bit 36
uint NUMBER_FORMS_LOW = 0x2150;
nCharsInRange = ur.GetRange(NUMBER_FORMS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000010, nCharsInRange, "Number Forms");
// bit 37
uint ARROWS_LOW = 0x2190;
uint SUPPLEMENTAL_ARROWS_A_LOW = 0x27F0;
uint SUPPLEMENTAL_ARROWS_B_LOW = 0x2900;
nCharsInRange = ur.GetRange(ARROWS_LOW).count
+ ur.GetRange(SUPPLEMENTAL_ARROWS_A_LOW).count
+ ur.GetRange(SUPPLEMENTAL_ARROWS_B_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x00000020, nCharsInRange, "Arrows, Supplementary Arrows A, Supplementary Arrows B");
// bit 38
uint MATH_OPERATORS_LOW = 0x2200;
uint SUPPLEMENTAL_MATH_OPERATORS_LOW = 0x2A00;
uint MISC_MATH_SYMBOLS_A_LOW = 0x27C0;
uint MISC_MATH_SYMBOLS_B_LOW = 0x2980;
nCharsInRange = ur.GetRange(MATH_OPERATORS_LOW).count
+ ur.GetRange(SUPPLEMENTAL_MATH_OPERATORS_LOW).count
+ ur.GetRange(MISC_MATH_SYMBOLS_A_LOW).count
+ ur.GetRange(MISC_MATH_SYMBOLS_B_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x00000040, nCharsInRange, "Mathematical Operators, Supplemental Mathematical Operators, Mathematical Symbols A, Mathematical Symbols B");
// bit 39
uint MISC_TECHNICAL_LOW = 0x2300;
nCharsInRange = ur.GetRange(MISC_TECHNICAL_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000080, nCharsInRange, "Miscellaneous Technical");
// bit 40
uint CONTROL_PICTURES_LOW = 0x2400;
nCharsInRange = ur.GetRange(CONTROL_PICTURES_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000100, nCharsInRange, "Control Pictures");
// bit 41
uint OCR_LOW = 0x2440;
nCharsInRange = ur.GetRange(OCR_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000200, nCharsInRange, "Optical Character Recognition");
// bit 42
uint ENCLOSED_ALPHANUMERICS_LOW = 0x2460;
nCharsInRange = ur.GetRange(ENCLOSED_ALPHANUMERICS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000400, nCharsInRange, "Enclosed Alphanumerics");
// bit 43
uint BOX_DRAWING_LOW = 0x2500;
nCharsInRange = ur.GetRange(BOX_DRAWING_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000800, nCharsInRange, "Box Drawing");
// bit 44
uint BLOCK_ELEMENTS_LOW = 0x2580;
nCharsInRange = ur.GetRange(BLOCK_ELEMENTS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00001000, nCharsInRange, "Block Elements");
// bit 45
uint GEOMETRIC_SHAPES_LOW = 0x25A0;
nCharsInRange = ur.GetRange(GEOMETRIC_SHAPES_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00002000, nCharsInRange, "Geometric Shapes");
// bit 46
uint MISC_SYMBOLS_LOW = 0x2600;
nCharsInRange = ur.GetRange(MISC_SYMBOLS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00004000, nCharsInRange, "Miscellaneous Symbols");
// bit 47
uint DINGBATS_LOW = 0x2700;
nCharsInRange = ur.GetRange(DINGBATS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00008000, nCharsInRange, "Dingbats");
// bit 48
uint CJK_SYMBOLS_PUNCTUATION_LOW = 0x3000;
nCharsInRange = ur.GetRange(CJK_SYMBOLS_PUNCTUATION_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00010000, nCharsInRange, "CJK Symbols and Punctuation");
// bit 49
uint HIRAGANA_LOW = 0x3040;
nCharsInRange = ur.GetRange(HIRAGANA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00020000, nCharsInRange, "Hiragana");
// bit 50
uint KATAKANA_LOW = 0x30A0;
nCharsInRange = ur.GetRange(KATAKANA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00040000, nCharsInRange, "Katakana");
// bit 51
uint BOPOMOFO_LOW = 0x3100;
uint BOPOMOFO_EXTENDED_LOW = 0x31A0;
nCharsInRange = ur.GetRange(BOPOMOFO_LOW).count
+ ur.GetRange(BOPOMOFO_EXTENDED_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x00080000, nCharsInRange, "Bopomofo, Bopomofo Extended");
// bit 52
uint HANGUL_COMPAT_JAMO_LOW = 0x3130;
nCharsInRange = ur.GetRange(HANGUL_COMPAT_JAMO_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00100000, nCharsInRange, "Hangul Compatibility Jamo");
// bit 53 !!! OT Spec 1.3 says bit 53 is CJK Misc !!!
// !!! Since there's no unicode range name !!!
// !!! that maps nicely to CJK Misc, !!!
// !!! I naturally just ignore it. !!!
/*
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00200000, nCharsInRange, "CJK Miscellaneous");
*/
// bit 54
uint ENCLOSED_CJK_LETTERS_MONTHS_LOW = 0x3200;
nCharsInRange = ur.GetRange(ENCLOSED_CJK_LETTERS_MONTHS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00400000, nCharsInRange, "Enclosed CJK Letters and Months");
// bit 55
uint CJK_COMPATIBILITY_LOW = 0x3300;
nCharsInRange = ur.GetRange(CJK_COMPATIBILITY_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00800000, nCharsInRange, "CJK Compatibility");
// bit 56
uint HANGUL_LOW = 0xAC00;
nCharsInRange = ur.GetRange(HANGUL_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x01000000, nCharsInRange, "Hangul");
// bit 57 surrogates
if (version > 1)
{
bool bSurrogatesCmapPresent = false;
Table_cmap cmapTable = (Table_cmap)fontOwner.GetTable("cmap");
if (cmapTable != null)
{
Table_cmap.EncodingTableEntry eteUniSurrogates = cmapTable.GetEncodingTableEntry(3,10);
if (eteUniSurrogates != null)
bSurrogatesCmapPresent = true;
}
if ((ulUnicodeRange2 & 0x02000000) == 0)
{
if (bSurrogatesCmapPresent)
{
v.Error(T.T_NULL, E.OS_2_E_SurrogatesBitClear, m_tag);
bOk = false;
}
}
else
{
if (!bSurrogatesCmapPresent)
{
v.Error(T.T_NULL, E.OS_2_E_SurrogatesBitSet, m_tag);
bOk = false;
}
}
}
// bit 58 reserved
if ((ulUnicodeRange2 & 0x04000000) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #58");
}
// bit 59
uint CJK_UNIFIED_IDEOGRAPHS_LOW = 0x4E00;
uint CJK_RADICALS_SUPPLEMENT_LOW = 0x2E80;
uint KANGXI_RADICALS_LOW = 0x2F00;
uint IDEOGRAPHIC_DESCRIPTION_CHARS_LOW = 0x2FF0;
uint CJK_UNIFIED_IDEOGRAPHS_EXT_A_LOW = 0x3400;
uint CJK_UNIFIED_IDEOGRAPHS_EXT_B_LOW = 0x20000;
uint KANBUN_LOW = 0x3190;
nCharsInRange = ur.GetRange(CJK_UNIFIED_IDEOGRAPHS_LOW).count
+ ur.GetRange(CJK_RADICALS_SUPPLEMENT_LOW).count
+ ur.GetRange(KANGXI_RADICALS_LOW).count
+ ur.GetRange(IDEOGRAPHIC_DESCRIPTION_CHARS_LOW).count
+ ur.GetRange(CJK_UNIFIED_IDEOGRAPHS_EXT_A_LOW).count
+ ur.GetRange(CJK_UNIFIED_IDEOGRAPHS_EXT_B_LOW).count
+ ur.GetRange(KANBUN_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x08000000, nCharsInRange, "CJK Unified Ideographs, CJK Radicals Supplement, Kangxi Radicals, Ideographic Description Chars, CJK Unified Ideographs Extended A, CJK Unified Ideographs Extended B, Kanbun");
// bit 60
uint PRIVATE_USE_AREA_LOW = 0xE000;
if (!fontOwner.ContainsMsSymbolEncodedCmap())
{
nCharsInRange = ur.GetRange(PRIVATE_USE_AREA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x10000000, nCharsInRange, "Private Use Area");
}
// bit 61
uint CJK_COMPATIBILITY_IDEOGRAPHS_LOW = 0xF900;
uint CJK_COMPATIBILITY_IDEO_SUPP_LOW = 0x2F800;
nCharsInRange = ur.GetRange(CJK_COMPATIBILITY_IDEOGRAPHS_LOW).count
+ ur.GetRange(CJK_COMPATIBILITY_IDEO_SUPP_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x20000000, nCharsInRange, "CJK Compatibility Ideographs, CJK Compatibility Ideographs Supplement");
// bit 62
uint ALPHABETIC_PRESENTATION_FORMS_LOW = 0xFB00;
nCharsInRange = ur.GetRange(ALPHABETIC_PRESENTATION_FORMS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x40000000, nCharsInRange, "Alphabetic Presentation Forms");
// bit 63
uint ARABIC_PRESENTATION_FORMS_A_LOW = 0xFB50;
nCharsInRange = ur.GetRange(ARABIC_PRESENTATION_FORMS_A_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x80000000, nCharsInRange, "Arabic Presentation Forms-A");
// bit 64
uint COMBINING_HALF_MARKS_LOW = 0xFE20;
nCharsInRange = ur.GetRange(COMBINING_HALF_MARKS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000001, nCharsInRange, "Combining Half Marks");
// bit 65
uint CJK_COMPATIBILITY_FORMS_LOW = 0xFE30;
nCharsInRange = ur.GetRange(CJK_COMPATIBILITY_FORMS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000002, nCharsInRange, "CJK Compatibility Forms");
// bit 66
uint SMALL_FORM_VARIANTS_LOW = 0xFE50;
nCharsInRange = ur.GetRange(SMALL_FORM_VARIANTS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000004, nCharsInRange, "Small Form Variants");
// bit 67
uint ARABIC_PRESENTATION_FORMS_B_LOW = 0xFE70;
nCharsInRange = ur.GetRange(ARABIC_PRESENTATION_FORMS_B_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000008, nCharsInRange, "Arabic Presentation Forms-B");
// bit 68
uint HALFWIDTH_FULLWIDTH_FORMS_LOW = 0xFF00;
nCharsInRange = ur.GetRange(HALFWIDTH_FULLWIDTH_FORMS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000010, nCharsInRange, "Halfwidth and Fullwidth Forms");
// bit 69
uint SPECIALS_LOW = 0xFFF0;
nCharsInRange = ur.GetRange(SPECIALS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000020, nCharsInRange, "Specials");
if (version < 2)
{
for (int bitpos = 6; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange3 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (64+bitpos)); }
}
for (int bitpos = 0; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange4 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (96+bitpos)); }
}
}
else
{
// bit 70
uint TIBETAN_LOW = 0x0F00;
nCharsInRange = ur.GetRange(TIBETAN_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000040, nCharsInRange, "Tibetan");
// bit 71
uint SYRIAC_LOW = 0x0700;
nCharsInRange = ur.GetRange(SYRIAC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000080, nCharsInRange, "Syriac");
// bit 72
uint THAANA_LOW = 0x0780;
nCharsInRange = ur.GetRange(THAANA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000100, nCharsInRange, "Thaana");
// bit 73
uint SINHALA_LOW = 0x0D80;
nCharsInRange = ur.GetRange(SINHALA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000200, nCharsInRange, "Sinhala");
// bit 74
uint MYANMAR_LOW = 0x1000;
nCharsInRange = ur.GetRange(MYANMAR_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000400, nCharsInRange, "Myanmar");
// bit 75
uint ETHIOPIC_LOW = 0x1200;
nCharsInRange = ur.GetRange(ETHIOPIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000800, nCharsInRange, "Ethiopic");
// bit 76
uint CHEROKEE_LOW = 0x13A0;
nCharsInRange = ur.GetRange(CHEROKEE_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00001000, nCharsInRange, "Cherokee");
// bit 77
uint UNIFIED_CANADIAN_AB_SYL_LOW = 0x1400;
nCharsInRange = ur.GetRange(UNIFIED_CANADIAN_AB_SYL_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00002000, nCharsInRange, "Unified Canadian Syllabics");
// bit 78
uint OGHAM_LOW = 0x1680;
nCharsInRange = ur.GetRange(OGHAM_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00004000, nCharsInRange, "Ogham");
// bit 79
uint RUNIC_LOW = 0x16A0;
nCharsInRange = ur.GetRange(RUNIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00008000, nCharsInRange, "Runic");
// bit 80
uint KHMER_LOW = 0x1780;
nCharsInRange = ur.GetRange(KHMER_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00010000, nCharsInRange, "Khmer");
// bit 81
uint MONGOLIAN_LOW = 0x1800;
nCharsInRange = ur.GetRange(MONGOLIAN_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00020000, nCharsInRange, "Mongolian");
// bit 82
uint BRAILLE_PATTERNS_LOW = 0x2800;
nCharsInRange = ur.GetRange(BRAILLE_PATTERNS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00040000, nCharsInRange, "Braille");
// bit 83
uint YI_LOW = 0xA000;
uint YI_RADICALS_LOW = 0xA490;
nCharsInRange = ur.GetRange(YI_LOW).count
+ ur.GetRange(YI_RADICALS_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange3, 0x00080000, nCharsInRange, "Yi, Yi Radicals");
if (version < 3)
{
for (int bitpos = 20; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange3 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (64+bitpos)); }
}
for (int bitpos = 0; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange4 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (96+bitpos)); }
}
}
else
{
// bit 84
uint TAGALOG_LOW = 0x1700;
uint HANUNOO_LOW = 0x1720;
uint BUHID_LOW = 0x1740;
uint TAGBANWA_LOW = 0x1760;
nCharsInRange = ur.GetRange(TAGALOG_LOW).count
+ ur.GetRange(HANUNOO_LOW).count
+ ur.GetRange(BUHID_LOW).count
+ ur.GetRange(TAGBANWA_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange3, 0x00100000, nCharsInRange, "Tagalog, Hanunoo, Buhid, Tagbanwa");
// bit 85
uint OLD_ITALIC_LOW = 0x10300;
nCharsInRange = ur.GetRange(OLD_ITALIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00200000, nCharsInRange, "Old Italic");
// bit 86
uint GOTHIC_LOW = 0x10330;
nCharsInRange = ur.GetRange(GOTHIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00400000, nCharsInRange, "Gothic");
// bit 87
uint DESERET_LOW = 0x10400;
nCharsInRange = ur.GetRange(DESERET_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00800000, nCharsInRange, "Deseret");
// bit 88
uint BYZANTINE_MUSICAL_SYMBOLS_LOW = 0x1D000;
uint MUSICAL_SYMBOLS_LOW = 0x1D100;
nCharsInRange = ur.GetRange(BYZANTINE_MUSICAL_SYMBOLS_LOW).count
+ ur.GetRange(MUSICAL_SYMBOLS_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange3, 0x01000000, nCharsInRange, "Byzantine Musical Symbols, Musical Symbols");
// bit 89
uint MATHEMATICAL_ALPHANUMERIC_LOW = 0x1D400;
nCharsInRange = ur.GetRange(MATHEMATICAL_ALPHANUMERIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x02000000, nCharsInRange, "Mathematical Alphanumeric Symbols");
// bit 90
uint PRIVATE_USE_15_LOW = 0xFFF80;
uint PRIVATE_USE_16_LOW = 0x10FF80;
nCharsInRange = ur.GetRange(PRIVATE_USE_15_LOW).count
+ ur.GetRange(PRIVATE_USE_16_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x04000000, nCharsInRange, "Private Use (Plane 15), Private Use (Plane 16)");
// bit 91
uint VARIATION_SELECTORS_LOW = 0xE0100;
nCharsInRange = ur.GetRange(VARIATION_SELECTORS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x08000000, nCharsInRange, "Variation Selectors");
// bit 92
uint TAGS_LOW = 0xE0000;
nCharsInRange = ur.GetRange(TAGS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x10000000, nCharsInRange, "Tags");
for (int bitpos = 29; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange3 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (64+bitpos)); }
}
for (int bitpos = 0; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange4 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (96+bitpos)); }
}
}
}
if (bOk)
{
v.Pass(P.OS_2_P_UnicodeRanges, m_tag);
}
else
{
bRet = false;
}
return bRet;
}
private bool CheckUnicodeRanges(Validator v, OTFont fontOwner)
{
bool bRet = true;
bool bOk = true;
if (version == 0)
{
if (ulUnicodeRange1 != 0)
{
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "Range[0] was undefined. All bits must be 0.");
bOk = false;
}
if (ulUnicodeRange2 != 0)
{
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "Range[1] was undefined. All bits must be 0.");
bOk = false;
}
if (ulUnicodeRange3 != 0)
{
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "Range[2] was undefined. All bits must be 0.");
bOk = false;
}
if (ulUnicodeRange4 != 0)
{
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "Range[3] was undefined. All bits must be 0.");
bOk = false;
}
if (bOk)
{
v.Pass(P.OS_2_P_UnicodeRanges, m_tag);
}
else
{
bRet = false;
}
return bRet;
}
// count the number of entries in each unicode range in the cmap subtable
UnicodeRanges ur = new UnicodeRanges();
for (uint c = 0; c < 0xffff; c++)
{
// check if c is mapped to a glyph
uint iGlyph = fontOwner.FastMapUnicodeToGlyphID((char)c);
if (iGlyph != 0)
{
UnicodeRanges.Range r = ur.GetRange(c);
if (r != null)
{
r.count++;
}
}
}
uint nCharsInRange;
// bit 0
uint BASIC_LATIN_LOW = 0x0020;
nCharsInRange = ur.GetRange(BASIC_LATIN_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000001, nCharsInRange, "Basic Latin");
// bit 1
uint LATIN_1_SUPPLEMENT_LOW = 0x00A0;
nCharsInRange = ur.GetRange(LATIN_1_SUPPLEMENT_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000002, nCharsInRange, "Latin-1 Supplement");
// bit 2
uint LATIN_EXTENDED_A_LOW = 0x0100;
nCharsInRange = ur.GetRange(LATIN_EXTENDED_A_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000004, nCharsInRange, "Latin Extended-A");
// bit 3
uint LATIN_EXTENDED_B_LOW = 0x0180;
nCharsInRange = ur.GetRange(LATIN_EXTENDED_B_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000008, nCharsInRange, "Latin Extended-B");
// bit 4
uint IPA_EXTENSIONS_LOW = 0x0250;
uint Phonetic_Extensions = 0x1D00;
uint Phonetic_Extensions_Supplement = 0x1D80;
nCharsInRange = ur.GetRange(IPA_EXTENSIONS_LOW).count
+ ur.GetRange(Phonetic_Extensions).count
+ ur.GetRange(Phonetic_Extensions_Supplement).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000010, nCharsInRange, "IPA Extensions");
// bit 5
uint SPACING_MODIFIER_LETTERS_LOW = 0x02B0;
uint Modifier_Tone_Letters = 0xA700;
nCharsInRange = ur.GetRange(SPACING_MODIFIER_LETTERS_LOW).count
+ ur.GetRange(Modifier_Tone_Letters).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000020, nCharsInRange, "Spacing Modifier Letters");
// bit 6
uint COMBINING_DIACRITICAL_MARKS_LOW = 0x0300;
uint Combining_Diacritical_Marks_Supplement = 0x1DC0;
nCharsInRange = ur.GetRange(COMBINING_DIACRITICAL_MARKS_LOW).count
+ ur.GetRange(Combining_Diacritical_Marks_Supplement).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000040, nCharsInRange, "Combining Diacritical Marks");
// bit 7
uint GREEK_LOW = 0x0370;
nCharsInRange = ur.GetRange(GREEK_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000080, nCharsInRange, "Greek");
// v1: Greek Symbols and Coptic
// v2/v3: Reserved for Unicode SubRanges
// v4: Coptic 2C80-2CFF
if (version > 1 && version < 4)
{
// bit 8 reserved
if ((ulUnicodeRange1 & 0x00000100) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #8");
}
}
// TODO: v4
// bit 9
uint CYRILLIC_LOW = 0x0400;
uint CYRILLIC_SUPPLEMENTARY_LOW = 0x0500;
uint Cyrillic_Extended_A = 0x2DE0;
uint Cyrillic_Extended_B = 0xA640;
nCharsInRange = ur.GetRange(CYRILLIC_LOW).count
+ ur.GetRange(CYRILLIC_SUPPLEMENTARY_LOW).count
+ ur.GetRange(Cyrillic_Extended_A).count
+ ur.GetRange(Cyrillic_Extended_B).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange1, 0x00000200, nCharsInRange, "Cyrillic, Cyrillic Supplementary");
// bit 10
uint ARMENIAN_LOW = 0x0530;
nCharsInRange = ur.GetRange(ARMENIAN_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000400, nCharsInRange, "Armenian");
// bit 11
uint HEBREW_LOW = 0x0590;
nCharsInRange = ur.GetRange(HEBREW_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00000800, nCharsInRange, "Hebrew");
// v1: Hebrew Extended (A and B blocks combined)
// v2/v3: Reserved for Unicode SubRanges
// v4: Vai A500-A63F
if (version > 1 && version < 4)
{
// bit 12 reserved
if ((ulUnicodeRange1 & 0x00001000) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #12");
}
}
// TODO: v4
// bit 13
uint ARABIC_LOW = 0x0600;
uint Arabic_Supplement = 0x0750;
nCharsInRange = ur.GetRange(ARABIC_LOW).count
+ ur.GetRange(Arabic_Supplement).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00002000, nCharsInRange, "Arabic");
// v1: Arabic Extended
// v2/v3: Reserved for Unicode SubRanges
// v4: NKo 07C0-07FF
if (version > 1 && version < 4)
{
// bit 14 reserved
if ((ulUnicodeRange1 & 0x00004000) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #14");
}
}
// TODO: v4
// bit 15
uint DEVANAGARI_LOW = 0x0900;
nCharsInRange = ur.GetRange(DEVANAGARI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00008000, nCharsInRange, "Devanagari");
// bit 16
uint BENGALI_LOW = 0x0980;
nCharsInRange = ur.GetRange(BENGALI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00010000, nCharsInRange, "Bengali");
// bit 17
uint GURMUKHI_LOW = 0x0A00;
nCharsInRange = ur.GetRange(GURMUKHI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00020000, nCharsInRange, "Gurmukhi");
// bit 18
uint GUJARATI_LOW = 0x0A80;
nCharsInRange = ur.GetRange(GUJARATI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00040000, nCharsInRange, "Gujarati");
// bit 19
uint ORIYA_LOW = 0x0B00;
nCharsInRange = ur.GetRange(ORIYA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00080000, nCharsInRange, "Oriya");
// bit 20
uint TAMIL_LOW = 0x0B80;
nCharsInRange = ur.GetRange(TAMIL_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00100000, nCharsInRange, "Tamil");
// bit 21
uint TELUGU_LOW = 0x0C00;
nCharsInRange = ur.GetRange(TELUGU_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00200000, nCharsInRange, "Telugu");
// bit 22
uint KANNADA_LOW = 0x0C80;
nCharsInRange = ur.GetRange(KANNADA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00400000, nCharsInRange, "Kannada");
// bit 23
uint MALAYALAM_LOW = 0x0D00;
nCharsInRange = ur.GetRange(MALAYALAM_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x00800000, nCharsInRange, "Malayalam");
// bit 24
uint THAI_LOW = 0x0E00;
nCharsInRange = ur.GetRange(THAI_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x01000000, nCharsInRange, "Thai");
// bit 25
uint LAO_LOW = 0x0E80;
nCharsInRange = ur.GetRange(LAO_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x02000000, nCharsInRange, "Lao");
// v3: Georgian
// v4: Georgian 10A0-10FF
// Georgian Supplement 2D00-2D2F
// TODO: v4
// bit 26
uint GEORGIAN_LOW = 0x10A0;
uint Georgian_Supplement = 0x2D00;
nCharsInRange = ur.GetRange(GEORGIAN_LOW).count
+ ur.GetRange(Georgian_Supplement).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x04000000, nCharsInRange, "Georgian");
// v1: Georgian Extended
// v2/v3: Reserved for Unicode SubRanges
// v4: Balinese 1B00-1B7F
if (version > 1 && version < 4)
{
// bit 27 reserved
if ((ulUnicodeRange1 & 0x08000000) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #27");
}
}
// TODO: v4
// bit 28
uint HANGUL_JAMO_LOW = 0x1100;
nCharsInRange = ur.GetRange(HANGUL_JAMO_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x10000000, nCharsInRange, "Hangul Jamo");
// bit 29
uint LATIN_EXTENDED_ADDITIONAL_LOW = 0x1E00;
uint Latin_Extended_C = 0x2C60;
uint Latin_Extended_D = 0xA720;
nCharsInRange = ur.GetRange(LATIN_EXTENDED_ADDITIONAL_LOW).count
+ ur.GetRange(Latin_Extended_C).count
+ ur.GetRange(Latin_Extended_D).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x20000000, nCharsInRange, "Latin Extended Additional");
// bit 30
uint GREEK_EXTENDED_LOW = 0x1F00;
nCharsInRange = ur.GetRange(GREEK_EXTENDED_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x40000000, nCharsInRange, "Greek Extended");
// bit 31
uint GENERAL_PUNCTUATION_LOW = 0x2000;
uint Supplemental_Punctuation = 0x2E00;
nCharsInRange = ur.GetRange(GENERAL_PUNCTUATION_LOW).count
+ ur.GetRange(Supplemental_Punctuation).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange1, 0x80000000, nCharsInRange, "General Punctuation");
// bit 32
uint SUPER_SUB_SCRIPTS_LOW = 0x2070;
nCharsInRange = ur.GetRange(SUPER_SUB_SCRIPTS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000001, nCharsInRange, "Superscripts and Subscripts");
// bit 33
uint CURRENCY_SYMBOLS_LOW = 0x20A0;
nCharsInRange = ur.GetRange(CURRENCY_SYMBOLS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000002, nCharsInRange, "Currency Symbols");
// bit 34
uint SYMBOL_COMBINING_MARKS_LOW = 0x20D0;
nCharsInRange = ur.GetRange(SYMBOL_COMBINING_MARKS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000004, nCharsInRange, "Combining Diacritical marks for symbols");
// bit 35
uint LETTERLIKE_SYMBOLS_LOW = 0x2100;
nCharsInRange = ur.GetRange(LETTERLIKE_SYMBOLS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000008, nCharsInRange, "Letterlike Symbols");
// bit 36
uint NUMBER_FORMS_LOW = 0x2150;
nCharsInRange = ur.GetRange(NUMBER_FORMS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000010, nCharsInRange, "Number Forms");
// bit 37
uint ARROWS_LOW = 0x2190;
uint SUPPLEMENTAL_ARROWS_A_LOW = 0x27F0;
uint SUPPLEMENTAL_ARROWS_B_LOW = 0x2900;
uint Miscellaneous_Symbols_and_Arrows = 0x2B00;
nCharsInRange = ur.GetRange(ARROWS_LOW).count
+ ur.GetRange(SUPPLEMENTAL_ARROWS_A_LOW).count
+ ur.GetRange(SUPPLEMENTAL_ARROWS_B_LOW).count
+ ur.GetRange(Miscellaneous_Symbols_and_Arrows).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x00000020, nCharsInRange, "Arrows, Supplementary Arrows A, Supplementary Arrows B");
// bit 38
uint MATH_OPERATORS_LOW = 0x2200;
uint SUPPLEMENTAL_MATH_OPERATORS_LOW = 0x2A00;
uint MISC_MATH_SYMBOLS_A_LOW = 0x27C0;
uint MISC_MATH_SYMBOLS_B_LOW = 0x2980;
nCharsInRange = ur.GetRange(MATH_OPERATORS_LOW).count
+ ur.GetRange(SUPPLEMENTAL_MATH_OPERATORS_LOW).count
+ ur.GetRange(MISC_MATH_SYMBOLS_A_LOW).count
+ ur.GetRange(MISC_MATH_SYMBOLS_B_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x00000040, nCharsInRange, "Mathematical Operators, Supplemental Mathematical Operators, Mathematical Symbols A, Mathematical Symbols B");
// bit 39
uint MISC_TECHNICAL_LOW = 0x2300;
nCharsInRange = ur.GetRange(MISC_TECHNICAL_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000080, nCharsInRange, "Miscellaneous Technical");
// bit 40
uint CONTROL_PICTURES_LOW = 0x2400;
nCharsInRange = ur.GetRange(CONTROL_PICTURES_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000100, nCharsInRange, "Control Pictures");
// bit 41
uint OCR_LOW = 0x2440;
nCharsInRange = ur.GetRange(OCR_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000200, nCharsInRange, "Optical Character Recognition");
// bit 42
uint ENCLOSED_ALPHANUMERICS_LOW = 0x2460;
nCharsInRange = ur.GetRange(ENCLOSED_ALPHANUMERICS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000400, nCharsInRange, "Enclosed Alphanumerics");
// bit 43
uint BOX_DRAWING_LOW = 0x2500;
nCharsInRange = ur.GetRange(BOX_DRAWING_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00000800, nCharsInRange, "Box Drawing");
// bit 44
uint BLOCK_ELEMENTS_LOW = 0x2580;
nCharsInRange = ur.GetRange(BLOCK_ELEMENTS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00001000, nCharsInRange, "Block Elements");
// bit 45
uint GEOMETRIC_SHAPES_LOW = 0x25A0;
nCharsInRange = ur.GetRange(GEOMETRIC_SHAPES_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00002000, nCharsInRange, "Geometric Shapes");
// bit 46
uint MISC_SYMBOLS_LOW = 0x2600;
nCharsInRange = ur.GetRange(MISC_SYMBOLS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00004000, nCharsInRange, "Miscellaneous Symbols");
// bit 47
uint DINGBATS_LOW = 0x2700;
nCharsInRange = ur.GetRange(DINGBATS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00008000, nCharsInRange, "Dingbats");
// bit 48
uint CJK_SYMBOLS_PUNCTUATION_LOW = 0x3000;
nCharsInRange = ur.GetRange(CJK_SYMBOLS_PUNCTUATION_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00010000, nCharsInRange, "CJK Symbols and Punctuation");
// bit 49
uint HIRAGANA_LOW = 0x3040;
nCharsInRange = ur.GetRange(HIRAGANA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00020000, nCharsInRange, "Hiragana");
// bit 50
uint KATAKANA_LOW = 0x30A0;
uint Katakana_Phonetic_Extensions = 0x31F0;
nCharsInRange = ur.GetRange(KATAKANA_LOW).count
+ ur.GetRange(Katakana_Phonetic_Extensions).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00040000, nCharsInRange, "Katakana");
// bit 51
uint BOPOMOFO_LOW = 0x3100;
uint BOPOMOFO_EXTENDED_LOW = 0x31A0;
nCharsInRange = ur.GetRange(BOPOMOFO_LOW).count
+ ur.GetRange(BOPOMOFO_EXTENDED_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x00080000, nCharsInRange, "Bopomofo, Bopomofo Extended");
// bit 52
uint HANGUL_COMPAT_JAMO_LOW = 0x3130;
nCharsInRange = ur.GetRange(HANGUL_COMPAT_JAMO_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00100000, nCharsInRange, "Hangul Compatibility Jamo");
// v2: CJK Miscellaneous
// v3: Reserved for Unicode SubRanges
// v4: Phags-pa A840-A87F
// bit 53 !!! OT Spec 1.3 says bit 53 is CJK Misc !!!
// !!! Since there's no unicode range name !!!
// !!! that maps nicely to CJK Misc, !!!
// !!! I naturally just ignore it. !!!
/*
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00200000, nCharsInRange, "CJK Miscellaneous");
*/
// TODO: v4
// bit 54
uint ENCLOSED_CJK_LETTERS_MONTHS_LOW = 0x3200;
nCharsInRange = ur.GetRange(ENCLOSED_CJK_LETTERS_MONTHS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00400000, nCharsInRange, "Enclosed CJK Letters and Months");
// bit 55
uint CJK_COMPATIBILITY_LOW = 0x3300;
nCharsInRange = ur.GetRange(CJK_COMPATIBILITY_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x00800000, nCharsInRange, "CJK Compatibility");
// bit 56
uint HANGUL_LOW = 0xAC00;
nCharsInRange = ur.GetRange(HANGUL_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x01000000, nCharsInRange, "Hangul");
// v1: Reserved for Unicode SubRanges
// v2: Surrogates
// v3: Non-Plane 0 *
// v4: Non-Plane 0 * D800-DFFF
// bit 57 surrogates
if (version > 1)
{
bool bSurrogatesCmapPresent = false;
Table_cmap cmapTable = (Table_cmap)fontOwner.GetTable("cmap");
if (cmapTable != null)
{
Table_cmap.EncodingTableEntry eteUniSurrogates = cmapTable.GetEncodingTableEntry(3,10);
if (eteUniSurrogates != null)
bSurrogatesCmapPresent = true;
}
if ((ulUnicodeRange2 & 0x02000000) == 0)
{
if (bSurrogatesCmapPresent)
{
v.Error(T.T_NULL, E.OS_2_E_SurrogatesBitClear, m_tag);
bOk = false;
}
}
else
{
if (!bSurrogatesCmapPresent)
{
v.Error(T.T_NULL, E.OS_2_E_SurrogatesBitSet, m_tag);
bOk = false;
}
}
}
// v1/v2/v3: Reserved for Unicode SubRanges
// v4: Phoenician 10900-1091F
if (version < 4)
{
// bit 58 reserved
if ((ulUnicodeRange2 & 0x04000000) != 0)
{
bOk = false;
v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #58");
}
}
// TODO: v4
// bit 59
uint CJK_UNIFIED_IDEOGRAPHS_LOW = 0x4E00;
uint CJK_RADICALS_SUPPLEMENT_LOW = 0x2E80;
uint KANGXI_RADICALS_LOW = 0x2F00;
uint IDEOGRAPHIC_DESCRIPTION_CHARS_LOW = 0x2FF0;
uint CJK_UNIFIED_IDEOGRAPHS_EXT_A_LOW = 0x3400;
uint CJK_UNIFIED_IDEOGRAPHS_EXT_B_LOW = 0x20000;
uint KANBUN_LOW = 0x3190;
nCharsInRange = ur.GetRange(CJK_UNIFIED_IDEOGRAPHS_LOW).count
+ ur.GetRange(CJK_RADICALS_SUPPLEMENT_LOW).count
+ ur.GetRange(KANGXI_RADICALS_LOW).count
+ ur.GetRange(IDEOGRAPHIC_DESCRIPTION_CHARS_LOW).count
+ ur.GetRange(CJK_UNIFIED_IDEOGRAPHS_EXT_A_LOW).count
+ ur.GetRange(CJK_UNIFIED_IDEOGRAPHS_EXT_B_LOW).count
+ ur.GetRange(KANBUN_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x08000000, nCharsInRange, "CJK Unified Ideographs, CJK Radicals Supplement, Kangxi Radicals, Ideographic Description Chars, CJK Unified Ideographs Extended A, CJK Unified Ideographs Extended B, Kanbun");
// bit 60
uint PRIVATE_USE_AREA_LOW = 0xE000;
if (!fontOwner.ContainsMsSymbolEncodedCmap())
{
nCharsInRange = ur.GetRange(PRIVATE_USE_AREA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x10000000, nCharsInRange, "Private Use Area");
}
// bit 61
uint CJK_Strokes = 0x31C0;
uint CJK_COMPATIBILITY_IDEOGRAPHS_LOW = 0xF900;
uint CJK_COMPATIBILITY_IDEO_SUPP_LOW = 0x2F800;
nCharsInRange = ur.GetRange(CJK_Strokes).count
+ ur.GetRange(CJK_COMPATIBILITY_IDEOGRAPHS_LOW).count
+ ur.GetRange(CJK_COMPATIBILITY_IDEO_SUPP_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange2, 0x20000000, nCharsInRange, "CJK Compatibility Ideographs, CJK Compatibility Ideographs Supplement");
// bit 62
uint ALPHABETIC_PRESENTATION_FORMS_LOW = 0xFB00;
nCharsInRange = ur.GetRange(ALPHABETIC_PRESENTATION_FORMS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x40000000, nCharsInRange, "Alphabetic Presentation Forms");
// bit 63
uint ARABIC_PRESENTATION_FORMS_A_LOW = 0xFB50;
nCharsInRange = ur.GetRange(ARABIC_PRESENTATION_FORMS_A_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange2, 0x80000000, nCharsInRange, "Arabic Presentation Forms-A");
// bit 64
uint COMBINING_HALF_MARKS_LOW = 0xFE20;
nCharsInRange = ur.GetRange(COMBINING_HALF_MARKS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000001, nCharsInRange, "Combining Half Marks");
// bit 65
uint Vertical_Forms = 0xFE10;
uint CJK_COMPATIBILITY_FORMS_LOW = 0xFE30;
nCharsInRange = ur.GetRange(Vertical_Forms).count
+ ur.GetRange(CJK_COMPATIBILITY_FORMS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000002, nCharsInRange, "CJK Compatibility Forms");
// bit 66
uint SMALL_FORM_VARIANTS_LOW = 0xFE50;
nCharsInRange = ur.GetRange(SMALL_FORM_VARIANTS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000004, nCharsInRange, "Small Form Variants");
// bit 67
uint ARABIC_PRESENTATION_FORMS_B_LOW = 0xFE70;
nCharsInRange = ur.GetRange(ARABIC_PRESENTATION_FORMS_B_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000008, nCharsInRange, "Arabic Presentation Forms-B");
// bit 68
uint HALFWIDTH_FULLWIDTH_FORMS_LOW = 0xFF00;
nCharsInRange = ur.GetRange(HALFWIDTH_FULLWIDTH_FORMS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000010, nCharsInRange, "Halfwidth and Fullwidth Forms");
// bit 69
uint SPECIALS_LOW = 0xFFF0;
nCharsInRange = ur.GetRange(SPECIALS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000020, nCharsInRange, "Specials");
// v1: 70-127 Reserved for Unicode SubRanges
if (version < 2)
{
for (int bitpos = 6; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange3 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (64+bitpos)); }
}
for (int bitpos = 0; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange4 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (96+bitpos)); }
}
}
else
{
// bit 70
uint TIBETAN_LOW = 0x0F00;
nCharsInRange = ur.GetRange(TIBETAN_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000040, nCharsInRange, "Tibetan");
// bit 71
uint SYRIAC_LOW = 0x0700;
nCharsInRange = ur.GetRange(SYRIAC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000080, nCharsInRange, "Syriac");
// bit 72
uint THAANA_LOW = 0x0780;
nCharsInRange = ur.GetRange(THAANA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000100, nCharsInRange, "Thaana");
// bit 73
uint SINHALA_LOW = 0x0D80;
nCharsInRange = ur.GetRange(SINHALA_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000200, nCharsInRange, "Sinhala");
// bit 74
uint MYANMAR_LOW = 0x1000;
nCharsInRange = ur.GetRange(MYANMAR_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000400, nCharsInRange, "Myanmar");
// bit 75
uint ETHIOPIC_LOW = 0x1200;
uint Ethiopic_Supplement = 0x1380;
uint Ethiopic_Extended = 0x2D80;
nCharsInRange = ur.GetRange(ETHIOPIC_LOW).count
+ ur.GetRange(Ethiopic_Supplement).count
+ ur.GetRange(Ethiopic_Extended).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00000800, nCharsInRange, "Ethiopic");
// bit 76
uint CHEROKEE_LOW = 0x13A0;
nCharsInRange = ur.GetRange(CHEROKEE_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00001000, nCharsInRange, "Cherokee");
// bit 77
uint UNIFIED_CANADIAN_AB_SYL_LOW = 0x1400;
nCharsInRange = ur.GetRange(UNIFIED_CANADIAN_AB_SYL_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00002000, nCharsInRange, "Unified Canadian Syllabics");
// bit 78
uint OGHAM_LOW = 0x1680;
nCharsInRange = ur.GetRange(OGHAM_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00004000, nCharsInRange, "Ogham");
// bit 79
uint RUNIC_LOW = 0x16A0;
nCharsInRange = ur.GetRange(RUNIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00008000, nCharsInRange, "Runic");
// bit 80
uint KHMER_LOW = 0x1780;
uint Khmer_Symbols = 0x19E0;
nCharsInRange = ur.GetRange(KHMER_LOW).count
+ ur.GetRange(Khmer_Symbols).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00010000, nCharsInRange, "Khmer");
// bit 81
uint MONGOLIAN_LOW = 0x1800;
nCharsInRange = ur.GetRange(MONGOLIAN_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00020000, nCharsInRange, "Mongolian");
// bit 82
uint BRAILLE_PATTERNS_LOW = 0x2800;
nCharsInRange = ur.GetRange(BRAILLE_PATTERNS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00040000, nCharsInRange, "Braille");
// bit 83
uint YI_LOW = 0xA000;
uint YI_RADICALS_LOW = 0xA490;
nCharsInRange = ur.GetRange(YI_LOW).count
+ ur.GetRange(YI_RADICALS_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange3, 0x00080000, nCharsInRange, "Yi, Yi Radicals");
// v2: 84-127 Reserved for Unicode SubRanges
if (version < 3)
{
for (int bitpos = 20; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange3 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (64+bitpos)); }
}
for (int bitpos = 0; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange4 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (96+bitpos)); }
}
}
else
{
// bit 84
uint TAGALOG_LOW = 0x1700;
uint HANUNOO_LOW = 0x1720;
uint BUHID_LOW = 0x1740;
uint TAGBANWA_LOW = 0x1760;
nCharsInRange = ur.GetRange(TAGALOG_LOW).count
+ ur.GetRange(HANUNOO_LOW).count
+ ur.GetRange(BUHID_LOW).count
+ ur.GetRange(TAGBANWA_LOW).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange3, 0x00100000, nCharsInRange, "Tagalog, Hanunoo, Buhid, Tagbanwa");
// bit 85
uint OLD_ITALIC_LOW = 0x10300;
nCharsInRange = ur.GetRange(OLD_ITALIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00200000, nCharsInRange, "Old Italic");
// bit 86
uint GOTHIC_LOW = 0x10330;
nCharsInRange = ur.GetRange(GOTHIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00400000, nCharsInRange, "Gothic");
// bit 87
uint DESERET_LOW = 0x10400;
nCharsInRange = ur.GetRange(DESERET_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x00800000, nCharsInRange, "Deseret");
// bit 88
uint BYZANTINE_MUSICAL_SYMBOLS_LOW = 0x1D000;
uint MUSICAL_SYMBOLS_LOW = 0x1D100;
uint Ancient_Greek_Musical_Notation = 0x1D200;
nCharsInRange = ur.GetRange(BYZANTINE_MUSICAL_SYMBOLS_LOW).count
+ ur.GetRange(MUSICAL_SYMBOLS_LOW).count
+ ur.GetRange(Ancient_Greek_Musical_Notation).count;
bOk &= VerifyUnicodeRanges(v, ulUnicodeRange3, 0x01000000, nCharsInRange, "Byzantine Musical Symbols, Musical Symbols");
// bit 89
uint MATHEMATICAL_ALPHANUMERIC_LOW = 0x1D400;
nCharsInRange = ur.GetRange(MATHEMATICAL_ALPHANUMERIC_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x02000000, nCharsInRange, "Mathematical Alphanumeric Symbols");
// bit 90
uint PRIVATE_USE_15_LOW = 0xFFF80;
uint PRIVATE_USE_16_LOW = 0x10FF80;
nCharsInRange = ur.GetRange(PRIVATE_USE_15_LOW).count
+ ur.GetRange(PRIVATE_USE_16_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x04000000, nCharsInRange, "Private Use (Plane 15), Private Use (Plane 16)");
// bit 91
uint Variation_Selectors = 0xFE00;
uint VARIATION_SELECTORS_SUPP = 0xE0100;
nCharsInRange = ur.GetRange(Variation_Selectors).count
+ ur.GetRange(VARIATION_SELECTORS_SUPP).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x08000000, nCharsInRange, "Variation Selectors");
// bit 92
uint TAGS_LOW = 0xE0000;
nCharsInRange = ur.GetRange(TAGS_LOW).count;
bOk &= VerifyUnicodeRange(v, ulUnicodeRange3, 0x10000000, nCharsInRange, "Tags");
// v3: 93-127 Reserved for Unicode SubRanges
if (version < 4)
{
for (int bitpos = 29; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange3 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (64+bitpos)); }
}
for (int bitpos = 0; bitpos < 32; bitpos++)
{
if ((ulUnicodeRange4 & (1<<bitpos)) != 0) { bOk = false; v.Error(T.T_NULL, E.OS_2_E_ReservedBitSet_Unicode, m_tag, "bit #" + (96+bitpos)); }
}
}
else
{
// v4 addition
/*
93 Limbu 1900-194F
94 Tai Le 1950-197F
95 New Tai Lue 1980-19DF
96 Buginese 1A00-1A1F
97 Glagolitic 2C00-2C5F
98 Tifinagh 2D30-2D7F
99 Yijing Hexagram Symbols 4DC0-4DFF
100 Syloti Nagri A800-A82F
101 Linear B Syllabary 10000-1007F
Linear B Ideograms 10080-100FF
Aegean Numbers 10100-1013F
102 Ancient Greek Numbers 10140-1018F
103 Ugaritic 10380-1039F
104 Old Persian 103A0-103DF
105 Shavian 10450-1047F
106 Osmanya 10480-104AF
107 Cypriot Syllabary 10800-1083F
108 Kharoshthi 10A00-10A5F
109 Tai Xuan Jing Symbols 1D300-1D35F
110 Cuneiform 12000-123FF
Cuneiform Numbers and Punctuation 12400-1247F
111 Counting Rod Numerals 1D360-1D37F
112 Sundanese 1B80-1BBF
113 Lepcha 1C00-1C4F
114 Ol Chiki 1C50-1C7F
115 Saurashtra A880-A8DF
116 Kayah Li A900-A92F
117 Rejang A930-A95F
118 Cham AA00-AA5F
119 Ancient Symbols 10190-101CF
120 Phaistos Disc 101D0-101FF
121 Carian 102A0-102DF
Lycian 10280-1029F
Lydian 10920-1093F
122 Domino Tiles 1F030-1F09F
Mahjong Tiles 1F000-1F02F
123-127 Reserved for process-internal usage
*/
// TODO: v4 addition
}
}
}
if (bOk)
{
v.Pass(P.OS_2_P_UnicodeRanges, m_tag);
}
else
{
bRet = false;
}
return bRet;
}
public GrmGrammar()
{
this.Resolve = ((IReadOnlyDictionary <SymbolId, string>) new Dictionary <SymbolId, string>()
{
{ SymbolId.Eof, "(EOF)" },
{ SymUnknown, "(Unknown)" },
{ SymWhitespace, "(Whitespace)" },
{ SymNewline, "(Newline)" },
{ SymLineComment, "(LineComment)" },
{ SymBlockComment, "(BlockComment)" },
{ SymParameterName, "ParameterName" },
{ SymNonterminal, "Nonterminal" },
{ SymTerminal, "Terminal" },
{ SymAssign, "=" },
{ SymDefine, "::=" },
{ SymQuestion, "?" },
{ SymStar, "*" },
{ SymParensOpen, "(" },
{ SymParensClose, ")" },
{ SymPlus, "+" },
{ SymMinus, "-" },
{ SymOr, "|" },
{ SymSetLiteral, "SetLiteral" },
{ SymSetName, "SetName" },
{ SymInit, "<Init>" },
{ SymGrammar, "<Grammar>" },
{ SymContent, "<Content>" },
{ SymDefinition, "<Definition>" },
{ SymNlOpt, "<NlOpt>" },
{ SymNl, "<Nl>" },
{ SymParameter, "<Parameter>" },
{ SymParameterBody, "<ParameterBody>" },
{ SymParameterItems, "<ParameterItems>" },
{ SymParameterItem, "<ParameterItem>" },
{ SymSetDecl, "<SetDecl>" },
{ SymSetExp, "<SetExp>" },
{ SymSetItem, "<SetItem>" },
{ SymTerminalDecl, "<TerminalDecl>" },
{ SymTerminalName, "<TerminalName>" },
{ SymRegExp, "<RegExp>" },
{ SymRegExpSeq, "<RegExpSeq>" },
{ SymRegExpItem, "<RegExpItem>" },
{ SymRegExp2, "<RegExp2>" },
{ SymKleeneOpt, "<KleeneOpt>" },
{ SymRuleDecl, "<RuleDecl>" },
{ SymHandles, "<Handles>" },
{ SymHandle, "<Handle>" },
{ SymSymbol, "<Symbol>" },
})
.CreateGetter();
var mapper = new UnicodeUtf16Mapper(false, false);
var charsetPrintable = Codepoints.ValidBmp - UnicodeRanges.FromUnicodeCategory(UnicodeCategory.Control) - UnicodeRanges.InCombiningDiacriticalMarks;
var charsetAlphanumeric = UnicodeRanges.Letter | UnicodeRanges.Number;
var charset = new UnicodeCharSetProvider(new Dictionary <string, RangeSet <Codepoint> >()
{
{ "Parameter Ch", charsetPrintable - (Codepoint)'\'' - (Codepoint)'"' },
{ "Nonterminal Ch", charsetAlphanumeric | '_' | '-' | '.' | ' ' },
{ "Terminal Ch", charsetAlphanumeric | '_' | '-' | '.' },
{ "Literal Ch", charsetPrintable - (Codepoint)'\'' },
{ "Set Literal Ch", charsetPrintable - (Codepoint)'[' - (Codepoint)']' - (Codepoint)'\'' },
{ "Set Name Ch", charsetPrintable - (Codepoint)'{' - (Codepoint)'}' },
{ "Whitespace Ch", UnicodeRanges.SpaceSeparator | '\t' | '\v' }
});
this.DfaStateMachine = new LexerBuilder <char>(mapper, Utf16Chars.EOF, charset)
{
{ SymParameterName, @"""{Parameter Ch}+""" },
{ SymNonterminal, @"<{Nonterminal Ch}+>" },
{ SymTerminal, @"{Terminal Ch}+|'{Literal Ch}*'" },
{ SymSetLiteral, @"\[({Set Literal Ch}+|'{Literal Ch}*')+\]" },
{ SymSetName, @"\{{Set Name Ch}+\}" },
{ SymWhitespace, @"{Whitespace Ch}+" },
{ SymNewline, @"\r\n?|\n\r?" },
{ SymLineComment, @"![^\r\n]*" },
{ SymBlockComment, @"!\*([^\*]|\*[^!])*\*!" },
{ SymAssign, @"=" },
{ SymDefine, @"::=" },
{ SymPlus, @"\+" },
{ SymMinus, @"\-" },
{ SymOr, @"\|" },
{ SymQuestion, @"\?" },
{ SymStar, @"\*" },
{ SymParensOpen, @"\(" },
{ SymParensClose, @"\)" }
}
.CreateStateMachine(out var dfaStartState)
.Compile();
this.DfaStartState = dfaStartState;
this.LalrTable = new LalrTableGenerator(new GrammarBuilder(SymUnknown, SymInit, SymGrammar)
{
{ SymGrammar, SymNlOpt, SymContent },
{ SymContent, SymContent, SymDefinition },
{ SymContent, SymDefinition },
{ SymDefinition, SymParameter },
{ SymDefinition, SymSetDecl },
{ SymDefinition, SymTerminalDecl },
{ SymDefinition, SymRuleDecl },
{ SymNlOpt, SymNewline, SymNlOpt },
{ SymNlOpt },
{ SymNl, SymNewline, SymNl },
{ SymNl, SymNewline },
{ SymParameter, SymParameterName, SymNlOpt, SymAssign, SymParameterBody, SymNl },
{ SymParameterBody, SymParameterBody, SymNlOpt, SymOr, SymParameterItems },
{ SymParameterBody, SymParameterItems },
{ SymParameterItems, SymParameterItems, SymParameterItem },
{ SymParameterItems, SymParameterItem },
{ SymParameterItem, SymParameterName },
{ SymParameterItem, SymTerminal },
{ SymParameterItem, SymSetLiteral },
{ SymParameterItem, SymSetName },
{ SymParameterItem, SymNonterminal },
{ SymSetDecl, SymSetName, SymNlOpt, SymAssign, SymSetExp, SymNl },
{ SymSetExp, SymSetExp, SymNlOpt, SymPlus, SymSetItem },
{ SymSetExp, SymSetExp, SymNlOpt, SymMinus, SymSetItem },
{ SymSetExp, SymSetItem },
{ SymSetItem, SymSetLiteral },
{ SymSetItem, SymSetName },
{ SymTerminalDecl, SymTerminalName, SymNlOpt, SymAssign, SymRegExp, SymNl },
{ SymTerminalName, SymTerminalName, SymTerminal },
{ SymTerminalName, SymTerminal },
{ SymRegExp, SymRegExp, SymNlOpt, SymOr, SymRegExpSeq },
{ SymRegExp, SymRegExpSeq },
{ SymRegExpSeq, SymRegExpSeq, SymRegExpItem },
{ SymRegExpSeq, SymRegExpItem },
{ SymRegExpItem, SymSetLiteral, SymKleeneOpt },
{ SymRegExpItem, SymSetName, SymKleeneOpt },
{ SymRegExpItem, SymTerminal, SymKleeneOpt },
{ SymRegExpItem, SymParensOpen, SymRegExp2, SymParensClose, SymKleeneOpt },
{ SymRegExp2, SymRegExp2, SymOr, SymRegExpSeq },
{ SymRegExp2, SymRegExpSeq },
{ SymKleeneOpt, SymPlus },
{ SymKleeneOpt, SymQuestion },
{ SymKleeneOpt, SymStar },
{ SymKleeneOpt },
{ SymRuleDecl, SymNonterminal, SymNlOpt, SymDefine, SymHandles, SymNl },
{ SymHandles, SymHandles, SymNlOpt, SymOr, SymHandle },
{ SymHandles, SymHandle },
{ SymHandle, SymHandle, SymSymbol },
{ SymHandle },
{ SymSymbol, SymTerminal },
{ SymSymbol, SymNonterminal }
})
.ComputeTable();
}
