mahta.fetrat
/
HomoFast-eSpeak-Persian
Watch 1
Star 0
Fork 0
Code Issues 0 Pull Requests 0 Releases 0 Wiki Activity
Browse Source

Merge commit 'ff018e33df5763dc0ff1644c9479ee2351bee45e'

master
Reece H. Dunn 8 years ago
parent
415868cc00 ff018e33df
commit
2615242f05
11 changed files with 2039 additions and 294 deletions
Split View Show Diff Stats
  1. 1
    0
      src/ucd-tools/CHANGELOG.md
  2. 1
    0
      src/ucd-tools/Makefile.am
  3. 5
    4
      src/ucd-tools/README.md
  4. 13
    285
      src/ucd-tools/src/ctype.c
  5. 96
    0
      src/ucd-tools/src/include/ucd/ucd.h
  6. 1826
    0
      src/ucd-tools/src/proplist.c
  7. 4
    1
      src/ucd-tools/tests/printcdata.c
  8. 4
    1
      src/ucd-tools/tests/printucddata.c
  9. 4
    1
      src/ucd-tools/tests/printucddata_cpp.cpp
  10. 39
    0
      src/ucd-tools/tools/mkencodingtable
  11. 46
    2
      src/ucd-tools/tools/printdata.py

+ 1
- 0
src/ucd-tools/CHANGELOG.md View File

@@ -4,6 +4,7 @@

* Add `iswblank` and `iswxdigit` compatibility.
* Improve ctype compatibility.
* PropList property lookup.

## 9.0.0 - 2016-12-28


+ 1
- 0
src/ucd-tools/Makefile.am View File

@@ -122,6 +122,7 @@ src_libucd_la_SOURCES = \
src/case.c \
src/categories.c \
src/ctype.c \
src/proplist.c \
src/scripts.c \
src/tostring.c


+ 5
- 4
src/ucd-tools/README.md View File

@@ -65,10 +65,11 @@ Doxygen documentation format.

The library exposes the following properties from the UCD data files:

| Property | Description |
|--------------------|-------------|
| `General_Category` | A [General Category Value](http://www.unicode.org/reports/tr44/#General_Category_Values), including the higher-level grouping. |
| `Script` | An [ISO 15924](http://www.unicode.org/iso15924/iso15924-codes.html) script code. |
| C API | C++ API | Data | Description |
|-----------------------|------------------------|-------------|-------------|
| `ucd_lookup_category` | `ucd::lookup_category` | UnicodeData | A [General Category Value](http://www.unicode.org/reports/tr44/#General_Category_Values). |
| `ucd_lookup_script` | `ucd::lookup_script` | Script | An [ISO 15924](http://www.unicode.org/iso15924/iso15924-codes.html) script code. |
| `ucd_properties` | `ucd::properties` | PropList | The code point properties from the PropList Unicode data file. |

### Case Conversion


+ 13
- 285
src/ucd-tools/src/ctype.c View File

@@ -20,262 +20,10 @@

#include "ucd/ucd.h"

static int other_alphabetic_MnMcSo(codepoint_t c)
{
switch (c & 0xFFFFFF00)
{
case 0x0300:
return c == 0x0345;
case 0x0500:
return (c >= 0x05B0 && c <= 0x05BD)
|| c == 0x05BF
|| (c >= 0x05C1 && c <= 0x05C2)
|| (c >= 0x05C4 && c <= 0x05C5)
|| c == 0x05C7;
case 0x0600:
return (c >= 0x0610 && c <= 0x061A)
|| (c >= 0x064B && c <= 0x0657)
|| (c >= 0x0659 && c <= 0x065F)
|| c == 0x0670
|| (c >= 0x06D6 && c <= 0x06DC)
|| (c >= 0x06E1 && c <= 0x06E4)
|| (c >= 0x06E7 && c <= 0x06E8)
|| c == 0x06ED;
case 0x0700:
return c == 0x0711
|| (c >= 0x0730 && c <= 0x073F)
|| (c >= 0x07A6 && c <= 0x07B0);
case 0x0800:
return (c >= 0x0816 && c <= 0x0817)
|| (c >= 0x081B && c <= 0x0823)
|| (c >= 0x0825 && c <= 0x0827)
|| (c >= 0x0829 && c <= 0x082C)
|| (c >= 0x08D4 && c <= 0x08DF)
|| (c >= 0x08E3 && c <= 0x08E9)
|| c >= 0x08F0;
case 0x0900:
return (c >= 0x0900 && c <= 0x0903)
|| (c >= 0x093A && c <= 0x093B)
|| (c >= 0x093E && c <= 0x094C)
|| (c >= 0x094E && c <= 0x094F)
|| (c >= 0x0955 && c <= 0x0957)
|| (c >= 0x0962 && c <= 0x0963)
|| (c >= 0x0981 && c <= 0x0983)
|| (c >= 0x0981 && c <= 0x0983)
|| (c >= 0x09BE && c <= 0x09C4)
|| (c >= 0x09C7 && c <= 0x09C8)
|| (c >= 0x09CB && c <= 0x09CC)
|| c == 0x09D7
|| (c >= 0x09E2 && c <= 0x09E3);
case 0x0A00:
return (c >= 0x0A01 && c <= 0x0A03)
|| (c >= 0x0A3E && c <= 0x0A42)
|| (c >= 0x0A47 && c <= 0x0A48)
|| (c >= 0x0A4B && c <= 0x0A4C)
|| c == 0x0A51
|| (c >= 0x0A70 && c <= 0x0A71)
|| c == 0x0A75
|| (c >= 0x0A81 && c <= 0x0A83)
|| (c >= 0x0ABE && c <= 0x0AC5)
|| (c >= 0x0AC7 && c <= 0x0AC9)
|| (c >= 0x0ACB && c <= 0x0ACC)
|| (c >= 0x0AE2 && c <= 0x0AE3);
case 0x0B00:
return (c >= 0x0B01 && c <= 0x0B03)
|| (c >= 0x0B3E && c <= 0x0B44)
|| (c >= 0x0B47 && c <= 0x0B48)
|| (c >= 0x0B4B && c <= 0x0B4C)
|| (c >= 0x0B56 && c <= 0x0B57)
|| (c >= 0x0B62 && c <= 0x0B63)
|| c == 0x0B82
|| (c >= 0x0BBE && c <= 0x0BC2)
|| (c >= 0x0BC6 && c <= 0x0BC8)
|| (c >= 0x0BCA && c <= 0x0BCC)
|| c == 0x0BD7;
case 0x0C00:
return (c >= 0x0C00 && c <= 0x0C03)
|| (c >= 0x0C3E && c <= 0x0C44)
|| (c >= 0x0C46 && c <= 0x0C48)
|| (c >= 0x0C4A && c <= 0x0C4C)
|| (c >= 0x0C55 && c <= 0x0C56)
|| (c >= 0x0C62 && c <= 0x0C63)
|| (c >= 0x0C81 && c <= 0x0C83)
|| (c >= 0x0CBE && c <= 0x0CBF)
|| (c >= 0x0CC0 && c <= 0x0CC4)
|| (c >= 0x0CC6 && c <= 0x0CC8)
|| (c >= 0x0CCA && c <= 0x0CCC)
|| (c >= 0x0CD5 && c <= 0x0CD6)
|| (c >= 0x0CE2 && c <= 0x0CE3);
case 0x0D00:
return (c >= 0x0D01 && c <= 0x0D03)
|| (c >= 0x0D3E && c <= 0x0D44)
|| (c >= 0x0D46 && c <= 0x0D48)
|| (c >= 0x0D4A && c <= 0x0D4C)
|| c == 0x0D57
|| (c >= 0x0D62 && c <= 0x0D63)
|| (c >= 0x0D82 && c <= 0x0D83)
|| (c >= 0x0DCF && c <= 0x0DD4)
|| c == 0x0DD6
|| (c >= 0x0DD8 && c <= 0x0DDF)
|| (c >= 0x0DF2 && c <= 0x0DF3);
case 0x0E00:
return c == 0x0E31
|| (c >= 0x0E34 && c <= 0x0E3A)
|| c == 0x0E4D
|| c == 0x0EB1
|| (c >= 0x0EB4 && c <= 0x0EB9)
|| (c >= 0x0EBB && c <= 0x0EBD)
|| c == 0x0ECD;
case 0x0F00:
return (c >= 0x0F71 && c <= 0x0F7F)
|| (c >= 0x0F80 && c <= 0x0F81)
|| (c >= 0x0F8D && c <= 0x0F97)
|| (c >= 0x0F99 && c <= 0x0FBC);
case 0x1000:
return (c >= 0x102B && c <= 0x1036)
|| c == 0x1038
|| (c >= 0x103B && c <= 0x103E)
|| (c >= 0x1056 && c <= 0x1059)
|| (c >= 0x105E && c <= 0x1060)
|| c == 0x1062
|| (c >= 0x1067 && c <= 0x1068)
|| (c >= 0x1071 && c <= 0x1074)
|| (c >= 0x1082 && c <= 0x1086)
|| (c >= 0x109C && c <= 0x109D);
case 0x1300:
return c == 0x135F;
case 0x1700:
return (c >= 0x1712 && c <= 0x1713)
|| (c >= 0x1732 && c <= 0x1733)
|| (c >= 0x1752 && c <= 0x1753)
|| (c >= 0x1772 && c <= 0x1773)
|| (c >= 0x17B6 && c <= 0x17C8);
case 0x1800:
return (c >= 0x1885 && c <= 0x1886)
|| c == 0x18A9;
case 0x1900:
return (c >= 0x1920 && c <= 0x192B)
|| (c >= 0x1930 && c <= 0x1938);
case 0x1A00:
return (c >= 0x1A17 && c <= 0x1A1B)
|| (c >= 0x1A55 && c <= 0x1A5E)
|| (c >= 0x1A61 && c <= 0x1A74);
case 0x1B00:
return (c >= 0x1B00 && c <= 0x1B04)
|| (c >= 0x1B35 && c <= 0x1B43)
|| (c >= 0x1B80 && c <= 0x1B82)
|| (c >= 0x1BA1 && c <= 0x1BA9)
|| (c >= 0x1BAC && c <= 0x1BAD)
|| (c >= 0x1BE7 && c <= 0x1BF1);
case 0x1C00:
return (c >= 0x1C24 && c <= 0x1C35)
|| (c >= 0x1CF2 && c <= 0x1CF3);
case 0x1D00:
return (c >= 0x1DE7 && c <= 0x1DF4);
case 0x2400:
return (c >= 0x24B6 && c <= 0x24E9);
case 0x2D00:
return (c >= 0x2DE0 && c <= 0x2DFF);
case 0xA600:
return (c >= 0xA674 && c <= 0xA67B)
|| (c >= 0xA69E && c <= 0xA69F);
case 0xA800:
return (c >= 0xA823 && c <= 0xA827)
|| (c >= 0xA880 && c <= 0xA881)
|| (c >= 0xA8B4 && c <= 0xA8C3)
|| c == 0xA8C5;
case 0xA900:
return (c >= 0xA926 && c <= 0xA92A)
|| (c >= 0xA947 && c <= 0xA952)
|| (c >= 0xA980 && c <= 0xA983)
|| (c >= 0xA9B4 && c <= 0xA9BF);
case 0xAA00:
return (c >= 0xAA29 && c <= 0xAA36)
|| c == 0xAA43
|| (c >= 0xAA4C && c <= 0xAA4D)
|| c == 0xAAB0
|| (c >= 0xAAB2 && c <= 0xAAB4)
|| (c >= 0xAAB7 && c <= 0xAAB8)
|| c == 0xAABE
|| (c >= 0xAAEB && c <= 0xAAEF)
|| c == 0xAAF5;
case 0xAB00:
return (c >= 0xABE3 && c <= 0xABEA);
case 0xFB00:
return c == 0xFB1E;
case 0x10300:
return (c >= 0x10376 && c <= 0x1037A);
case 0x10A00:
return (c >= 0x10A01 && c <= 0x10A03)
|| (c >= 0x10A05 && c <= 0x10A06)
|| (c >= 0x10A0C && c <= 0x10A0F);
case 0x11000:
return (c >= 0x11000 && c <= 0x11002)
|| (c >= 0x11038 && c <= 0x11045)
|| c == 0x11082
|| (c >= 0x110B0 && c <= 0x110B8);
case 0x11100:
return (c >= 0x11100 && c <= 0x11102)
|| (c >= 0x11127 && c <= 0x11132)
|| (c >= 0x11180 && c <= 0x11182)
|| (c >= 0x111B3 && c <= 0x111BF);
case 0x11200:
return (c >= 0x1122C && c <= 0x11234)
|| c == 0x11237
|| c == 0x1123E
|| (c >= 0x112DF && c <= 0x112E8);
case 0x11300:
return (c >= 0x11300 && c <= 0x11303)
|| (c >= 0x1133E && c <= 0x11344)
|| (c >= 0x11347 && c <= 0x11348)
|| (c >= 0x1134B && c <= 0x1134C)
|| c == 0x11357
|| (c >= 0x11362 && c <= 0x11363);
case 0x11400:
return (c >= 0x11435 && c <= 0x11441)
|| (c >= 0x11443 && c <= 0x11445)
|| (c >= 0x114B0 && c <= 0x114C1);
case 0x11500:
return (c >= 0x115AF && c <= 0x115B5)
|| (c >= 0x115B8 && c <= 0x115BE)
|| (c >= 0x115DC && c <= 0x115DD);
case 0x11600:
return (c >= 0x11630 && c <= 0x1163E)
|| c == 0x11640
|| (c >= 0x116AB && c <= 0x116B5);
case 0x11700:
return (c >= 0x1171D && c <= 0x1172A);
case 0x11C00:
return (c >= 0x11C2F && c <= 0x11C36)
|| (c >= 0x11C38 && c <= 0x11C3E)
|| (c >= 0x11C92 && c <= 0x11CA7)
|| (c >= 0x11CA9 && c <= 0x11CB6);
case 0x16B00:
return (c >= 0x16B30 && c <= 0x16B36);
case 0x16F00:
return (c >= 0x16F51 && c <= 0x16F7E);
case 0x1BC00:
return c == 0x1BC9E;
case 0x1E000:
return (c >= 0x1E000 && c <= 0x1E006)
|| (c >= 0x1E008 && c <= 0x1E018)
|| (c >= 0x1E01B && c <= 0x1E021)
|| (c >= 0x1E023 && c <= 0x1E024)
|| (c >= 0x1E026 && c <= 0x1E02A);
case 0x1E900:
return c == 0x1E947;
case 0x1F100:
return (c >= 0x01F130 && c <= 0x01F149)
|| (c >= 0x01F150 && c <= 0x01F169)
|| (c >= 0x01F170 && c <= 0x01F189);
default:
return 0;
}
}

int ucd_isalnum(codepoint_t c)
{
switch (ucd_lookup_category(c))
ucd_category cat = ucd_lookup_category(c);
switch (cat)
{
case UCD_CATEGORY_Lu:
case UCD_CATEGORY_Ll:
@@ -289,7 +37,7 @@ int ucd_isalnum(codepoint_t c)
case UCD_CATEGORY_Mn:
case UCD_CATEGORY_Mc:
case UCD_CATEGORY_So:
return other_alphabetic_MnMcSo(c);
return (ucd_properties(c, cat) & UCD_PROPERTY_OTHER_ALPHABETIC) == UCD_PROPERTY_OTHER_ALPHABETIC;
default:
return 0;
}
@@ -297,7 +45,8 @@ int ucd_isalnum(codepoint_t c)

int ucd_isalpha(codepoint_t c)
{
switch (ucd_lookup_category(c))
ucd_category cat = ucd_lookup_category(c);
switch (cat)
{
case UCD_CATEGORY_Lu:
case UCD_CATEGORY_Ll:
@@ -309,7 +58,7 @@ int ucd_isalpha(codepoint_t c)
case UCD_CATEGORY_Mn:
case UCD_CATEGORY_Mc:
case UCD_CATEGORY_So:
return other_alphabetic_MnMcSo(c);
return (ucd_properties(c, cat) & UCD_PROPERTY_OTHER_ALPHABETIC) == UCD_PROPERTY_OTHER_ALPHABETIC;
default:
return 0;
}
@@ -366,37 +115,19 @@ int ucd_isgraph(codepoint_t c)

int ucd_islower(codepoint_t c)
{
switch (ucd_lookup_category(c))
ucd_category cat = ucd_lookup_category(c);
switch (cat)
{
case UCD_CATEGORY_Ll:
return 1;
case UCD_CATEGORY_Lt:
return ucd_toupper(c) != c;
case UCD_CATEGORY_Lo:
return c == 0xAA // Other_Lowercase : FEMININE ORDINAL INDICATOR
|| c == 0xBA; // Other_Lowercase : MASCULINE ORDINAL INDICATOR
case UCD_CATEGORY_Lm:
return (c >= 0x02B0 && c <= 0x02B8) // Other_Lowercase
|| (c >= 0x02C0 && c <= 0x02C1) // Other_Lowercase
|| (c >= 0x02E0 && c <= 0x02E4) // Other_Lowercase
|| c == 0x037A // Other_Lowercase
|| (c >= 0x1D2C && c <= 0x1D6A) // Other_Lowercase
|| c == 0x1D78 // Other_Lowercase
|| (c >= 0x1D9B && c <= 0x1DBF) // Other_Lowercase
|| c == 0x2071 // Other_Lowercase
|| c == 0x207F // Other_Lowercase
|| (c >= 0x2090 && c <= 0x209C) // Other_Lowercase
|| (c >= 0x2C7C && c <= 0x2C7D) // Other_Lowercase
|| (c >= 0xA69C && c <= 0xA69D) // Other_Lowercase
|| c == 0xA770 // Other_Lowercase
|| (c >= 0xA7F8 && c <= 0xA7F9) // Other_Lowercase
|| (c >= 0xAB5C && c <= 0xAB5F); // Other_Lowercase
case UCD_CATEGORY_Lo:
case UCD_CATEGORY_Mn:
return c == 0x0345; // Other_Lowercase : COMBINING GREEK YPOGEGRAMMENI
case UCD_CATEGORY_Nl:
return (c >= 0x2170 && c <= 0x217F); // Other_Lowercase
case UCD_CATEGORY_So:
return (c >= 0x24D0 && c <= 0x24E9); // Other_Lowercase
return (ucd_properties(c, cat) & UCD_PROPERTY_OTHER_LOWERCASE) == UCD_PROPERTY_OTHER_LOWERCASE;
default:
return 0;
}
@@ -469,19 +200,16 @@ int ucd_isspace(codepoint_t c)

int ucd_isupper(codepoint_t c)
{
switch (ucd_lookup_category(c))
ucd_category cat = ucd_lookup_category(c);
switch (cat)
{
case UCD_CATEGORY_Lu:
return 1;
case UCD_CATEGORY_Lt:
return ucd_tolower(c) != c;
case UCD_CATEGORY_Nl:
return (c >= 0x002160 && c <= 0x00216F); // Other_Uppercase
case UCD_CATEGORY_So:
return (c >= 0x0024B6 && c <= 0x0024CF) // Other_Uppercase
|| (c >= 0x01F130 && c <= 0x01F149) // Other_Uppercase
|| (c >= 0x01F150 && c <= 0x01F169) // Other_Uppercase
|| (c >= 0x01F170 && c <= 0x01F189); // Other_Uppercase
return (ucd_properties(c, cat) & UCD_PROPERTY_OTHER_UPPERCASE) == UCD_PROPERTY_OTHER_UPPERCASE;
default:
return 0;
}

+ 96
- 0
src/ucd-tools/src/include/ucd/ucd.h View File

@@ -325,6 +325,51 @@ const char *ucd_get_script_string(ucd_script s);
*/
ucd_script ucd_lookup_script(codepoint_t c);

/** @brief Properties
*/
typedef uint64_t ucd_property;
static const ucd_property UCD_PROPERTY_WHITE_SPACE = 0x0000000000000001ull; /**< @brief White_Space */
static const ucd_property UCD_PROPERTY_BIDI_CONTROL = 0x0000000000000002ull; /**< @brief Bidi_Control */
static const ucd_property UCD_PROPERTY_JOIN_CONTROL = 0x0000000000000004ull; /**< @brief Join_Control */
static const ucd_property UCD_PROPERTY_DASH = 0x0000000000000008ull; /**< @brief Dash */
static const ucd_property UCD_PROPERTY_HYPHEN = 0x0000000000000010ull; /**< @brief Hyphen */
static const ucd_property UCD_PROPERTY_QUOTATION_MARK = 0x0000000000000020ull; /**< @brief Quotation_Mark */
static const ucd_property UCD_PROPERTY_TERMINAL_PUNCTUATION = 0x0000000000000040ull; /**< @brief Terminal_Punctuation */
static const ucd_property UCD_PROPERTY_OTHER_MATH = 0x0000000000000080ull; /**< @brief Other_Math */
static const ucd_property UCD_PROPERTY_HEX_DIGIT = 0x0000000000000100ull; /**< @brief Hex_Digit */
static const ucd_property UCD_PROPERTY_ASCII_HEX_DIGIT = 0x0000000000000200ull; /**< @brief ASCII_Hex_Digit */
static const ucd_property UCD_PROPERTY_OTHER_ALPHABETIC = 0x0000000000000400ull; /**< @brief Other_Alphabetic */
static const ucd_property UCD_PROPERTY_IDEOGRAPHIC = 0x0000000000000800ull; /**< @brief Ideographic */
static const ucd_property UCD_PROPERTY_DIACRITIC = 0x0000000000001000ull; /**< @brief Diacritic */
static const ucd_property UCD_PROPERTY_EXTENDER = 0x0000000000002000ull; /**< @brief Extender */
static const ucd_property UCD_PROPERTY_OTHER_LOWERCASE = 0x0000000000004000ull; /**< @brief Other_Lowercase */
static const ucd_property UCD_PROPERTY_OTHER_UPPERCASE = 0x0000000000008000ull; /**< @brief Other_Uppercase */
static const ucd_property UCD_PROPERTY_NONCHARACTER_CODE_POINT = 0x0000000000010000ull; /**< @brief Noncharacter_Code_Point */
static const ucd_property UCD_PROPERTY_OTHER_GRAPHEME_EXTEND = 0x0000000000020000ull; /**< @brief Other_Grapheme_Extend */
static const ucd_property UCD_PROPERTY_IDS_BINARY_OPERATOR = 0x0000000000040000ull; /**< @brief IDS_Binary_Operator */
static const ucd_property UCD_PROPERTY_IDS_TRINARY_OPERATOR = 0x0000000000080000ull; /**< @brief IDS_Trinary_Operator */
static const ucd_property UCD_PROPERTY_RADICAL = 0x0000000000100000ull; /**< @brief Radical */
static const ucd_property UCD_PROPERTY_UNIFIED_IDEOGRAPH = 0x0000000000200000ull; /**< @brief Unified_Ideograph */
static const ucd_property UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT = 0x0000000000400000ull; /**< @brief Other_Default_Ignorable_Code_Point */
static const ucd_property UCD_PROPERTY_DEPRECATED = 0x0000000000800000ull; /**< @brief Deprecated */
static const ucd_property UCD_PROPERTY_SOFT_DOTTED = 0x0000000001000000ull; /**< @brief Soft_Dotted */
static const ucd_property UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION = 0x0000000002000000ull; /**< @brief Logical_Order_Exception */
static const ucd_property UCD_PROPERTY_OTHER_ID_START = 0x0000000004000000ull; /**< @brief Other_ID_Start */
static const ucd_property UCD_PROPERTY_OTHER_ID_CONTINUE = 0x0000000008000000ull; /**< @brief Other_ID_Continue */
static const ucd_property UCD_PROPERTY_SENTENCE_TERMINAL = 0x0000000010000000ull; /**< @brief Sentence_Terminal */
static const ucd_property UCD_PROPERTY_VARIATION_SELECTOR = 0x0000000020000000ull; /**< @brief Variation_Selector */
static const ucd_property UCD_PROPERTY_PATTERN_WHITE_SPACE = 0x0000000040000000ull; /**< @brief Pattern_White_Space */
static const ucd_property UCD_PROPERTY_PATTERN_SYNTAX = 0x0000000080000000ull; /**< @brief Pattern_Syntax */
static const ucd_property UCD_PROPERTY_PREPENDED_CONCATENATION_MARK = 0x0000000100000000ull; /**< @brief Prepended_Concatenation_Mark */

/** @brief Return the properties of the specified codepoint.
*
* @param c The Unicode codepoint to lookup.
* @param category The General Category of the codepoint.
* @return The properties associated with the codepoint.
*/
ucd_property ucd_properties(codepoint_t c, ucd_category category);

/** @brief Is the codepoint in the 'alnum' class?
*
* @param c The Unicode codepoint to check.
@@ -770,6 +815,57 @@ namespace ucd
return (script)ucd_lookup_script(c);
}

/** @brief Properties
*/
typedef ucd_property property;
enum
{
White_Space = UCD_PROPERTY_WHITE_SPACE, /**< @brief White_Space */
Bidi_Control = UCD_PROPERTY_BIDI_CONTROL, /**< @brief Bidi_Control */
Join_Control = UCD_PROPERTY_JOIN_CONTROL, /**< @brief Join_Control */
Dash = UCD_PROPERTY_DASH, /**< @brief Dash */
Hyphen = UCD_PROPERTY_HYPHEN, /**< @brief Hyphen */
Quotation_Mark = UCD_PROPERTY_QUOTATION_MARK, /**< @brief Quotation_Mark */
Terminal_Punctuation = UCD_PROPERTY_TERMINAL_PUNCTUATION, /**< @brief Terminal_Punctuation */
Other_Math = UCD_PROPERTY_OTHER_MATH, /**< @brief Other_Math */
Hex_Digit = UCD_PROPERTY_HEX_DIGIT, /**< @brief Hex_Digit */
ASCII_Hex_Digit = UCD_PROPERTY_ASCII_HEX_DIGIT, /**< @brief ASCII_Hex_Digit */
Other_Alphabetic = UCD_PROPERTY_OTHER_ALPHABETIC, /**< @brief Other_Alphabetic */
Ideographic = UCD_PROPERTY_IDEOGRAPHIC, /**< @brief Ideographic */
Diacritic = UCD_PROPERTY_DIACRITIC, /**< @brief Diacritic */
Extender = UCD_PROPERTY_EXTENDER, /**< @brief Extender */
Other_Lowercase = UCD_PROPERTY_OTHER_LOWERCASE, /**< @brief Other_Lowercase */
Other_Uppercase = UCD_PROPERTY_OTHER_UPPERCASE, /**< @brief Other_Uppercase */
Noncharacter_Code_Point = UCD_PROPERTY_NONCHARACTER_CODE_POINT, /**< @brief Noncharacter_Code_Point */
Other_Grapheme_Extend = UCD_PROPERTY_OTHER_GRAPHEME_EXTEND, /**< @brief Other_Grapheme_Extend */
IDS_Binary_Operator = UCD_PROPERTY_IDS_BINARY_OPERATOR, /**< @brief IDS_Binary_Operator */
IDS_Trinary_Operator = UCD_PROPERTY_IDS_TRINARY_OPERATOR, /**< @brief IDS_Trinary_Operator */
Radical = UCD_PROPERTY_RADICAL, /**< @brief Radical */
Unified_Ideograph = UCD_PROPERTY_UNIFIED_IDEOGRAPH, /**< @brief Unified_Ideograph */
Other_Default_Ignorable_Code_Point = UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT, /**< @brief Other_Default_Ignorable_Code_Point */
Deprecated = UCD_PROPERTY_DEPRECATED, /**< @brief Deprecated */
Soft_Dotted = UCD_PROPERTY_SOFT_DOTTED, /**< @brief Soft_Dotted */
Logical_Order_Exception = UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION, /**< @brief Logical_Order_Exception */
Other_ID_Start = UCD_PROPERTY_OTHER_ID_START, /**< @brief Other_ID_Start */
Other_ID_Continue = UCD_PROPERTY_OTHER_ID_CONTINUE, /**< @brief Other_ID_Continue */
Sentence_Terminal = UCD_PROPERTY_SENTENCE_TERMINAL, /**< @brief Sentence_Terminal */
Variation_Selector = UCD_PROPERTY_VARIATION_SELECTOR, /**< @brief Variation_Selector */
Pattern_White_Space = UCD_PROPERTY_PATTERN_WHITE_SPACE, /**< @brief Pattern_White_Space */
Pattern_Syntax = UCD_PROPERTY_PATTERN_SYNTAX, /**< @brief Pattern_Syntax */
Prepended_Concatenation_Mark = UCD_PROPERTY_PREPENDED_CONCATENATION_MARK, /**< @brief Prepended_Concatenation_Mark */
};

/** @brief Return the properties of the specified codepoint.
*
* @param c The Unicode codepoint to lookup.
* @param cat The General Category of the codepoint.
* @return The properties associated with the codepoint.
*/
inline property properties(codepoint_t c, category cat)
{
return (property)ucd_properties(c, (ucd_category)cat);
}

/** @brief Is the codepoint in the 'alnum' class?
*
* @param c The Unicode codepoint to check.

+ 1826
- 0
src/ucd-tools/src/proplist.c
File diff suppressed because it is too large
View File


+ 4
- 1
src/ucd-tools/tests/printcdata.c View File

@@ -163,6 +163,9 @@ void uprintf(FILE *out, codepoint_t c, const char *format)
case 'p': // codepoint
uprintf_codepoint(out, c, *++format);
break;
case 'P': // properties
fprintf(out, "%016llx", ucd_properties(c, ucd_lookup_category(c)));
break;
case 'i': // is*
uprintf_is(out, c, *++format);
break;
@@ -249,7 +252,7 @@ int main(int argc, char **argv)
{
for (codepoint_t c = 0; c <= 0x10FFFF; ++c)
uprintf(stdout, c, format ? format :
"%pH %s %C %c %UH %LH %TH %id %ix %ic %is %ib %ip %iP %ig %iA %ia %iu %il\n");
"%pH %s %C %c %UH %LH %TH %id %ix %ic %is %ib %ip %iP %ig %iA %ia %iu %il %P\n");
}
return 0;
}

+ 4
- 1
src/ucd-tools/tests/printucddata.c View File

@@ -160,6 +160,9 @@ void uprintf(FILE *out, codepoint_t c, const char *format)
case 'p': // codepoint
uprintf_codepoint(out, c, *++format);
break;
case 'P': // properties
fprintf(out, "%016llx", ucd_properties(c, ucd_lookup_category(c)));
break;
case 'i': // is*
uprintf_is(out, c, *++format);
break;
@@ -244,7 +247,7 @@ int main(int argc, char **argv)
{
for (codepoint_t c = 0; c <= 0x10FFFF; ++c)
uprintf(stdout, c, format ? format :
"%pH %s %C %c %UH %LH %TH %id %ix %ic %is %ib %ip %iP %ig %iA %ia %iu %il\n");
"%pH %s %C %c %UH %LH %TH %id %ix %ic %is %ib %ip %iP %ig %iA %ia %iu %il %P\n");
}
return 0;
}

+ 4
- 1
src/ucd-tools/tests/printucddata_cpp.cpp View File

@@ -160,6 +160,9 @@ void uprintf(FILE *out, ucd::codepoint_t c, const char *format)
case 'p': // codepoint
uprintf_codepoint(out, c, *++format);
break;
case 'P': // properties
fprintf(out, "%016llx", ucd::properties(c, ucd::lookup_category(c)));
break;
case 'i': // is*
uprintf_is(out, c, *++format);
break;
@@ -244,7 +247,7 @@ int main(int argc, char **argv)
{
for (ucd::codepoint_t c = 0; c <= 0x10FFFF; ++c)
uprintf(stdout, c, format ? format :
"%pH %s %C %c %UH %LH %TH %id %ix %ic %is %ib %ip %iP %ig %iA %ia %iu %il\n");
"%pH %s %C %c %UH %LH %TH %id %ix %ic %is %ib %ip %iP %ig %iA %ia %iu %il %P\n");
}
return 0;
}

+ 39
- 0
src/ucd-tools/tools/mkencodingtable View File

@@ -0,0 +1,39 @@
#!/usr/bin/python3

# Copyright (C) 2017 Reece H. Dunn
#
# This file is part of ucd-tools.
#
# ucd-tools is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# ucd-tools is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with ucd-tools. If not, see <http://www.gnu.org/licenses/>.

import os
import sys

def readtable(filename):
with open(filename) as f:
for line in f:
if line == '\n' or line.startswith('#'):
continue
yield line.split()[0:2]

table = {}
for code, mapped in readtable(sys.argv[1]):
table[int(code, 0)] = mapped.lower()

for code in range(0x80, 0x100):
if (code % 8 == 0):
print('\t', end='')
print('%s, ' % table.get(code, '0xfffd'), end='')
if (code % 8 == 7):
print('// %02x\n' % (code - 7), end='')

+ 46
- 2
src/ucd-tools/tools/printdata.py View File

@@ -113,6 +113,49 @@ def islower(data):
else:
return 0

def decomposition_type(data, dtype):
value = data.get('DecompositionType', None)
if value and value.startswith(dtype):
return value
return None

def properties(data):
props = 0
props += (2 ** 0) * data.get('White_Space', 0)
props += (2 ** 1) * data.get('Bidi_Control', 0)
props += (2 ** 2) * data.get('Join_Control', 0)
props += (2 ** 3) * data.get('Dash', 0)
props += (2 ** 4) * data.get('Hyphen', 0)
props += (2 ** 5) * data.get('Quotation_Mark', 0)
props += (2 ** 6) * data.get('Terminal_Punctuation', 0)
props += (2 ** 7) * data.get('Other_Math', 0)
props += (2 ** 8) * data.get('Hex_Digit', 0)
props += (2 ** 9) * data.get('ASCII_Hex_Digit', 0)
props += (2 ** 10) * data.get('Other_Alphabetic', 0)
props += (2 ** 11) * data.get('Ideographic', 0)
props += (2 ** 12) * data.get('Diacritic', 0)
props += (2 ** 13) * data.get('Extender', 0)
props += (2 ** 14) * data.get('Other_Lowercase', 0)
props += (2 ** 15) * data.get('Other_Uppercase', 0)
props += (2 ** 16) * data.get('Noncharacter_Code_Point', 0)
props += (2 ** 17) * data.get('Other_Grapheme_Extend', 0)
props += (2 ** 18) * data.get('IDS_Binary_Operator', 0)
props += (2 ** 19) * data.get('IDS_Trinary_Operator', 0)
props += (2 ** 20) * data.get('Radical', 0)
props += (2 ** 21) * data.get('Unified_Ideograph', 0)
props += (2 ** 22) * data.get('Other_Default_Ignorable_Code_Point', 0)
props += (2 ** 23) * data.get('Deprecated', 0)
props += (2 ** 24) * data.get('Soft_Dotted', 0)
props += (2 ** 25) * data.get('Logical_Order_Exception', 0)
props += (2 ** 26) * data.get('Other_ID_Start', 0)
props += (2 ** 27) * data.get('Other_ID_Continue', 0)
props += (2 ** 28) * data.get('Sentence_Terminal', 0)
props += (2 ** 29) * data.get('Variation_Selector', 0)
props += (2 ** 30) * data.get('Pattern_White_Space', 0)
props += (2 ** 31) * data.get('Pattern_Syntax', 0)
props += (2 ** 32) * data.get('Prepended_Concatenation_Mark', 0)
return props

if __name__ == '__main__':
for codepoint in ucd.CodeRange('000000..10FFFF'):
try:
@@ -126,10 +169,11 @@ if __name__ == '__main__':
if title == null: title = codepoint
if upper == null: upper = codepoint
if lower == null: lower = codepoint
print('%s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s' % (
print('%s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %016x' % (
codepoint, script,
data.get('GeneralCategory', 'Cn')[0], data.get('GeneralCategory', 'Cn'),
upper, lower, title,
isdigit(data), isxdigit(data),
iscntrl(data), isspace(data), isblank(data), ispunct(data),
isprint(data), isgraph(data), isalnum(data), isalpha(data), isupper(data), islower(data)))
isprint(data), isgraph(data), isalnum(data), isalpha(data), isupper(data), islower(data),
properties(data)))

Write Preview
Loading…
Cancel
Save