Rewrite the generator script of unicode ranges. (#1583)
The script generates the source file instead of copy the tables and paste these manually. JerryScript-DCO-1.0-Signed-off-by: Robert Sipka rsipka.uszeged@partner.samsung.com
This commit is contained in:
Robert Sipka
committed by
Dániel Bátyai
Dániel Bátyai
parent
799726aa42
commit
188dc46fe0
@@ -136,9 +136,11 @@ lit_char_is_white_space (ecma_char_t c) /**< code unit */
|
||||
{
|
||||
return (c == LIT_CHAR_NBSP
|
||||
|| c == LIT_CHAR_BOM
|
||||
|| (c >= unicode_separator_char_interv_sps[0]
|
||||
&& c <= unicode_separator_char_interv_sps[0] + unicode_separator_char_interv_lens[0])
|
||||
|| search_char_in_char_array (c, unicode_separator_chars, NUM_OF_ELEMENTS (unicode_separator_chars)));
|
||||
|| (c >= jerry_unicode_separator_char_interval_sps[0]
|
||||
&& c <= jerry_unicode_separator_char_interval_sps[0] + jerry_unicode_separator_char_interval_lengths[0])
|
||||
|| search_char_in_char_array (c,
|
||||
jerry_unicode_separator_chars,
|
||||
NUM_OF_ELEMENTS (jerry_unicode_separator_chars)));
|
||||
}
|
||||
} /* lit_char_is_white_space */
|
||||
|
||||
@@ -178,9 +180,11 @@ lit_char_is_line_terminator (ecma_char_t c) /**< code unit */
|
||||
static bool
|
||||
lit_char_is_unicode_letter (ecma_char_t c) /**< code unit */
|
||||
{
|
||||
return (search_char_in_interval_array (c, unicode_letter_interv_sps, unicode_letter_interv_lens,
|
||||
NUM_OF_ELEMENTS (unicode_letter_interv_sps))
|
||||
|| search_char_in_char_array (c, unicode_letter_chars, NUM_OF_ELEMENTS (unicode_letter_chars)));
|
||||
return (search_char_in_interval_array (c,
|
||||
jerry_unicode_letter_interval_sps,
|
||||
jerry_unicode_letter_interval_lengths,
|
||||
NUM_OF_ELEMENTS (jerry_unicode_letter_interval_sps))
|
||||
|| search_char_in_char_array (c, jerry_unicode_letter_chars, NUM_OF_ELEMENTS (jerry_unicode_letter_chars)));
|
||||
} /* lit_char_is_unicode_letter */
|
||||
|
||||
/**
|
||||
@@ -200,11 +204,13 @@ lit_char_is_unicode_letter (ecma_char_t c) /**< code unit */
|
||||
static bool
|
||||
lit_char_is_unicode_non_letter_ident_part (ecma_char_t c) /**< code unit */
|
||||
{
|
||||
return (search_char_in_interval_array (c, unicode_non_letter_ident_part_interv_sps,
|
||||
unicode_non_letter_ident_part_interv_lens,
|
||||
NUM_OF_ELEMENTS (unicode_non_letter_ident_part_interv_sps))
|
||||
|| search_char_in_char_array (c, unicode_non_letter_ident_part_chars,
|
||||
NUM_OF_ELEMENTS (unicode_non_letter_ident_part_chars)));
|
||||
return (search_char_in_interval_array (c,
|
||||
jerry_unicode_non_letter_ident_part_interval_sps,
|
||||
jerry_unicode_non_letter_ident_part_interval_lengths,
|
||||
NUM_OF_ELEMENTS (jerry_unicode_non_letter_ident_part_interval_sps))
|
||||
|| search_char_in_char_array (c,
|
||||
jerry_unicode_non_letter_ident_part_chars,
|
||||
NUM_OF_ELEMENTS (jerry_unicode_non_letter_ident_part_chars)));
|
||||
} /* lit_char_is_unicode_non_letter_ident_part */
|
||||
|
||||
/**
|
||||
|
||||
@@ -12,169 +12,152 @@
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*
|
||||
*
|
||||
* Unicode characters and ranges generated by tools/print-unicode-ranges.sh
|
||||
* from UnicodeData-3.0.0.txt.
|
||||
* See also:
|
||||
* http://www.unicode.org/Public/3.0-Update/UnicodeData-3.0.0.txt
|
||||
* http://www.unicode.org/Public/3.0-Update/UnicodeData-3.0.0.html
|
||||
* This file is automatically generated by the unicode_ranges.py script
|
||||
* from UnicodeData-3.0.0.txt. Do not edit!
|
||||
*/
|
||||
|
||||
#ifndef LIT_UNICODE_RANGES_INC_H_
|
||||
#define LIT_UNICODE_RANGES_INC_H_
|
||||
|
||||
/**
|
||||
* Character interval starting points for the unicode letters.
|
||||
*
|
||||
* The characters covered by these intervalse are from
|
||||
* The characters covered by these intervals are from
|
||||
* the following Unicode categories: Lu, Ll, Lt, Lm, Lo, Nl
|
||||
*/
|
||||
static const uint16_t unicode_letter_interv_sps[] JERRY_CONST_DATA =
|
||||
static const uint16_t jerry_unicode_letter_interval_sps[] JERRY_CONST_DATA =
|
||||
{
|
||||
/*
|
||||
* these are handled separetely
|
||||
* 0x0041, len 25
|
||||
* 0x0061, len 25
|
||||
*/
|
||||
0x00C0, 0x00D8, 0XF8, 0X1F8, 0x0222, 0x0250, 0x02B0, 0x02BB,
|
||||
0x02D0, 0x02E0, 0x0388, 0x038E, 0x03A3, 0x03D0, 0x03DA, 0x0400, 0x048C, 0x04C7,
|
||||
0x04CB, 0x04D0, 0x04F8, 0x0531, 0x0561, 0x05D0, 0x05F0, 0x0621, 0x0640, 0x0671,
|
||||
0x06E5, 0x06FA, 0x0712, 0x0780, 0x0905, 0x0958, 0x0985, 0x098F, 0x0993, 0x09AA,
|
||||
0x09B6, 0x09DC, 0x09DF, 0x09F0, 0x0A05, 0x0A0F, 0x0A13, 0x0A2A, 0x0A32, 0x0A35,
|
||||
0x0A38, 0x0A59, 0x0A72, 0x0A85, 0x0A8F, 0x0A93, 0x0AAA, 0x0AB2, 0x0AB5, 0x0B05,
|
||||
0x0B0F, 0x0B13, 0x0B2A, 0x0B32, 0x0B36, 0x0B5C, 0x0B5F, 0x0B85, 0x0B8E, 0x0B92,
|
||||
0x0B99, 0x0B9E, 0x0BA3, 0x0BA8, 0x0BAE, 0x0BB7, 0x0C05, 0x0C0E, 0x0C12, 0x0C2A,
|
||||
0x0C35, 0x0C60, 0x0C85, 0x0C8E, 0x0C92, 0x0CAA, 0x0CB5, 0x0CE0, 0x0D05, 0x0D0E,
|
||||
0x0D12, 0x0D2A, 0x0D60, 0x0D85, 0x0D9A, 0x0DB3, 0x0DC0, 0x0E01, 0x0E32, 0x0E40,
|
||||
0x0E81, 0x0E87, 0x0E94, 0x0E99, 0x0EA1, 0x0EAA, 0x0EAD, 0x0EB2, 0x0EC0, 0x0EDC,
|
||||
0x0F40, 0x0F49, 0x0F88, 0x1000, 0x1023, 0x1029, 0x1050, 0x10A0, 0x10D0, 0x1100,
|
||||
0x115F, 0x11A8, 0x1200, 0x1208, 0x124A, 0x1250, 0x125A, 0x1260, 0x128A, 0x1290,
|
||||
0x12B2, 0x12B8, 0x12C2, 0x12C8, 0x12D0, 0x12D8, 0x12F0, 0x1312, 0x1318, 0x1320,
|
||||
0x1348, 0x13A0, 0X1401, 0X1501, 0X1601, 0x166F, 0x1681, 0x16A0, 0x1780, 0x1820,
|
||||
0x1880, 0x1E00, 0x1EA0, 0x1F00, 0x1F18, 0x1F20, 0x1F48, 0x1F50, 0x1F5F, 0x1F80,
|
||||
0x1FB6, 0x1FC2, 0x1FC6, 0x1FD0, 0x1FD6, 0x1FE0, 0x1FF2, 0x1FF6, 0x210A, 0x2119,
|
||||
0x212A, 0x212F, 0x2133, 0x2160, 0x3005, 0x3021, 0x3031, 0x3038, 0x3041, 0x309D,
|
||||
0x30A1, 0x30FC, 0x3105, 0x3131, 0x31A0, 0XA000, 0XA100, 0XA200, 0XA300, 0XA400,
|
||||
0XF900, 0XFA00, 0xFB00, 0xFB13, 0xFB1F, 0xFB2A, 0xFB38, 0xFB40, 0xFB43, 0xFB46,
|
||||
0XFBD3, 0XFCD3, 0xFD50, 0xFD92, 0xFDF0, 0xFE70, 0xFE76, 0xFF21, 0xFF41, 0xFF66,
|
||||
0xFFC2, 0xFFCA, 0xFFD2, 0xFFDA
|
||||
0x00c0, 0x00d8, 0x00f8, 0x01f8, 0x0222, 0x0250, 0x02b0, 0x02bb, 0x02d0, 0x02e0,
|
||||
0x0388, 0x038e, 0x03a3, 0x03d0, 0x03da, 0x0400, 0x048c, 0x04c7, 0x04cb, 0x04d0,
|
||||
0x04f8, 0x0531, 0x0561, 0x05d0, 0x05f0, 0x0621, 0x0640, 0x0671, 0x06e5, 0x06fa,
|
||||
0x0712, 0x0780, 0x0905, 0x0958, 0x0985, 0x098f, 0x0993, 0x09aa, 0x09b6, 0x09dc,
|
||||
0x09df, 0x09f0, 0x0a05, 0x0a0f, 0x0a13, 0x0a2a, 0x0a32, 0x0a35, 0x0a38, 0x0a59,
|
||||
0x0a72, 0x0a85, 0x0a8f, 0x0a93, 0x0aaa, 0x0ab2, 0x0ab5, 0x0b05, 0x0b0f, 0x0b13,
|
||||
0x0b2a, 0x0b32, 0x0b36, 0x0b5c, 0x0b5f, 0x0b85, 0x0b8e, 0x0b92, 0x0b99, 0x0b9e,
|
||||
0x0ba3, 0x0ba8, 0x0bae, 0x0bb7, 0x0c05, 0x0c0e, 0x0c12, 0x0c2a, 0x0c35, 0x0c60,
|
||||
0x0c85, 0x0c8e, 0x0c92, 0x0caa, 0x0cb5, 0x0ce0, 0x0d05, 0x0d0e, 0x0d12, 0x0d2a,
|
||||
0x0d60, 0x0d85, 0x0d9a, 0x0db3, 0x0dc0, 0x0e01, 0x0e32, 0x0e40, 0x0e81, 0x0e87,
|
||||
0x0e94, 0x0e99, 0x0ea1, 0x0eaa, 0x0ead, 0x0eb2, 0x0ec0, 0x0edc, 0x0f40, 0x0f49,
|
||||
0x0f88, 0x1000, 0x1023, 0x1029, 0x1050, 0x10a0, 0x10d0, 0x1100, 0x115f, 0x11a8,
|
||||
0x1200, 0x1208, 0x124a, 0x1250, 0x125a, 0x1260, 0x128a, 0x1290, 0x12b2, 0x12b8,
|
||||
0x12c2, 0x12c8, 0x12d0, 0x12d8, 0x12f0, 0x1312, 0x1318, 0x1320, 0x1348, 0x13a0,
|
||||
0x1401, 0x1501, 0x1601, 0x166f, 0x1681, 0x16a0, 0x1780, 0x1820, 0x1880, 0x1e00,
|
||||
0x1ea0, 0x1f00, 0x1f18, 0x1f20, 0x1f48, 0x1f50, 0x1f5f, 0x1f80, 0x1fb6, 0x1fc2,
|
||||
0x1fc6, 0x1fd0, 0x1fd6, 0x1fe0, 0x1ff2, 0x1ff6, 0x210a, 0x2119, 0x212a, 0x212f,
|
||||
0x2133, 0x2160, 0x3005, 0x3021, 0x3031, 0x3038, 0x3041, 0x309d, 0x30a1, 0x30fc,
|
||||
0x3105, 0x3131, 0x31a0, 0xa000, 0xa100, 0xa200, 0xa300, 0xa400, 0xf900, 0xfa00,
|
||||
0xfb00, 0xfb13, 0xfb1f, 0xfb2a, 0xfb38, 0xfb40, 0xfb43, 0xfb46, 0xfbd3, 0xfcd3,
|
||||
0xfd50, 0xfd92, 0xfdf0, 0xfe70, 0xfe76, 0xff21, 0xff41, 0xff66, 0xffc2, 0xffca,
|
||||
0xffd2, 0xffda
|
||||
};
|
||||
|
||||
/**
|
||||
* Character lengths for the unicode letters.
|
||||
*
|
||||
* The characters covered by these intervalse are from
|
||||
* The characters covered by these intervals are from
|
||||
* the following Unicode categories: Lu, Ll, Lt, Lm, Lo, Nl
|
||||
*/
|
||||
static const uint8_t unicode_letter_interv_lens[] JERRY_CONST_DATA =
|
||||
static const uint8_t jerry_unicode_letter_interval_lengths[] JERRY_CONST_DATA =
|
||||
{
|
||||
22, 30, 255, 39, 17, 93, 8, 6,
|
||||
1, 4, 2, 19, 43, 7, 25, 129, 56, 1,
|
||||
1, 37, 1, 37, 38, 26, 2, 25, 10, 98,
|
||||
1, 2, 26, 37, 52, 9, 7, 1, 21, 6,
|
||||
3, 1, 2, 1, 5, 1, 21, 6, 1, 1,
|
||||
1, 3, 2, 6, 2, 21, 6, 1, 4, 7,
|
||||
1, 21, 6, 1, 3, 1, 2, 5, 2, 3,
|
||||
1, 1, 1, 2, 7, 2, 7, 2, 22, 9,
|
||||
4, 1, 7, 2, 22, 9, 4, 1, 7, 2,
|
||||
22, 15, 1, 17, 23, 8, 6, 47, 1, 6,
|
||||
1, 1, 3, 6, 2, 1, 3, 1, 4, 1,
|
||||
7, 33, 3, 33, 4, 1, 5, 37, 38, 89,
|
||||
67, 81, 6, 62, 3, 6, 3, 38, 3, 30,
|
||||
3, 6, 3, 6, 6, 22, 30, 3, 6, 38,
|
||||
18, 84, 255, 255, 107, 7, 25, 74, 51, 87,
|
||||
40, 155, 89, 21, 5, 37, 5, 7, 30, 52,
|
||||
6, 2, 6, 3, 5, 12, 2, 6, 9, 4,
|
||||
3, 2, 6, 35, 2, 8, 4, 2, 83, 1,
|
||||
89, 2, 39, 93, 23, 255, 255, 255, 255, 140,
|
||||
255, 45, 6, 4, 9, 12, 4, 1, 1, 107,
|
||||
255, 106, 63, 53, 11, 2, 134, 25, 25, 88,
|
||||
5, 5, 5, 2
|
||||
0x0016, 0x001e, 0x00ff, 0x0027, 0x0011, 0x005d, 0x0008, 0x0006, 0x0001, 0x0004,
|
||||
0x0002, 0x0013, 0x002b, 0x0007, 0x0019, 0x0081, 0x0038, 0x0001, 0x0001, 0x0025,
|
||||
0x0001, 0x0025, 0x0026, 0x001a, 0x0002, 0x0019, 0x000a, 0x0062, 0x0001, 0x0002,
|
||||
0x001a, 0x0025, 0x0034, 0x0009, 0x0007, 0x0001, 0x0015, 0x0006, 0x0003, 0x0001,
|
||||
0x0002, 0x0001, 0x0005, 0x0001, 0x0015, 0x0006, 0x0001, 0x0001, 0x0001, 0x0003,
|
||||
0x0002, 0x0006, 0x0002, 0x0015, 0x0006, 0x0001, 0x0004, 0x0007, 0x0001, 0x0015,
|
||||
0x0006, 0x0001, 0x0003, 0x0001, 0x0002, 0x0005, 0x0002, 0x0003, 0x0001, 0x0001,
|
||||
0x0001, 0x0002, 0x0007, 0x0002, 0x0007, 0x0002, 0x0016, 0x0009, 0x0004, 0x0001,
|
||||
0x0007, 0x0002, 0x0016, 0x0009, 0x0004, 0x0001, 0x0007, 0x0002, 0x0016, 0x000f,
|
||||
0x0001, 0x0011, 0x0017, 0x0008, 0x0006, 0x002f, 0x0001, 0x0006, 0x0001, 0x0001,
|
||||
0x0003, 0x0006, 0x0002, 0x0001, 0x0003, 0x0001, 0x0004, 0x0001, 0x0007, 0x0021,
|
||||
0x0003, 0x0021, 0x0004, 0x0001, 0x0005, 0x0025, 0x0026, 0x0059, 0x0043, 0x0051,
|
||||
0x0006, 0x003e, 0x0003, 0x0006, 0x0003, 0x0026, 0x0003, 0x001e, 0x0003, 0x0006,
|
||||
0x0003, 0x0006, 0x0006, 0x0016, 0x001e, 0x0003, 0x0006, 0x0026, 0x0012, 0x0054,
|
||||
0x00ff, 0x00ff, 0x006b, 0x0007, 0x0019, 0x004a, 0x0033, 0x0057, 0x0028, 0x009b,
|
||||
0x0059, 0x0015, 0x0005, 0x0025, 0x0005, 0x0007, 0x001e, 0x0034, 0x0006, 0x0002,
|
||||
0x0006, 0x0003, 0x0005, 0x000c, 0x0002, 0x0006, 0x0009, 0x0004, 0x0003, 0x0002,
|
||||
0x0006, 0x0023, 0x0002, 0x0008, 0x0004, 0x0002, 0x0053, 0x0001, 0x0059, 0x0002,
|
||||
0x0027, 0x005d, 0x0017, 0x00ff, 0x00ff, 0x00ff, 0x00ff, 0x008c, 0x00ff, 0x002d,
|
||||
0x0006, 0x0004, 0x0009, 0x000c, 0x0004, 0x0001, 0x0001, 0x006b, 0x00ff, 0x006a,
|
||||
0x003f, 0x0035, 0x000b, 0x0002, 0x0086, 0x0019, 0x0019, 0x0058, 0x0005, 0x0005,
|
||||
0x0005, 0x0002
|
||||
};
|
||||
|
||||
/**
|
||||
* Those unicode letter characters that are not inside any of
|
||||
* the intervals specified in unicode_letter_intervals array.
|
||||
* the intervals specified in jerry_unicode_letter_interval_sps array.
|
||||
*
|
||||
* The characters are from the following Unicode categories:
|
||||
* Lu, Ll, Lt, Lm, Lo, Nl
|
||||
*/
|
||||
static const uint16_t unicode_letter_chars[] JERRY_CONST_DATA =
|
||||
static const uint16_t jerry_unicode_letter_chars[] JERRY_CONST_DATA =
|
||||
{
|
||||
0x00AA, 0x00B5, 0x00BA, 0x02EE, 0x037A, 0x0386, 0x038C, 0x0559, 0x06D5, 0x0710,
|
||||
0x093D, 0x0950, 0x09B2, 0x0A5E, 0x0A8D, 0x0ABD, 0x0AD0, 0x0AE0, 0x0B3D, 0x0B9C,
|
||||
0x0CDE, 0x0DBD, 0x0E84, 0x0E8A, 0x0E8D, 0x0EA5, 0x0EA7, 0x0EBD, 0x0EC6, 0x0F00,
|
||||
0x1248, 0x1258, 0x1288, 0x12B0, 0x12C0, 0x1310, 0x1F59, 0x1F5B, 0x1F5D, 0x1FBE,
|
||||
0x207F, 0x2102, 0x2107, 0x2115, 0x2124, 0x2126, 0x2128, 0x3400, 0x4DB5, 0x4E00,
|
||||
0x9FA5, 0xAC00, 0xD7A3, 0xFB1D, 0xFB3E, 0xFE74
|
||||
0x00aa, 0x00b5, 0x00ba, 0x02ee, 0x037a, 0x0386, 0x038c, 0x0559, 0x06d5, 0x0710,
|
||||
0x093d, 0x0950, 0x09b2, 0x0a5e, 0x0a8d, 0x0abd, 0x0ad0, 0x0ae0, 0x0b3d, 0x0b9c,
|
||||
0x0cde, 0x0dbd, 0x0e84, 0x0e8a, 0x0e8d, 0x0ea5, 0x0ea7, 0x0ebd, 0x0ec6, 0x0f00,
|
||||
0x1248, 0x1258, 0x1288, 0x12b0, 0x12c0, 0x1310, 0x1f59, 0x1f5b, 0x1f5d, 0x1fbe,
|
||||
0x207f, 0x2102, 0x2107, 0x2115, 0x2124, 0x2126, 0x2128, 0x3400, 0x4db5, 0x4e00,
|
||||
0x9fa5, 0xac00, 0xd7a3, 0xfb1d, 0xfb3e, 0xfe74
|
||||
};
|
||||
|
||||
/**
|
||||
* Character interval starting points for non-letter character
|
||||
* that can be used as a non-first character of an identifier.
|
||||
*
|
||||
* The characters covered by these intervalse are from
|
||||
* The characters covered by these intervals are from
|
||||
* the following Unicode categories: Nd, Mn, Mc, Pc
|
||||
*/
|
||||
static const uint16_t unicode_non_letter_ident_part_interv_sps[] JERRY_CONST_DATA =
|
||||
static const uint16_t jerry_unicode_non_letter_ident_part_interval_sps[] JERRY_CONST_DATA =
|
||||
{
|
||||
/*
|
||||
* decimal digits: handled separately
|
||||
* 0x0030, len: 9
|
||||
*/
|
||||
0x0300, 0x0360, 0x0483, 0x0591, 0x05A3, 0x05BB, 0x05C1, 0x064B, 0x0660,
|
||||
0x06D6, 0x06DF, 0x06E7, 0x06EA, 0x06F0, 0x0730, 0x07A6, 0x0901, 0x093E, 0x0951,
|
||||
0x0962, 0x0966, 0x0981, 0x09BE, 0x09C7, 0x09CB, 0x09E2, 0x09E6, 0x0A3E, 0x0A47,
|
||||
0x0A4B, 0x0A66, 0x0A81, 0x0ABE, 0x0AC7, 0x0ACB, 0x0AE6, 0x0B01, 0x0B3E, 0x0B47,
|
||||
0x0B4B, 0x0B56, 0x0B66, 0x0B82, 0x0BBE, 0x0BC6, 0x0BCA, 0x0BE7, 0x0C01, 0x0C3E,
|
||||
0x0C46, 0x0C4A, 0x0C55, 0x0C66, 0x0C82, 0x0CBE, 0x0CC6, 0x0CCA, 0x0CD5, 0x0CE6,
|
||||
0x0D02, 0x0D3E, 0x0D46, 0x0D4A, 0x0D66, 0x0D82, 0x0DCF, 0x0DD8, 0x0DF2, 0x0E34,
|
||||
0x0E47, 0x0E50, 0x0EB4, 0x0EBB, 0x0EC8, 0x0ED0, 0x0F18, 0x0F20, 0x0F3E, 0x0F71,
|
||||
0x0F86, 0x0F90, 0x0F99, 0x102C, 0x1036, 0x1040, 0x1056, 0x1369, 0x17B4, 0x17E0,
|
||||
0x1810, 0x203F, 0x20D0, 0x302A, 0x3099, 0xFE20, 0xFE33, 0xFE4D, 0xFF10
|
||||
0x0300, 0x0360, 0x0483, 0x0591, 0x05a3, 0x05bb, 0x05c1, 0x064b, 0x0660, 0x06d6,
|
||||
0x06df, 0x06e7, 0x06ea, 0x06f0, 0x0730, 0x07a6, 0x0901, 0x093e, 0x0951, 0x0962,
|
||||
0x0966, 0x0981, 0x09be, 0x09c7, 0x09cb, 0x09e2, 0x09e6, 0x0a3e, 0x0a47, 0x0a4b,
|
||||
0x0a66, 0x0a81, 0x0abe, 0x0ac7, 0x0acb, 0x0ae6, 0x0b01, 0x0b3e, 0x0b47, 0x0b4b,
|
||||
0x0b56, 0x0b66, 0x0b82, 0x0bbe, 0x0bc6, 0x0bca, 0x0be7, 0x0c01, 0x0c3e, 0x0c46,
|
||||
0x0c4a, 0x0c55, 0x0c66, 0x0c82, 0x0cbe, 0x0cc6, 0x0cca, 0x0cd5, 0x0ce6, 0x0d02,
|
||||
0x0d3e, 0x0d46, 0x0d4a, 0x0d66, 0x0d82, 0x0dcf, 0x0dd8, 0x0df2, 0x0e34, 0x0e47,
|
||||
0x0e50, 0x0eb4, 0x0ebb, 0x0ec8, 0x0ed0, 0x0f18, 0x0f20, 0x0f3e, 0x0f71, 0x0f86,
|
||||
0x0f90, 0x0f99, 0x102c, 0x1036, 0x1040, 0x1056, 0x1369, 0x17b4, 0x17e0, 0x1810,
|
||||
0x203f, 0x20d0, 0x302a, 0x3099, 0xfe20, 0xfe33, 0xfe4d, 0xff10
|
||||
};
|
||||
|
||||
/**
|
||||
* Character interval lengths for non-letter character
|
||||
* that can be used as a non-first character of an identifier.
|
||||
*
|
||||
* The characters covered by these intervalse are from
|
||||
* The characters covered by these intervals are from
|
||||
* the following Unicode categories: Nd, Mn, Mc, Pc
|
||||
*/
|
||||
static const uint8_t unicode_non_letter_ident_part_interv_lens[] =
|
||||
static const uint8_t jerry_unicode_non_letter_ident_part_interval_lengths[] JERRY_CONST_DATA =
|
||||
{
|
||||
78, 2, 3, 16, 22, 2, 1, 10, 9,
|
||||
6, 5, 1, 3, 9, 26, 10, 2, 15, 3,
|
||||
1, 9, 2, 6, 1, 2, 1, 9, 4, 1,
|
||||
2, 11, 2, 7, 2, 2, 9, 2, 5, 1,
|
||||
2, 1, 9, 1, 4, 2, 3, 8, 2, 6,
|
||||
2, 3, 1, 9, 1, 6, 2, 3, 1, 9,
|
||||
1, 5, 2, 3, 9, 1, 5, 7, 1, 6,
|
||||
7, 9, 5, 1, 5, 9, 1, 9, 1, 19,
|
||||
1, 7, 35, 6, 3, 9, 3, 8, 31, 9,
|
||||
9, 1, 12, 5, 1, 3, 1, 2, 9
|
||||
0x004e, 0x0002, 0x0003, 0x0010, 0x0016, 0x0002, 0x0001, 0x000a, 0x0009, 0x0006,
|
||||
0x0005, 0x0001, 0x0003, 0x0009, 0x001a, 0x000a, 0x0002, 0x000f, 0x0003, 0x0001,
|
||||
0x0009, 0x0002, 0x0006, 0x0001, 0x0002, 0x0001, 0x0009, 0x0004, 0x0001, 0x0002,
|
||||
0x000b, 0x0002, 0x0007, 0x0002, 0x0002, 0x0009, 0x0002, 0x0005, 0x0001, 0x0002,
|
||||
0x0001, 0x0009, 0x0001, 0x0004, 0x0002, 0x0003, 0x0008, 0x0002, 0x0006, 0x0002,
|
||||
0x0003, 0x0001, 0x0009, 0x0001, 0x0006, 0x0002, 0x0003, 0x0001, 0x0009, 0x0001,
|
||||
0x0005, 0x0002, 0x0003, 0x0009, 0x0001, 0x0005, 0x0007, 0x0001, 0x0006, 0x0007,
|
||||
0x0009, 0x0005, 0x0001, 0x0005, 0x0009, 0x0001, 0x0009, 0x0001, 0x0013, 0x0001,
|
||||
0x0007, 0x0023, 0x0006, 0x0003, 0x0009, 0x0003, 0x0008, 0x001f, 0x0009, 0x0009,
|
||||
0x0001, 0x000c, 0x0005, 0x0001, 0x0003, 0x0001, 0x0002, 0x0009
|
||||
};
|
||||
|
||||
/**
|
||||
* Those non-letter characters that can be used as a non-first
|
||||
* character of an identifier and not included in any of the intervals
|
||||
* specified in unicode_non_letter_ident_part_intervals array.
|
||||
* specified in jerry_unicode_non_letter_ident_part_interval_sps array.
|
||||
*
|
||||
* The characters are from the following Unicode categories:
|
||||
* Nd, Mn, Mc, Pc
|
||||
*/
|
||||
static const uint16_t unicode_non_letter_ident_part_chars[] =
|
||||
static const uint16_t jerry_unicode_non_letter_ident_part_chars[] JERRY_CONST_DATA =
|
||||
{
|
||||
0x005F, 0x05BF, 0x05C4, 0x0670, 0x0711, 0x093C, 0x09BC, 0x09D7, 0x0A02, 0x0A3C,
|
||||
0x0ABC, 0x0B3C, 0x0BD7, 0x0D57, 0x0DCA, 0x0DD6, 0x0E31, 0x0EB1, 0x0F35, 0x0F37,
|
||||
0x0F39, 0x0FC6, 0x18A9, 0x20E1, 0x30FB, 0xFB1E, 0xFF3F, 0xFF65
|
||||
0x05bf, 0x05c4, 0x0670, 0x0711, 0x093c, 0x09bc, 0x09d7, 0x0a02, 0x0a3c, 0x0abc,
|
||||
0x0b3c, 0x0bd7, 0x0d57, 0x0dca, 0x0dd6, 0x0e31, 0x0eb1, 0x0f35, 0x0f37, 0x0f39,
|
||||
0x0fc6, 0x18a9, 0x20e1, 0x30fb, 0xfb1e, 0xff3f, 0xff65
|
||||
};
|
||||
|
||||
|
||||
/**
|
||||
* Unicode separator character interval strting points from Unicode category: Zs
|
||||
* Unicode separator character interval starting points from Unicode category: Zs
|
||||
*/
|
||||
static const uint16_t unicode_separator_char_interv_sps[] =
|
||||
static const uint16_t jerry_unicode_separator_char_interval_sps[] JERRY_CONST_DATA =
|
||||
{
|
||||
0x2000
|
||||
};
|
||||
@@ -182,29 +165,19 @@ static const uint16_t unicode_separator_char_interv_sps[] =
|
||||
/**
|
||||
* Unicode separator character interval lengths from Unicode category: Zs
|
||||
*/
|
||||
static const uint8_t unicode_separator_char_interv_lens[] =
|
||||
static const uint8_t jerry_unicode_separator_char_interval_lengths[] JERRY_CONST_DATA =
|
||||
{
|
||||
11
|
||||
0x000b
|
||||
};
|
||||
|
||||
/**
|
||||
* Unicode separator characters that are not in the
|
||||
* unicode_separator_char_intervals array.
|
||||
* jerry_unicode_separator_char_intervals array.
|
||||
*
|
||||
* Unicode category: Zs
|
||||
*/
|
||||
static const uint16_t unicode_separator_chars[] =
|
||||
static const uint16_t jerry_unicode_separator_chars[] JERRY_CONST_DATA =
|
||||
{
|
||||
/*
|
||||
* these two chars are handled separatly @see lit_char_is_space_separator
|
||||
* 0x0020, space
|
||||
* 0x00A0, non-braking space
|
||||
*/
|
||||
0x1680, \
|
||||
0x180E, /* manually added */ \
|
||||
0x202F, /* manually added */ \
|
||||
0x205F, \
|
||||
0x3000
|
||||
0x1680, 0x180e, 0x202f, 0x205f, 0x3000
|
||||
};
|
||||
|
||||
#endif
|
||||
|
||||
Reference in New Issue
Block a user