Update RegExp unicode mode case folding to conform to the standard (#4004)

JerryScript-DCO-1.0-Signed-off-by: Dániel Bátyai daniel.batyai@h-lab.eu
This commit is contained in:
Dániel Bátyai
2020-07-20 15:51:43 +02:00
committed by GitHub
parent 33359ac506
commit 321215fdbb
9 changed files with 284 additions and 84 deletions
+31 -38
View File
@@ -96,68 +96,61 @@ static const uint8_t lit_unicode_upper_case_special_range_lengths[] JERRY_ATTR_C
/* Contains start points of lowercase ranges. */
static const uint16_t lit_unicode_lower_case_ranges[] JERRY_ATTR_CONST_DATA =
{
0x1e96, 0x1e96, 0x1f80, 0x1f80, 0x1f88, 0x1f80, 0x1f90, 0x1f90, 0x1f98, 0x1f90,
0x1fa0, 0x1fa0, 0x1fa8, 0x1fa0, 0x1fb2, 0x1fb2, 0x1fb6, 0x1fb6, 0x1fc2, 0x1fc2,
0x1fc6, 0x1fc6, 0x1fd2, 0x1fd2, 0x1fd6, 0x1fd6, 0x1fe2, 0x1fe2, 0x1fe6, 0x1fe6,
0x1ff2, 0x1ff2, 0x1ff6, 0x1ff6, 0xfb00, 0xfb00, 0xfb13, 0xfb13
0x1f88, 0x1f80, 0x1f98, 0x1f90, 0x1fa8, 0x1fa0
};
/* Interval lengths for start points in `lower_case_ranges` table. */
static const uint8_t lit_unicode_lower_case_range_lengths[] JERRY_ATTR_CONST_DATA =
{
0x0005, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0003, 0x0002, 0x0003,
0x0002, 0x0002, 0x0002, 0x0003, 0x0002, 0x0003, 0x0002, 0x0007, 0x0005
0x0008, 0x0008, 0x0008
};
/* The remaining lowercase conversions. The lowercase variant can be one-to-three character long. */
static const uint16_t lit_unicode_lower_case_conversions[] JERRY_ATTR_CONST_DATA =
{
0x00df, 0x00df, 0x0149, 0x0149, 0x01c5, 0x01c6, 0x01c8, 0x01c9, 0x01cb, 0x01cc,
0x01f0, 0x01f0, 0x01f2, 0x01f3, 0x0390, 0x0390, 0x03b0, 0x03b0, 0x03f4, 0x03b8,
0x0587, 0x0587, 0x1e9e, 0x00df, 0x1f50, 0x1f50, 0x1f52, 0x1f52, 0x1f54, 0x1f54,
0x1f56, 0x1f56, 0x1fbc, 0x1fb3, 0x1fcc, 0x1fc3, 0x1ffc, 0x1ff3, 0x2126, 0x03c9,
0x01c5, 0x01c6, 0x01c8, 0x01c9, 0x01cb, 0x01cc, 0x01f2, 0x01f3, 0x03f4, 0x03b8,
0x1e9e, 0x00df, 0x1fbc, 0x1fb3, 0x1fcc, 0x1fc3, 0x1ffc, 0x1ff3, 0x2126, 0x03c9,
0x212a, 0x006b, 0x212b, 0x00e5, 0x0130, 0x0069, 0x0307
};
/* Number of one-to-one, one-to-two, and one-to-three lowercase conversions. */
static const uint8_t lit_unicode_lower_case_conversion_counters[] JERRY_ATTR_CONST_DATA =
{
0x0016, 0x0001, 0x0000
0x000c, 0x0001, 0x0000
};
/* The remaining uppercase conversions. The uppercase variant can be one-to-three character long. */
static const uint16_t lit_unicode_upper_case_conversions[] JERRY_ATTR_CONST_DATA =
{
0x00b5, 0x039c, 0x0130, 0x0130, 0x0131, 0x0049, 0x017f, 0x0053, 0x01c5, 0x01c4,
0x01c8, 0x01c7, 0x01cb, 0x01ca, 0x01f2, 0x01f1, 0x0345, 0x0399, 0x03c2, 0x03a3,
0x03d0, 0x0392, 0x03d1, 0x0398, 0x03d5, 0x03a6, 0x03d6, 0x03a0, 0x03f0, 0x039a,
0x03f1, 0x03a1, 0x03f5, 0x0395, 0x1c80, 0x0412, 0x1c81, 0x0414, 0x1c82, 0x041e,
0x1c83, 0x0421, 0x1c84, 0x0422, 0x1c85, 0x0422, 0x1c86, 0x042a, 0x1c87, 0x0462,
0x1c88, 0xa64a, 0x1e9b, 0x1e60, 0x1fbe, 0x0399, 0x00df, 0x0053, 0x0053, 0x0149,
0x02bc, 0x004e, 0x01f0, 0x004a, 0x030c, 0x0587, 0x0535, 0x0552, 0x1e96, 0x0048,
0x0331, 0x1e97, 0x0054, 0x0308, 0x1e98, 0x0057, 0x030a, 0x1e99, 0x0059, 0x030a,
0x1e9a, 0x0041, 0x02be, 0x1f50, 0x03a5, 0x0313, 0x1f87, 0x1f0f, 0x0399, 0x1f8f,
0x1f0f, 0x0399, 0x1f97, 0x1f2f, 0x0399, 0x1f9f, 0x1f2f, 0x0399, 0x1fa7, 0x1f6f,
0x0399, 0x1faf, 0x1f6f, 0x0399, 0x1fb2, 0x1fba, 0x0399, 0x1fb3, 0x0391, 0x0399,
0x1fb4, 0x0386, 0x0399, 0x1fb6, 0x0391, 0x0342, 0x1fbc, 0x0391, 0x0399, 0x1fc2,
0x1fca, 0x0399, 0x1fc3, 0x0397, 0x0399, 0x1fc4, 0x0389, 0x0399, 0x1fc6, 0x0397,
0x0342, 0x1fcc, 0x0397, 0x0399, 0x1fd6, 0x0399, 0x0342, 0x1fe4, 0x03a1, 0x0313,
0x1fe6, 0x03a5, 0x0342, 0x1ff2, 0x1ffa, 0x0399, 0x1ff3, 0x03a9, 0x0399, 0x1ff4,
0x038f, 0x0399, 0x1ff6, 0x03a9, 0x0342, 0x1ffc, 0x03a9, 0x0399, 0xfb00, 0x0046,
0x0046, 0xfb01, 0x0046, 0x0049, 0xfb02, 0x0046, 0x004c, 0xfb05, 0x0053, 0x0054,
0xfb06, 0x0053, 0x0054, 0xfb13, 0x0544, 0x0546, 0xfb14, 0x0544, 0x0535, 0xfb15,
0x0544, 0x053b, 0xfb16, 0x054e, 0x0546, 0xfb17, 0x0544, 0x053d, 0x0390, 0x0399,
0x0308, 0x0301, 0x03b0, 0x03a5, 0x0308, 0x0301, 0x1f52, 0x03a5, 0x0313, 0x0300,
0x1f54, 0x03a5, 0x0313, 0x0301, 0x1f56, 0x03a5, 0x0313, 0x0342, 0x1fb7, 0x0391,
0x0342, 0x0399, 0x1fc7, 0x0397, 0x0342, 0x0399, 0x1fd2, 0x0399, 0x0308, 0x0300,
0x1fd3, 0x0399, 0x0308, 0x0301, 0x1fd7, 0x0399, 0x0308, 0x0342, 0x1fe2, 0x03a5,
0x0308, 0x0300, 0x1fe3, 0x03a5, 0x0308, 0x0301, 0x1fe7, 0x03a5, 0x0308, 0x0342,
0x1ff7, 0x03a9, 0x0342, 0x0399, 0xfb03, 0x0046, 0x0046, 0x0049, 0xfb04, 0x0046,
0x0046, 0x004c
0x00b5, 0x039c, 0x0131, 0x0049, 0x017f, 0x0053, 0x01c5, 0x01c4, 0x01c8, 0x01c7,
0x01cb, 0x01ca, 0x01f2, 0x01f1, 0x0345, 0x0399, 0x03c2, 0x03a3, 0x03d0, 0x0392,
0x03d1, 0x0398, 0x03d5, 0x03a6, 0x03d6, 0x03a0, 0x03f0, 0x039a, 0x03f1, 0x03a1,
0x03f5, 0x0395, 0x1c80, 0x0412, 0x1c81, 0x0414, 0x1c82, 0x041e, 0x1c83, 0x0421,
0x1c84, 0x0422, 0x1c85, 0x0422, 0x1c86, 0x042a, 0x1c87, 0x0462, 0x1c88, 0xa64a,
0x1e9b, 0x1e60, 0x1fbe, 0x0399, 0x00df, 0x0053, 0x0053, 0x0149, 0x02bc, 0x004e,
0x01f0, 0x004a, 0x030c, 0x0587, 0x0535, 0x0552, 0x1e96, 0x0048, 0x0331, 0x1e97,
0x0054, 0x0308, 0x1e98, 0x0057, 0x030a, 0x1e99, 0x0059, 0x030a, 0x1e9a, 0x0041,
0x02be, 0x1f50, 0x03a5, 0x0313, 0x1f87, 0x1f0f, 0x0399, 0x1f8f, 0x1f0f, 0x0399,
0x1f97, 0x1f2f, 0x0399, 0x1f9f, 0x1f2f, 0x0399, 0x1fa7, 0x1f6f, 0x0399, 0x1faf,
0x1f6f, 0x0399, 0x1fb2, 0x1fba, 0x0399, 0x1fb3, 0x0391, 0x0399, 0x1fb4, 0x0386,
0x0399, 0x1fb6, 0x0391, 0x0342, 0x1fbc, 0x0391, 0x0399, 0x1fc2, 0x1fca, 0x0399,
0x1fc3, 0x0397, 0x0399, 0x1fc4, 0x0389, 0x0399, 0x1fc6, 0x0397, 0x0342, 0x1fcc,
0x0397, 0x0399, 0x1fd6, 0x0399, 0x0342, 0x1fe4, 0x03a1, 0x0313, 0x1fe6, 0x03a5,
0x0342, 0x1ff2, 0x1ffa, 0x0399, 0x1ff3, 0x03a9, 0x0399, 0x1ff4, 0x038f, 0x0399,
0x1ff6, 0x03a9, 0x0342, 0x1ffc, 0x03a9, 0x0399, 0xfb00, 0x0046, 0x0046, 0xfb01,
0x0046, 0x0049, 0xfb02, 0x0046, 0x004c, 0xfb05, 0x0053, 0x0054, 0xfb06, 0x0053,
0x0054, 0xfb13, 0x0544, 0x0546, 0xfb14, 0x0544, 0x0535, 0xfb15, 0x0544, 0x053b,
0xfb16, 0x054e, 0x0546, 0xfb17, 0x0544, 0x053d, 0x0390, 0x0399, 0x0308, 0x0301,
0x03b0, 0x03a5, 0x0308, 0x0301, 0x1f52, 0x03a5, 0x0313, 0x0300, 0x1f54, 0x03a5,
0x0313, 0x0301, 0x1f56, 0x03a5, 0x0313, 0x0342, 0x1fb7, 0x0391, 0x0342, 0x0399,
0x1fc7, 0x0397, 0x0342, 0x0399, 0x1fd2, 0x0399, 0x0308, 0x0300, 0x1fd3, 0x0399,
0x0308, 0x0301, 0x1fd7, 0x0399, 0x0308, 0x0342, 0x1fe2, 0x03a5, 0x0308, 0x0300,
0x1fe3, 0x03a5, 0x0308, 0x0301, 0x1fe7, 0x03a5, 0x0308, 0x0342, 0x1ff7, 0x03a9,
0x0342, 0x0399, 0xfb03, 0x0046, 0x0046, 0x0049, 0xfb04, 0x0046, 0x0046, 0x004c
};
/* Number of one-to-one, one-to-two, and one-to-three uppercase conversions. */
static const uint8_t lit_unicode_upper_case_conversion_counters[] JERRY_ATTR_CONST_DATA =
{
0x001c, 0x002c, 0x0010
0x001b, 0x002c, 0x0010
};