summaryrefslogtreecommitdiffstats
path: root/js/src/irregexp/RegExpCharacters.cpp
diff options
context:
space:
mode:
authorwolfbeast <mcwerewolf@gmail.com>2018-03-12 14:15:56 +0100
committerwolfbeast <mcwerewolf@gmail.com>2018-03-12 14:15:56 +0100
commitb392e6d2ea60191615771900690b37b52b47bcd3 (patch)
tree64904a45624d866ff0232dc2f5b50478b17d69ba /js/src/irregexp/RegExpCharacters.cpp
parentba74a4174d2cd6ccbabbc5aa6f4ffdf74b48f45c (diff)
downloadUXP-b392e6d2ea60191615771900690b37b52b47bcd3.tar
UXP-b392e6d2ea60191615771900690b37b52b47bcd3.tar.gz
UXP-b392e6d2ea60191615771900690b37b52b47bcd3.tar.lz
UXP-b392e6d2ea60191615771900690b37b52b47bcd3.tar.xz
UXP-b392e6d2ea60191615771900690b37b52b47bcd3.zip
Generate irregexp character tables with make_unicode.py.
Diffstat (limited to 'js/src/irregexp/RegExpCharacters.cpp')
-rw-r--r--js/src/irregexp/RegExpCharacters.cpp135
1 files changed, 135 insertions, 0 deletions
diff --git a/js/src/irregexp/RegExpCharacters.cpp b/js/src/irregexp/RegExpCharacters.cpp
new file mode 100644
index 000000000..096c02760
--- /dev/null
+++ b/js/src/irregexp/RegExpCharacters.cpp
@@ -0,0 +1,135 @@
+/* Generated by make_unicode.py DO NOT MODIFY */
+/* Unicode version: 9.0.0 */
+#include "irregexp/RegExpCharacters.h"
+
+#include "mozilla/Assertions.h"
+
+char16_t
+js::irregexp::ConvertNonLatin1ToLatin1(char16_t c, bool unicode)
+{
+ MOZ_ASSERT(c > 0xFF, "Character mustn't be Latin1");
+ if (unicode) {
+ // "LATIN SMALL LETTER LONG S" case folds to "LATIN SMALL LETTER S".
+ if (c == 0x017F)
+ return 0x73;
+ // "LATIN CAPITAL LETTER SHARP S" case folds to "LATIN SMALL LETTER SHARP S".
+ if (c == 0x1E9E)
+ return 0xDF;
+ // "KELVIN SIGN" case folds to "LATIN SMALL LETTER K".
+ if (c == 0x212A)
+ return 0x6B;
+ // "ANGSTROM SIGN" case folds to "LATIN SMALL LETTER A WITH RING ABOVE".
+ if (c == 0x212B)
+ return 0xE5;
+ }
+
+ // "GREEK CAPITAL LETTER MU" case maps to "MICRO SIGN".
+ // "GREEK SMALL LETTER MU" case maps to "MICRO SIGN".
+ if (c == 0x039C || c == 0x03BC)
+ return 0xB5;
+ // "LATIN CAPITAL LETTER Y WITH DIAERESIS" case maps to "LATIN SMALL LETTER Y WITH DIAERESIS".
+ if (c == 0x0178)
+ return 0xFF;
+ return 0;
+}
+
+const int js::irregexp::kSpaceRanges[] = {
+ 0x0009, 0x000D + 1, // CHARACTER TABULATION..CARRIAGE RETURN (CR)
+ 0x0020, 0x0020 + 1, // SPACE
+ 0x00A0, 0x00A0 + 1, // NO-BREAK SPACE
+ 0x1680, 0x1680 + 1, // OGHAM SPACE MARK
+ 0x2000, 0x200A + 1, // EN QUAD..HAIR SPACE
+ 0x2028, 0x2029 + 1, // LINE SEPARATOR..PARAGRAPH SEPARATOR
+ 0x202F, 0x202F + 1, // NARROW NO-BREAK SPACE
+ 0x205F, 0x205F + 1, // MEDIUM MATHEMATICAL SPACE
+ 0x3000, 0x3000 + 1, // IDEOGRAPHIC SPACE
+ 0xFEFF, 0xFEFF + 1, // ZERO WIDTH NO-BREAK SPACE
+ 0xFFFF + 1
+};
+const int js::irregexp::kSpaceRangeCount = 21;
+
+const int js::irregexp::kSpaceAndSurrogateRanges[] = {
+ 0x0009, 0x000D + 1, // CHARACTER TABULATION..CARRIAGE RETURN (CR)
+ 0x0020, 0x0020 + 1, // SPACE
+ 0x00A0, 0x00A0 + 1, // NO-BREAK SPACE
+ 0x1680, 0x1680 + 1, // OGHAM SPACE MARK
+ 0x2000, 0x200A + 1, // EN QUAD..HAIR SPACE
+ 0x2028, 0x2029 + 1, // LINE SEPARATOR..PARAGRAPH SEPARATOR
+ 0x202F, 0x202F + 1, // NARROW NO-BREAK SPACE
+ 0x205F, 0x205F + 1, // MEDIUM MATHEMATICAL SPACE
+ 0x3000, 0x3000 + 1, // IDEOGRAPHIC SPACE
+ 0xD800, 0xDFFF + 1, // <Lead Surrogate Min>..<Trail Surrogate Max>
+ 0xFEFF, 0xFEFF + 1, // ZERO WIDTH NO-BREAK SPACE
+ 0xFFFF + 1
+};
+const int js::irregexp::kSpaceAndSurrogateRangeCount = 23;
+
+const int js::irregexp::kWordRanges[] = {
+ 0x0030, 0x0039 + 1, // DIGIT ZERO..DIGIT NINE
+ 0x0041, 0x005A + 1, // LATIN CAPITAL LETTER A..LATIN CAPITAL LETTER Z
+ 0x005F, 0x005F + 1, // LOW LINE
+ 0x0061, 0x007A + 1, // LATIN SMALL LETTER A..LATIN SMALL LETTER Z
+ 0xFFFF + 1
+};
+const int js::irregexp::kWordRangeCount = 9;
+
+const int js::irregexp::kIgnoreCaseWordRanges[] = {
+ 0x0030, 0x0039 + 1, // DIGIT ZERO..DIGIT NINE
+ 0x0041, 0x005A + 1, // LATIN CAPITAL LETTER A..LATIN CAPITAL LETTER Z
+ 0x005F, 0x005F + 1, // LOW LINE
+ 0x0061, 0x007A + 1, // LATIN SMALL LETTER A..LATIN SMALL LETTER Z
+ 0x017F, 0x017F + 1, // LATIN SMALL LETTER LONG S
+ 0x212A, 0x212A + 1, // KELVIN SIGN
+ 0xFFFF + 1
+};
+const int js::irregexp::kIgnoreCaseWordRangeCount = 13;
+
+const int js::irregexp::kWordAndSurrogateRanges[] = {
+ 0x0030, 0x0039 + 1, // DIGIT ZERO..DIGIT NINE
+ 0x0041, 0x005A + 1, // LATIN CAPITAL LETTER A..LATIN CAPITAL LETTER Z
+ 0x005F, 0x005F + 1, // LOW LINE
+ 0x0061, 0x007A + 1, // LATIN SMALL LETTER A..LATIN SMALL LETTER Z
+ 0xD800, 0xDFFF + 1, // <Lead Surrogate Min>..<Trail Surrogate Max>
+ 0xFFFF + 1
+};
+const int js::irregexp::kWordAndSurrogateRangeCount = 11;
+
+const int js::irregexp::kNegatedIgnoreCaseWordAndSurrogateRanges[] = {
+ 0x0000, 0x002F + 1, // NULL..SOLIDUS
+ 0x003A, 0x0040 + 1, // COLON..COMMERCIAL AT
+ 0x005B, 0x005E + 1, // LEFT SQUARE BRACKET..CIRCUMFLEX ACCENT
+ 0x0060, 0x0060 + 1, // GRAVE ACCENT
+ 0x007B, 0x017E + 1, // LEFT CURLY BRACKET..LATIN SMALL LETTER Z WITH CARON
+ 0x0180, 0x2129 + 1, // LATIN SMALL LETTER B WITH STROKE..TURNED GREEK SMALL LETTER IOTA
+ 0x212B, 0xD7FF + 1, // ANGSTROM SIGN..<Unused>
+ 0xE000, 0xFFFF + 1, // Private Use..<Unused>
+ 0xFFFF + 1
+};
+const int js::irregexp::kNegatedIgnoreCaseWordAndSurrogateRangeCount = 17;
+
+const int js::irregexp::kDigitRanges[] = {
+ 0x0030, 0x0039 + 1, // DIGIT ZERO..DIGIT NINE
+ 0xFFFF + 1
+};
+const int js::irregexp::kDigitRangeCount = 3;
+
+const int js::irregexp::kDigitAndSurrogateRanges[] = {
+ 0x0030, 0x0039 + 1, // DIGIT ZERO..DIGIT NINE
+ 0xD800, 0xDFFF + 1, // <Lead Surrogate Min>..<Trail Surrogate Max>
+ 0xFFFF + 1
+};
+const int js::irregexp::kDigitAndSurrogateRangeCount = 5;
+
+const int js::irregexp::kSurrogateRanges[] = {
+ 0xD800, 0xDFFF + 1, // <Lead Surrogate Min>..<Trail Surrogate Max>
+ 0xFFFF + 1
+};
+const int js::irregexp::kSurrogateRangeCount = 3;
+
+const int js::irregexp::kLineTerminatorRanges[] = {
+ 0x000A, 0x000A + 1, // LINE FEED (LF)
+ 0x000D, 0x000D + 1, // CARRIAGE RETURN (CR)
+ 0x2028, 0x2029 + 1, // LINE SEPARATOR..PARAGRAPH SEPARATOR
+ 0xFFFF + 1
+};
+const int js::irregexp::kLineTerminatorRangeCount = 7;