summaryrefslogtreecommitdiffstats
path: root/js/src/irregexp/RegExpEngine.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'js/src/irregexp/RegExpEngine.cpp')
-rw-r--r--js/src/irregexp/RegExpEngine.cpp140
1 files changed, 25 insertions, 115 deletions
diff --git a/js/src/irregexp/RegExpEngine.cpp b/js/src/irregexp/RegExpEngine.cpp
index 2e19065fd..4d691a5dc 100644
--- a/js/src/irregexp/RegExpEngine.cpp
+++ b/js/src/irregexp/RegExpEngine.cpp
@@ -31,10 +31,14 @@
#include "irregexp/RegExpEngine.h"
#include "irregexp/NativeRegExpMacroAssembler.h"
+#include "irregexp/RegExpCharacters.h"
#include "irregexp/RegExpMacroAssembler.h"
#include "jit/ExecutableAllocator.h"
#include "jit/JitCommon.h"
+// Generated table
+#include "irregexp/RegExpCharacters-inl.h"
+
using namespace js;
using namespace js::irregexp;
@@ -61,61 +65,6 @@ RegExpNode::RegExpNode(LifoAlloc* alloc)
bm_info_[0] = bm_info_[1] = nullptr;
}
-// -------------------------------------------------------------------
-// CharacterRange
-
-// The '2' variant has inclusive from and exclusive to.
-// This covers \s as defined in ECMA-262 5.1, 15.10.2.12,
-// which include WhiteSpace (7.2) or LineTerminator (7.3) values.
-static const int kSpaceRanges[] = { '\t', '\r' + 1, ' ', ' ' + 1,
- 0x00A0, 0x00A1, 0x1680, 0x1681, 0x180E, 0x180F, 0x2000, 0x200B,
- 0x2028, 0x202A, 0x202F, 0x2030, 0x205F, 0x2060, 0x3000, 0x3001,
- 0xFEFF, 0xFF00, 0x10000 };
-static const int kSpaceRangeCount = ArrayLength(kSpaceRanges);
-
-static const int kSpaceAndSurrogateRanges[] = { '\t', '\r' + 1, ' ', ' ' + 1,
- 0x00A0, 0x00A1, 0x1680, 0x1681, 0x180E, 0x180F, 0x2000, 0x200B,
- 0x2028, 0x202A, 0x202F, 0x2030, 0x205F, 0x2060, 0x3000, 0x3001,
- unicode::LeadSurrogateMin, unicode::TrailSurrogateMax + 1,
- 0xFEFF, 0xFF00, 0x10000 };
-static const int kSpaceAndSurrogateRangeCount = ArrayLength(kSpaceAndSurrogateRanges);
-static const int kWordRanges[] = {
- '0', '9' + 1, 'A', 'Z' + 1, '_', '_' + 1, 'a', 'z' + 1, 0x10000 };
-static const int kWordRangeCount = ArrayLength(kWordRanges);
-static const int kIgnoreCaseWordRanges[] = {
- '0', '9' + 1, 'A', 'Z' + 1, '_', '_' + 1, 'a', 'z' + 1,
- 0x017F, 0x017F + 1, 0x212A, 0x212A + 1,
- 0x10000 };
-static const int kIgnoreCaseWordCount = ArrayLength(kIgnoreCaseWordRanges);
-static const int kWordAndSurrogateRanges[] = {
- '0', '9' + 1, 'A', 'Z' + 1, '_', '_' + 1, 'a', 'z' + 1,
- unicode::LeadSurrogateMin, unicode::TrailSurrogateMax + 1,
- 0x10000 };
-static const int kWordAndSurrogateRangeCount = ArrayLength(kWordAndSurrogateRanges);
-static const int kNegatedIgnoreCaseWordAndSurrogateRanges[] = {
- 0, '0', '9' + 1, 'A',
- 'Z' + 1, '_', '_' + 1, 'a',
- 'z' + 1, 0x017F,
- 0x017F + 1, 0x212A,
- 0x212A + 1, unicode::LeadSurrogateMin,
- unicode::TrailSurrogateMax + 1, 0x10000,
- 0x10000 };
-static const int kNegatedIgnoreCaseWordAndSurrogateRangeCount =
- ArrayLength(kNegatedIgnoreCaseWordAndSurrogateRanges);
-static const int kDigitRanges[] = { '0', '9' + 1, 0x10000 };
-static const int kDigitRangeCount = ArrayLength(kDigitRanges);
-static const int kDigitAndSurrogateRanges[] = {
- '0', '9' + 1,
- unicode::LeadSurrogateMin, unicode::TrailSurrogateMax + 1,
- 0x10000 };
-static const int kDigitAndSurrogateRangeCount = ArrayLength(kDigitAndSurrogateRanges);
-static const int kSurrogateRanges[] = {
- unicode::LeadSurrogateMin, unicode::TrailSurrogateMax + 1,
- 0x10000 };
-static const int kSurrogateRangeCount = ArrayLength(kSurrogateRanges);
-static const int kLineTerminatorRanges[] = { 0x000A, 0x000B, 0x000D, 0x000E,
- 0x2028, 0x202A, 0x10000 };
-static const int kLineTerminatorRangeCount = ArrayLength(kLineTerminatorRanges);
static const int kMaxOneByteCharCode = 0xff;
static const int kMaxUtf16CodeUnit = 0xffff;
@@ -213,7 +162,7 @@ CharacterRange::AddClassEscapeUnicode(LifoAlloc* alloc, char16_t type,
break;
case 'w':
if (ignore_case)
- AddClass(kIgnoreCaseWordRanges, kIgnoreCaseWordCount, ranges);
+ AddClass(kIgnoreCaseWordRanges, kIgnoreCaseWordRangeCount, ranges);
else
AddClassEscape(alloc, type, ranges);
break;
@@ -233,33 +182,6 @@ CharacterRange::AddClassEscapeUnicode(LifoAlloc* alloc, char16_t type,
}
}
-#define FOR_EACH_NON_ASCII_TO_ASCII_FOLDING(macro) \
- /* LATIN CAPITAL LETTER Y WITH DIAERESIS */ \
- macro(0x0178, 0x00FF) \
- /* LATIN SMALL LETTER LONG S */ \
- macro(0x017F, 0x0073) \
- /* LATIN CAPITAL LETTER SHARP S */ \
- macro(0x1E9E, 0x00DF) \
- /* KELVIN SIGN */ \
- macro(0x212A, 0x006B) \
- /* ANGSTROM SIGN */ \
- macro(0x212B, 0x00E5)
-
-// We need to check for the following characters: 0x39c 0x3bc 0x178.
-static inline bool
-RangeContainsLatin1Equivalents(CharacterRange range, bool unicode)
-{
- /* TODO(dcarney): this could be a lot more efficient. */
- if (unicode) {
-#define CHECK_RANGE(C, F) \
- if (range.Contains(C)) return true;
-FOR_EACH_NON_ASCII_TO_ASCII_FOLDING(CHECK_RANGE)
-#undef CHECK_RANGE
- }
-
- return range.Contains(0x39c) || range.Contains(0x3bc) || range.Contains(0x178);
-}
-
static bool
RangesContainLatin1Equivalents(const CharacterRangeVector& ranges, bool unicode)
{
@@ -336,7 +258,7 @@ GetCaseIndependentLetters(char16_t character,
// step 3.g.
// The standard requires that non-ASCII characters cannot have ASCII
// character codes in their equivalence class, even though this
- // situation occurs multiple times in the unicode tables.
+ // situation occurs multiple times in the Unicode tables.
static const unsigned kMaxAsciiCharCode = 127;
if (upper <= kMaxAsciiCharCode) {
if (character > kMaxAsciiCharCode) {
@@ -365,31 +287,6 @@ GetCaseIndependentLetters(char16_t character,
choices, ArrayLength(choices), letters);
}
-static char16_t
-ConvertNonLatin1ToLatin1(char16_t c, bool unicode)
-{
- MOZ_ASSERT(c > kMaxOneByteCharCode);
- if (unicode) {
- switch (c) {
-#define CONVERT(C, F) case C: return F;
-FOR_EACH_NON_ASCII_TO_ASCII_FOLDING(CONVERT)
-#undef CONVERT
- }
- }
-
- switch (c) {
- // This are equivalent characters in unicode.
- case 0x39c:
- case 0x3bc:
- return 0xb5;
- // This is an uppercase of a Latin-1 character
- // outside of Latin-1.
- case 0x178:
- return 0xff;
- }
- return 0;
-}
-
void
CharacterRange::AddCaseEquivalents(bool is_ascii, bool unicode, CharacterRangeVector* ranges)
{
@@ -2358,7 +2255,10 @@ void
BoyerMoorePositionInfo::SetInterval(const Interval& interval)
{
s_ = AddRange(s_, kSpaceRanges, kSpaceRangeCount, interval);
- w_ = AddRange(w_, kWordRanges, kWordRangeCount, interval);
+ if (unicode_ignore_case_)
+ w_ = AddRange(w_, kIgnoreCaseWordRanges, kIgnoreCaseWordRangeCount, interval);
+ else
+ w_ = AddRange(w_, kWordRanges, kWordRangeCount, interval);
d_ = AddRange(d_, kDigitRanges, kDigitRangeCount, interval);
surrogate_ =
AddRange(surrogate_, kSurrogateRanges, kSurrogateRangeCount, interval);
@@ -2395,11 +2295,12 @@ BoyerMoorePositionInfo::SetAll()
BoyerMooreLookahead::BoyerMooreLookahead(LifoAlloc* alloc, size_t length, RegExpCompiler* compiler)
: length_(length), compiler_(compiler), bitmaps_(*alloc)
{
+ bool unicode_ignore_case = compiler->unicode() && compiler->ignore_case();
max_char_ = MaximumCharacter(compiler->ascii());
bitmaps_.reserve(length);
for (size_t i = 0; i < length; i++)
- bitmaps_.append(alloc->newInfallible<BoyerMoorePositionInfo>(alloc));
+ bitmaps_.append(alloc->newInfallible<BoyerMoorePositionInfo>(alloc, unicode_ignore_case));
}
// Find the longest range of lookahead that has the fewest number of different
@@ -3065,15 +2966,22 @@ EmitNotInSurrogatePair(RegExpCompiler* compiler, RegExpNode* on_success, Trace*
// Check for [0-9A-Z_a-z].
static void
EmitWordCheck(RegExpMacroAssembler* assembler,
- jit::Label* word, jit::Label* non_word, bool fall_through_on_word)
+ jit::Label* word, jit::Label* non_word, bool fall_through_on_word,
+ bool unicode_ignore_case)
{
- if (assembler->CheckSpecialCharacterClass(fall_through_on_word ? 'w' : 'W',
+ if (!unicode_ignore_case &&
+ assembler->CheckSpecialCharacterClass(fall_through_on_word ? 'w' : 'W',
fall_through_on_word ? non_word : word))
{
// Optimized implementation available.
return;
}
+ if (unicode_ignore_case) {
+ assembler->CheckCharacter(0x017F, word);
+ assembler->CheckCharacter(0x212A, word);
+ }
+
assembler->CheckCharacterGT('z', non_word);
assembler->CheckCharacterLT('0', non_word);
assembler->CheckCharacterGT('a' - 1, word);
@@ -3122,7 +3030,8 @@ AssertionNode::EmitBoundaryCheck(RegExpCompiler* compiler, Trace* trace)
assembler->LoadCurrentCharacter(trace->cp_offset(), &before_non_word);
}
// Fall through on non-word.
- EmitWordCheck(assembler, &before_word, &before_non_word, false);
+ EmitWordCheck(assembler, &before_word, &before_non_word, false,
+ compiler->unicode() && compiler->ignore_case());
// Next character is not a word character.
assembler->Bind(&before_non_word);
jit::Label ok;
@@ -3162,7 +3071,8 @@ AssertionNode::BacktrackIfPrevious(RegExpCompiler* compiler,
// We already checked that we are not at the start of input so it must be
// OK to load the previous character.
assembler->LoadCurrentCharacter(new_trace.cp_offset() - 1, &dummy, false);
- EmitWordCheck(assembler, word, non_word, backtrack_if_previous == kIsNonWord);
+ EmitWordCheck(assembler, word, non_word, backtrack_if_previous == kIsNonWord,
+ compiler->unicode() && compiler->ignore_case());
assembler->Bind(&fall_through);
on_success()->Emit(compiler, &new_trace);
generated by cgit v1.2.3 (git 2.25.1) at 2024-07-10 09:27:33 +0000