| 1 | // Copyright (c) 2014, the Dart project authors. Please see the AUTHORS file |
| 2 | // for details. All rights reserved. Use of this source code is governed by a |
| 3 | // BSD-style license that can be found in the LICENSE file. |
| 4 | |
| 5 | #include "vm/regexp_assembler.h" |
| 6 | |
| 7 | #include "unicode/uchar.h" |
| 8 | |
| 9 | #include "platform/unicode.h" |
| 10 | |
| 11 | #include "vm/flags.h" |
| 12 | #include "vm/regexp.h" |
| 13 | #include "vm/runtime_entry.h" |
| 14 | #include "vm/unibrow-inl.h" |
| 15 | |
| 16 | namespace dart { |
| 17 | |
| 18 | void PrintUtf16(uint16_t c) { |
| 19 | const char* format = |
| 20 | (0x20 <= c && c <= 0x7F) ? "%c" : (c <= 0xff) ? "\\x%02x" : "\\u%04x" ; |
| 21 | OS::PrintErr(format, c); |
| 22 | } |
| 23 | |
| 24 | uword /*BoolPtr*/ CaseInsensitiveCompareUCS2(uword /*StringPtr*/ str_raw, |
| 25 | uword /*SmiPtr*/ lhs_index_raw, |
| 26 | uword /*SmiPtr*/ rhs_index_raw, |
| 27 | uword /*SmiPtr*/ length_raw) { |
| 28 | const String& str = String::Handle(static_cast<StringPtr>(str_raw)); |
| 29 | const Smi& lhs_index = Smi::Handle(static_cast<SmiPtr>(lhs_index_raw)); |
| 30 | const Smi& rhs_index = Smi::Handle(static_cast<SmiPtr>(rhs_index_raw)); |
| 31 | const Smi& length = Smi::Handle(static_cast<SmiPtr>(length_raw)); |
| 32 | |
| 33 | // TODO(zerny): Optimize as single instance. V8 has this as an |
| 34 | // isolate member. |
| 35 | unibrow::Mapping<unibrow::Ecma262Canonicalize> canonicalize; |
| 36 | |
| 37 | for (intptr_t i = 0; i < length.Value(); i++) { |
| 38 | int32_t c1 = str.CharAt(lhs_index.Value() + i); |
| 39 | int32_t c2 = str.CharAt(rhs_index.Value() + i); |
| 40 | if (c1 != c2) { |
| 41 | int32_t s1[1] = {c1}; |
| 42 | canonicalize.get(c1, '\0', s1); |
| 43 | if (s1[0] != c2) { |
| 44 | int32_t s2[1] = {c2}; |
| 45 | canonicalize.get(c2, '\0', s2); |
| 46 | if (s1[0] != s2[0]) { |
| 47 | return static_cast<uword>(Bool::False().raw()); |
| 48 | } |
| 49 | } |
| 50 | } |
| 51 | } |
| 52 | return static_cast<uword>(Bool::True().raw()); |
| 53 | } |
| 54 | |
| 55 | uword /*BoolPtr*/ CaseInsensitiveCompareUTF16(uword /*StringPtr*/ str_raw, |
| 56 | uword /*SmiPtr*/ lhs_index_raw, |
| 57 | uword /*SmiPtr*/ rhs_index_raw, |
| 58 | uword /*SmiPtr*/ length_raw) { |
| 59 | const String& str = String::Handle(static_cast<StringPtr>(str_raw)); |
| 60 | const Smi& lhs_index = Smi::Handle(static_cast<SmiPtr>(lhs_index_raw)); |
| 61 | const Smi& rhs_index = Smi::Handle(static_cast<SmiPtr>(rhs_index_raw)); |
| 62 | const Smi& length = Smi::Handle(static_cast<SmiPtr>(length_raw)); |
| 63 | |
| 64 | for (intptr_t i = 0; i < length.Value(); i++) { |
| 65 | int32_t c1 = str.CharAt(lhs_index.Value() + i); |
| 66 | int32_t c2 = str.CharAt(rhs_index.Value() + i); |
| 67 | if (Utf16::IsLeadSurrogate(c1)) { |
| 68 | // Non-BMP characters do not have case-equivalents in the BMP. |
| 69 | // Both have to be non-BMP for them to be able to match. |
| 70 | if (!Utf16::IsLeadSurrogate(c2)) |
| 71 | return static_cast<uword>(Bool::False().raw()); |
| 72 | if (i + 1 < length.Value()) { |
| 73 | uint16_t c1t = str.CharAt(lhs_index.Value() + i + 1); |
| 74 | uint16_t c2t = str.CharAt(rhs_index.Value() + i + 1); |
| 75 | if (Utf16::IsTrailSurrogate(c1t) && Utf16::IsTrailSurrogate(c2t)) { |
| 76 | c1 = Utf16::Decode(c1, c1t); |
| 77 | c2 = Utf16::Decode(c2, c2t); |
| 78 | i++; |
| 79 | } |
| 80 | } |
| 81 | } |
| 82 | c1 = u_foldCase(c1, U_FOLD_CASE_DEFAULT); |
| 83 | c2 = u_foldCase(c2, U_FOLD_CASE_DEFAULT); |
| 84 | if (c1 != c2) return static_cast<uword>(Bool::False().raw()); |
| 85 | } |
| 86 | return static_cast<uword>(Bool::True().raw()); |
| 87 | } |
| 88 | |
| 89 | DEFINE_RAW_LEAF_RUNTIME_ENTRY( |
| 90 | CaseInsensitiveCompareUCS2, |
| 91 | 4, |
| 92 | false /* is_float */, |
| 93 | reinterpret_cast<RuntimeFunction>(&CaseInsensitiveCompareUCS2)); |
| 94 | |
| 95 | DEFINE_RAW_LEAF_RUNTIME_ENTRY( |
| 96 | CaseInsensitiveCompareUTF16, |
| 97 | 4, |
| 98 | false /* is_float */, |
| 99 | reinterpret_cast<RuntimeFunction>(&CaseInsensitiveCompareUTF16)); |
| 100 | |
| 101 | BlockLabel::BlockLabel() { |
| 102 | #if !defined(DART_PRECOMPILED_RUNTIME) |
| 103 | if (!FLAG_interpret_irregexp) { |
| 104 | // Only needed by the compiled IR backend. |
| 105 | block_ = |
| 106 | new JoinEntryInstr(-1, -1, CompilerState::Current().GetNextDeoptId()); |
| 107 | } |
| 108 | #endif |
| 109 | } |
| 110 | |
| 111 | RegExpMacroAssembler::RegExpMacroAssembler(Zone* zone) |
| 112 | : slow_safe_compiler_(false), global_mode_(NOT_GLOBAL), zone_(zone) {} |
| 113 | |
| 114 | RegExpMacroAssembler::~RegExpMacroAssembler() {} |
| 115 | |
| 116 | void RegExpMacroAssembler::CheckNotInSurrogatePair(intptr_t cp_offset, |
| 117 | BlockLabel* on_failure) { |
| 118 | BlockLabel ok; |
| 119 | // Check that current character is not a trail surrogate. |
| 120 | LoadCurrentCharacter(cp_offset, &ok); |
| 121 | CheckCharacterNotInRange(Utf16::kTrailSurrogateStart, |
| 122 | Utf16::kTrailSurrogateEnd, &ok); |
| 123 | // Check that previous character is not a lead surrogate. |
| 124 | LoadCurrentCharacter(cp_offset - 1, &ok); |
| 125 | CheckCharacterInRange(Utf16::kLeadSurrogateStart, Utf16::kLeadSurrogateEnd, |
| 126 | on_failure); |
| 127 | BindBlock(&ok); |
| 128 | } |
| 129 | |
| 130 | } // namespace dart |
| 131 | |