diff options
author | wolfbeast <mcwerewolf@wolfbeast.com> | 2019-08-18 16:25:15 +0200 |
---|---|---|
committer | wolfbeast <mcwerewolf@wolfbeast.com> | 2019-08-18 16:26:16 +0200 |
commit | 202296d02aa99afa90581333ab059c050b9c3ade (patch) | |
tree | 6ed8a1289bed1bb06d2703ea77cba5fccd3bdffa /js/src/frontend/TokenStream.cpp | |
parent | 1530f48c27fb13d7cbd2708c9f0fcf2dabc6ed6a (diff) | |
parent | ab6242a93b849b0a3c7525b16bc01dd3172fc167 (diff) | |
download | UXP-202296d02aa99afa90581333ab059c050b9c3ade.tar UXP-202296d02aa99afa90581333ab059c050b9c3ade.tar.gz UXP-202296d02aa99afa90581333ab059c050b9c3ade.tar.lz UXP-202296d02aa99afa90581333ab059c050b9c3ade.tar.xz UXP-202296d02aa99afa90581333ab059c050b9c3ade.zip |
Pull Basilisk-release forward.
Diffstat (limited to 'js/src/frontend/TokenStream.cpp')
-rw-r--r-- | js/src/frontend/TokenStream.cpp | 478 |
1 files changed, 303 insertions, 175 deletions
diff --git a/js/src/frontend/TokenStream.cpp b/js/src/frontend/TokenStream.cpp index 8438ff7c5..b8623d545 100644 --- a/js/src/frontend/TokenStream.cpp +++ b/js/src/frontend/TokenStream.cpp @@ -8,6 +8,7 @@ #include "frontend/TokenStream.h" +#include "mozilla/ArrayUtils.h" #include "mozilla/IntegerTypeTraits.h" #include "mozilla/PodOperations.h" @@ -23,80 +24,81 @@ #include "jsnum.h" #include "frontend/BytecodeCompiler.h" +#include "frontend/ReservedWords.h" #include "js/CharacterEncoding.h" #include "js/UniquePtr.h" #include "vm/HelperThreads.h" -#include "vm/Keywords.h" #include "vm/StringBuffer.h" #include "vm/Unicode.h" using namespace js; using namespace js::frontend; +using mozilla::ArrayLength; using mozilla::Maybe; using mozilla::PodAssign; using mozilla::PodCopy; using mozilla::PodZero; -struct KeywordInfo { - const char* chars; // C string with keyword text +struct ReservedWordInfo { + const char* chars; // C string with reserved word text TokenKind tokentype; }; -static const KeywordInfo keywords[] = { -#define KEYWORD_INFO(keyword, name, type) \ - {js_##keyword##_str, type}, - FOR_EACH_JAVASCRIPT_KEYWORD(KEYWORD_INFO) -#undef KEYWORD_INFO +static const ReservedWordInfo reservedWords[] = { +#define RESERVED_WORD_INFO(word, name, type) \ + {js_##word##_str, type}, + FOR_EACH_JAVASCRIPT_RESERVED_WORD(RESERVED_WORD_INFO) +#undef RESERVED_WORD_INFO }; -// Returns a KeywordInfo for the specified characters, or nullptr if the string -// is not a keyword. +// Returns a ReservedWordInfo for the specified characters, or nullptr if the +// string is not a reserved word. template <typename CharT> -static const KeywordInfo* -FindKeyword(const CharT* s, size_t length) +static const ReservedWordInfo* +FindReservedWord(const CharT* s, size_t length) { MOZ_ASSERT(length != 0); size_t i; - const KeywordInfo* kw; + const ReservedWordInfo* rw; const char* chars; -#define JSKW_LENGTH() length -#define JSKW_AT(column) s[column] -#define JSKW_GOT_MATCH(index) i = (index); goto got_match; -#define JSKW_TEST_GUESS(index) i = (index); goto test_guess; -#define JSKW_NO_MATCH() goto no_match; -#include "jsautokw.h" -#undef JSKW_NO_MATCH -#undef JSKW_TEST_GUESS -#undef JSKW_GOT_MATCH -#undef JSKW_AT -#undef JSKW_LENGTH +#define JSRW_LENGTH() length +#define JSRW_AT(column) s[column] +#define JSRW_GOT_MATCH(index) i = (index); goto got_match; +#define JSRW_TEST_GUESS(index) i = (index); goto test_guess; +#define JSRW_NO_MATCH() goto no_match; +#include "frontend/ReservedWordsGenerated.h" +#undef JSRW_NO_MATCH +#undef JSRW_TEST_GUESS +#undef JSRW_GOT_MATCH +#undef JSRW_AT +#undef JSRW_LENGTH got_match: - return &keywords[i]; + return &reservedWords[i]; test_guess: - kw = &keywords[i]; - chars = kw->chars; + rw = &reservedWords[i]; + chars = rw->chars; do { if (*s++ != (unsigned char)(*chars++)) goto no_match; } while (--length != 0); - return kw; + return rw; no_match: return nullptr; } -static const KeywordInfo* -FindKeyword(JSLinearString* str) +static const ReservedWordInfo* +FindReservedWord(JSLinearString* str) { JS::AutoCheckCannotGC nogc; return str->hasLatin1Chars() - ? FindKeyword(str->latin1Chars(nogc), str->length()) - : FindKeyword(str->twoByteChars(nogc), str->length()); + ? FindReservedWord(str->latin1Chars(nogc), str->length()) + : FindReservedWord(str->twoByteChars(nogc), str->length()); } template <typename CharT> @@ -186,7 +188,68 @@ frontend::IsIdentifier(const char16_t* chars, size_t length) bool frontend::IsKeyword(JSLinearString* str) { - return FindKeyword(str) != nullptr; + if (const ReservedWordInfo* rw = FindReservedWord(str)) + return TokenKindIsKeyword(rw->tokentype); + + return false; +} + +bool +frontend::IsFutureReservedWord(JSLinearString* str) +{ + if (const ReservedWordInfo* rw = FindReservedWord(str)) + return TokenKindIsFutureReservedWord(rw->tokentype); + + return false; +} + +bool +frontend::IsStrictReservedWord(JSLinearString* str) +{ + if (const ReservedWordInfo* rw = FindReservedWord(str)) + return TokenKindIsStrictReservedWord(rw->tokentype); + + return false; +} + +bool +frontend::IsReservedWordLiteral(JSLinearString* str) +{ + if (const ReservedWordInfo* rw = FindReservedWord(str)) + return TokenKindIsReservedWordLiteral(rw->tokentype); + + return false; +} + +const char* +frontend::ReservedWordToCharZ(PropertyName* str) +{ + const ReservedWordInfo* rw = FindReservedWord(str); + if (rw == nullptr) + return nullptr; + + switch (rw->tokentype) { +#define EMIT_CASE(word, name, type) case type: return js_##word##_str; + FOR_EACH_JAVASCRIPT_RESERVED_WORD(EMIT_CASE) +#undef EMIT_CASE + default: + MOZ_ASSERT_UNREACHABLE("Not a reserved word PropertyName."); + } + return nullptr; +} + +PropertyName* +TokenStream::reservedWordToPropertyName(TokenKind tt) const +{ + MOZ_ASSERT(tt != TOK_NAME); + switch (tt) { +#define EMIT_CASE(word, name, type) case type: return cx->names().name; + FOR_EACH_JAVASCRIPT_RESERVED_WORD(EMIT_CASE) +#undef EMIT_CASE + default: + MOZ_ASSERT_UNREACHABLE("Not a reserved word TokenKind."); + } + return nullptr; } TokenStream::SourceCoords::SourceCoords(ExclusiveContext* cx, uint32_t ln) @@ -223,8 +286,13 @@ TokenStream::SourceCoords::add(uint32_t lineNum, uint32_t lineStartOffset) // only if lineStartOffsets_.append succeeds, to keep sentinel. // Otherwise return false to tell TokenStream about OOM. uint32_t maxPtr = MAX_PTR; - if (!lineStartOffsets_.append(maxPtr)) + if (!lineStartOffsets_.append(maxPtr)) { + static_assert(mozilla::IsSame<decltype(lineStartOffsets_.allocPolicy()), + TempAllocPolicy&>::value, + "this function's caller depends on it reporting an " + "error on failure, as TempAllocPolicy ensures"); return false; + } lineStartOffsets_[lineIndex] = lineStartOffset; } else { @@ -554,8 +622,9 @@ TokenStream::advance(size_t position) MOZ_MAKE_MEM_UNDEFINED(&cur->type, sizeof(cur->type)); lookahead = 0; - if (flags.hitOOM) - return reportError(JSMSG_OUT_OF_MEMORY); + if (flags.hitOOM) { + return false; + } return true; } @@ -599,8 +668,8 @@ TokenStream::seek(const Position& pos, const TokenStream& other) } bool -TokenStream::reportStrictModeErrorNumberVA(uint32_t offset, bool strictMode, unsigned errorNumber, - va_list args) +TokenStream::reportStrictModeErrorNumberVA(UniquePtr<JSErrorNotes> notes, uint32_t offset, + bool strictMode, unsigned errorNumber, va_list args) { // In strict mode code, this is an error, not merely a warning. unsigned flags; @@ -611,7 +680,7 @@ TokenStream::reportStrictModeErrorNumberVA(uint32_t offset, bool strictMode, uns else return true; - return reportCompileErrorNumberVA(offset, flags, errorNumber, args); + return reportCompileErrorNumberVA(Move(notes), offset, flags, errorNumber, args); } void @@ -637,8 +706,8 @@ CompileError::throwError(JSContext* cx) } bool -TokenStream::reportCompileErrorNumberVA(uint32_t offset, unsigned flags, unsigned errorNumber, - va_list args) +TokenStream::reportCompileErrorNumberVA(UniquePtr<JSErrorNotes> notes, uint32_t offset, + unsigned flags, unsigned errorNumber, va_list args) { bool warning = JSREPORT_IS_WARNING(flags); @@ -655,6 +724,7 @@ TokenStream::reportCompileErrorNumberVA(uint32_t offset, unsigned flags, unsigne return false; CompileError& err = *tempErrPtr; + err.notes = Move(notes); err.flags = flags; err.errorNumber = errorNumber; err.filename = filename; @@ -746,7 +816,7 @@ TokenStream::reportStrictModeError(unsigned errorNumber, ...) { va_list args; va_start(args, errorNumber); - bool result = reportStrictModeErrorNumberVA(currentToken().pos.begin, strictMode(), + bool result = reportStrictModeErrorNumberVA(nullptr, currentToken().pos.begin, strictMode(), errorNumber, args); va_end(args); return result; @@ -757,8 +827,8 @@ TokenStream::reportError(unsigned errorNumber, ...) { va_list args; va_start(args, errorNumber); - bool result = reportCompileErrorNumberVA(currentToken().pos.begin, JSREPORT_ERROR, errorNumber, - args); + bool result = reportCompileErrorNumberVA(nullptr, currentToken().pos.begin, JSREPORT_ERROR, + errorNumber, args); va_end(args); return result; } @@ -768,30 +838,32 @@ TokenStream::reportErrorNoOffset(unsigned errorNumber, ...) { va_list args; va_start(args, errorNumber); - bool result = reportCompileErrorNumberVA(NoOffset, JSREPORT_ERROR, errorNumber, - args); + bool result = reportCompileErrorNumberVA(nullptr, NoOffset, JSREPORT_ERROR, + errorNumber, args); va_end(args); return result; } bool -TokenStream::reportWarning(unsigned errorNumber, ...) +TokenStream::warning(unsigned errorNumber, ...) { va_list args; va_start(args, errorNumber); - bool result = reportCompileErrorNumberVA(currentToken().pos.begin, JSREPORT_WARNING, + bool result = reportCompileErrorNumberVA(nullptr, currentToken().pos.begin, JSREPORT_WARNING, errorNumber, args); va_end(args); return result; } bool -TokenStream::reportExtraWarningErrorNumberVA(uint32_t offset, unsigned errorNumber, va_list args) +TokenStream::reportExtraWarningErrorNumberVA(UniquePtr<JSErrorNotes> notes, uint32_t offset, + unsigned errorNumber, va_list args) { if (!options().extraWarningsOption) return true; - return reportCompileErrorNumberVA(offset, JSREPORT_STRICT|JSREPORT_WARNING, errorNumber, args); + return reportCompileErrorNumberVA(Move(notes), offset, JSREPORT_STRICT|JSREPORT_WARNING, + errorNumber, args); } void @@ -802,7 +874,34 @@ TokenStream::reportAsmJSError(uint32_t offset, unsigned errorNumber, ...) unsigned flags = options().throwOnAsmJSValidationFailureOption ? JSREPORT_ERROR : JSREPORT_WARNING; - reportCompileErrorNumberVA(offset, flags, errorNumber, args); + reportCompileErrorNumberVA(nullptr, offset, flags, errorNumber, args); + va_end(args); +} + +void +TokenStream::error(unsigned errorNumber, ...) +{ + va_list args; + va_start(args, errorNumber); +#ifdef DEBUG + bool result = +#endif + reportCompileErrorNumberVA(nullptr, currentToken().pos.begin, JSREPORT_ERROR, + errorNumber, args); + MOZ_ASSERT(!result, "reporting an error returned true?"); + va_end(args); +} + +void +TokenStream::errorAt(uint32_t offset, unsigned errorNumber, ...) +{ + va_list args; + va_start(args, errorNumber); +#ifdef DEBUG + bool result = +#endif + reportCompileErrorNumberVA(nullptr, offset, JSREPORT_ERROR, errorNumber, args); + MOZ_ASSERT(!result, "reporting an error returned true?"); va_end(args); } @@ -934,34 +1033,49 @@ TokenStream::getDirectives(bool isMultiline, bool shouldWarnDeprecated) bool TokenStream::getDirective(bool isMultiline, bool shouldWarnDeprecated, - const char* directive, int directiveLength, + const char* directive, uint8_t directiveLength, const char* errorMsgPragma, UniqueTwoByteChars* destination) { MOZ_ASSERT(directiveLength <= 18); char16_t peeked[18]; - int32_t c; if (peekChars(directiveLength, peeked) && CharsMatch(peeked, directive)) { - if (shouldWarnDeprecated && - !reportWarning(JSMSG_DEPRECATED_PRAGMA, errorMsgPragma)) - return false; + if (shouldWarnDeprecated) { + if (!warning(JSMSG_DEPRECATED_PRAGMA, errorMsgPragma)) + return false; + } skipChars(directiveLength); tokenbuf.clear(); - while ((c = peekChar()) && c != EOF && !unicode::IsSpaceOrBOM2(c)) { - getChar(); + do { + int32_t c; + if (!peekChar(&c)) + return false; + + if (c == EOF || unicode::IsSpaceOrBOM2(c)) + break; + + consumeKnownChar(c); + // Debugging directives can occur in both single- and multi-line // comments. If we're currently inside a multi-line comment, we also // need to recognize multi-line comment terminators. - if (isMultiline && c == '*' && peekChar() == '/') { - ungetChar('*'); - break; + if (isMultiline && c == '*') { + int32_t c2; + if (!peekChar(&c2)) + return false; + + if (c2 == '/') { + ungetChar('*'); + break; + } } + if (!tokenbuf.append(c)) return false; - } + } while (true); if (tokenbuf.empty()) { // The directive's URL was missing, but this is not quite an @@ -993,7 +1107,10 @@ TokenStream::getDisplayURL(bool isMultiline, bool shouldWarnDeprecated) // developer would like to refer to the source as from the source's actual // URL. - return getDirective(isMultiline, shouldWarnDeprecated, " sourceURL=", 11, + static const char sourceURLDirective[] = " sourceURL="; + constexpr uint8_t sourceURLDirectiveLength = ArrayLength(sourceURLDirective) - 1; + return getDirective(isMultiline, shouldWarnDeprecated, + sourceURLDirective, sourceURLDirectiveLength, "sourceURL", &displayURL_); } @@ -1003,7 +1120,10 @@ TokenStream::getSourceMappingURL(bool isMultiline, bool shouldWarnDeprecated) // Match comments of the form "//# sourceMappingURL=<url>" or // "/\* //# sourceMappingURL=<url> *\/" - return getDirective(isMultiline, shouldWarnDeprecated, " sourceMappingURL=", 18, + static const char sourceMappingURLDirective[] = " sourceMappingURL="; + constexpr uint8_t sourceMappingURLDirectiveLength = ArrayLength(sourceMappingURLDirective) - 1; + return getDirective(isMultiline, shouldWarnDeprecated, + sourceMappingURLDirective, sourceMappingURLDirectiveLength, "sourceMappingURL", &sourceMapURL_); } @@ -1110,36 +1230,6 @@ TokenStream::putIdentInTokenbuf(const char16_t* identStart) return true; } -bool -TokenStream::checkForKeyword(const KeywordInfo* kw, TokenKind* ttp) -{ - if (!awaitIsKeyword && kw->tokentype == TOK_AWAIT) { - if (ttp) - *ttp = TOK_NAME; - return true; - } - - if (kw->tokentype == TOK_RESERVED) - return reportError(JSMSG_RESERVED_ID, kw->chars); - - if (kw->tokentype == TOK_STRICT_RESERVED) - return reportStrictModeError(JSMSG_RESERVED_ID, kw->chars); - - // Working keyword. - *ttp = kw->tokentype; - return true; -} - -bool -TokenStream::checkForKeyword(JSAtom* atom, TokenKind* ttp) -{ - const KeywordInfo* kw = FindKeyword(atom); - if (!kw) - return true; - - return checkForKeyword(kw, ttp); -} - enum FirstCharKind { // A char16_t has the 'OneChar' kind if it, by itself, constitutes a valid // token that cannot also be a prefix of a longer token. E.g. ';' has the @@ -1363,36 +1453,18 @@ TokenStream::getTokenInternal(TokenKind* ttp, Modifier modifier) length = userbuf.addressOfNextRawChar() - identStart; } - // Represent keywords as keyword tokens unless told otherwise. - if (modifier != KeywordIsName) { - if (const KeywordInfo* kw = FindKeyword(chars, length)) { - // That said, keywords can't contain escapes. (Contexts where - // keywords are treated as names, that also sometimes treat - // keywords as keywords, must manually check this requirement.) - // There are two exceptions - // 1) StrictReservedWords: These keywords need to be treated as - // names in non-strict mode. - // 2) yield is also treated as a name if it contains an escape - // sequence. The parser must handle this case separately. - if (hadUnicodeEscape && !( - (kw->tokentype == TOK_STRICT_RESERVED && !strictMode()) || - kw->tokentype == TOK_YIELD)) - { - reportError(JSMSG_ESCAPED_KEYWORD); - goto error; - } - - tp->type = TOK_NAME; - if (!checkForKeyword(kw, &tp->type)) - goto error; - if (tp->type != TOK_NAME && !hadUnicodeEscape) - goto out; + // Represent reserved words as reserved word tokens. + if (!hadUnicodeEscape) { + if (const ReservedWordInfo* rw = FindReservedWord(chars, length)) { + tp->type = rw->tokentype; + goto out; } } JSAtom* atom = AtomizeChars(cx, chars, length); - if (!atom) + if (!atom) { goto error; + } tp->type = TOK_NAME; tp->setName(atom->asPropertyName()); goto out; @@ -1538,10 +1610,11 @@ TokenStream::getTokenInternal(TokenKind* ttp, Modifier modifier) // grammar. We might not always be so permissive, so we warn // about it. if (c >= '8') { - if (!reportWarning(JSMSG_BAD_OCTAL, c == '8' ? "08" : "09")) { + if (!warning(JSMSG_BAD_OCTAL, c == '8' ? "08" : "09")) goto error; - } - goto decimal; // use the decimal scanner for the rest of the number + + // Use the decimal scanner for the rest of the number. + goto decimal; } c = getCharIgnoreEOL(); } @@ -1690,7 +1763,8 @@ TokenStream::getTokenInternal(TokenKind* ttp, Modifier modifier) case '/': // Look for a single-line comment. if (matchChar('/')) { - c = peekChar(); + if (!peekChar(&c)) + goto error; if (c == '@' || c == '#') { bool shouldWarn = getChar() == '@'; if (!getDirectives(false, shouldWarn)) @@ -1757,7 +1831,8 @@ TokenStream::getTokenInternal(TokenKind* ttp, Modifier modifier) RegExpFlag reflags = NoFlags; unsigned length = tokenbuf.length() + 1; while (true) { - c = peekChar(); + if (!peekChar(&c)) + goto error; if (c == 'g' && !(reflags & GlobalFlag)) reflags = RegExpFlag(reflags | GlobalFlag); else if (c == 'i' && !(reflags & IgnoreCaseFlag)) @@ -1774,7 +1849,8 @@ TokenStream::getTokenInternal(TokenKind* ttp, Modifier modifier) length++; } - c = peekChar(); + if (!peekChar(&c)) + goto error; if (JS7_ISLET(c)) { char buf[2] = { '\0', '\0' }; tp->pos.begin += length + 1; @@ -1797,8 +1873,13 @@ TokenStream::getTokenInternal(TokenKind* ttp, Modifier modifier) case '-': if (matchChar('-')) { - if (peekChar() == '>' && !flags.isDirtyLine) + int32_t c2; + if (!peekChar(&c2)) + goto error; + + if (c2 == '>' && !flags.isDirtyLine) goto skipline; + tp->type = TOK_DEC; } else { tp->type = matchChar('=') ? TOK_SUBASSIGN : TOK_SUB; @@ -1814,8 +1895,9 @@ TokenStream::getTokenInternal(TokenKind* ttp, Modifier modifier) MOZ_CRASH("should have jumped to |out| or |error|"); out: - if (flags.hitOOM) - return reportError(JSMSG_OUT_OF_MEMORY); + if (flags.hitOOM) { + return false; + } flags.isDirtyLine = true; tp->pos.end = userbuf.offset(); @@ -1831,8 +1913,9 @@ TokenStream::getTokenInternal(TokenKind* ttp, Modifier modifier) return true; error: - if (flags.hitOOM) - return reportError(JSMSG_OUT_OF_MEMORY); + if (flags.hitOOM) { + return false; + } flags.isDirtyLine = true; tp->pos.end = userbuf.offset(); @@ -1850,37 +1933,6 @@ TokenStream::getTokenInternal(TokenKind* ttp, Modifier modifier) } bool -TokenStream::getBracedUnicode(uint32_t* cp) -{ - consumeKnownChar('{'); - - bool first = true; - int32_t c; - uint32_t code = 0; - while (true) { - c = getCharIgnoreEOL(); - if (c == EOF) - return false; - if (c == '}') { - if (first) - return false; - break; - } - - if (!JS7_ISHEX(c)) - return false; - - code = (code << 4) | JS7_UNHEX(c); - if (code > unicode::NonBMPMax) - return false; - first = false; - } - - *cp = code; - return true; -} - -bool TokenStream::getStringOrTemplateToken(int untilChar, Token** tp) { int c; @@ -1897,11 +1949,15 @@ TokenStream::getStringOrTemplateToken(int untilChar, Token** tp) while ((c = getCharIgnoreEOL()) != untilChar) { if (c == EOF) { ungetCharIgnoreEOL(c); - reportError(JSMSG_UNTERMINATED_STRING); + error(JSMSG_UNTERMINATED_STRING); return false; } if (c == '\\') { + // When parsing templates, we don't immediately report errors for + // invalid escapes; these are handled by the parser. + // In those cases we don't append to tokenbuf, since it won't be + // read. switch (c = getChar()) { case 'b': c = '\b'; break; case 'f': c = '\f'; break; @@ -1917,12 +1973,73 @@ TokenStream::getStringOrTemplateToken(int untilChar, Token** tp) // Unicode character specification. case 'u': { - if (peekChar() == '{') { - uint32_t code; - if (!getBracedUnicode(&code)) { - reportError(JSMSG_MALFORMED_ESCAPE, "Unicode"); - return false; - } + uint32_t code = 0; + + int32_t c2; + if (!peekChar(&c2)) + return false; + + uint32_t start = userbuf.offset() - 2; + + if (c2 == '{') { + consumeKnownChar('{'); + + bool first = true; + bool valid = true; + do { + int32_t c = getCharIgnoreEOL(); + if (c == EOF) { + if (parsingTemplate) { + setInvalidTemplateEscape(start, InvalidEscapeType::Unicode); + valid = false; + break; + } + reportInvalidEscapeError(start, InvalidEscapeType::Unicode); + return false; + } + if (c == '}') { + if (first) { + if (parsingTemplate) { + setInvalidTemplateEscape(start, InvalidEscapeType::Unicode); + valid = false; + break; + } + reportInvalidEscapeError(start, InvalidEscapeType::Unicode); + return false; + } + break; + } + + if (!JS7_ISHEX(c)) { + if (parsingTemplate) { + // We put the character back so that we read + // it on the next pass, which matters if it + // was '`' or '\'. + ungetCharIgnoreEOL(c); + setInvalidTemplateEscape(start, InvalidEscapeType::Unicode); + valid = false; + break; + } + reportInvalidEscapeError(start, InvalidEscapeType::Unicode); + return false; + } + + code = (code << 4) | JS7_UNHEX(c); + if (code > unicode::NonBMPMax) { + if (parsingTemplate) { + setInvalidTemplateEscape(start + 3, InvalidEscapeType::UnicodeOverflow); + valid = false; + break; + } + reportInvalidEscapeError(start + 3, InvalidEscapeType::UnicodeOverflow); + return false; + } + + first = false; + } while (true); + + if (!valid) + continue; MOZ_ASSERT(code <= unicode::NonBMPMax); if (code < unicode::NonBMPMin) { @@ -1945,7 +2062,11 @@ TokenStream::getStringOrTemplateToken(int untilChar, Token** tp) c = (c << 4) + JS7_UNHEX(cp[3]); skipChars(4); } else { - reportError(JSMSG_MALFORMED_ESCAPE, "Unicode"); + if (parsingTemplate) { + setInvalidTemplateEscape(start, InvalidEscapeType::Unicode); + continue; + } + reportInvalidEscapeError(start, InvalidEscapeType::Unicode); return false; } break; @@ -1958,7 +2079,12 @@ TokenStream::getStringOrTemplateToken(int untilChar, Token** tp) c = (JS7_UNHEX(cp[0]) << 4) + JS7_UNHEX(cp[1]); skipChars(2); } else { - reportError(JSMSG_MALFORMED_ESCAPE, "hexadecimal"); + uint32_t start = userbuf.offset() - 2; + if (parsingTemplate) { + setInvalidTemplateEscape(start, InvalidEscapeType::Hexadecimal); + continue; + } + reportInvalidEscapeError(start, InvalidEscapeType::Hexadecimal); return false; } break; @@ -1969,13 +2095,14 @@ TokenStream::getStringOrTemplateToken(int untilChar, Token** tp) if (JS7_ISOCT(c)) { int32_t val = JS7_UNOCT(c); - c = peekChar(); + if (!peekChar(&c)) + return false; // Strict mode code allows only \0, then a non-digit. if (val != 0 || JS7_ISDEC(c)) { if (parsingTemplate) { - reportError(JSMSG_DEPRECATED_OCTAL); - return false; + setInvalidTemplateEscape(userbuf.offset() - 2, InvalidEscapeType::Octal); + continue; } if (!reportStrictModeError(JSMSG_DEPRECATED_OCTAL)) return false; @@ -1985,7 +2112,8 @@ TokenStream::getStringOrTemplateToken(int untilChar, Token** tp) if (JS7_ISOCT(c)) { val = 8 * val + JS7_UNOCT(c); getChar(); - c = peekChar(); + if (!peekChar(&c)) + return false; if (JS7_ISOCT(c)) { int32_t save = val; val = 8 * val + JS7_UNOCT(c); @@ -2003,7 +2131,7 @@ TokenStream::getStringOrTemplateToken(int untilChar, Token** tp) } else if (TokenBuf::isRawEOLChar(c)) { if (!parsingTemplate) { ungetCharIgnoreEOL(c); - reportError(JSMSG_UNTERMINATED_STRING); + error(JSMSG_UNTERMINATED_STRING); return false; } if (c == '\r') { |