X-Git-Url: https://git.saurik.com/cycript.git/blobdiff_plain/8c191568adc897a5ac3ec251e17a48a2c9deb65e..7c4c728ddba1afa034d842723580e8ead8f71ac3:/Cycript.l.in diff --git a/Cycript.l.in b/Cycript.l.in index 85db0e8..fa27d9f 100644 --- a/Cycript.l.in +++ b/Cycript.l.in @@ -21,13 +21,20 @@ /* XXX: supposedly I will be screwed on very very long multi-line comments and need to replace these with a manual lexer. http://websrv.cs.fsu.edu/~engelen/courses/COP5621/Pr2.pdf */ -%{ - +%top{ #if defined(__clang__) #pragma clang diagnostic push -#pragma clang diagnostic ignored "-Wunknown-pragmas" +#pragma clang diagnostic ignored "-Wunused-variable" #pragma clang diagnostic ignored "-Wdeprecated-register" +#else +#pragma GCC diagnostic push +#pragma GCC diagnostic ignored "-Wsign-compare" +#pragma GCC diagnostic ignored "-Wunused-function" +#pragma GCC diagnostic ignored "-Wunused-variable" #endif +} + +%{ #define YYLTYPE CYLocation #include "Cycript.tab.hh" @@ -49,8 +56,9 @@ typedef cy::parser::token tk; return token; \ } while (false) -#define A new($pool) -#define Y $pool.strmemdup(yytext, yyleng) +#define P yyextra->pool_ +#define A new(P) +#define Y P.strmemdup(yytext, yyleng) #define I(type, Type, value, highlight) do { \ yylval->type ## _ = A CY ## Type; \ @@ -116,6 +124,46 @@ int H(char c) { return -1; } +static void U(char *&local, unsigned point) { + if (false) { + } else if (point < 0x000080) { + *local++ = point; + } else if (point < 0x000800) { + *local++ = 0xc0 | point >> 0x06 & 0x1f; + goto one; + } else if (point < 0x010000) { + *local++ = 0xe0 | point >> 0x0c & 0x0f; + goto two; + } else if (point < 0x110000) { + *local++ = 0xf0 | point >> 0x12 & 0x07; + *local++ = 0x80 | point >> 0x0c & 0x3f; + two: + *local++ = 0x80 | point >> 0x06 & 0x3f; + one: + *local++ = 0x80 | point >> 0x00 & 0x3f; + } else _assert(false); +} + +static void U(char *&local, const char *text, yy_size_t &i) { + unsigned point; + + char next(text[++i]); + if (next != '{') { + point = H(text[i + 0]) << 12 | H(text[i + 1]) << 8 | H(text[i + 2]) << 4 | H(text[i + 3]); + i += 3; + } else { + point = 0; + for (;;) { + next = text[++i]; + if (next == '}') + break; + point = (point << 4) | H(next); + } + } + + U(local, point); +} + #define YY_INPUT(data, value, size) { \ if (yyextra->data_.eof()) \ value = YY_NULL; \ @@ -146,18 +194,45 @@ int H(char c) { %option full %option ecs -%option align -Escape \\[\\'"bfnrtv]|\\0|\\x[0-9a-fA-F]{2}|\\u[0-9a-fA-F]{4}|\\\n +U1 [\x00-\x7f] +U0 [\x80-\xbf] +U2 [\xc2-\xdf] +U3 [\xe0-\xef] +U4 [\xf0-\xf4] + +HexDigit [0-9a-fA-F] +LineTerminatorSequence \r?\n|\r|\xe2\x80[\xa8\xa9] +WhiteSpace [\x09\x0b\x0c\x20]|\xc2\xa0|\xef\xbb\xbf +UnicodeEscape \\u({HexDigit}{4}|\{{HexDigit}+\}) + +OctalEscape \\[1-7]|\\[4-7][0-7]|\\[0-3][0-7][0-7]? +StringEscape \\['"\\bfnrtv]|\\0|{OctalEscape}|\\x{HexDigit}{2}|{UnicodeEscape} +StringExtra {StringEscape}|\\{LineTerminatorSequence} +SingleString ([^'\\\n]|{StringExtra})* +DoubleString ([^"\\\n]|{StringExtra})* +StringPrefix '{SingleString}|\"{DoubleString} + +@include UnicodeIDStart.l +@include UnicodeIDContinue.l -IdentifierStart [a-zA-Z$_] -IdentifierPart [a-zA-Z$_0-9] +IdentifierMore [$_] + +UnicodeStart {IdentifierMore}|{UnicodeIDStart} +UnicodePart {IdentifierMore}|\xe2\x80[\x8c\x8d]|{UnicodeIDContinue} +UnicodeFail {U2}|{U3}|{U3}{U0}|{U4}|{U4}{U0}|{U4}{U0}{U0} +UnicodeScrap {UnicodePart}*{UnicodeFail}? + +IdentifierStart {UnicodeStart}|{UnicodeEscape} +IdentifierPart {UnicodePart}|{UnicodeEscape} +IdentifierFail {UnicodeFail}|\\(u({HexDigit}{0,3}|\{{HexDigit}*))? +IdentifierScrap {IdentifierPart}*{IdentifierFail}? NonTerminator [^\n] BackslashSequence \\{NonTerminator} RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence} RegularExpressionChar [^\n\\/]|{BackslashSequence} -RegularExpressionFlags {IdentifierPart}* +RegularExpressionFlags {UnicodePart}* RegularExpressionChars {RegularExpressionChar}* RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars} @@ -177,19 +252,22 @@ XMLName {XMLNameStart}{XMLNamePart}* %% + /* RegEx {{{ */ \/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant); +\/{RegularExpressionBody}\/{RegularExpressionFlags}{UnicodeFail} L E("invalid flags") \/{RegularExpressionBody}?\\? L E("unterminated regex") - + /* }}} */ + /* Comment {{{ */ #![^\n]* L M - \/\/[^\n]* L M - /* http://ostermiller.org/findcomment.html */ - /* XXX: unify these two rules using !? */ + /* http://ostermiller.org/findcomment.html */ + /* XXX: unify these two rules using !? */ \/\*!([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V() C I(comment, Comment(Y), tk::Comment, hi::Comment); \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+\/ V(N) M \/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\** V() E("invalid comment") - + /* }}} */ + /* Element {{{ */ @begin E4X "<>" L F(tk::LeftRight, hi::Structure); "" L F(tk::LeftSlashRight, hi::Structure); @@ -211,7 +289,8 @@ XMLName {XMLNameStart}{XMLNamePart}* "<" L F(tk::Left, hi::Structure); "" L C F(tk::EqualRight, hi::Operator); +"=>" L C F(yylval->newline_ ? tk::EqualRight_ : tk::EqualRight, hi::Operator); "!" L C F(tk::Exclamation, hi::Operator); "!=" L C F(tk::ExclamationEqual, hi::Operator); "!==" L C F(tk::ExclamationEqualEqual, hi::Operator); @@ -280,133 +359,164 @@ XMLName {XMLNameStart}{XMLNamePart}* "[" L C F(tk::OpenBracket, hi::Structure); "]" L C F(tk::CloseBracket, hi::Structure); - + /* }}} */ + /* Keyword {{{ */ "@error" L C F(tk::AtError, hi::Error); @begin Java "@class" L C F(tk::AtClass, hi::Meta); @end -@begin C -"typedef" L C I(identifier, Identifier("typedef"), tk::Typedef, hi::Meta); -"unsigned" L C I(identifier, Identifier("unsigned"), tk::Unsigned, hi::Type); -"signed" L C I(identifier, Identifier("signed"), tk::Signed, hi::Type); -"extern" L C I(identifier, Identifier("extern"), tk::Extern, hi::Type); -@end - @begin C "@encode" L C F(tk::AtEncode, hi::Meta); @end @begin ObjectiveC "@end" L C F(tk::AtEnd, hi::Meta); +"@false" L C F(tk::AtFalse, hi::Constant); "@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta); "@import" L C F(tk::AtImport, hi::Special); +"@NO" L C F(tk::AtNo, hi::Constant); +"@null" L C F(tk::AtNull, hi::Constant); "@selector" L C F(tk::AtSelector, hi::Meta); +"@true" L C F(tk::AtTrue, hi::Constant); +"@YES" L C F(tk::AtYes, hi::Constant); +@end -"NULL" L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant); -"nil" L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant); -"YES" L C I(identifier, Identifier("YES"), tk::Yes, hi::Constant); -"NO" L C I(identifier, Identifier("NO"), tk::No, hi::Constant); +@({UnicodeStart}{UnicodeScrap}|{UnicodeFail}) L E("invalid keyword") + /* }}} */ + /* Highlight {{{ */ +"undefined" L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator); -"bool" L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type); -"BOOL" L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type); -"id" L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type); -"SEL" L C I(identifier, Identifier("SEL"), tk::Identifier_, hi::Type); +@begin ObjectiveC +"bool" L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type); +"BOOL" L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type); +"id" L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type); +"nil" L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant); +"NULL" L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant); +"SEL" L C I(identifier, Identifier("SEL"), tk::Identifier_, hi::Type); @end + /* }}} */ + /* Reserved {{{ */ +"abstract" L C /*FII*/ I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta); +"await" L C /*II?*/ I(identifier, Identifier("await"), tk::Await, hi::Meta); +"boolean" L C /*FII*/ I(identifier, Identifier("boolean"), tk::Boolean, hi::Type); +"break" L R /*KKK*/ F(tk::Break, hi::Control); +"byte" L C /*FII*/ I(identifier, Identifier("byte"), tk::Byte, hi::Type); +"case" L C /*KKK*/ F(tk::Case, hi::Control); +"catch" L C /*KKK*/ F(tk::Catch, hi::Control); +"char" L C /*FII*/ I(identifier, Identifier("char"), tk::Char, hi::Type); +"class" L C /*FFK*/ F(tk::Class, hi::Meta); +"const" L C /*FFK*/ F(tk::Const, hi::Meta); +"continue" L R /*KKK*/ F(tk::Continue, hi::Control); +"debugger" L C /*FKK*/ F(tk::Debugger, hi::Meta); +"default" L C /*KKK*/ F(tk::Default, hi::Control); +"delete" L C /*KKK*/ F(tk::Delete, hi::Operator); +"do" L C /*KKK*/ F(tk::Do, hi::Control); +"double" L C /*FII*/ I(identifier, Identifier("double"), tk::Double, hi::Type); +"else" L C /*KKK*/ F(tk::Else, hi::Control); +"enum" L C /*FFF*/ F(tk::Enum, hi::Meta); +"export" L C /*FFK*/ F(tk::Export, hi::Meta); +"extends" L C /*FFK*/ F(tk::Extends, hi::Meta); +"false" L C /*LLL*/ F(tk::False, hi::Constant); +"final" L C /*FII*/ I(identifier, Identifier("final"), tk::Final, hi::Meta); +"finally" L C /*KKK*/ F(tk::Finally, hi::Control); +"float" L C /*FII*/ I(identifier, Identifier("float"), tk::Float, hi::Type); +"for" L C /*KKK*/ F(tk::For, hi::Control); +"function" L C /*KKK*/ F(yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta); +"goto" L C /*FII*/ I(identifier, Identifier("goto"), tk::Goto, hi::Control); +"if" L C /*KKK*/ F(tk::If, hi::Control); +"implements" L C /*FSS*/ I(identifier, Identifier("implements"), tk::Implements, hi::Meta); +"import" L C /*FFK*/ F(tk::Import, hi::Meta); +"in" L C /*KKK*/ F(yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator); +"instanceof" L C /*KKK*/ F(tk::InstanceOf, hi::Operator); +"int" L C /*FII*/ I(identifier, Identifier("int"), tk::Int, hi::Type); +"interface" L C /*FSS*/ I(identifier, Identifier("interface"), tk::Interface, hi::Meta); +"let" L C /*IS?*/ I(identifier, Identifier("let"), tk::Let, hi::Meta); +"long" L C /*FII*/ I(identifier, Identifier("long"), tk::Long, hi::Type); +"native" L C /*FII*/ I(identifier, Identifier("native"), tk::Native, hi::Meta); +"new" L C /*KKK*/ F(tk::New, hi::Operator); +"null" L C /*LLL*/ F(tk::Null, hi::Constant); +"package" L C /*FSS*/ I(identifier, Identifier("package"), tk::Package, hi::Meta); +"private" L C /*FSS*/ I(identifier, Identifier("private"), tk::Private, hi::Meta); +"protected" L C /*FSS*/ I(identifier, Identifier("protected"), tk::Protected, hi::Meta); +"public" L C /*FSS*/ I(identifier, Identifier("public"), tk::Public, hi::Meta); +"return" L R /*KKK*/ F(tk::Return, hi::Control); +"short" L C /*FII*/ I(identifier, Identifier("short"), tk::Short, hi::Type); +"static" L C /*FS?*/ I(identifier, Identifier("static"), tk::Static, hi::Meta); +"super" L C /*FFK*/ F(tk::Super, hi::Constant); +"switch" L C /*KKK*/ F(tk::Switch, hi::Control); +"synchronized" L C /*FII*/ I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta); +"this" L C /*KKK*/ F(tk::This, hi::Constant); +"throw" L R /*KKK*/ F(tk::Throw, hi::Control); +"throws" L C /*FII*/ I(identifier, Identifier("throws"), tk::Throws, hi::Meta); +"transient" L C /*FII*/ I(identifier, Identifier("transient"), tk::Transient, hi::Meta); +"true" L C /*LLL*/ F(tk::True, hi::Constant); +"try" L C /*KKK*/ F(tk::Try, hi::Control); +"typeof" L C /*KKK*/ F(tk::TypeOf, hi::Operator); +"var" L C /*KKK*/ F(tk::Var, hi::Meta); +"void" L C /*KKK*/ F(tk::Void, hi::Operator); +"volatile" L C /*FII*/ I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta); +"while" L C /*KKK*/ F(tk::While, hi::Control); +"with" L C /*KKK*/ F(tk::With, hi::Control); +"yield" L R /*IS?*/ I(identifier, Identifier("yield"), tk::Yield, hi::Control); + +"auto" L C F(tk::Auto, hi::Meta); +"each" L C I(identifier, Identifier("each"), tk::Each, hi::Control); +"of" L C I(identifier, Identifier("of"), tk::Of, hi::Operator); -"undefined" L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator); - -"false" L C I(false, False(), tk::False, hi::Constant); -"null" L C I(null, Null(), tk::Null, hi::Constant); -"true" L C I(true, True(), tk::True, hi::Constant); - -"auto" L C I(word, Word("auto"), tk::Auto, hi::Meta); -"break" L R I(word, Word("break"), tk::Break, hi::Control); -"case" L C I(word, Word("case"), tk::Case, hi::Control); -"catch" L C I(word, Word("catch"), tk::Catch, hi::Control); -"continue" L R I(word, Word("continue"), tk::Continue, hi::Control); -"default" L C I(word, Word("default"), tk::Default, hi::Control); -"delete" L C I(word, Word("delete"), tk::Delete, hi::Operator); -"do" L C I(word, Word("do"), tk::Do, hi::Control); -"else" L C I(word, Word("else"), tk::Else, hi::Control); -"finally" L C I(word, Word("finally"), tk::Finally, hi::Control); -"for" L C I(word, Word("for"), tk::For, hi::Control); -"function" L C I(word, Word("function"), yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta); -"if" L C I(word, Word("if"), tk::If, hi::Control); -"in" L C I(word, Word("in"), yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator); -"instanceof" L C I(word, Word("instanceof"), tk::InstanceOf, hi::Operator); -"new" L C I(word, Word("new"), tk::New, hi::Operator); -"return" L R I(word, Word("return"), tk::Return, hi::Control); -"switch" L C I(word, Word("switch"), tk::Switch, hi::Control); -"this" L C I(this, This(), tk::This, hi::Constant); -"throw" L R I(word, Word("throw"), tk::Throw, hi::Control); -"try" L C I(word, Word("try"), tk::Try, hi::Control); -"typeof" L C I(word, Word("typeof"), tk::TypeOf, hi::Operator); -"var" L C I(word, Word("var"), tk::Var, hi::Meta); -"void" L C I(word, Word("void"), tk::Void, hi::Operator); -"while" L C I(word, Word("while"), tk::While, hi::Control); -"with" L C I(word, Word("with"), tk::With, hi::Control); - -"debugger" L C I(word, Word("debugger"), tk::Debugger, hi::Meta); - -"const" L C I(word, Word("const"), tk::Const, hi::Meta); - -"class" L C I(word, Word("class"), tk::Class, hi::Meta); -"enum" L C I(word, Word("enum"), tk::Enum, hi::Meta); -"export" L C I(word, Word("export"), tk::Export, hi::Meta); -"extends" L C I(word, Word("extends"), tk::Extends, hi::Meta); -"import" L C I(word, Word("import"), tk::Import, hi::Meta); -"super" L C I(word, Word("super"), tk::Super, hi::Constant); - -"implements" L C I(identifier, Identifier("implements"), tk::Implements, hi::Meta); -"interface" L C I(identifier, Identifier("interface"), tk::Interface, hi::Meta); -"package" L C I(identifier, Identifier("package"), tk::Package, hi::Meta); -"private" L C I(identifier, Identifier("private"), tk::Private, hi::Meta); -"protected" L C I(identifier, Identifier("protected"), tk::Protected, hi::Meta); -"public" L C I(identifier, Identifier("public"), tk::Public, hi::Meta); -"static" L C I(identifier, Identifier("static"), tk::Static, hi::Meta); - -"abstract" L C I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta); -"boolean" L C I(identifier, Identifier("boolean"), tk::Boolean, hi::Type); -"byte" L C I(identifier, Identifier("byte"), tk::Byte, hi::Type); -"char" L C I(identifier, Identifier("char"), tk::Char, hi::Type); -"double" L C I(identifier, Identifier("double"), tk::Double, hi::Type); -"final" L C I(identifier, Identifier("final"), tk::Final, hi::Meta); -"float" L C I(identifier, Identifier("float"), tk::Float, hi::Type); -"goto" L C I(identifier, Identifier("goto"), tk::Goto, hi::Control); -"int" L C I(identifier, Identifier("int"), tk::Int, hi::Type); -"long" L C I(identifier, Identifier("long"), tk::Long, hi::Type); -"native" L C I(identifier, Identifier("native"), tk::Native, hi::Meta); -"short" L C I(identifier, Identifier("short"), tk::Short, hi::Type); -"synchronized" L C I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta); -"throws" L C I(identifier, Identifier("throws"), tk::Throws, hi::Meta); -"transient" L C I(identifier, Identifier("transient"), tk::Transient, hi::Meta); -"volatile" L C I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta); - -"let" L C I(identifier, Identifier("let"), tk::Let, hi::Meta); -"yield" L C I(identifier, Identifier("yield"), tk::Yield, hi::Control); - -"each" L C I(identifier, Identifier("each"), tk::Each, hi::Control); -"of" L C I(identifier, Identifier("of"), tk::Of, hi::Operator); +@begin C +"extern" L C I(identifier, Identifier("extern"), tk::Extern, hi::Type); +"signed" L C I(identifier, Identifier("signed"), tk::Signed, hi::Type); +"typedef" L C I(identifier, Identifier("typedef"), tk::Typedef, hi::Meta); +"unsigned" L C I(identifier, Identifier("unsigned"), tk::Unsigned, hi::Type); +@end + +@begin ObjectiveC +"NO" L C I(identifier, Identifier("NO"), tk::No, hi::Constant); +"YES" L C I(identifier, Identifier("YES"), tk::Yes, hi::Constant); +@end @begin E4X -"namespace" L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta); -"xml" L C I(identifier, Identifier("xml"), tk::XML, hi::Meta); +"namespace" L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta); +"xml" L C I(identifier, Identifier("xml"), tk::XML, hi::Meta); @end + /* }}} */ + /* Identifier {{{ */ +{UnicodeStart}{UnicodePart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier); -{IdentifierStart}{IdentifierPart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier); +{IdentifierStart}{IdentifierPart}* L C { + char *value(A char[yyleng + 1]); + char *local(value); + for (yy_size_t i(0), e(yyleng); i != e; ++i) { + char next(yytext[i]); + if (next != '\\') + *local++ = next; + else + U(local, yytext, ++i); + } -0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant); + *local = '\0'; + I(identifier, Identifier(value), tk::Identifier_, hi::Identifier); +} + +({IdentifierStart}{IdentifierPart}*)?{IdentifierFail} L E("invalid identifier") + /* }}} */ + /* Number {{{ */ 0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant); +0[0-9]+ L C I(number, Number(strtoull(yytext + 1, NULL, 10)), tk::NumericLiteral, hi::Constant); + +0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant); +0[oO][0-7]+ L C I(number, Number(strtoull(yytext + 2, NULL, 8)), tk::NumericLiteral, hi::Constant); 0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant); (\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)([eE][+-]?[0-9]+)? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant); -(\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)[eE][+-]?{IdentifierPart}* L E("invalid exponent") -(\.?[0-9]|(0|[1-9][0-9]*)\.){IdentifierPart}* L E("invalid number") - -\"([^"\\\n]|{Escape})*\"|'([^'\\\n]|{Escape})*' L C { +(\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?)[eE][+-]?{IdentifierScrap} L E("invalid exponent") +(\.?[0-9]|(0|[1-9][0-9]*)\.){IdentifierScrap} L E("invalid number") + /* }}} */ + /* String {{{ */ +'{SingleString}'|\"{DoubleString}\" L C { char *value(A char[yyleng]); char *local(value); @@ -414,8 +524,13 @@ XMLName {XMLNameStart}{XMLNamePart}* char next(yytext[i]); if (yytext[i] == '\\') - switch (next = yytext[++i]) { - case '\n': continue; + // XXX: support more line continuation characters + if (false) line: { + yylloc->end.lines(1); + yylloc->end.columns(yyleng - i); + } else switch (next = yytext[++i]) { + case '\n': goto line; + case '\\': next = '\\'; break; case '\'': next = '\''; break; case '"': next = '"'; break; @@ -425,12 +540,31 @@ XMLName {XMLNameStart}{XMLNamePart}* case 'r': next = '\r'; break; case 't': next = '\t'; break; case 'v': next = '\v'; break; - case '0': next = '\0'; break; + + case '0': case '1': case '2': case '3': + if (yytext[i + 1] < '0' || yytext[i + 1] > '7') + next = H(yytext[i]), i += 0; + else if (yytext[i + 2] < '0' || yytext[i + 2] > '7') + next = H(yytext[i]) << 3 | H(yytext[i + 1]), i += 1; + else + next = H(yytext[i]) << 6 | H(yytext[i + 1]) << 3 | H(yytext[i + 2]), i += 2; + break; + + case '4': case '5': case '6': case '7': + if (yytext[i + 1] < '0' || yytext[i + 1] > '7') + next = H(yytext[i]), i += 0; + else + next = H(yytext[i]) << 3 | H(yytext[i + 1]), i += 1; + break; case 'x': - next = H(yytext[i + 1]) << 4 | H(yytext[i + 2]); + U(local, H(yytext[i + 1]) << 4 | H(yytext[i + 2])); i += 2; - break; + continue; + + case 'u': + U(local, yytext, i); + continue; } *local++ = next; @@ -440,15 +574,16 @@ XMLName {XMLNameStart}{XMLNamePart}* I(string, String(value, local - value), tk::StringLiteral, hi::Constant); } -(\"([^"\\\n]|{Escape})*|'([^'\\\n]|{Escape})*)(\\(x.{0,2}|u.{0,4})?)? L E("invalid escape") - -\r?\n|\r|\xe2\x80[\xa8\xa9] yylloc->step(); yylloc->end.lines(); N +{StringPrefix}\\(x.{0,2}|u([^{].{0,3}|\{[^}]*)?|{UnicodeFail})? L E("invalid escape") +{StringPrefix} L E("invalid string") + /* }}} */ -[ \t] L +{LineTerminatorSequence} yylloc->step(); yylloc->end.lines(); N +{WhiteSpace} L <> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate(); -@{IdentifierPart}+|\xe2.|. L E("unknown token") +. L E("invalid character") %% @@ -520,4 +655,7 @@ void CYDriver::PopCondition() { #if defined(__clang__) #pragma clang diagnostic pop +#else +// must not pop -Wunused-function +//#pragma GCC diagnostic pop #endif