X-Git-Url: https://git.saurik.com/cycript.git/blobdiff_plain/c1d3e52e58e86c49f9d04e06ae8e0ece4b98250c..a7d8b4130acfb550e437d8cf1f6d3d81cb0cf6ee:/Cycript.l.in

diff --git a/Cycript.l.in b/Cycript.l.in
index 4878a25..42c45e2 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"
@@ -43,33 +50,32 @@ typedef cy::parser::token tk;
 @begin ObjectiveC
     yyextra->no_.AtImplementation = false; \
 @end
+    yyextra->no_.Class = false; \
     yyextra->no_.Function = false; \
+    yyextra->no_.NewLine = false; \
     yyextra->no_.OpenBrace = false; \
     yylval->highlight_ = highlight; \
     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; \
+    yylval->semantic_.type ## _ = A CY ## Type; \
     F(value, highlight); \
 } while (false)
 
-#define T yylval->newline_ = yyextra->state_ == CYNewLine; BEGIN(Div);
-#define C T yyextra->state_ = CYClear;
-#define R T yyextra->state_ = CYRestricted;
+#define C \
+    yyextra->newline_ = yyextra->last_; \
+    yyextra->last_ = false; \
+    BEGIN(Div);
 
 #define N \
-    if (yyextra->state_ != CYNewLine) { \
-        if (yyextra->state_ != CYRestricted) \
-            yyextra->state_ = CYNewLine; \
-        else { \
-            yyextra->state_ = CYClear; \
-            F(tk::NewLine, hi::Nothing); \
-        } \
-    }
+    yyextra->last_ = true; \
+    if (yyextra->no_.NewLine) \
+        F(tk::NewLine, hi::Nothing);
 
 #define V(more) { \
     if (const char *nl = reinterpret_cast<const char *>(memchr(yytext, '\n', yyleng))) { \
@@ -98,6 +104,14 @@ typedef cy::parser::token tk;
     } \
 }
 
+#define E(message) { \
+    CYDriver::Error error; \
+    error.location_ = *yylloc; \
+    error.message_ = "syntax error, " message; \
+    yyextra->errors_.push_back(error); \
+    yyterminate(); \
+}
+
 int H(char c) {
     if (c >= '0' && c <= '9')
         return c - '0';
@@ -108,6 +122,46 @@ 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; \
@@ -123,25 +177,60 @@ int H(char c) {
 %option prefix="cy"
 %option bison-bridge
 %option bison-locations
+%option nodefault
 %option noyywrap
-%option yylineno
+%option noyylineno
 %option nounput
+%option nounistd
+%option 8bit
+%option backup
 %option batch
 %option never-interactive
+%option pointer
 %option reentrant
 %option stack
 
-Exponent [eE][+-]?[0-9]+
-Escape   \\[\\'"bfnrtv]|\\0|\\x[0-9a-fA-F]{2}|\\u[0-9a-fA-F]{4}|\\\n
+%option full
+%option ecs
+
+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
+
+IdentifierMore [$_]
 
-IdentifierStart [a-zA-Z$_]
-IdentifierPart  [a-zA-Z$_0-9]
+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}
 
@@ -161,17 +250,22 @@ XMLName {XMLNameStart}{XMLNamePart}*
 
 %%
 
+    /* RegEx {{{ */
 <RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);
-
+<RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags}{UnicodeFail} L E("invalid flags")
+<RegExp>\/{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
 <RegExp>"<>"      L F(tk::LeftRight, hi::Structure);
 <XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
@@ -193,8 +287,10 @@ XMLName {XMLNameStart}{XMLNamePart}*
 <XMLContent>"<"  L F(tk::Left, hi::Structure);
 <XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
 @end
-
+    /* }}} */
+    /* Operator {{{ */
 "..."  L C F(tk::PeriodPeriodPeriod, hi::Meta);
+".."   L E("invalid operator")
 
 @begin E4X
 "::"   L C F(tk::ColonColon, hi::Operator);
@@ -214,13 +310,13 @@ XMLName {XMLNameStart}{XMLNamePart}*
 "="    L C F(tk::Equal, hi::Operator);
 "=="   L C F(tk::EqualEqual, hi::Operator);
 "==="  L C F(tk::EqualEqualEqual, hi::Operator);
-"=>"   L C F(tk::EqualRight, hi::Operator);
+"=>"   L C F(yyextra->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);
 "-"    L C F(tk::Hyphen, hi::Operator);
 "-="   L C F(tk::HyphenEqual, hi::Operator);
-"--"   L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
+"--"   L C F(yyextra->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
 "->"   L C F(tk::HyphenRight, hi::Operator);
 "<"    L C F(tk::Left, hi::Operator);
 "<="   L C F(tk::LeftEqual, hi::Operator);
@@ -234,7 +330,7 @@ XMLName {XMLNameStart}{XMLNamePart}*
 "||"   L C F(tk::PipePipe, hi::Operator);
 "+"    L C F(tk::Plus, hi::Operator);
 "+="   L C F(tk::PlusEqual, hi::Operator);
-"++"   L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
+"++"   L C F(yyextra->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
 ">"    L C F(tk::Right, hi::Operator);
 ">="   L C F(tk::RightEqual, hi::Operator);
 ">>"   L C F(tk::RightRight, hi::Operator);
@@ -256,135 +352,169 @@ XMLName {XMLNameStart}{XMLNamePart}*
 "("    L C F(tk::OpenParen, hi::Structure);
 ")"    L C F(tk::CloseParen, hi::Structure);
 
-"{"    L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
+"{"    L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yyextra->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
 "}"    L C F(tk::CloseBrace, hi::Structure);
 
 "["    L C F(tk::OpenBracket, hi::Structure);
 "]"    L C F(tk::CloseBracket, hi::Structure);
-
-"@error"          L C F(tk::AtError, hi::Error);
+    /* }}} */
+    /* Keyword {{{ */
+"@error"          L C F(tk::At_error_, hi::Error);
 
 @begin Java
-"@class"          L C F(tk::AtClass, hi::Meta);
+"@class"          L C F(tk::At_class_, 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);
+"@encode"         L C F(tk::At_encode_, hi::Meta);
 @end
 
-@begin C
-"@encode"         L C F(tk::AtEncode, hi::Meta);
+@begin ObjectiveC
+"@end"            L C F(tk::At_end_, hi::Meta);
+"@false"          L C F(tk::At_false_, hi::Constant);
+"@implementation" L C F(yyextra->no_.AtImplementation ? tk::At_implementation__ : tk::At_implementation_, hi::Meta);
+"@import"         L C F(tk::At_import_, hi::Special);
+"@NO"             L C F(tk::At_NO_, hi::Constant);
+"@null"           L C F(tk::At_null_, hi::Constant);
+"@selector"       L C F(tk::At_selector_, hi::Meta);
+"@true"           L C F(tk::At_true_, hi::Constant);
+"@YES"            L C F(tk::At_YES_, hi::Constant);
 @end
 
+@({UnicodeStart}{UnicodeScrap}|{UnicodeFail}) L E("invalid keyword")
+    /* }}} */
+    /* Highlight {{{ */
+"undefined"       L C F(tk::_undefined_, hi::Operator);
+
 @begin ObjectiveC
-"@end"            L C F(tk::AtEnd, hi::Meta);
-"@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta);
-"@import"         L C F(tk::AtImport, hi::Special);
-"@selector"       L C F(tk::AtSelector, hi::Meta);
-
-"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);
-
-"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);
+"bool"            L C F(tk::_bool_, hi::Type);
+"BOOL"            L C F(tk::_BOOL_, hi::Type);
+"id"              L C F(tk::_id_, hi::Type);
+"nil"             L C F(tk::_nil_, hi::Constant);
+"NULL"            L C F(tk::_NULL_, hi::Constant);
+"SEL"             L C F(tk::_SEL_, hi::Type);
 @end
+    /* }}} */
+    /* Reserved {{{ */
+"abstract"        L C /*FII*/ F(tk::_abstract_, hi::Meta);
+"await"           L C /*II?*/ F(tk::_await_, hi::Meta);
+"boolean"         L C /*FII*/ F(tk::_boolean_, hi::Type);
+"break"           L C /*KKK*/ F(tk::_break_, hi::Control);
+"byte"            L C /*FII*/ F(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*/ F(tk::_char_, hi::Type);
+"class"           L C /*FFK*/ F(yyextra->no_.Class ? tk::_class__ : tk::_class_, hi::Meta);
+"const"           L C /*FFK*/ F(tk::_const_, hi::Meta);
+"continue"        L C /*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*/ F(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*/ F(tk::_final_, hi::Meta);
+"finally"         L C /*KKK*/ F(tk::_finally_, hi::Control);
+"float"           L C /*FII*/ F(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*/ F(tk::_goto_, hi::Control);
+"if"              L C /*KKK*/ F(tk::_if_, hi::Control);
+"implements"      L C /*FSS*/ F(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*/ F(tk::_int_, hi::Type);
+"interface"       L C /*FSS*/ F(tk::_interface_, hi::Meta);
+"let"             L C /*IS?*/ F(tk::_let_, hi::Meta);
+"long"            L C /*FII*/ F(tk::_long_, hi::Type);
+"native"          L C /*FII*/ F(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*/ F(tk::_package_, hi::Meta);
+"private"         L C /*FSS*/ F(tk::_private_, hi::Meta);
+"protected"       L C /*FSS*/ F(tk::_protected_, hi::Meta);
+"public"          L C /*FSS*/ F(tk::_public_, hi::Meta);
+"return"          L C /*KKK*/ F(tk::_return_, hi::Control);
+"short"           L C /*FII*/ F(tk::_short_, hi::Type);
+"static"          L C /*FS?*/ F(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*/ F(tk::_synchronized_, hi::Meta);
+"this"            L C /*KKK*/ F(tk::_this_, hi::Constant);
+"throw"           L C /*KKK*/ F(tk::_throw_, hi::Control);
+"throws"          L C /*FII*/ F(tk::_throws_, hi::Meta);
+"transient"       L C /*FII*/ F(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*/ F(tk::_volatile_, hi::Meta);
+"while"           L C /*KKK*/ F(tk::_while_, hi::Control);
+"with"            L C /*KKK*/ F(tk::_with_, hi::Control);
+"yield"           L C /*IS?*/ F(tk::_yield_, hi::Control);
+
+"auto"            L C F(tk::_auto_, hi::Meta);
+"each"            L C F(tk::_each_, hi::Control);
+"of"              L C F(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 F(tk::_extern_, hi::Type);
+"signed"          L C F(tk::_signed_, hi::Type);
+"typedef"         L C F(tk::_typedef_, hi::Meta);
+"unsigned"        L C F(tk::_unsigned_, hi::Type);
+@end
+
+@begin ObjectiveC
+"NO"              L C F(tk::_NO_, hi::Constant);
+"YES"             L C F(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 F(tk::_namespace_, hi::Meta);
+"xml"             L C F(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);
 
-(\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?){Exponent}? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
+    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);
 
-\"([^"\\\n]|{Escape})*\"|'([^'\\\n]|{Escape})*' L C {
+(\.[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][+-]?{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);
 
@@ -392,8 +522,13 @@ 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;
@@ -403,12 +538,31 @@ 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;
@@ -418,19 +572,16 @@ XMLName {XMLNameStart}{XMLNamePart}*
     I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
 }
 
-\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
 
 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
 
-. L {
-    CYDriver::Error error;
-    error.location_ = *yylloc;
-    error.message_ = "syntax error, unknown token";
-    yyextra->errors_.push_back(error);
-    yyterminate();
-}
+. L E("invalid character")
 
 %%
 
@@ -502,4 +653,7 @@ void CYDriver::PopCondition() {
 
 #if defined(__clang__)
 #pragma clang diagnostic pop
+#else
+// must not pop -Wunused-function
+//#pragma GCC diagnostic pop
 #endif