The arguments variable needs to be declared Magic.

[cycript.git] / Scanner.lpp.in

diff --git a/Scanner.lpp.in b/Scanner.lpp.in
index 72011986e4b26010b69b1e62f74cba2b6d4ce192..1fb16aa5362c3a521b8662ec9efc3b1c08b584a0 100644 (file)
--- a/Scanner.lpp.in
+++ b/Scanner.lpp.in
@@ -41,6 +41,9 @@ typedef cy::parser::token tk;
 
 #include "Highlight.hpp"
 
 
 #include "Highlight.hpp"
 

+#include "IdentifierStart.h"
+#include "IdentifierContinue.h"
+

 #define YY_EXTRA_TYPE CYDriver *
 
 #define F(value, highlight) do { \
 #define YY_EXTRA_TYPE CYDriver *
 
 #define F(value, highlight) do { \


@@ -140,7 +143,7 @@ static void U(Type_ &local, unsigned point) {
     } else _assert(false);
 }
 
     } else _assert(false);
 }
 

-static void U(char *&local, const char *text, yy_size_t &i) {
+static unsigned U(char *&local, const char *text, yy_size_t &i) {

     unsigned point;
 
     char next(text[++i]);
     unsigned point;
 
     char next(text[++i]);


@@ -158,6 +161,7 @@ static void U(char *&local, const char *text, yy_size_t &i) {
     }
 
     U(local, point);
     }
 
     U(local, point);

+    return point;

 }
 
 #define CYLexBufferPoint(point) do { \
 }
 
 #define CYLexBufferPoint(point) do { \


@@ -217,6 +221,7 @@ U2 [\xc2-\xdf]
 U3 [\xe0-\xef]
 U4 [\xf0-\xf4]
 UN [\xc0-\xc1\xf5-\xff]
 U3 [\xe0-\xef]
 U4 [\xf0-\xf4]
 UN [\xc0-\xc1\xf5-\xff]

+UE {U1}|{U2}|{U3}|{U4}|{UN}

 
 HexDigit [0-9a-fA-F]
 LineTerminatorSequence \r?\n|\r|\xe2\x80[\xa8\xa9]
 
 HexDigit [0-9a-fA-F]
 LineTerminatorSequence \r?\n|\r|\xe2\x80[\xa8\xa9]


@@ -238,12 +243,12 @@ IdentifierMore [$_]
 
 UnicodeStart {IdentifierMore}|{UnicodeIDStart}
 UnicodePart {IdentifierMore}|\xe2\x80[\x8c\x8d]|{UnicodeIDContinue}
 
 UnicodeStart {IdentifierMore}|{UnicodeIDStart}
 UnicodePart {IdentifierMore}|\xe2\x80[\x8c\x8d]|{UnicodeIDContinue}

-UnicodeFail {U2}|{U3}|{U3}{U0}|{U4}|{U4}{U0}|{U4}{U0}{U0}|{UN}|{U0}
-UnicodeScrap {UnicodePart}*{UnicodeFail}?
+UnicodeScrap {U2}|{U3}{U0}{0,1}|{U4}{U0}{0,2}|{UN}|{U0}
+UnicodeError ({U2}|{U3}{U0}{0,1}|{U4}{U0}{0,2}){UE}|{UN}|{U0}

 
 IdentifierStart {UnicodeStart}|{UnicodeEscape}
 IdentifierPart {UnicodePart}|{UnicodeEscape}
 
 IdentifierStart {UnicodeStart}|{UnicodeEscape}
 IdentifierPart {UnicodePart}|{UnicodeEscape}

-IdentifierFail {UnicodeFail}|\\(u({HexDigit}{0,3}|\{{HexDigit}*))?
+IdentifierFail {UnicodeError}|\\(u({HexDigit}{0,3}|\{{HexDigit}*))?

 IdentifierScrap {IdentifierPart}*{IdentifierFail}?
 
 RegularExpressionBackslashSequence \\{NoneTerminatorCharacter}
 IdentifierScrap {IdentifierPart}*{IdentifierFail}?
 
 RegularExpressionBackslashSequence \\{NoneTerminatorCharacter}


@@ -278,26 +283,30 @@ XMLName {XMLNameStart}{XMLNamePart}*
     /* RegEx {{{ */
 <RegularExpression>{
     \/{UnicodePart}* R CYLexBufferUnits(yytext, yyleng); CYLexBufferEnd(literal, RegEx, tk::RegularExpressionLiteral, hi::Constant);
     /* RegEx {{{ */
 <RegularExpression>{
     \/{UnicodePart}* R CYLexBufferUnits(yytext, yyleng); CYLexBufferEnd(literal, RegEx, tk::RegularExpressionLiteral, hi::Constant);

-    \/{UnicodePart}*{UnicodeFail} R E("invalid flags")
+    \/{UnicodePart}*{UnicodeError} R E("invalid character");

 
     {RegExCharacter}+ R CYLexBufferUnits(yytext, yyleng);
 
     {RegExCharacter}+ R CYLexBufferUnits(yytext, yyleng);

-    {RegExCharacter}*{UnicodeFail} R E("invalid character");

 
     {RegularExpressionBackslashSequence} R CYLexBufferUnits(yytext, yyleng);
 
     {RegularExpressionBackslashSequence} R CYLexBufferUnits(yytext, yyleng);

-    \\{UnicodeFail}? R E("invalid escape")
+    \\ R E("invalid escape")
+
+    (\\|{RegExCharacter}+)?{LineTerminatorSequence} R E("invalid newline");
+    (\\|{RegExCharacter}+)?{UnicodeScrap} R E("invalid character");

 
     "["{RegularExpressionClassChars}"]" R CYLexBufferUnits(yytext, yyleng);
     "["{RegularExpressionClassChars}\\? R E("invalid class");
 
     "["{RegularExpressionClassChars}"]" R CYLexBufferUnits(yytext, yyleng);
     "["{RegularExpressionClassChars}\\? R E("invalid class");

-    "["{RegularExpressionClassChars}\\?{UnicodeFail} R E("invalid character");
+

     "["{RegularExpressionClassChars}\\?{LineTerminatorSequence} R E("invalid newline");
     "["{RegularExpressionClassChars}\\?{LineTerminatorSequence} R E("invalid newline");

+    "["{RegularExpressionClassChars}\\?{UnicodeScrap} R E("invalid character");

 
 

-    (\\|{RegExCharacter}+)?{LineTerminatorSequence} R E("invalid newline");

     <<EOF>> R E("unterminated regex")
 }
     /* }}} */
     /* Comment {{{ */
     <<EOF>> R E("unterminated regex")
 }
     /* }}} */
     /* Comment {{{ */

-#![^\n]* L M
-\/\/[^\n]* L M
+    /* XXX: maybe fold LineTerminatorSequence into these definitions */
+#!{NoneTerminatorCharacter}* L M
+\/\/{NoneTerminatorCharacter}* L M
+(#!|\/\/){NoneTerminatorCharacter}*{UnicodeError} L E("invalid character");

 
 \/\* L yy_push_state(MultiLine, yyscanner);
 
 
 \/\* L yy_push_state(MultiLine, yyscanner);
 


@@ -305,7 +314,10 @@ XMLName {XMLNameStart}{XMLNamePart}*
     \**\*\/ R yy_pop_state(yyscanner); M N
     \**{LineTerminatorSequence} yylloc->end.Lines(); yyextra->last_ = true;
     \**{CommentCharacter}|\/ R
     \**\*\/ R yy_pop_state(yyscanner); M N
     \**{LineTerminatorSequence} yylloc->end.Lines(); yyextra->last_ = true;
     \**{CommentCharacter}|\/ R

-    \**({UnicodeFail}|\*) R E("invalid comment");
+
+    \**{UnicodeScrap} R E("invalid character");
+    \**\* R E("invalid comment");
+

     <<EOF>> R E("invalid comment")
 }
     /* }}} */
     <<EOF>> R E("invalid comment")
 }
     /* }}} */


@@ -425,7 +437,7 @@ XMLName {XMLNameStart}{XMLNamePart}*
 "@YES"            L F(tk::At_YES_, hi::Constant);
 @end
 
 "@YES"            L F(tk::At_YES_, hi::Constant);
 @end
 

-@({UnicodeStart}{UnicodeScrap}|{UnicodeFail}) L E("invalid keyword")
+@({UnicodeStart}{UnicodePart}*{UnicodeError}?|{UnicodeError}) L E("invalid keyword")

     /* }}} */
     /* Highlight {{{ */
 "undefined"       L F(tk::_undefined_, hi::Operator);
     /* }}} */
     /* Highlight {{{ */
 "undefined"       L F(tk::_undefined_, hi::Operator);


@@ -475,6 +487,7 @@ XMLName {XMLNameStart}{XMLNamePart}*
 "implements"      L /*FSS*/ F(tk::_implements_, hi::Meta);
 "import"          L /*FFK*/ F(tk::_import_, hi::Meta);
 "in"              L /*KKK*/ F(yyextra->in_.top() ? tk::_in__ : tk::_in_, hi::Operator);
 "implements"      L /*FSS*/ F(tk::_implements_, hi::Meta);
 "import"          L /*FFK*/ F(tk::_import_, hi::Meta);
 "in"              L /*KKK*/ F(yyextra->in_.top() ? tk::_in__ : tk::_in_, hi::Operator);

+"Infinity"        L /*III*/ F(tk::_Infinity_, hi::Constant);

 "instanceof"      L /*KKK*/ F(tk::_instanceof_, hi::Operator);
 "int"             L /*FII*/ F(tk::_int_, hi::Type);
 "interface"       L /*FSS*/ F(tk::_interface_, hi::Meta);
 "instanceof"      L /*KKK*/ F(tk::_instanceof_, hi::Operator);
 "int"             L /*FII*/ F(tk::_int_, hi::Type);
 "interface"       L /*FSS*/ F(tk::_interface_, hi::Meta);


@@ -492,9 +505,10 @@ XMLName {XMLNameStart}{XMLNamePart}*
 "set"             L /*III*/ F(tk::_set_, hi::Meta);
 "short"           L /*FII*/ F(tk::_short_, hi::Type);
 "static"          L /*FS?*/ F(tk::_static_, hi::Meta);
 "set"             L /*III*/ F(tk::_set_, hi::Meta);
 "short"           L /*FII*/ F(tk::_short_, hi::Type);
 "static"          L /*FS?*/ F(tk::_static_, hi::Meta);

-"super"           L /*FFK*/ F(yyextra->super_.top() ? tk::_super__ : tk::_super_, hi::Constant);
+"super"           L /*FFK*/ F(tk::_super_, hi::Constant);

 "switch"          L /*KKK*/ F(tk::_switch_, hi::Control);
 "synchronized"    L /*FII*/ F(tk::_synchronized_, hi::Meta);
 "switch"          L /*KKK*/ F(tk::_switch_, hi::Control);
 "synchronized"    L /*FII*/ F(tk::_synchronized_, hi::Meta);

+"target"          L /*III*/ F(tk::_target_, hi::Identifier);

 "this"            L /*KKK*/ F(tk::_this_, hi::Constant);
 "throw"           L /*KKK*/ F(tk::_throw_, hi::Control);
 "throws"          L /*FII*/ F(tk::_throws_, hi::Meta);
 "this"            L /*KKK*/ F(tk::_this_, hi::Constant);
 "throw"           L /*KKK*/ F(tk::_throw_, hi::Control);
 "throws"          L /*FII*/ F(tk::_throws_, hi::Meta);


@@ -541,8 +555,12 @@ XMLName {XMLNameStart}{XMLNamePart}*
         char next(yytext[i]);
         if (next != '\\')
             *local++ = next;
         char next(yytext[i]);
         if (next != '\\')
             *local++ = next;

-        else
-            U(local, yytext, ++i);
+        else {
+            bool (*is)(unsigned) = (i == 0 ? &IsIdentifierStart : &IsIdentifierContinue);
+            unsigned point(U(local, yytext, ++i));
+            if (!is(point))
+                E("invalid character");
+        }

     }
 
     *local = '\0';
     }
 
     *local = '\0';


@@ -568,16 +586,16 @@ XMLName {XMLNameStart}{XMLNamePart}*
 <LegacySingleString,StrictSingleString>{
     \' R CYLexBufferEnd(string, String, tk::StringLiteral, hi::Constant);
     {SingleCharacter}+ R CYLexBufferUnits(yytext, yyleng);
 <LegacySingleString,StrictSingleString>{
     \' R CYLexBufferEnd(string, String, tk::StringLiteral, hi::Constant);
     {SingleCharacter}+ R CYLexBufferUnits(yytext, yyleng);

-    {SingleCharacter}*{UnicodeFail} R E("invalid character");
-    {LineTerminatorSequence} R E("invalid newline");
+    {SingleCharacter}*{LineTerminatorSequence} R E("invalid newline");
+    {SingleCharacter}*{UnicodeScrap} R E("invalid character");

 }
 
 \" L CYLexBufferStart(LegacyDoubleString);
 <LegacyDoubleString,StrictDoubleString>{
     \" R CYLexBufferEnd(string, String, tk::StringLiteral, hi::Constant);
     {DoubleCharacter}+ R CYLexBufferUnits(yytext, yyleng);
 }
 
 \" L CYLexBufferStart(LegacyDoubleString);
 <LegacyDoubleString,StrictDoubleString>{
     \" R CYLexBufferEnd(string, String, tk::StringLiteral, hi::Constant);
     {DoubleCharacter}+ R CYLexBufferUnits(yytext, yyleng);

-    {DoubleCharacter}*{UnicodeFail} R E("invalid character");
-    {LineTerminatorSequence} R E("invalid newline");
+    {DoubleCharacter}*{LineTerminatorSequence} R E("invalid newline");
+    {DoubleCharacter}*{UnicodeScrap} R E("invalid character");

 }
     /* }}} */
     /* Template {{{ */
 }
     /* }}} */
     /* Template {{{ */


@@ -591,8 +609,8 @@ XMLName {XMLNameStart}{XMLNamePart}*
     "$" R CYLexBufferUnit('$');
 
     {PlateCharacter}+ R CYLexBufferUnits(yytext, yyleng);
     "$" R CYLexBufferUnit('$');
 
     {PlateCharacter}+ R CYLexBufferUnits(yytext, yyleng);

-    {PlateCharacter}*{UnicodeFail} R E("invalid character");
-    {LineTerminatorSequence} R E("invalid newline");
+    {PlateCharacter}*{LineTerminatorSequence} R E("invalid newline");
+    {PlateCharacter}*{UnicodeScrap} R E("invalid character");

 }
     /* }}} */
     /* Escapes {{{ */
 }
     /* }}} */
     /* Escapes {{{ */


@@ -627,20 +645,20 @@ XMLName {XMLNameStart}{XMLNamePart}*
     }
 
     \\{LineTerminatorSequence} yylloc->end.Lines();
     }
 
     \\{LineTerminatorSequence} yylloc->end.Lines();

-    \\(.|{NotLineTerminator}) R CYLexBufferUnits(yytext + 1, yyleng - 1);
+    \\{NoneTerminatorCharacter} R CYLexBufferUnits(yytext + 1, yyleng - 1);
+    \\{UnicodeScrap} R E("invalid character");

 
 

-    \\(x{HexDigit}{0,1}|u({HexDigit}{0,3}|\{{HexDigit}*)|{UnicodeFail})? R E("invalid escape");
+    \\(x{HexDigit}{0,1}|u({HexDigit}{0,3}|\{{HexDigit}*))? R E("invalid escape");

     <<EOF>> R E("invalid string");
 }
     /* }}} */
 
 {LineTerminatorSequence} yylloc->step(); yylloc->end.Lines(); yyextra->last_ = true; N
 {WhiteSpace} L
     <<EOF>> R E("invalid string");
 }
     /* }}} */
 
 {LineTerminatorSequence} yylloc->step(); yylloc->end.Lines(); yyextra->last_ = true; N
 {WhiteSpace} L

+{U1}|{UnicodeScrap} L E("invalid character");

 
 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
 
 
 <<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
 

-. L E("invalid character")
-

 %%
 
 #undef yyextra
 %%
 
 #undef yyextra