Allow scanner to backtrack (for UTF-8 whitespace).

[cycript.git] / Scanner.lpp.in

diff --git a/Scanner.lpp.in b/Scanner.lpp.in
index e34f67519ee3b4f3c79ac3fd756e3f3762193884..96599f4da781a093d3007cfa108447c6f2b92860 100644 (file)
--- a/Scanner.lpp.in
+++ b/Scanner.lpp.in
@@ -217,6 +217,7 @@ 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 +239,12 @@ 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}


@@ -278,20 +279,22 @@ 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")
 }
     /* }}} */
     <<EOF>> R E("unterminated regex")
 }
     /* }}} */


@@ -305,7 +308,10 @@ 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 +431,7 @@ 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);


@@ -569,16 +575,16 @@ 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 {{{ */


@@ -592,8 +598,8 @@ 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 {{{ */


@@ -628,20 +634,20 @@ 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