#include "Highlight.hpp"
+#include "IdentifierStart.h"
+#include "IdentifierContinue.h"
+
#define YY_EXTRA_TYPE CYDriver *
#define F(value, highlight) do { \
} 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]);
}
U(local, point);
+ return point;
}
#define CYLexBufferPoint(point) do { \
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]
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}
-IdentifierFail {UnicodeFail}|\\(u({HexDigit}{0,3}|\{{HexDigit}*))?
+IdentifierFail {UnicodeError}|\\(u({HexDigit}{0,3}|\{{HexDigit}*))?
IdentifierScrap {IdentifierPart}*{IdentifierFail}?
RegularExpressionBackslashSequence \\{NoneTerminatorCharacter}
/* 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}*{UnicodeFail} R E("invalid character");
{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}\\?{UnicodeFail} R E("invalid character");
+
"["{RegularExpressionClassChars}\\?{LineTerminatorSequence} R E("invalid newline");
+ "["{RegularExpressionClassChars}\\?{UnicodeScrap} R E("invalid character");
- (\\|{RegExCharacter}+)?{LineTerminatorSequence} R E("invalid newline");
<<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);
\**\*\/ 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")
}
/* }}} */
"@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);
"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);
"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);
+"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);
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';
<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);
- {DoubleCharacter}*{UnicodeFail} R E("invalid character");
- {LineTerminatorSequence} R E("invalid newline");
+ {DoubleCharacter}*{LineTerminatorSequence} R E("invalid newline");
+ {DoubleCharacter}*{UnicodeScrap} R E("invalid character");
}
/* }}} */
/* Template {{{ */
"$" R CYLexBufferUnit('$');
{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 {{{ */
}
\\{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
+{U1}|{UnicodeScrap} L E("invalid character");
<<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
-. L E("invalid character")
-
%%
#undef yyextra