[cycript.git] / Cycript.l.in

/* Cycript - Optimizing JavaScript Compiler/Runtime
 * Copyright (C) 2009-2013  Jay Freeman (saurik)
*/

/* GNU General Public License, Version 3 {{{ */
/*
 * Cycript is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published
 * by the Free Software Foundation, either version 3 of the License,
 * or (at your option) any later version.
 *
 * Cycript is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with Cycript.  If not, see <http://www.gnu.org/licenses/>.
**/
/* }}} */

/* 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 */

%{
#define YYLTYPE cy::location
#include "Cycript.tab.hh"
typedef cy::parser::token tk;

#include "Highlight.hpp"

#define YY_EXTRA_TYPE CYDriver *

// do /not/ fold token to the return: this is a macro and the ordering is dependent
#define F(value, highlight) do { \
    int token(value); \
@begin ObjectiveC
    yyextra->no_.AtImplementation = false; \
@end
    yyextra->no_.Function = false; \
    yyextra->no_.OpenBrace = false; \
    yylval->highlight_ = highlight; \
    return token; \
} while (false)

#define A new($pool)
#define Y $pool.strmemdup(yytext, yyleng)

#define I(type, Type, value, highlight) do { \
    yylval->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 N \
    if (yyextra->state_ != CYNewLine) { \
        if (yyextra->state_ != CYRestricted) \
            yyextra->state_ = CYNewLine; \
        else { \
            yyextra->state_ = CYClear; \
            F(tk::NewLine, hi::Nothing); \
        } \
    }

#define V(more) { \
    if (const char *nl = reinterpret_cast<const char *>(memchr(yytext, '\n', yyleng))) { \
        unsigned lines(0); \
        size_t left; \
        do { \
            ++lines; \
            left = yyleng - (nl - yytext) - 1; \
            nl = reinterpret_cast<const char *>(memchr(nl + 1, '\n', left)); \
        } while (nl != NULL); \
        yylloc->end.lines(lines); \
        yylloc->end.columns(left); \
        yylloc->step(); \
        more \
    } else L \
}

#define L { \
    yylloc->step(); \
    yylloc->columns(yyleng); \
}

#define M { \
    if (yyextra->commented_) { \
        I(comment, Comment(Y), tk::Comment, hi::Comment); \
    } \
}

int H(char c) {
    if (c >= '0' && c <= '9')
        return c - '0';
    if (c >= 'a' && c <= 'f')
        return c - 'a' + 10;
    if (c >= 'A' && c <= 'F')
        return c - 'A' + 10;
    return -1;
}

#define YY_INPUT(data, value, size) { \
    if (yyextra->data_.eof()) \
        value = YY_NULL; \
    else { \
        yyextra->data_.read(data, size); \
        size_t copy(yyextra->data_.gcount()); \
        value = copy == 0 ? YY_NULL : copy; \
    } \
}

%}

%option prefix="cy"
%option bison-bridge
%option bison-locations
%option noyywrap
%option yylineno
%option nounput
%option batch
%option never-interactive
%option reentrant
%option stack

Exponent [eE][+-]?[0-9]+
Escape   \\[\\'"bfnrtv]|\\0|\\x[0-9a-fA-F]{2}|\\u[0-9a-fA-F]{4}|\\\n

IdentifierStart [a-zA-Z$_]
IdentifierPart  [a-zA-Z$_0-9]

NonTerminator [^\n]
BackslashSequence \\{NonTerminator}
RegularExpressionFirstChar [^\n*\\/]|{BackslashSequence}
RegularExpressionChar [^\n\\/]|{BackslashSequence}
RegularExpressionFlags {IdentifierPart}*
RegularExpressionChars {RegularExpressionChar}*
RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}

@begin E4X
XMLNameStart [a-zA-Z_:]
XMLNamePart [a-zA-Z0-9.-_:]
XMLName {XMLNameStart}{XMLNamePart}*
@end

%s Div
%s RegExp

@begin E4X
%x XMLContent
%x XMLTag
@end

%%

<RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);

#![^\n]* L M

\/\/[^\n]* L M

	/* 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

@begin E4X
<RegExp>"<>"      L F(tk::LeftRight, hi::Structure);
<XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);

<RegExp,XMLContent>\<!\[CDATA\[(\n|[^[]|\[[^[]|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
<RegExp,XMLContent>\<!--(\n|[^-]|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
<RegExp,XMLContent>\<?(\n|[^?]|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);

<XMLTag>"="  L F(tk::Equal, hi::Structure);
<XMLTag>">"  L F(tk::Right, hi::Structure);
<XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
<XMLTag>"{"  L F(tk::OpenBrace, hi::Structure);

<XMLTag>\"(\n|[^"])*\"|'(\n|[^'])*' V() F(tk::XMLAttributeValue, hi::Constant);
<XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
<XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);

<XMLContent>"{"  L F(tk::OpenBrace, hi::Structure);
<XMLContent>"<"  L F(tk::Left, hi::Structure);
<XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
@end

"..."  L C F(tk::PeriodPeriodPeriod, hi::Meta);

@begin E4X
"::"   L C F(tk::ColonColon, hi::Operator);
".."   L C F(tk::PeriodPeriod, hi::Operator);
@end

@begin E4X ObjectiveC
"@"    L C F(tk::At, hi::Operator);
"#"    L C F(tk::Pound, hi::Operator);
@end

"&"    L C F(tk::Ampersand, hi::Operator);
"&&"   L C F(tk::AmpersandAmpersand, hi::Operator);
"&="   L C F(tk::AmpersandEqual, hi::Operator);
"^"    L C F(tk::Carrot, hi::Operator);
"^="   L C F(tk::CarrotEqual, hi::Operator);
"="    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(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(tk::HyphenRight, hi::Operator);
"<"    L C F(tk::Left, hi::Operator);
"<="   L C F(tk::LeftEqual, hi::Operator);
"<<"   L C F(tk::LeftLeft, hi::Operator);
"<<="  L C F(tk::LeftLeftEqual, hi::Operator);
"%"    L C F(tk::Percent, hi::Operator);
"%="   L C F(tk::PercentEqual, hi::Operator);
"."    L C F(tk::Period, hi::Operator);
"|"    L C F(tk::Pipe, hi::Operator);
"|="   L C F(tk::PipeEqual, hi::Operator);
"||"   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(tk::Right, hi::Operator);
">="   L C F(tk::RightEqual, hi::Operator);
">>"   L C F(tk::RightRight, hi::Operator);
">>="  L C F(tk::RightRightEqual, hi::Operator);
">>>"  L C F(tk::RightRightRight, hi::Operator);
">>>=" L C F(tk::RightRightRightEqual, hi::Operator);
"*"    L C F(tk::Star, hi::Operator);
"*="   L C F(tk::StarEqual, hi::Operator);
"~"    L C F(tk::Tilde, hi::Operator);

<Div>"/"  L C F(tk::Slash, hi::Operator);
<Div>"/=" L C F(tk::SlashEqual, hi::Operator);

":"    L C F(tk::Colon, hi::Structure);
","    L C F(tk::Comma, hi::Structure);
"?"    L C F(tk::Question, hi::Structure);
";"    L C F(tk::SemiColon, hi::Structure);

"("    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(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);

@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);
@end

@begin C
"@encode"         L C F(tk::AtEncode, hi::Meta);
@end

@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);
@end

"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 E4X
"namespace"    L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta);
"xml"          L C I(identifier, Identifier("xml"), tk::XML, hi::Meta);
@end

{IdentifierStart}{IdentifierPart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);

(\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?){Exponent}? L C I(number, Number(strtod(yytext, NULL)), 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[0-7]+ L C I(number, Number(strtoull(yytext + 1, 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 {
    char *value(A char[yyleng]);
    char *local(value);

    for (yy_size_t i(1), e(yyleng - 1); i != e; ++i) {
        char next(yytext[i]);

        if (yytext[i] == '\\')
            switch (next = yytext[++i]) {
                case '\n': continue;
                case '\\': next = '\\'; break;
                case '\'': next = '\''; break;
                case '"': next = '"'; break;
                case 'b': next = '\b'; break;
                case 'f': next = '\f'; break;
                case 'n': next = '\n'; break;
                case 'r': next = '\r'; break;
                case 't': next = '\t'; break;
                case 'v': next = '\v'; break;
                case '0': next = '\0'; break;

                case 'x':
                    next = H(yytext[i + 1]) << 4 | H(yytext[i + 2]);
                    i += 2;
                break;
            }

        *local++ = next;
    }

    *local = '\0';
    I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
}

\r?\n|\r|\xe2\x80[\xa8\xa9] yylloc->end.lines(); yylloc->step(); N

[ \t] 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();
}

%%

void CYDriver::ScannerInit() {
    cylex_init(&scanner_);
    cyset_extra(this, scanner_);
}

void CYDriver::ScannerDestroy() {
    cylex_destroy(scanner_);
}

CYDriver::Condition CYDriver::GetCondition() {
    switch (yy_top_state(scanner_)) {
        case RegExp:
            return RegExpCondition;
@begin E4X
        case XMLContent:
            return XMLContentCondition;
        case XMLTag:
            return XMLTagCondition;
@end
        default:
            _assert(false);
    }
}

void CYDriver::SetCondition(Condition condition) {
    struct yyguts_t *yyg(reinterpret_cast<struct yyguts_t *>(scanner_));

    switch (condition) {
        case RegExpCondition:
            BEGIN(RegExp);
            break;
@begin E4X
        case XMLContentCondition:
            BEGIN(XMLContent);
            break;
        case XMLTagCondition:
            BEGIN(XMLTag);
            break;
@end
        default:
            _assert(false);
    }
}

void CYDriver::PushCondition(Condition condition) {
    switch (condition) {
        case RegExpCondition:
            yy_push_state(RegExp, scanner_);
            break;
@begin E4X
        case XMLContentCondition:
            yy_push_state(XMLContent, scanner_);
            break;
        case XMLTagCondition:
            yy_push_state(XMLTag, scanner_);
            break;
@end
        default:
            _assert(false);
    }
}

void CYDriver::PopCondition() {
    yy_pop_state(scanner_);
}
Commit	Line	Data
b3378a02	1	/* Cycript - Optimizing JavaScript Compiler/Runtime
c15969fd	2	* Copyright (C) 2009-2013 Jay Freeman (saurik)
d15b59f5 JF	3	*/
d15b59f5 JF	4
c15969fd	5	/* GNU General Public License, Version 3 {{{ */
d15b59f5	6	/*
c15969fd JF	7	* Cycript is free software: you can redistribute it and/or modify
	8	* it under the terms of the GNU General Public License as published
	9	* by the Free Software Foundation, either version 3 of the License,
	10	* or (at your option) any later version.
d15b59f5	11	*
c15969fd JF	12	* Cycript is distributed in the hope that it will be useful, but
	13	* WITHOUT ANY WARRANTY; without even the implied warranty of
	14	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	15	* GNU General Public License for more details.
d15b59f5	16	*
c15969fd	17	* You should have received a copy of the GNU General Public License
b3378a02 JF	18	* along with Cycript. If not, see <http://www.gnu.org/licenses/>.
b3378a02 JF	19	**/
d15b59f5 JF	20	/* }}} */
d15b59f5 JF	21
2155ab92	22	/* 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 */
cb02f8ae	23
d15b59f5	24	%{
cac61857	25	#define YYLTYPE cy::location
63b4c5a8 JF	26	#include "Cycript.tab.hh"
63b4c5a8 JF	27	typedef cy::parser::token tk;
693d501b	28
82a02ede JF	29	#include "Highlight.hpp"
82a02ede JF	30
5999c315	31	#define YY_EXTRA_TYPE CYDriver *
db5e2840	32
153c6dec	33	// do /not/ fold token to the return: this is a macro and the ordering is dependent
82a02ede	34	#define F(value, highlight) do { \
153c6dec	35	int token(value); \
39ed6a51	36	@begin ObjectiveC
5602b1ee	37	yyextra->no_.AtImplementation = false; \
39ed6a51	38	@end
5602b1ee JF	39	yyextra->no_.Function = false; \
5602b1ee JF	40	yyextra->no_.OpenBrace = false; \
82a02ede	41	yylval->highlight_ = highlight; \
153c6dec	42	return token; \
3ea7eed0 JF	43	} while (false)
3ea7eed0 JF	44
2eb8215d	45	#define A new($pool)
0cbeddf8	46	#define Y $pool.strmemdup(yytext, yyleng)
2eb8215d	47
82a02ede	48	#define I(type, Type, value, highlight) do { \
2eb8215d	49	yylval->type ## _ = A CY ## Type; \
82a02ede	50	F(value, highlight); \
2eb8215d JF	51	} while (false)
2eb8215d JF	52
697d6fd2	53	#define T yylval->newline_ = yyextra->state_ == CYNewLine; BEGIN(Div);
db5e2840 JF	54	#define C T yyextra->state_ = CYClear;
db5e2840 JF	55	#define R T yyextra->state_ = CYRestricted;
5befe15e JF	56
	57	#define N \
	58	if (yyextra->state_ != CYNewLine) { \
2eb8215d JF	59	if (yyextra->state_ != CYRestricted) \
	60	yyextra->state_ = CYNewLine; \
	61	else { \
5befe15e	62	yyextra->state_ = CYClear; \
82a02ede	63	F(tk::NewLine, hi::Nothing); \
2eb8215d	64	} \
5befe15e JF	65	}
5befe15e JF	66
691e4717	67	#define V(more) { \
cb02f8ae JF	68	if (const char nl = reinterpret_cast<const char >(memchr(yytext, '\n', yyleng))) { \
	69	unsigned lines(0); \
	70	size_t left; \
	71	do { \
	72	++lines; \
	73	left = yyleng - (nl - yytext) - 1; \
	74	nl = reinterpret_cast<const char *>(memchr(nl + 1, '\n', left)); \
	75	} while (nl != NULL); \
	76	yylloc->end.lines(lines); \
	77	yylloc->end.columns(left); \
	78	yylloc->step(); \
691e4717	79	more \
cb02f8ae JF	80	} else L \
	81	}
	82
5befe15e JF	83	#define L { \
	84	yylloc->step(); \
	85	yylloc->columns(yyleng); \
	86	}
e7ed5354	87
abc79d6e JF	88	#define M { \
	89	if (yyextra->commented_) { \
	90	I(comment, Comment(Y), tk::Comment, hi::Comment); \
	91	} \
	92	}
	93
931b816a JF	94	int H(char c) {
	95	if (c >= '0' && c <= '9')
	96	return c - '0';
	97	if (c >= 'a' && c <= 'f')
	98	return c - 'a' + 10;
	99	if (c >= 'A' && c <= 'F')
	100	return c - 'A' + 10;
	101	return -1;
	102	}
	103
e7ed5354	104	#define YY_INPUT(data, value, size) { \
d3b63265	105	if (yyextra->data_.eof()) \
e7ed5354 JF	106	value = YY_NULL; \
e7ed5354 JF	107	else { \
d3b63265 JF	108	yyextra->data_.read(data, size); \
	109	size_t copy(yyextra->data_.gcount()); \
	110	value = copy == 0 ? YY_NULL : copy; \
e7ed5354 JF	111	} \
	112	}
	113
e5332278 JF	114	%}
	115
	116	%option prefix="cy"
	117	%option bison-bridge
	118	%option bison-locations
	119	%option noyywrap
	120	%option yylineno
	121	%option nounput
6c962f8b JF	122	%option batch
6c962f8b JF	123	%option never-interactive
924f67b2	124	%option reentrant
691e4717	125	%option stack
e5332278	126
2bf24581	127	Exponent [eE][+-]?[0-9]+
367eebb1	128	Escape \\[\\'"bfnrtv]\|\\0\|\\x[0-9a-fA-F]{2}\|\\u[0-9a-fA-F]{4}\|\\\n
e5332278	129
63cd45c9 JF	130	IdentifierStart [a-zA-Z$_]
	131	IdentifierPart [a-zA-Z$_0-9]
	132
	133	NonTerminator [^\n]
	134	BackslashSequence \\{NonTerminator}
	135	RegularExpressionFirstChar [^\n*\\/]\|{BackslashSequence}
	136	RegularExpressionChar [^\n\\/]\|{BackslashSequence}
	137	RegularExpressionFlags {IdentifierPart}*
	138	RegularExpressionChars {RegularExpressionChar}*
697d6fd2	139	RegularExpressionBody {RegularExpressionFirstChar}{RegularExpressionChars}
63cd45c9	140
691e4717 JF	141	@begin E4X
	142	XMLNameStart [a-zA-Z_:]
	143	XMLNamePart [a-zA-Z0-9.-_:]
	144	XMLName {XMLNameStart}{XMLNamePart}*
	145	@end
	146
697d6fd2 JF	147	%s Div
697d6fd2 JF	148	%s RegExp
63cd45c9	149
691e4717 JF	150	@begin E4X
	151	%x XMLContent
	152	%x XMLTag
	153	@end
	154
e5332278 JF	155	%%
e5332278 JF	156
82a02ede	157	<RegExp>\/{RegularExpressionBody}\/{RegularExpressionFlags} L C I(literal, RegEx(Y), tk::RegularExpressionLiteral, hi::Constant);
63cd45c9	158
fc44232b JF	159	#![^\n]* L M
fc44232b JF	160
abc79d6e	161	\/\/[^\n]* L M
fe123f47 JF	162
	163	/* http://ostermiller.org/findcomment.html */
	164	/* XXX: unify these two rules using !? */
82a02ede	165	\/\!([^]\|[\r\n]\|(\+([^/]\|[\r\n])))\+\/ V() C I(comment, Comment(Y), tk::Comment, hi::Comment);
abc79d6e	166	\/\([^]\|[\r\n]\|(\+([^/]\|[\r\n])))\+\/ V(N) M
66fb559f	167
cb02f8ae	168	@begin E4X
82a02ede JF	169	<RegExp>"<>" L F(tk::LeftRight, hi::Structure);
82a02ede JF	170	<XMLContent>"</>" L F(tk::LeftSlashRight, hi::Structure);
691e4717	171
82a02ede JF	172	<RegExp,XMLContent>\<!\[CDATA\[(\n\|[^[]\|\[[^[]\|\[\[[^>])*]]> V() F(tk::XMLCDATA, hi::Constant);
	173	<RegExp,XMLContent>\<!--(\n\|[^-]\|-[^-])*--> V() F(tk::XMLComment, hi::Comment);
	174	<RegExp,XMLContent>\<?(\n\|[^?]\|\?[^>])*?> V() F(tk::XMLPI, hi::Meta);
691e4717	175
82a02ede JF	176	<XMLTag>"=" L F(tk::Equal, hi::Structure);
	177	<XMLTag>">" L F(tk::Right, hi::Structure);
	178	<XMLTag>"/>" L F(tk::SlashRight, hi::Structure);
	179	<XMLTag>"{" L F(tk::OpenBrace, hi::Structure);
691e4717	180
82a02ede JF	181	<XMLTag>\"(\n\|[^"])\"\|'(\n\|[^'])' V() F(tk::XMLAttributeValue, hi::Constant);
	182	<XMLTag>{XMLName} L F(tk::XMLName, hi::Identifier);
	183	<XMLTag>[ \t\r\n] V() F(tk::XMLWhitespace, hi::Nothing);
db5e2840	184
82a02ede JF	185	<XMLContent>"{" L F(tk::OpenBrace, hi::Structure);
	186	<XMLContent>"<" L F(tk::Left, hi::Structure);
	187	<XMLContent>"</" L F(tk::LeftSlash, hi::Structure);
691e4717 JF	188	@end
691e4717 JF	189
82a02ede	190	"..." L C F(tk::PeriodPeriodPeriod, hi::Meta);
c8a0500b	191
691e4717	192	@begin E4X
82a02ede JF	193	"::" L C F(tk::ColonColon, hi::Operator);
82a02ede JF	194	".." L C F(tk::PeriodPeriod, hi::Operator);
cb02f8ae	195	@end
ac9a5ce1	196
313708a9	197	@begin E4X ObjectiveC
82a02ede	198	"@" L C F(tk::At, hi::Operator);
ac9d4181	199	"#" L C F(tk::Pound, hi::Operator);
313708a9 JF	200	@end
313708a9 JF	201
82a02ede JF	202	"&" L C F(tk::Ampersand, hi::Operator);
	203	"&&" L C F(tk::AmpersandAmpersand, hi::Operator);
	204	"&=" L C F(tk::AmpersandEqual, hi::Operator);
	205	"^" L C F(tk::Carrot, hi::Operator);
	206	"^=" L C F(tk::CarrotEqual, hi::Operator);
	207	"=" L C F(tk::Equal, hi::Operator);
	208	"==" L C F(tk::EqualEqual, hi::Operator);
	209	"===" L C F(tk::EqualEqualEqual, hi::Operator);
	210	"=>" L C F(tk::EqualRight, hi::Operator);
	211	"!" L C F(tk::Exclamation, hi::Operator);
	212	"!=" L C F(tk::ExclamationEqual, hi::Operator);
	213	"!==" L C F(tk::ExclamationEqualEqual, hi::Operator);
	214	"-" L C F(tk::Hyphen, hi::Operator);
	215	"-=" L C F(tk::HyphenEqual, hi::Operator);
	216	"--" L C F(yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen, hi::Operator);
	217	"->" L C F(tk::HyphenRight, hi::Operator);
	218	"<" L C F(tk::Left, hi::Operator);
	219	"<=" L C F(tk::LeftEqual, hi::Operator);
	220	"<<" L C F(tk::LeftLeft, hi::Operator);
	221	"<<=" L C F(tk::LeftLeftEqual, hi::Operator);
	222	"%" L C F(tk::Percent, hi::Operator);
	223	"%=" L C F(tk::PercentEqual, hi::Operator);
	224	"." L C F(tk::Period, hi::Operator);
	225	"\|" L C F(tk::Pipe, hi::Operator);
	226	"\|=" L C F(tk::PipeEqual, hi::Operator);
	227	"\|\|" L C F(tk::PipePipe, hi::Operator);
	228	"+" L C F(tk::Plus, hi::Operator);
	229	"+=" L C F(tk::PlusEqual, hi::Operator);
	230	"++" L C F(yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus, hi::Operator);
	231	">" L C F(tk::Right, hi::Operator);
	232	">=" L C F(tk::RightEqual, hi::Operator);
	233	">>" L C F(tk::RightRight, hi::Operator);
	234	">>=" L C F(tk::RightRightEqual, hi::Operator);
	235	">>>" L C F(tk::RightRightRight, hi::Operator);
	236	">>>=" L C F(tk::RightRightRightEqual, hi::Operator);
	237	"*" L C F(tk::Star, hi::Operator);
	238	"*=" L C F(tk::StarEqual, hi::Operator);
	239	"~" L C F(tk::Tilde, hi::Operator);
	240
	241	<Div>"/" L C F(tk::Slash, hi::Operator);
	242	<Div>"/=" L C F(tk::SlashEqual, hi::Operator);
	243
	244	":" L C F(tk::Colon, hi::Structure);
	245	"," L C F(tk::Comma, hi::Structure);
	246	"?" L C F(tk::Question, hi::Structure);
	247	";" L C F(tk::SemiColon, hi::Structure);
	248
	249	"(" L C F(tk::OpenParen, hi::Structure);
	250	")" L C F(tk::CloseParen, hi::Structure);
	251
	252	"{" L C F(yyextra->no_.OpenBrace ? tk::OpenBrace__ : yylval->newline_ ? tk::OpenBrace_ : tk::OpenBrace, hi::Structure);
	253	"}" L C F(tk::CloseBrace, hi::Structure);
	254
	255	"[" L C F(tk::OpenBracket, hi::Structure);
	256	"]" L C F(tk::CloseBracket, hi::Structure);
5befe15e	257
dc5d7cf4 JF	258	"@error" L C F(tk::AtError, hi::Error);
dc5d7cf4 JF	259
1ba6903e	260	@begin Java
82a02ede	261	"@class" L C F(tk::AtClass, hi::Meta);
1ba6903e JF	262	@end
1ba6903e JF	263
60097023 JF	264	@begin C
60097023 JF	265	"typedef" L C I(identifier, Identifier("typedef"), tk::Typedef, hi::Meta);
8aebacfe JF	266	"unsigned" L C I(identifier, Identifier("unsigned"), tk::Unsigned, hi::Type);
8aebacfe JF	267	"signed" L C I(identifier, Identifier("signed"), tk::Signed, hi::Type);
60097023 JF	268	@end
60097023 JF	269
8a2eb1be	270	@begin C
46f4f308	271	"@encode" L C F(tk::AtEncode, hi::Meta);
8a2eb1be JF	272	@end
	273
	274	@begin ObjectiveC
82a02ede JF	275	"@end" L C F(tk::AtEnd, hi::Meta);
82a02ede JF	276	"@implementation" L C F(yyextra->no_.AtImplementation ? tk::AtImplementation_ : tk::AtImplementation, hi::Meta);
73439a14	277	"@import" L C F(tk::AtImport, hi::Special);
82a02ede	278	"@selector" L C F(tk::AtSelector, hi::Meta);
cf19d1c1	279
355cc1eb JF	280	"NULL" L C I(identifier, Identifier("NULL"), tk::Identifier_, hi::Constant);
355cc1eb JF	281	"nil" L C I(identifier, Identifier("nil"), tk::Identifier_, hi::Constant);
60496dd5 JF	282	"YES" L C I(identifier, Identifier("YES"), tk::Yes, hi::Constant);
60496dd5 JF	283	"NO" L C I(identifier, Identifier("NO"), tk::No, hi::Constant);
8b820c00	284
aa82b83b JF	285	"bool" L C I(identifier, Identifier("bool"), tk::Identifier_, hi::Type);
aa82b83b JF	286	"BOOL" L C I(identifier, Identifier("BOOL"), tk::Identifier_, hi::Type);
8b820c00	287	"id" L C I(identifier, Identifier("id"), tk::Identifier_, hi::Type);
aa8f2b2c	288	"SEL" L C I(identifier, Identifier("SEL"), tk::Identifier_, hi::Type);
cb02f8ae	289	@end
d35a3b07	290
355cc1eb JF	291	"undefined" L C I(identifier, Identifier("undefined"), tk::Identifier_, hi::Operator);
355cc1eb JF	292
82a02ede JF	293	"false" L C I(false, False(), tk::False, hi::Constant);
	294	"null" L C I(null, Null(), tk::Null, hi::Constant);
	295	"true" L C I(true, True(), tk::True, hi::Constant);
	296
	297	"auto" L C I(word, Word("auto"), tk::Auto, hi::Meta);
	298	"break" L R I(word, Word("break"), tk::Break, hi::Control);
	299	"case" L C I(word, Word("case"), tk::Case, hi::Control);
	300	"catch" L C I(word, Word("catch"), tk::Catch, hi::Control);
	301	"continue" L R I(word, Word("continue"), tk::Continue, hi::Control);
	302	"default" L C I(word, Word("default"), tk::Default, hi::Control);
	303	"delete" L C I(word, Word("delete"), tk::Delete, hi::Operator);
	304	"do" L C I(word, Word("do"), tk::Do, hi::Control);
	305	"else" L C I(word, Word("else"), tk::Else, hi::Control);
	306	"finally" L C I(word, Word("finally"), tk::Finally, hi::Control);
	307	"for" L C I(word, Word("for"), tk::For, hi::Control);
	308	"function" L C I(word, Word("function"), yyextra->no_.Function ? tk::Function_ : tk::Function, hi::Meta);
	309	"if" L C I(word, Word("if"), tk::If, hi::Control);
	310	"in" L C I(word, Word("in"), yyextra->in_.top() ? tk::In_ : tk::In, hi::Operator);
	311	"instanceof" L C I(word, Word("instanceof"), tk::InstanceOf, hi::Operator);
	312	"new" L C I(word, Word("new"), tk::New, hi::Operator);
	313	"return" L R I(word, Word("return"), tk::Return, hi::Control);
	314	"switch" L C I(word, Word("switch"), tk::Switch, hi::Control);
	315	"this" L C I(this, This(), tk::This, hi::Constant);
	316	"throw" L R I(word, Word("throw"), tk::Throw, hi::Control);
	317	"try" L C I(word, Word("try"), tk::Try, hi::Control);
	318	"typeof" L C I(word, Word("typeof"), tk::TypeOf, hi::Operator);
	319	"var" L C I(word, Word("var"), tk::Var, hi::Meta);
	320	"void" L C I(word, Word("void"), tk::Void, hi::Operator);
	321	"while" L C I(word, Word("while"), tk::While, hi::Control);
	322	"with" L C I(word, Word("with"), tk::With, hi::Control);
	323
	324	"debugger" L C I(word, Word("debugger"), tk::Debugger, hi::Meta);
	325
	326	"const" L C I(word, Word("const"), tk::Const, hi::Meta);
	327
	328	"class" L C I(word, Word("class"), tk::Class, hi::Meta);
	329	"enum" L C I(word, Word("enum"), tk::Enum, hi::Meta);
	330	"export" L C I(word, Word("export"), tk::Export, hi::Meta);
	331	"extends" L C I(word, Word("extends"), tk::Extends, hi::Meta);
	332	"import" L C I(word, Word("import"), tk::Import, hi::Meta);
	333	"super" L C I(word, Word("super"), tk::Super, hi::Constant);
	334
	335	"implements" L C I(identifier, Identifier("implements"), tk::Implements, hi::Meta);
	336	"interface" L C I(identifier, Identifier("interface"), tk::Interface, hi::Meta);
	337	"package" L C I(identifier, Identifier("package"), tk::Package, hi::Meta);
	338	"private" L C I(identifier, Identifier("private"), tk::Private, hi::Meta);
	339	"protected" L C I(identifier, Identifier("protected"), tk::Protected, hi::Meta);
	340	"public" L C I(identifier, Identifier("public"), tk::Public, hi::Meta);
	341	"static" L C I(identifier, Identifier("static"), tk::Static, hi::Meta);
	342
	343	"abstract" L C I(identifier, Identifier("abstract"), tk::Abstract, hi::Meta);
	344	"boolean" L C I(identifier, Identifier("boolean"), tk::Boolean, hi::Type);
	345	"byte" L C I(identifier, Identifier("byte"), tk::Byte, hi::Type);
	346	"char" L C I(identifier, Identifier("char"), tk::Char, hi::Type);
	347	"double" L C I(identifier, Identifier("double"), tk::Double, hi::Type);
	348	"final" L C I(identifier, Identifier("final"), tk::Final, hi::Meta);
	349	"float" L C I(identifier, Identifier("float"), tk::Float, hi::Type);
	350	"goto" L C I(identifier, Identifier("goto"), tk::Goto, hi::Control);
	351	"int" L C I(identifier, Identifier("int"), tk::Int, hi::Type);
	352	"long" L C I(identifier, Identifier("long"), tk::Long, hi::Type);
	353	"native" L C I(identifier, Identifier("native"), tk::Native, hi::Meta);
	354	"short" L C I(identifier, Identifier("short"), tk::Short, hi::Type);
	355	"synchronized" L C I(identifier, Identifier("synchronized"), tk::Synchronized, hi::Meta);
	356	"throws" L C I(identifier, Identifier("throws"), tk::Throws, hi::Meta);
357	"transient" L C I(identifier, Identifier("transient"), tk::Transient, hi::Meta);
358	"volatile" L C I(identifier, Identifier("volatile"), tk::Volatile, hi::Meta);
359
360	"let" L C I(identifier, Identifier("let"), tk::Let, hi::Meta);
361	"yield" L C I(identifier, Identifier("yield"), tk::Yield, hi::Control);
362
363	"each" L C I(identifier, Identifier("each"), tk::Each, hi::Control);
364	"of" L C I(identifier, Identifier("of"), tk::Of, hi::Operator);
5d646fb5	365
691e4717	366	@begin E4X
82a02ede JF	367	"namespace" L C I(identifier, Identifier("namespace"), tk::Namespace, hi::Meta);
82a02ede JF	368	"xml" L C I(identifier, Identifier("xml"), tk::XML, hi::Meta);
691e4717 JF	369	@end
691e4717 JF	370
82a02ede	371	{IdentifierStart}{IdentifierPart}* L C I(identifier, Identifier(Y), tk::Identifier_, hi::Identifier);
5d646fb5	372
82a02ede	373	(\.[0-9]+\|(0\|[1-9][0-9])(\.[0-9])?){Exponent}? L C I(number, Number(strtod(yytext, NULL)), tk::NumericLiteral, hi::Constant);
5d646fb5	374
82a02ede JF	375	0[xX][0-9a-fA-F]+ L C I(number, Number(strtoull(yytext + 2, NULL, 16)), tk::NumericLiteral, hi::Constant);
	376	0[0-7]+ L C I(number, Number(strtoull(yytext + 1, NULL, 8)), tk::NumericLiteral, hi::Constant);
	377	0[bB][0-1]+ L C I(number, Number(strtoull(yytext + 2, NULL, 2)), tk::NumericLiteral, hi::Constant);
5befe15e	378
931b816a	379	\"([^"\\\n]\|{Escape})\"\|'([^'\\\n]\|{Escape})' L C {
2eb8215d	380	char *value(A char[yyleng]);
931b816a JF	381	char *local(value);
931b816a JF	382
66f8d960	383	for (yy_size_t i(1), e(yyleng - 1); i != e; ++i) {
931b816a JF	384	char next(yytext[i]);
	385
	386	if (yytext[i] == '\\')
	387	switch (next = yytext[++i]) {
367eebb1	388	case '\n': continue;
931b816a JF	389	case '\\': next = '\\'; break;
	390	case '\'': next = '\''; break;
	391	case '"': next = '"'; break;
	392	case 'b': next = '\b'; break;
	393	case 'f': next = '\f'; break;
	394	case 'n': next = '\n'; break;
	395	case 'r': next = '\r'; break;
	396	case 't': next = '\t'; break;
	397	case 'v': next = '\v'; break;
	398	case '0': next = '\0'; break;
	399
	400	case 'x':
	401	next = H(yytext[i + 1]) << 4 \| H(yytext[i + 2]);
	402	i += 2;
	403	break;
	404	}
	405
	406	*local++ = next;
	407	}
	408
	409	*local = '\0';
82a02ede	410	I(string, String(value, local - value), tk::StringLiteral, hi::Constant);
931b816a	411	}
5befe15e	412
3ea7eed0	413	\r?\n\|\r\|\xe2\x80[\xa8\xa9] yylloc->end.lines(); yylloc->step(); N
5befe15e JF	414
5befe15e JF	415	[ \t] L
7e5391fd	416
82a02ede	417	<<EOF>> if (yyextra->auto_) { yyextra->auto_ = false; F(tk::AutoComplete, hi::Nothing); } L yyterminate();
94d55b5c	418
48e3be8a	419	. L {
94d55b5c JF	420	CYDriver::Error error;
	421	error.location_ = *yylloc;
	422	error.message_ = "syntax error, unknown token";
	423	yyextra->errors_.push_back(error);
	424	yyterminate();
	425	}
924f67b2 JF	426
	427	%%
	428
5999c315	429	void CYDriver::ScannerInit() {
924f67b2 JF	430	cylex_init(&scanner_);
	431	cyset_extra(this, scanner_);
	432	}
	433
5999c315	434	void CYDriver::ScannerDestroy() {
924f67b2 JF	435	cylex_destroy(scanner_);
924f67b2 JF	436	}
63cd45c9	437
691e4717 JF	438	CYDriver::Condition CYDriver::GetCondition() {
	439	switch (yy_top_state(scanner_)) {
	440	case RegExp:
	441	return RegExpCondition;
	442	@begin E4X
	443	case XMLContent:
	444	return XMLContentCondition;
	445	case XMLTag:
	446	return XMLTagCondition;
	447	@end
	448	default:
	449	_assert(false);
	450	}
	451	}
	452
	453	void CYDriver::SetCondition(Condition condition) {
63cd45c9 JF	454	struct yyguts_t yyg(reinterpret_cast<struct yyguts_t >(scanner_));
	455
	456	switch (condition) {
697d6fd2 JF	457	case RegExpCondition:
697d6fd2 JF	458	BEGIN(RegExp);
63cd45c9	459	break;
691e4717 JF	460	@begin E4X
	461	case XMLContentCondition:
	462	BEGIN(XMLContent);
	463	break;
	464	case XMLTagCondition:
	465	BEGIN(XMLTag);
	466	break;
	467	@end
63cd45c9 JF	468	default:
	469	_assert(false);
	470	}
	471	}
691e4717 JF	472
	473	void CYDriver::PushCondition(Condition condition) {
	474	switch (condition) {
	475	case RegExpCondition:
	476	yy_push_state(RegExp, scanner_);
	477	break;
	478	@begin E4X
	479	case XMLContentCondition:
	480	yy_push_state(XMLContent, scanner_);
	481	break;
	482	case XMLTagCondition:
	483	yy_push_state(XMLTag, scanner_);
	484	break;
	485	@end
	486	default:
	487	_assert(false);
	488	}
	489	}
	490
	491	void CYDriver::PopCondition() {
	492	yy_pop_state(scanner_);
	493	}