[cycript.git] / Cycript.l

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

#define YY_EXTRA_TYPE CYDriver *
#define YYLTYPE cy::location

#define T yylval->newline_ = yyextra->state_ == CYNewLine;
#define C T yyextra->state_ = CYClear;
#define R T yyextra->state_ = CYRestricted;

#define N \
    if (yyextra->state_ != CYNewLine) { \
        bool restricted(yyextra->state_ == CYRestricted); \
        if (restricted) { \
            yyextra->state_ = CYClear; \
            return tk::NewLine; \
        } else \
            yyextra->state_ = CYNewLine; \
    }

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

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->size_ == 0) \
        value = YY_NULL; \
    else { \
        size_t copy(std::min(size, yyextra->size_)); \
        memcpy(data, yyextra->data_, copy); \
        yyextra->data_ += copy; \
        yyextra->size_ -= copy; \
        value = copy; \
    } \
}

%}

%option prefix="cy"
%option bison-bridge
%option bison-locations
%option noyywrap
%option yylineno
%option nounput
%option interactive
%option reentrant

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

%%

\/\/[^\n]* ;
\/\*(\n|[^\*]|\*[^/])\*\/ if (memchr(yytext, '\n', yyleng) != NULL) N // XXX: supposedly I will be screwed on very very long multi-line comments and need to replace this with a manual lexer. http://websrv.cs.fsu.edu/~engelen/courses/COP5621/Pr2.pdf ; XXX: this rule doesn't work anyway, fucking A :(

"&"    L C return tk::Ampersand;
"&&"   L C return tk::AmpersandAmpersand;
"&="   L C return tk::AmpersandEqual;
"^"    L C return tk::Carrot;
"^="   L C return tk::CarrotEqual;
"="    L C return tk::Equal;
"=="   L C return tk::EqualEqual;
"==="  L C return tk::EqualEqualEqual;
"!"    L C return tk::Exclamation;
"!="   L C return tk::ExclamationEqual;
"!=="  L C return tk::ExclamationEqualEqual;
"-"    L C return tk::Hyphen;
"-="   L C return tk::HyphenEqual;
"--"   L C return yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen;
"->"   L C return tk::HyphenRight;
"<"    L C return tk::Left;
"<="   L C return tk::LeftEqual;
"<<"   L C return tk::LeftLeft;
"<<="  L C return tk::LeftLeftEqual;
"%"    L C return tk::Percent;
"%="   L C return tk::PercentEqual;
"."    L C return tk::Period;
"|"    L C return tk::Pipe;
"|="   L C return tk::PipeEqual;
"||"   L C return tk::PipePipe;
"+"    L C return tk::Plus;
"+="   L C return tk::PlusEqual;
"++"   L C return yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus;
">"    L C return tk::Right;
">="   L C return tk::RightEqual;
">>"   L C return tk::RightRight;
">>="  L C return tk::RightRightEqual;
">>>"  L C return tk::RightRightRight;
">>>=" L C return tk::RightRightRightEqual;
"/"    L C return tk::Slash;
"/="   L C return tk::SlashEqual;
"*"    L C return tk::Star;
"*="   L C return tk::StarEqual;
"~"    L C return tk::Tilde;

":"    L C return tk::Colon;
","    L C return tk::Comma;
"?"    L C return tk::Question;
";"    L C return tk::SemiColon;

"("    L C return tk::OpenParen;
")"    L C return tk::CloseParen;

"{"    L C return tk::OpenBrace;
"}"    L C return tk::CloseBrace;

"["    L C return tk::OpenBracket;
"]"    L C return tk::CloseBracket;

"@selector"  L C return tk::AtSelector;

"break"      L R yylval->word_ = new CYWord("break"); return tk::Break;
"case"       L C yylval->word_ = new CYWord("case"); return tk::Case;
"catch"      L C yylval->word_ = new CYWord("catch"); return tk::Catch;
"continue"   L R yylval->word_ = new CYWord("continue"); return tk::Continue;
"default"    L C yylval->word_ = new CYWord("default"); return tk::Default;
"delete"     L C yylval->word_ = new CYWord("delete"); return tk::Delete;
"do"         L C yylval->word_ = new CYWord("do"); return tk::Do;
"else"       L C yylval->word_ = new CYWord("else"); return tk::Else;
"false"      L C yylval->false_ = new CYFalse(); return tk::False;
"finally"    L C yylval->word_ = new CYWord("finally"); return tk::Finally;
"for"        L C yylval->word_ = new CYWord("for"); return tk::For;
"function"   L C yylval->word_ = new CYWord("function"); return tk::Function;
"if"         L C yylval->word_ = new CYWord("if"); return tk::If;
"in"         L C yylval->word_ = new CYWord("in"); return tk::In;
"instanceof" L C yylval->word_ = new CYWord("instanceof"); return tk::InstanceOf;
"new"        L C yylval->word_ = new CYWord("new"); return tk::New;
"null"       L C yylval->null_ = new CYNull(); return tk::Null;
"return"     L R yylval->word_ = new CYWord("return"); return tk::Return;
"switch"     L C yylval->word_ = new CYWord("switch"); return tk::Switch;
"this"       L C yylval->this_ = new CYThis(); return tk::This;
"throw"      L R yylval->word_ = new CYWord("throw"); return tk::Throw;
"true"       L C yylval->true_ = new CYTrue(); return tk::True;
"try"        L C yylval->word_ = new CYWord("try"); return tk::Try;
"typeof"     L C yylval->word_ = new CYWord("typeof"); return tk::TypeOf;
"var"        L C yylval->word_ = new CYWord("var"); return tk::Var;
"void"       L C yylval->word_ = new CYWord("void"); return tk::Void;
"while"      L C yylval->word_ = new CYWord("while"); return tk::While;
"with"       L C yylval->word_ = new CYWord("with"); return tk::With;

[a-zA-Z$_][a-zA-Z$_0-9]* yylval->identifier_ = new CYIdentifier(apr_pstrmemdup(yyextra->pool_, yytext, yyleng)); L C return tk::Identifier;

(\.[0-9]+|(0|[1-9][0-9]*)(\.[0-9]*)?){Exponent}? yylval->number_ = new CYNumber(strtod(yytext, NULL)); L C return tk::NumericLiteral;

0[xX][0-9a-fA-F]+ L C yylval->number_ = new CYNumber(strtoull(yytext + 2, NULL, 16)); return tk::NumericLiteral;

0[bB][0-1]+ L C yylval->number_ = new CYNumber(strtoull(yytext + 2, NULL, 2)); return tk::NumericLiteral;

\"([^"\\\n]|{Escape})*\"|'([^'\\\n]|{Escape})*' L C {
    char *value(reinterpret_cast<char *>(apr_palloc(yyextra->pool_, yyleng)));
    char *local(value);

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

        if (yytext[i] == '\\')
            switch (next = yytext[++i]) {
                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';
    yylval->string_ = new CYString(value, local - value);
    return tk::StringLiteral;
}

\n yylloc->end.lines(); yylloc->step(); N

[ \t] L
<<EOF>> L yyterminate();

. {
    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_);
}
Commit	Line	Data
e5332278	1	%{
63b4c5a8 JF	2	#include "Cycript.tab.hh"
63b4c5a8 JF	3	typedef cy::parser::token tk;
693d501b	4
5999c315	5	#define YY_EXTRA_TYPE CYDriver *
693d501b	6	#define YYLTYPE cy::location
db5e2840 JF	7
	8	#define T yylval->newline_ = yyextra->state_ == CYNewLine;
	9	#define C T yyextra->state_ = CYClear;
	10	#define R T yyextra->state_ = CYRestricted;
5befe15e JF	11
	12	#define N \
	13	if (yyextra->state_ != CYNewLine) { \
	14	bool restricted(yyextra->state_ == CYRestricted); \
	15	if (restricted) { \
	16	yyextra->state_ = CYClear; \
	17	return tk::NewLine; \
	18	} else \
	19	yyextra->state_ = CYNewLine; \
	20	}
	21
	22	#define L { \
	23	yylloc->step(); \
	24	yylloc->columns(yyleng); \
	25	}
e7ed5354	26
931b816a JF	27	int H(char c) {
	28	if (c >= '0' && c <= '9')
	29	return c - '0';
	30	if (c >= 'a' && c <= 'f')
	31	return c - 'a' + 10;
	32	if (c >= 'A' && c <= 'F')
	33	return c - 'A' + 10;
	34	return -1;
	35	}
	36
e7ed5354 JF	37	#define YY_INPUT(data, value, size) { \
	38	if (yyextra->size_ == 0) \
	39	value = YY_NULL; \
	40	else { \
	41	size_t copy(std::min(size, yyextra->size_)); \
	42	memcpy(data, yyextra->data_, copy); \
	43	yyextra->data_ += copy; \
	44	yyextra->size_ -= copy; \
	45	value = copy; \
	46	} \
	47	}
	48
e5332278 JF	49	%}
	50
	51	%option prefix="cy"
	52	%option bison-bridge
	53	%option bison-locations
	54	%option noyywrap
	55	%option yylineno
	56	%option nounput
	57	%option interactive
924f67b2	58	%option reentrant
e5332278	59
2bf24581	60	Exponent [eE][+-]?[0-9]+
931b816a	61	Escape \\[\\'"bfnrtv]\|\\0\|\\x[0-9a-fA-F]{2}\|\\u[0-9a-fA-F]{4}
e5332278 JF	62
	63	%%
	64
c3c20102	65	\/\/[^\n]* ;
db5e2840 JF	66	\/\(\n\|[^\]\|\[^/])\\/ if (memchr(yytext, '\n', yyleng) != NULL) N // XXX: supposedly I will be screwed on very very long multi-line comments and need to replace this with a manual lexer. http://websrv.cs.fsu.edu/~engelen/courses/COP5621/Pr2.pdf ; XXX: this rule doesn't work anyway, fucking A :(
db5e2840 JF	67
5befe15e JF	68	"&" L C return tk::Ampersand;
	69	"&&" L C return tk::AmpersandAmpersand;
	70	"&=" L C return tk::AmpersandEqual;
	71	"^" L C return tk::Carrot;
	72	"^=" L C return tk::CarrotEqual;
	73	"=" L C return tk::Equal;
	74	"==" L C return tk::EqualEqual;
	75	"===" L C return tk::EqualEqualEqual;
	76	"!" L C return tk::Exclamation;
	77	"!=" L C return tk::ExclamationEqual;
	78	"!==" L C return tk::ExclamationEqualEqual;
	79	"-" L C return tk::Hyphen;
	80	"-=" L C return tk::HyphenEqual;
	81	"--" L C return yylval->newline_ ? tk::HyphenHyphen_ : tk::HyphenHyphen;
	82	"->" L C return tk::HyphenRight;
	83	"<" L C return tk::Left;
	84	"<=" L C return tk::LeftEqual;
	85	"<<" L C return tk::LeftLeft;
	86	"<<=" L C return tk::LeftLeftEqual;
	87	"%" L C return tk::Percent;
	88	"%=" L C return tk::PercentEqual;
	89	"." L C return tk::Period;
	90	"\|" L C return tk::Pipe;
	91	"\|=" L C return tk::PipeEqual;
	92	"\|\|" L C return tk::PipePipe;
	93	"+" L C return tk::Plus;
	94	"+=" L C return tk::PlusEqual;
	95	"++" L C return yylval->newline_ ? tk::PlusPlus_ : tk::PlusPlus;
	96	">" L C return tk::Right;
	97	">=" L C return tk::RightEqual;
	98	">>" L C return tk::RightRight;
	99	">>=" L C return tk::RightRightEqual;
	100	">>>" L C return tk::RightRightRight;
	101	">>>=" L C return tk::RightRightRightEqual;
	102	"/" L C return tk::Slash;
	103	"/=" L C return tk::SlashEqual;
	104	"*" L C return tk::Star;
	105	"*=" L C return tk::StarEqual;
	106	"~" L C return tk::Tilde;
	107
	108	":" L C return tk::Colon;
	109	"," L C return tk::Comma;
	110	"?" L C return tk::Question;
	111	";" L C return tk::SemiColon;
	112
	113	"(" L C return tk::OpenParen;
	114	")" L C return tk::CloseParen;
	115
	116	"{" L C return tk::OpenBrace;
	117	"}" L C return tk::CloseBrace;
	118
	119	"[" L C return tk::OpenBracket;
	120	"]" L C return tk::CloseBracket;
	121
	122	"@selector" L C return tk::AtSelector;
	123
	124	"break" L R yylval->word_ = new CYWord("break"); return tk::Break;
	125	"case" L C yylval->word_ = new CYWord("case"); return tk::Case;
	126	"catch" L C yylval->word_ = new CYWord("catch"); return tk::Catch;
	127	"continue" L R yylval->word_ = new CYWord("continue"); return tk::Continue;
	128	"default" L C yylval->word_ = new CYWord("default"); return tk::Default;
	129	"delete" L C yylval->word_ = new CYWord("delete"); return tk::Delete;
	130	"do" L C yylval->word_ = new CYWord("do"); return tk::Do;
	131	"else" L C yylval->word_ = new CYWord("else"); return tk::Else;
132	"false" L C yylval->false_ = new CYFalse(); return tk::False;
133	"finally" L C yylval->word_ = new CYWord("finally"); return tk::Finally;
134	"for" L C yylval->word_ = new CYWord("for"); return tk::For;
135	"function" L C yylval->word_ = new CYWord("function"); return tk::Function;
136	"if" L C yylval->word_ = new CYWord("if"); return tk::If;
137	"in" L C yylval->word_ = new CYWord("in"); return tk::In;
138	"instanceof" L C yylval->word_ = new CYWord("instanceof"); return tk::InstanceOf;
139	"new" L C yylval->word_ = new CYWord("new"); return tk::New;
140	"null" L C yylval->null_ = new CYNull(); return tk::Null;
141	"return" L R yylval->word_ = new CYWord("return"); return tk::Return;
142	"switch" L C yylval->word_ = new CYWord("switch"); return tk::Switch;
143	"this" L C yylval->this_ = new CYThis(); return tk::This;
144	"throw" L R yylval->word_ = new CYWord("throw"); return tk::Throw;
145	"true" L C yylval->true_ = new CYTrue(); return tk::True;
146	"try" L C yylval->word_ = new CYWord("try"); return tk::Try;
147	"typeof" L C yylval->word_ = new CYWord("typeof"); return tk::TypeOf;
148	"var" L C yylval->word_ = new CYWord("var"); return tk::Var;
149	"void" L C yylval->word_ = new CYWord("void"); return tk::Void;
150	"while" L C yylval->word_ = new CYWord("while"); return tk::While;
151	"with" L C yylval->word_ = new CYWord("with"); return tk::With;
152
153	[a-zA-Z$_][a-zA-Z$_0-9]* yylval->identifier_ = new CYIdentifier(apr_pstrmemdup(yyextra->pool_, yytext, yyleng)); L C return tk::Identifier;
154
155	(\.[0-9]+\|(0\|[1-9][0-9])(\.[0-9])?){Exponent}? yylval->number_ = new CYNumber(strtod(yytext, NULL)); L C return tk::NumericLiteral;
156
157	0[xX][0-9a-fA-F]+ L C yylval->number_ = new CYNumber(strtoull(yytext + 2, NULL, 16)); return tk::NumericLiteral;
158
159	0[bB][0-1]+ L C yylval->number_ = new CYNumber(strtoull(yytext + 2, NULL, 2)); return tk::NumericLiteral;
160
931b816a JF	161	\"([^"\\\n]\|{Escape})\"\|'([^'\\\n]\|{Escape})' L C {
	162	char value(reinterpret_cast<char >(apr_palloc(yyextra->pool_, yyleng)));
	163	char *local(value);
	164
	165	for (int i(1); i != yyleng - 1; ++i) {
	166	char next(yytext[i]);
	167
	168	if (yytext[i] == '\\')
	169	switch (next = yytext[++i]) {
	170	case '\\': next = '\\'; break;
	171	case '\'': next = '\''; break;
	172	case '"': next = '"'; break;
	173	case 'b': next = '\b'; break;
	174	case 'f': next = '\f'; break;
	175	case 'n': next = '\n'; break;
	176	case 'r': next = '\r'; break;
	177	case 't': next = '\t'; break;
	178	case 'v': next = '\v'; break;
	179	case '0': next = '\0'; break;
	180
	181	case 'x':
	182	next = H(yytext[i + 1]) << 4 \| H(yytext[i + 2]);
	183	i += 2;
	184	break;
	185	}
	186
	187	*local++ = next;
	188	}
	189
	190	*local = '\0';
	191	yylval->string_ = new CYString(value, local - value);
	192	return tk::StringLiteral;
	193	}
5befe15e JF	194
	195	\n yylloc->end.lines(); yylloc->step(); N
	196
	197	[ \t] L
	198	<<EOF>> L yyterminate();
94d55b5c JF	199
	200	. {
	201	CYDriver::Error error;
	202	error.location_ = *yylloc;
	203	error.message_ = "syntax error, unknown token";
	204	yyextra->errors_.push_back(error);
	205	yyterminate();
	206	}
924f67b2 JF	207
	208	%%
	209
5999c315	210	void CYDriver::ScannerInit() {
924f67b2 JF	211	cylex_init(&scanner_);
	212	cyset_extra(this, scanner_);
	213	}
	214
5999c315	215	void CYDriver::ScannerDestroy() {
924f67b2 JF	216	cylex_destroy(scanner_);
924f67b2 JF	217	}