3 Routines for compiling Flash2 AVM2 ABC Actionscript
5 Extension module for the rfxswf library.
6 Part of the swftools package.
8 Copyright (c) 2008 Matthias Kramm <kramm@quiss.org>
10 This program is free software; you can redistribute it and/or modify
11 it under the terms of the GNU General Public License as published by
12 the Free Software Foundation; either version 2 of the License, or
13 (at your option) any later version.
15 This program is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 GNU General Public License for more details.
20 You should have received a copy of the GNU General Public License
21 along with this program; if not, write to the Free Software
22 Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
31 #include "tokenizer.h"
34 int as3_verbosity = 1;
35 void as3_error(const char*format, ...)
42 va_start(arglist, format);
43 vsprintf(buf, format, arglist);
45 fprintf(stderr, "%s:%d:%d: error: %s\n", current_filename_short, current_line, current_column, buf);
49 void as3_warning(const char*format, ...)
56 va_start(arglist, format);
57 vsprintf(buf, format, arglist);
59 fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
62 void as3_softwarning(const char*format, ...)
69 va_start(arglist, format);
70 vsprintf(buf, format, arglist);
72 fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
75 static void dbg(const char*format, ...)
82 va_start(arglist, format);
83 vsprintf(buf, format, arglist);
86 while(l && buf[l-1]=='\n') {
90 printf("(tokenizer) ");
97 #ifndef YY_CURRENT_BUFFER
98 #define YY_CURRENT_BUFFER yy_current_buffer
101 void handleInclude(char*text, int len, char quotes)
105 char*p1 = strchr(text, '"');
106 char*p2 = strrchr(text, '"');
107 if(!p1 || !p2 || p1==p2) {
108 syntaxerror("Invalid include in line %d\n", current_line);
111 filename = strdup(p1+1);
115 while(!strchr(" \n\r\t", text[i1])) i1++;
117 while(strchr(" \n\r\t", text[i1])) i1++;
118 while(strchr(" \n\r\t", text[i2-1])) i2--;
119 if(i2!=len) text[i2]=0;
120 filename = strdup(&text[i1]);
123 char*fullfilename = enter_file(filename, YY_CURRENT_BUFFER);
124 yyin = fopen(fullfilename, "rb");
126 syntaxerror("Couldn't open include file \"%s\"\n", fullfilename);
129 yy_switch_to_buffer(yy_create_buffer( yyin, YY_BUF_SIZE ) );
130 //BEGIN(INITIAL); keep context
133 static int do_unescape(const char*s, const char*end, char*n)
139 if(o) o[len] = *s;len++;
144 if(s==end) syntaxerror("invalid \\ at end of string");
146 /* handle the various line endings (mac, dos, unix) */
159 case '\\': if(o) o[len] = '\\';s++;len++; break;
160 case '"': if(o) o[len] = '"';s++;len++; break;
161 case '\'': if(o) o[len] = '\'';s++;len++; break;
162 case 'b': if(o) o[len] = '\b';s++;len++; break;
163 case 'f': if(o) o[len] = '\f';s++;len++; break;
164 case 'n': if(o) o[len] = '\n';s++;len++; break;
165 case 'r': if(o) o[len] = '\r';s++;len++; break;
166 case 't': if(o) o[len] = '\t';s++;len++; break;
167 case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': {
170 while(strchr("01234567", *s) && nr<3 && s<end) {
177 syntaxerror("octal number out of range (0-255): %d", num);
178 if(o) o[len] = num;len++;
181 case 'x': case 'u': {
190 if(s==end) syntaxerror("invalid \\u or \\x at end of string");
193 if(s==end) syntaxerror("invalid \\u{ at end of string");
198 while(strchr("0123456789abcdefABCDEF", *s) && (bracket || nr < max) && s<end) {
200 if(*s>='0' && *s<='9') num |= *s - '0';
201 if(*s>='a' && *s<='f') num |= *s - 'a' + 10;
202 if(*s>='A' && *s<='F') num |= *s - 'A' + 10;
207 if(*s=='}' && s<end) {
210 syntaxerror("missing terminating '}'");
214 char*utf8 = getUTF8(num);
216 if(o) o[len] = *utf8;utf8++;len++;
220 syntaxerror("byte out of range (0-255): %d", num);
221 if(o) o[len] = num;len++;
226 syntaxerror("unknown escape sequence: \"\\%c\"", *s);
233 static string_t string_unescape(const char*in, int l)
236 const char*end = &in[l];
238 int len = do_unescape(s, end, 0);
239 char*n = (char*)malloc(len+1);
240 do_unescape(s, end, n);
241 string_t out = string_new(n, len);
245 static void handleString(char*s, int len)
248 if(s[len-1]!='"') syntaxerror("String doesn't end with '\"'");
251 else if(s[0]=='\'') {
252 if(s[len-1]!='\'') syntaxerror("String doesn't end with '\"'");
255 else syntaxerror("String incorrectly terminated");
258 avm2_lval.str = string_unescape(s, len);
262 char start_of_expression;
264 static inline int mkid(int type)
266 char*s = malloc(yyleng+1);
267 memcpy(s, yytext, yyleng);
273 static inline int m(int type)
275 avm2_lval.token = type;
280 static char numberbuf[64];
283 if(yyleng>sizeof(numberbuf)-1)
284 syntaxerror("decimal number overflow");
286 memcpy(s, yytext, yyleng);
291 static inline int setint(int v)
293 avm2_lval.number_int = v;
301 static inline int setuint(unsigned int v)
303 avm2_lval.number_uint = v;
311 static inline int setfloat(double v)
313 avm2_lval.number_float = v;
317 static inline int handlefloat()
320 avm2_lval.number_float = atof(s);
324 static inline int handleint()
327 char l = (yytext[0]=='-');
329 char*max = l?"1073741824":"2147483647";
331 as3_warning("integer overflow: %s (converted to Number)", s);
332 return handlefloat();
336 for(t=0;t<yyleng-l;t++) {
337 if(yytext[l+t]>max[t]) {
338 as3_warning("integer overflow: %s (converted to Number)", s);
339 return handlefloat();
341 else if(yytext[l+t]<max[t])
351 for(t=0;t<yyleng;t++) {
359 static inline int handlehexfloat()
361 char l = (yytext[0]=='-')+2;
366 for(t=l;t<yyleng;t++) {
379 else if((c>='a' && c<='f') || (c>='A' && c<='F'))
380 d+=((c&0x0f)+9)*base;
384 static inline int handlehex()
386 char l = (yytext[0]=='-')+2;
391 syntaxerror("integer overflow %s", s);
401 else if((c>='a' && c<='f') || (c>='A' && c<='F'))
404 if(l && v>1073741824) {
406 as3_warning("signed integer overflow: %s (converted to Number)", s);
409 if(!l && v>2147483647) {
411 as3_warning("unsigned integer overflow: %s (converted to Number)", s);
416 return setint(-(int)v);
422 void handleLabel(char*text, int len)
425 for(t=len-1;t>=0;--t) {
430 char*s = malloc(t+1);
431 memcpy(s, yytext, t);
436 static int handleregexp()
438 char*s = malloc(yyleng);
440 memcpy(s, yytext+1, len);
443 for(t=len;t>=0;--t) {
449 avm2_lval.regexp.pattern = s;
451 avm2_lval.regexp.options = 0;
453 avm2_lval.regexp.options = s+t+1;
458 void initialize_scanner();
459 #define YY_USER_INIT initialize_scanner();
461 /* count the number of lines+columns consumed by this token */
462 static inline void l() {
464 for(t=0;t<yyleng;t++) {
465 if(yytext[t]=='\n') {
473 /* count the number of columns consumed by this token */
474 static inline void c() {
475 current_column+=yyleng;
478 //Boolean {c();return m(KW_BOOLEAN);}
479 //int {c();return m(KW_INT);}
480 //uint {c();return m(KW_UINT);}
481 //Number {c();return m(KW_NUMBER);}
489 NAME [a-zA-Z_][a-zA-Z0-9_\\]*
492 HEXINT 0x[a-zA-Z0-9]+
493 HEXFLOAT 0x[a-zA-Z0-9]*\.[a-zA-Z0-9]*
495 FLOAT [0-9]+(\.[0-9]*)?|\.[0-9]+
497 HEXWITHSIGN [+-]?({HEXINT})
498 HEXFLOATWITHSIGN [+-]?({HEXFLOAT})
499 INTWITHSIGN [+-]?({INT})
500 FLOATWITHSIGN [+-]?({FLOAT})
502 STRING ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*[']
504 MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[^*][/]|[\x00-\x1f])*[*]+[/]
505 SINGLELINE_COMMENT \/\/[^\n]*\n
506 REGEXP [/]([^/\n]|\\[/])*[/][a-zA-Z]*
510 {SINGLELINE_COMMENT} {l(); /* single line comment */}
511 {MULTILINE_COMMENT} {l(); /* multi line comment */}
512 [/][*] {syntaxerror("syntax error: unterminated comment", yytext);}
514 ^include{S}+{STRING}{S}*/\n {l();handleInclude(yytext, yyleng, 1);}
515 ^include{S}+[^" \t\r\n][\x20-\xff]*{S}*/\n {l();handleInclude(yytext, yyleng, 0);}
516 {STRING} {l(); BEGIN(INITIAL);handleString(yytext, yyleng);return T_STRING;}
518 <BEGINNING,REGEXPOK>{
519 {REGEXP} {c(); BEGIN(INITIAL);return handleregexp();}
520 {HEXWITHSIGN} {c(); BEGIN(INITIAL);return handlehex();}
521 {HEXFLOATWITHSIGN} {c(); BEGIN(INITIAL);return handlehexfloat();}
522 {INTWITHSIGN} {c(); BEGIN(INITIAL);return handleint();}
523 {FLOATWITHSIGN} {c(); BEGIN(INITIAL);return handlefloat();}
526 \xef\xbb\xbf {/* utf 8 bom */}
529 {HEXINT} {c(); BEGIN(INITIAL);return handlehex();}
530 {HEXFLOAT} {c(); BEGIN(INITIAL);return handlehexfloat();}
531 {INT} {c(); BEGIN(INITIAL);return handleint();}
532 {FLOAT} {c(); BEGIN(INITIAL);return handlefloat();}
534 3rr0r {/* for debugging: generates a tokenizer-level error */
535 syntaxerror("3rr0r");}
537 {NAME}{S}*:{S}*for/{_} {l();handleLabel(yytext, yyleng-3);return T_FOR;}
538 {NAME}{S}*:{S}*do/{_} {l();handleLabel(yytext, yyleng-2);return T_DO;}
539 {NAME}{S}*:{S}*while/{_} {l();handleLabel(yytext, yyleng-5);return T_WHILE;}
540 {NAME}{S}*:{S}*switch/{_} {l();handleLabel(yytext, yyleng-6);return T_SWITCH;}
541 for {c();avm2_lval.id="";return T_FOR;}
542 do {c();avm2_lval.id="";return T_DO;}
543 while {c();avm2_lval.id="";return T_WHILE;}
544 switch {c();avm2_lval.id="";return T_SWITCH;}
546 [&][&] {c();BEGIN(REGEXPOK);return m(T_ANDAND);}
547 [|][|] {c();BEGIN(REGEXPOK);return m(T_OROR);}
548 [!][=] {c();BEGIN(REGEXPOK);return m(T_NE);}
549 [!][=][=] {c();BEGIN(REGEXPOK);return m(T_NEE);}
550 [=][=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQEQ);}
551 [=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQ);}
552 [>][=] {c();return m(T_GE);}
553 [<][=] {c();return m(T_LE);}
554 [-][-] {c();BEGIN(INITIAL);return m(T_MINUSMINUS);}
555 [+][+] {c();BEGIN(INITIAL);return m(T_PLUSPLUS);}
556 [+][=] {c();return m(T_PLUSBY);}
557 [-][=] {c();return m(T_MINUSBY);}
558 [/][=] {c();return m(T_DIVBY);}
559 [%][=] {c();return m(T_MODBY);}
560 [*][=] {c();return m(T_MULBY);}
561 [|][=] {c();return m(T_ORBY);}
562 [>][>][=] {c();return m(T_SHRBY);}
563 [<][<][=] {c();return m(T_SHLBY);}
564 [>][>][>][=] {c();return m(T_USHRBY);}
565 [<][<] {c();return m(T_SHL);}
566 [>][>][>] {c();return m(T_USHR);}
567 [>][>] {c();return m(T_SHR);}
568 \.\.\. {c();return m(T_DOTDOTDOT);}
569 \.\. {c();return m(T_DOTDOT);}
570 \. {c();return m('.');}
571 :: {c();return m(T_COLONCOLON);}
572 : {c();return m(':');}
573 instanceof {c();return m(KW_INSTANCEOF);}
574 implements {c();return m(KW_IMPLEMENTS);}
575 interface {c();return m(KW_INTERFACE);}
576 namespace {c();return m(KW_NAMESPACE);}
577 protected {c();return m(KW_PROTECTED);}
578 undefined {c();return m(KW_UNDEFINED);}
579 continue {c();return m(KW_CONTINUE);}
580 override {c();return m(KW_OVERRIDE);}
581 internal {c();return m(KW_INTERNAL);}
582 function {c();return m(KW_FUNCTION);}
583 finally {c();return m(KW_FINALLY);}
584 default {c();return m(KW_DEFAULT);}
585 package {c();return m(KW_PACKAGE);}
586 private {c();return m(KW_PRIVATE);}
587 dynamic {c();return m(KW_DYNAMIC);}
588 extends {c();return m(KW_EXTENDS);}
589 delete {c();return m(KW_DELETE);}
590 return {c();return m(KW_RETURN);}
591 public {c();return m(KW_PUBLIC);}
592 native {c();return m(KW_NATIVE);}
593 static {c();return m(KW_STATIC);}
594 import {c();return m(KW_IMPORT);}
595 typeof {c();return m(KW_TYPEOF);}
596 throw {c();return m(KW_THROW);}
597 class {c();return m(KW_CLASS);}
598 const {c();return m(KW_CONST);}
599 catch {c();return m(KW_CATCH);}
600 final {c();return m(KW_FINAL);}
601 false {c();return m(KW_FALSE);}
602 break {c();return m(KW_BREAK);}
603 super {c();return m(KW_SUPER);}
604 each {c();return m(KW_EACH);}
605 void {c();return m(KW_VOID);}
606 true {c();return m(KW_TRUE);}
607 null {c();return m(KW_NULL);}
608 else {c();return m(KW_ELSE);}
609 case {c();return m(KW_CASE);}
610 with {c();return m(KW_WITH);}
611 use {c();return m(KW_USE);}
612 new {c();return m(KW_NEW);}
613 get {c();return m(KW_GET);}
614 set {c();return m(KW_SET);}
615 var {c();return m(KW_VAR);}
616 try {c();return m(KW_TRY);}
617 is {c();return m(KW_IS) ;}
618 in {c();return m(KW_IN) ;}
619 if {c();return m(KW_IF) ;}
620 as {c();return m(KW_AS);}
621 {NAME} {c();BEGIN(INITIAL);return mkid(T_IDENTIFIER);}
623 [+-\/*^~@$!%&\(=\[\]\{\}|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
624 [\)\]] {c();BEGIN(INITIAL);return m(yytext[0]);}
626 . {char c1=yytext[0];
631 char c = buf[t]=input();
632 if(c=='\n' || c==EOF) {
637 if(c1>='0' && c1<='9')
638 syntaxerror("syntax error: %s (identifiers must not start with a digit)");
640 syntaxerror("syntax error: %s", buf);
646 void*b = leave_file();
649 yy_delete_buffer(YY_CURRENT_BUFFER);
652 yy_delete_buffer(YY_CURRENT_BUFFER);
653 yy_switch_to_buffer(b);
664 static char mbuf[256];
665 char*token2string(enum yytokentype nr, YYSTYPE v)
667 if(nr==T_STRING) return "<string>";
668 else if(nr==T_INT) return "<int>";
669 else if(nr==T_UINT) return "<uint>";
670 else if(nr==T_BYTE) return "<byte>";
671 else if(nr==T_FLOAT) return "<float>";
672 else if(nr==T_REGEXP) return "REGEXP";
673 else if(nr==T_EOF) return "***END***";
674 else if(nr==T_GE) return ">=";
675 else if(nr==T_LE) return "<=";
676 else if(nr==T_MINUSMINUS) return "--";
677 else if(nr==T_PLUSPLUS) return "++";
678 else if(nr==KW_IMPLEMENTS) return "implements";
679 else if(nr==KW_INTERFACE) return "interface";
680 else if(nr==KW_NAMESPACE) return "namespace";
681 else if(nr==KW_PROTECTED) return "protected";
682 else if(nr==KW_OVERRIDE) return "override";
683 else if(nr==KW_INTERNAL) return "internal";
684 else if(nr==KW_FUNCTION) return "function";
685 else if(nr==KW_PACKAGE) return "package";
686 else if(nr==KW_PRIVATE) return "private";
687 else if(nr==KW_BOOLEAN) return "Boolean";
688 else if(nr==KW_DYNAMIC) return "dynamic";
689 else if(nr==KW_EXTENDS) return "extends";
690 else if(nr==KW_PUBLIC) return "public";
691 else if(nr==KW_NATIVE) return "native";
692 else if(nr==KW_STATIC) return "static";
693 else if(nr==KW_IMPORT) return "import";
694 else if(nr==KW_NUMBER) return "number";
695 else if(nr==KW_CLASS) return "class";
696 else if(nr==KW_CONST) return "const";
697 else if(nr==KW_FINAL) return "final";
698 else if(nr==KW_FALSE) return "False";
699 else if(nr==KW_TRUE) return "True";
700 else if(nr==KW_UINT) return "uint";
701 else if(nr==KW_NULL) return "null";
702 else if(nr==KW_ELSE) return "else";
703 else if(nr==KW_USE) return "use";
704 else if(nr==KW_INT) return "int";
705 else if(nr==KW_NEW) return "new";
706 else if(nr==KW_GET) return "get";
707 else if(nr==KW_SET) return "set";
708 else if(nr==KW_VAR) return "var";
709 else if(nr==KW_IS) return "is";
710 else if(nr==KW_AS) return "as";
711 else if(nr==T_IDENTIFIER) return "ID";
713 sprintf(mbuf, "%d", nr);
718 void initialize_scanner()