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"
36 int as3_verbosity = 1;
37 void as3_error(const char*format, ...)
44 va_start(arglist, format);
45 vsprintf(buf, format, arglist);
47 fprintf(stderr, "%s:%d:%d: error: %s\n", current_filename_short, current_line, current_column, buf);
51 void as3_warning(const char*format, ...)
58 va_start(arglist, format);
59 vsprintf(buf, format, arglist);
61 fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
64 void as3_softwarning(const char*format, ...)
71 va_start(arglist, format);
72 vsprintf(buf, format, arglist);
74 fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
77 static void dbg(const char*format, ...)
84 va_start(arglist, format);
85 vsprintf(buf, format, arglist);
88 while(l && buf[l-1]=='\n') {
92 printf("(tokenizer) ");
99 #ifndef YY_CURRENT_BUFFER
100 #define YY_CURRENT_BUFFER yy_current_buffer
103 void handleInclude(char*text, int len, char quotes)
107 char*p1 = strchr(text, '"');
108 char*p2 = strrchr(text, '"');
109 if(!p1 || !p2 || p1==p2) {
110 syntaxerror("Invalid include in line %d\n", current_line);
113 filename = strdup(p1+1);
117 while(!strchr(" \n\r\t", text[i1])) i1++;
119 while(strchr(" \n\r\t", text[i1])) i1++;
120 while(strchr(" \n\r\t", text[i2-1])) i2--;
121 if(i2!=len) text[i2]=0;
122 filename = strdup(&text[i1]);
125 char*fullfilename = enter_file(filename, YY_CURRENT_BUFFER);
126 yyin = fopen(fullfilename, "rb");
128 syntaxerror("Couldn't open include file \"%s\"\n", fullfilename);
131 yy_switch_to_buffer(yy_create_buffer( yyin, YY_BUF_SIZE ) );
132 //BEGIN(INITIAL); keep context
135 static int do_unescape(const char*s, const char*end, char*n)
141 if(o) o[len] = *s;len++;
146 if(s==end) syntaxerror("invalid \\ at end of string");
148 /* handle the various line endings (mac, dos, unix) */
161 case '\\': if(o) o[len] = '\\';s++;len++; break;
162 case '"': if(o) o[len] = '"';s++;len++; break;
163 case '\'': if(o) o[len] = '\'';s++;len++; break;
164 case 'b': if(o) o[len] = '\b';s++;len++; break;
165 case 'f': if(o) o[len] = '\f';s++;len++; break;
166 case 'n': if(o) o[len] = '\n';s++;len++; break;
167 case 'r': if(o) o[len] = '\r';s++;len++; break;
168 case 't': if(o) o[len] = '\t';s++;len++; break;
169 case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': {
172 while(strchr("01234567", *s) && nr<3 && s<end) {
179 syntaxerror("octal number out of range (0-255): %d", num);
180 if(o) o[len] = num;len++;
183 case 'x': case 'u': {
192 if(s==end) syntaxerror("invalid \\u or \\x at end of string");
195 if(s==end) syntaxerror("invalid \\u{ at end of string");
200 while(strchr("0123456789abcdefABCDEF", *s) && (bracket || nr < max) && s<end) {
202 if(*s>='0' && *s<='9') num |= *s - '0';
203 if(*s>='a' && *s<='f') num |= *s - 'a' + 10;
204 if(*s>='A' && *s<='F') num |= *s - 'A' + 10;
209 if(*s=='}' && s<end) {
212 syntaxerror("missing terminating '}'");
216 char*utf8 = getUTF8(num);
218 if(o) o[len] = *utf8;utf8++;len++;
222 syntaxerror("byte out of range (0-255): %d", num);
223 if(o) o[len] = num;len++;
228 syntaxerror("unknown escape sequence: \"\\%c\"", *s);
235 static string_t string_unescape(const char*in, int l)
238 const char*end = &in[l];
240 int len = do_unescape(s, end, 0);
241 char*n = (char*)malloc(len+1);
242 do_unescape(s, end, n);
243 string_t out = string_new(n, len);
247 static void handleString(char*s, int len)
250 if(s[len-1]!='"') syntaxerror("String doesn't end with '\"'");
253 else if(s[0]=='\'') {
254 if(s[len-1]!='\'') syntaxerror("String doesn't end with '\"'");
257 else syntaxerror("String incorrectly terminated");
260 a3_lval.str = string_unescape(s, len);
264 char start_of_expression;
266 static inline int mkid(int type)
268 char*s = malloc(yyleng+1);
269 memcpy(s, yytext, yyleng);
275 static inline int m(int type)
277 a3_lval.token = type;
282 static char numberbuf[64];
285 if(yyleng>sizeof(numberbuf)-1)
286 syntaxerror("decimal number overflow");
288 memcpy(s, yytext, yyleng);
293 static inline int setint(int v)
295 a3_lval.number_int = v;
303 static inline int setuint(unsigned int v)
305 a3_lval.number_uint = v;
313 static inline int setfloat(double v)
315 a3_lval.number_float = v;
319 static inline int handlefloat()
322 a3_lval.number_float = atof(s);
326 static inline int handleint()
329 char l = (yytext[0]=='-');
331 char*max = l?"1073741824":"2147483647";
333 as3_warning("integer overflow: %s (converted to Number)", s);
334 return handlefloat();
338 for(t=0;t<yyleng-l;t++) {
339 if(yytext[l+t]>max[t]) {
340 as3_warning("integer overflow: %s (converted to Number)", s);
341 return handlefloat();
343 else if(yytext[l+t]<max[t])
353 for(t=0;t<yyleng;t++) {
361 static inline int handlehexfloat()
363 char l = (yytext[0]=='-')+2;
368 for(t=l;t<yyleng;t++) {
381 else if((c>='a' && c<='f') || (c>='A' && c<='F'))
382 d+=((c&0x0f)+9)*base;
386 static inline int handlehex()
388 char l = (yytext[0]=='-')+2;
393 syntaxerror("integer overflow %s", s);
403 else if((c>='a' && c<='f') || (c>='A' && c<='F'))
406 if(l && v>1073741824) {
408 as3_warning("signed integer overflow: %s (converted to Number)", s);
411 if(!l && v>2147483647) {
413 as3_warning("unsigned integer overflow: %s (converted to Number)", s);
418 return setint(-(int)v);
424 void handleLabel(char*text, int len)
427 for(t=len-1;t>=0;--t) {
432 char*s = malloc(t+1);
433 memcpy(s, yytext, t);
438 static int handleregexp()
440 char*s = malloc(yyleng);
442 memcpy(s, yytext+1, len);
445 for(t=len;t>=0;--t) {
451 a3_lval.regexp.pattern = s;
453 a3_lval.regexp.options = 0;
455 a3_lval.regexp.options = s+t+1;
460 void initialize_scanner();
461 #define YY_USER_INIT initialize_scanner();
463 /* count the number of lines+columns consumed by this token */
464 static inline void l() {
466 for(t=0;t<yyleng;t++) {
467 if(yytext[t]=='\n') {
475 /* count the number of columns consumed by this token */
476 static inline void c() {
477 current_column+=yyleng;
480 //Boolean {c();return m(KW_BOOLEAN);}
481 //int {c();return m(KW_INT);}
482 //uint {c();return m(KW_UINT);}
483 //Number {c();return m(KW_NUMBER);}
491 NAME [a-zA-Z_][a-zA-Z0-9_\\]*
494 HEXINT 0x[a-zA-Z0-9]+
495 HEXFLOAT 0x[a-zA-Z0-9]*\.[a-zA-Z0-9]*
497 FLOAT [0-9]+(\.[0-9]*)?|\.[0-9]+
499 HEXWITHSIGN [+-]?({HEXINT})
500 HEXFLOATWITHSIGN [+-]?({HEXFLOAT})
501 INTWITHSIGN [+-]?({INT})
502 FLOATWITHSIGN [+-]?({FLOAT})
504 STRING ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*[']
506 MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[^*][/]|[\x00-\x1f])*[*]+[/]
507 SINGLELINE_COMMENT \/\/[^\n]*\n
508 REGEXP [/]([^/\n]|\\[/])*[/][a-zA-Z]*
512 {SINGLELINE_COMMENT} {l(); /* single line comment */}
513 {MULTILINE_COMMENT} {l(); /* multi line comment */}
514 [/][*] {syntaxerror("syntax error: unterminated comment", yytext);}
516 ^include{S}+{STRING}{S}*/\n {l();handleInclude(yytext, yyleng, 1);}
517 ^include{S}+[^" \t\r\n][\x20-\xff]*{S}*/\n {l();handleInclude(yytext, yyleng, 0);}
518 {STRING} {l(); BEGIN(INITIAL);handleString(yytext, yyleng);return T_STRING;}
520 <BEGINNING,REGEXPOK>{
521 {REGEXP} {c(); BEGIN(INITIAL);return handleregexp();}
522 {HEXWITHSIGN} {c(); BEGIN(INITIAL);return handlehex();}
523 {HEXFLOATWITHSIGN} {c(); BEGIN(INITIAL);return handlehexfloat();}
524 {INTWITHSIGN} {c(); BEGIN(INITIAL);return handleint();}
525 {FLOATWITHSIGN} {c(); BEGIN(INITIAL);return handlefloat();}
528 \xef\xbb\xbf {/* utf 8 bom */}
531 {HEXINT} {c(); BEGIN(INITIAL);return handlehex();}
532 {HEXFLOAT} {c(); BEGIN(INITIAL);return handlehexfloat();}
533 {INT} {c(); BEGIN(INITIAL);return handleint();}
534 {FLOAT} {c(); BEGIN(INITIAL);return handlefloat();}
536 3rr0r {/* for debugging: generates a tokenizer-level error */
537 syntaxerror("3rr0r");}
539 {NAME}{S}*:{S}*for/{_} {l();handleLabel(yytext, yyleng-3);return T_FOR;}
540 {NAME}{S}*:{S}*do/{_} {l();handleLabel(yytext, yyleng-2);return T_DO;}
541 {NAME}{S}*:{S}*while/{_} {l();handleLabel(yytext, yyleng-5);return T_WHILE;}
542 {NAME}{S}*:{S}*switch/{_} {l();handleLabel(yytext, yyleng-6);return T_SWITCH;}
543 for {c();a3_lval.id="";return T_FOR;}
544 do {c();a3_lval.id="";return T_DO;}
545 while {c();a3_lval.id="";return T_WHILE;}
546 switch {c();a3_lval.id="";return T_SWITCH;}
548 [&][&] {c();BEGIN(REGEXPOK);return m(T_ANDAND);}
549 [|][|] {c();BEGIN(REGEXPOK);return m(T_OROR);}
550 [!][=] {c();BEGIN(REGEXPOK);return m(T_NE);}
551 [!][=][=] {c();BEGIN(REGEXPOK);return m(T_NEE);}
552 [=][=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQEQ);}
553 [=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQ);}
554 [>][=] {c();return m(T_GE);}
555 [<][=] {c();return m(T_LE);}
556 [-][-] {c();BEGIN(INITIAL);return m(T_MINUSMINUS);}
557 [+][+] {c();BEGIN(INITIAL);return m(T_PLUSPLUS);}
558 [+][=] {c();return m(T_PLUSBY);}
559 [-][=] {c();return m(T_MINUSBY);}
560 [/][=] {c();return m(T_DIVBY);}
561 [%][=] {c();return m(T_MODBY);}
562 [*][=] {c();return m(T_MULBY);}
563 [|][=] {c();return m(T_ORBY);}
564 [>][>][=] {c();return m(T_SHRBY);}
565 [<][<][=] {c();return m(T_SHLBY);}
566 [>][>][>][=] {c();return m(T_USHRBY);}
567 [<][<] {c();return m(T_SHL);}
568 [>][>][>] {c();return m(T_USHR);}
569 [>][>] {c();return m(T_SHR);}
570 \.\.\. {c();return m(T_DOTDOTDOT);}
571 \.\. {c();return m(T_DOTDOT);}
572 \. {c();return m('.');}
573 :: {c();return m(T_COLONCOLON);}
574 : {c();return m(':');}
575 instanceof {c();return m(KW_INSTANCEOF);}
576 implements {c();return m(KW_IMPLEMENTS);}
577 interface {c();return m(KW_INTERFACE);}
578 namespace {c();return m(KW_NAMESPACE);}
579 protected {c();return m(KW_PROTECTED);}
580 undefined {c();return m(KW_UNDEFINED);}
581 continue {c();return m(KW_CONTINUE);}
582 override {c();return m(KW_OVERRIDE);}
583 internal {c();return m(KW_INTERNAL);}
584 function {c();return m(KW_FUNCTION);}
585 finally {c();return m(KW_FINALLY);}
586 default {c();return m(KW_DEFAULT);}
587 package {c();return m(KW_PACKAGE);}
588 private {c();return m(KW_PRIVATE);}
589 dynamic {c();return m(KW_DYNAMIC);}
590 extends {c();return m(KW_EXTENDS);}
591 delete {c();return m(KW_DELETE);}
592 return {c();return m(KW_RETURN);}
593 public {c();return m(KW_PUBLIC);}
594 native {c();return m(KW_NATIVE);}
595 static {c();return m(KW_STATIC);}
596 import {c();return m(KW_IMPORT);}
597 typeof {c();return m(KW_TYPEOF);}
598 throw {c();return m(KW_THROW);}
599 class {c();return m(KW_CLASS);}
600 const {c();return m(KW_CONST);}
601 catch {c();return m(KW_CATCH);}
602 final {c();return m(KW_FINAL);}
603 false {c();return m(KW_FALSE);}
604 break {c();return m(KW_BREAK);}
605 super {c();return m(KW_SUPER);}
606 each {c();return m(KW_EACH);}
607 void {c();return m(KW_VOID);}
608 true {c();return m(KW_TRUE);}
609 null {c();return m(KW_NULL);}
610 else {c();return m(KW_ELSE);}
611 case {c();return m(KW_CASE);}
612 with {c();return m(KW_WITH);}
613 use {c();return m(KW_USE);}
614 new {c();return m(KW_NEW);}
615 get {c();return m(KW_GET);}
616 set {c();return m(KW_SET);}
617 var {c();return m(KW_VAR);}
618 try {c();return m(KW_TRY);}
619 is {c();return m(KW_IS) ;}
620 in {c();return m(KW_IN) ;}
621 if {c();return m(KW_IF) ;}
622 as {c();return m(KW_AS);}
623 {NAME} {c();BEGIN(INITIAL);return mkid(T_IDENTIFIER);}
625 [+-\/*^~@$!%&\(=\[\]\{\}|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
626 [\)\]] {c();BEGIN(INITIAL);return m(yytext[0]);}
628 . {char c1=yytext[0];
633 char c = buf[t]=input();
634 if(c=='\n' || c==EOF) {
639 if(c1>='0' && c1<='9')
640 syntaxerror("syntax error: %s (identifiers must not start with a digit)");
642 syntaxerror("syntax error: %s", buf);
648 void*b = leave_file();
651 yy_delete_buffer(YY_CURRENT_BUFFER);
654 yy_delete_buffer(YY_CURRENT_BUFFER);
655 yy_switch_to_buffer(b);
666 static char mbuf[256];
667 char*token2string(enum yytokentype nr, YYSTYPE v)
669 if(nr==T_STRING) return "<string>";
670 else if(nr==T_INT) return "<int>";
671 else if(nr==T_UINT) return "<uint>";
672 else if(nr==T_BYTE) return "<byte>";
673 else if(nr==T_FLOAT) return "<float>";
674 else if(nr==T_REGEXP) return "REGEXP";
675 else if(nr==T_EOF) return "***END***";
676 else if(nr==T_GE) return ">=";
677 else if(nr==T_LE) return "<=";
678 else if(nr==T_MINUSMINUS) return "--";
679 else if(nr==T_PLUSPLUS) return "++";
680 else if(nr==KW_IMPLEMENTS) return "implements";
681 else if(nr==KW_INTERFACE) return "interface";
682 else if(nr==KW_NAMESPACE) return "namespace";
683 else if(nr==KW_PROTECTED) return "protected";
684 else if(nr==KW_OVERRIDE) return "override";
685 else if(nr==KW_INTERNAL) return "internal";
686 else if(nr==KW_FUNCTION) return "function";
687 else if(nr==KW_PACKAGE) return "package";
688 else if(nr==KW_PRIVATE) return "private";
689 else if(nr==KW_BOOLEAN) return "Boolean";
690 else if(nr==KW_DYNAMIC) return "dynamic";
691 else if(nr==KW_EXTENDS) return "extends";
692 else if(nr==KW_PUBLIC) return "public";
693 else if(nr==KW_NATIVE) return "native";
694 else if(nr==KW_STATIC) return "static";
695 else if(nr==KW_IMPORT) return "import";
696 else if(nr==KW_NUMBER) return "number";
697 else if(nr==KW_CLASS) return "class";
698 else if(nr==KW_CONST) return "const";
699 else if(nr==KW_FINAL) return "final";
700 else if(nr==KW_FALSE) return "False";
701 else if(nr==KW_TRUE) return "True";
702 else if(nr==KW_UINT) return "uint";
703 else if(nr==KW_NULL) return "null";
704 else if(nr==KW_ELSE) return "else";
705 else if(nr==KW_USE) return "use";
706 else if(nr==KW_INT) return "int";
707 else if(nr==KW_NEW) return "new";
708 else if(nr==KW_GET) return "get";
709 else if(nr==KW_SET) return "set";
710 else if(nr==KW_VAR) return "var";
711 else if(nr==KW_IS) return "is";
712 else if(nr==KW_AS) return "as";
713 else if(nr==T_IDENTIFIER) return "ID";
715 sprintf(mbuf, "%d", nr);
720 void initialize_scanner()