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 static int verbose = 1;
35 static void dbg(const char*format, ...)
42 va_start(arglist, format);
43 vsprintf(buf, format, arglist);
46 while(l && buf[l-1]=='\n') {
50 printf("(tokenizer) ");
55 void syntaxerror(const char*format, ...)
62 va_start(arglist, format);
63 vsprintf(buf, format, arglist);
65 fprintf(stderr, "%s:%d:%d: error: %s\n", current_filename_short, current_line, current_column, buf);
69 void warning(const char*format, ...)
77 va_start(arglist, format);
78 vsprintf(buf, format, arglist);
80 fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
85 #ifndef YY_CURRENT_BUFFER
86 #define YY_CURRENT_BUFFER yy_current_buffer
89 void handleInclude(char*text, int len, char quotes)
93 char*p1 = strchr(text, '"');
94 char*p2 = strrchr(text, '"');
95 if(!p1 || !p2 || p1==p2) {
96 syntaxerror("Invalid include in line %d\n", current_line);
99 filename = strdup(p1+1);
103 while(!strchr(" \n\r\t", text[i1])) i1++;
105 while(strchr(" \n\r\t", text[i1])) i1++;
106 while(strchr(" \n\r\t", text[i2-1])) i2--;
107 if(i2!=len) text[i2]=0;
108 filename = strdup(&text[i1]);
111 char*fullfilename = enter_file(filename, YY_CURRENT_BUFFER);
112 yyin = fopen(fullfilename, "rb");
114 syntaxerror("Couldn't open include file \"%s\"\n", fullfilename);
117 yy_switch_to_buffer(yy_create_buffer( yyin, YY_BUF_SIZE ) );
118 //BEGIN(INITIAL); keep context
121 static int do_unescape(const char*s, const char*end, char*n)
127 if(o) o[len] = *s;len++;
132 if(s==end) syntaxerror("invalid \\ at end of string");
134 /* handle the various line endings (mac, dos, unix) */
147 case '\\': if(o) o[len] = '\\';s++;len++; break;
148 case '"': if(o) o[len] = '"';s++;len++; break;
149 case '\'': if(o) o[len] = '\'';s++;len++; break;
150 case 'b': if(o) o[len] = '\b';s++;len++; break;
151 case 'f': if(o) o[len] = '\f';s++;len++; break;
152 case 'n': if(o) o[len] = '\n';s++;len++; break;
153 case 'r': if(o) o[len] = '\r';s++;len++; break;
154 case 't': if(o) o[len] = '\t';s++;len++; break;
155 case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': {
158 while(strchr("01234567", *s) && nr<3 && s<end) {
165 syntaxerror("octal number out of range (0-255): %d", num);
166 if(o) o[len] = num;len++;
169 case 'x': case 'u': {
178 if(s==end) syntaxerror("invalid \\u or \\x at end of string");
181 if(s==end) syntaxerror("invalid \\u{ at end of string");
186 while(strchr("0123456789abcdefABCDEF", *s) && (bracket || nr < max) && s<end) {
188 if(*s>='0' && *s<='9') num |= *s - '0';
189 if(*s>='a' && *s<='f') num |= *s - 'a' + 10;
190 if(*s>='A' && *s<='F') num |= *s - 'A' + 10;
195 if(*s=='}' && s<end) {
198 syntaxerror("missing terminating '}'");
202 char*utf8 = getUTF8(num);
204 if(o) o[len] = *utf8;utf8++;len++;
208 syntaxerror("byte out of range (0-255): %d", num);
209 if(o) o[len] = num;len++;
214 syntaxerror("unknown escape sequence: \"\\%c\"", *s);
221 static string_t string_unescape(const char*in, int l)
224 const char*end = &in[l];
226 int len = do_unescape(s, end, 0);
227 char*n = (char*)malloc(len+1);
228 do_unescape(s, end, n);
229 string_t out = string_new(n, len);
233 static void handleString(char*s, int len)
236 if(s[len-1]!='"') syntaxerror("String doesn't end with '\"'");
239 else if(s[0]=='\'') {
240 if(s[len-1]!='\'') syntaxerror("String doesn't end with '\"'");
243 else syntaxerror("String incorrectly terminated");
246 avm2_lval.str = string_unescape(s, len);
250 char start_of_expression;
252 static inline int mkid(int type)
254 char*s = malloc(yyleng+1);
255 memcpy(s, yytext, yyleng);
261 static inline int m(int type)
263 avm2_lval.token = type;
268 static char numberbuf[64];
271 if(yyleng>sizeof(numberbuf)-1)
272 syntaxerror("decimal number overflow");
274 memcpy(s, yytext, yyleng);
279 static inline int setint(int v)
281 avm2_lval.number_int = v;
289 static inline int setuint(unsigned int v)
291 avm2_lval.number_uint = v;
299 static inline int setfloat(double v)
301 avm2_lval.number_float = v;
305 static inline int handlefloat()
308 avm2_lval.number_float = atof(s);
312 static inline int handleint()
315 char l = (yytext[0]=='-');
317 char*max = l?"1073741824":"2147483647";
319 warning("integer overflow: %s", s);
320 return handlefloat();
324 for(t=0;t<yyleng-l;t++) {
325 if(yytext[l+t]>max[t]) {
326 warning("integer overflow: %s", s);
327 return handlefloat();
329 else if(yytext[l+t]<max[t])
339 for(t=0;t<yyleng;t++) {
347 static inline int handlehex()
349 char l = (yytext[0]=='-')+2;
353 syntaxerror("integer overflow %s", s);
358 for(t=l;t<yyleng;t++) {
363 else if(c>='a' && c<='f' ||
367 if(l && v>1073741824) {
369 warning("signed integer overflow: %s", s);
372 if(!l && v>2147483647) {
374 warning("unsigned integer overflow: %s", s);
379 return setint(-(int)v);
385 void handleLabel(char*text, int len)
388 for(t=len-1;t>=0;--t) {
393 char*s = malloc(t+1);
394 memcpy(s, yytext, t);
399 static int handleregexp()
401 char*s = malloc(yyleng);
403 memcpy(s, yytext+1, len);
406 for(t=len;t>=0;--t) {
412 avm2_lval.regexp.pattern = s;
414 avm2_lval.regexp.options = 0;
416 avm2_lval.regexp.options = s+t+1;
421 void initialize_scanner();
422 #define YY_USER_INIT initialize_scanner();
424 /* count the number of lines+columns consumed by this token */
425 static inline void l() {
427 for(t=0;t<yyleng;t++) {
428 if(yytext[t]=='\n') {
436 /* count the number of columns consumed by this token */
437 static inline void c() {
438 current_column+=yyleng;
441 //Boolean {c();return m(KW_BOOLEAN);}
442 //int {c();return m(KW_INT);}
443 //uint {c();return m(KW_UINT);}
444 //Number {c();return m(KW_NUMBER);}
452 NAME [a-zA-Z_][a-zA-Z0-9_\\]*
455 HEXINT 0x[a-zA-Z0-9]+
457 FLOAT [0-9]+(\.[0-9]*)?|\.[0-9]+
459 HEXWITHSIGN [+-]?({HEXINT})
460 INTWITHSIGN [+-]?({INT})
461 FLOATWITHSIGN [+-]?({FLOAT})
463 STRING ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*[']
465 MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[^*][/]|[\x00-\x1f])*[*]+[/]
466 SINGLELINE_COMMENT \/\/[^\n]*\n
467 REGEXP [/]([^/\n]|\\[/])*[/][a-zA-Z]*
471 {SINGLELINE_COMMENT} {l(); /* single line comment */}
472 {MULTILINE_COMMENT} {l(); /* multi line comment */}
473 [/][*] {syntaxerror("syntax error: unterminated comment", yytext);}
475 ^include{S}+{STRING}{S}*/\n {l();handleInclude(yytext, yyleng, 1);}
476 ^include{S}+[^" \t\r\n][\x20-\xff]*{S}*/\n {l();handleInclude(yytext, yyleng, 0);}
477 {STRING} {l(); BEGIN(INITIAL);handleString(yytext, yyleng);return T_STRING;}
479 <BEGINNING,REGEXPOK>{
480 {REGEXP} {c(); BEGIN(INITIAL);return handleregexp();}
481 {HEXWITHSIGN} {c(); BEGIN(INITIAL);return handlehex();}
482 {INTWITHSIGN} {c(); BEGIN(INITIAL);return handleint();}
483 {FLOATWITHSIGN} {c(); BEGIN(INITIAL);return handlefloat();}
486 \xef\xbb\xbf {/* utf 8 bom */}
489 {HEXINT} {c(); BEGIN(INITIAL);return handlehex();}
490 {INT} {c(); BEGIN(INITIAL);return handleint();}
491 {FLOAT} {c(); BEGIN(INITIAL);return handlefloat();}
493 3rr0r {/* for debugging: generates a tokenizer-level error */
494 syntaxerror("3rr0r");}
496 {NAME}{S}*:{S}*for/{_} {l();handleLabel(yytext, yyleng-3);return T_FOR;}
497 {NAME}{S}*:{S}*do/{_} {l();handleLabel(yytext, yyleng-2);return T_DO;}
498 {NAME}{S}*:{S}*while/{_} {l();handleLabel(yytext, yyleng-5);return T_WHILE;}
499 {NAME}{S}*:{S}*switch/{_} {l();handleLabel(yytext, yyleng-6);return T_SWITCH;}
500 for {c();avm2_lval.id="";return T_FOR;}
501 do {c();avm2_lval.id="";return T_DO;}
502 while {c();avm2_lval.id="";return T_WHILE;}
503 switch {c();avm2_lval.id="";return T_SWITCH;}
505 [&][&] {c();BEGIN(REGEXPOK);return m(T_ANDAND);}
506 [|][|] {c();BEGIN(REGEXPOK);return m(T_OROR);}
507 [!][=] {c();BEGIN(REGEXPOK);return m(T_NE);}
508 [!][=][=] {c();BEGIN(REGEXPOK);return m(T_NEE);}
509 [=][=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQEQ);}
510 [=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQ);}
511 [>][=] {c();return m(T_GE);}
512 [<][=] {c();return m(T_LE);}
513 [-][-] {c();BEGIN(INITIAL);return m(T_MINUSMINUS);}
514 [+][+] {c();BEGIN(INITIAL);return m(T_PLUSPLUS);}
515 [+][=] {c();return m(T_PLUSBY);}
516 [-][=] {c();return m(T_MINUSBY);}
517 [/][=] {c();return m(T_DIVBY);}
518 [%][=] {c();return m(T_MODBY);}
519 [*][=] {c();return m(T_MULBY);}
520 [|][=] {c();return m(T_ORBY);}
521 [>][>][=] {c();return m(T_SHRBY);}
522 [<][<][=] {c();return m(T_SHLBY);}
523 [>][>][>][=] {c();return m(T_USHRBY);}
524 [<][<] {c();return m(T_SHL);}
525 [>][>][>] {c();return m(T_USHR);}
526 [>][>] {c();return m(T_SHR);}
527 \.\.\. {c();return m(T_DOTDOTDOT);}
528 \.\. {c();return m(T_DOTDOT);}
529 \. {c();return m('.');}
530 :: {c();return m(T_COLONCOLON);}
531 : {c();return m(':');}
532 instanceof {c();return m(KW_INSTANCEOF);}
533 implements {c();return m(KW_IMPLEMENTS);}
534 interface {c();return m(KW_INTERFACE);}
535 namespace {c();return m(KW_NAMESPACE);}
536 protected {c();return m(KW_PROTECTED);}
537 undefined {c();return m(KW_UNDEFINED);}
538 continue {c();return m(KW_CONTINUE);}
539 override {c();return m(KW_OVERRIDE);}
540 internal {c();return m(KW_INTERNAL);}
541 function {c();return m(KW_FUNCTION);}
542 default {c();return m(KW_DEFAULT);}
543 package {c();return m(KW_PACKAGE);}
544 private {c();return m(KW_PRIVATE);}
545 dynamic {c();return m(KW_DYNAMIC);}
546 extends {c();return m(KW_EXTENDS);}
547 delete {c();return m(KW_DELETE);}
548 return {c();return m(KW_RETURN);}
549 public {c();return m(KW_PUBLIC);}
550 native {c();return m(KW_NATIVE);}
551 static {c();return m(KW_STATIC);}
552 import {c();return m(KW_IMPORT);}
553 typeof {c();return m(KW_TYPEOF);}
554 throw {c();return m(KW_THROW);}
555 class {c();return m(KW_CLASS);}
556 const {c();return m(KW_CONST);}
557 catch {c();return m(KW_CATCH);}
558 final {c();return m(KW_FINAL);}
559 false {c();return m(KW_FALSE);}
560 break {c();return m(KW_BREAK);}
561 super {c();return m(KW_SUPER);}
562 each {c();return m(KW_EACH);}
563 void {c();return m(KW_VOID);}
564 true {c();return m(KW_TRUE);}
565 null {c();return m(KW_NULL);}
566 else {c();return m(KW_ELSE);}
567 case {c();return m(KW_CASE);}
568 with {c();return m(KW_WITH);}
569 use {c();return m(KW_USE);}
570 new {c();return m(KW_NEW);}
571 get {c();return m(KW_GET);}
572 set {c();return m(KW_SET);}
573 var {c();return m(KW_VAR);}
574 try {c();return m(KW_TRY);}
575 is {c();return m(KW_IS) ;}
576 in {c();return m(KW_IN) ;}
577 if {c();return m(KW_IF) ;}
578 as {c();return m(KW_AS);}
579 {NAME} {c();BEGIN(INITIAL);return mkid(T_IDENTIFIER);}
581 [+-\/*^~@$!%&\(=\[\]\{\}|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
582 [\)\]] {c();BEGIN(INITIAL);return m(yytext[0]);}
584 . {char c1=yytext[0];
589 char c = buf[t]=input();
590 if(c=='\n' || c==EOF) {
595 if(c1>='0' && c1<='9')
596 syntaxerror("syntax error: %s (identifiers must not start with a digit)");
598 syntaxerror("syntax error: %s", buf);
604 void*b = leave_file();
607 yy_delete_buffer(YY_CURRENT_BUFFER);
610 yy_delete_buffer(YY_CURRENT_BUFFER);
611 yy_switch_to_buffer(b);
622 static char mbuf[256];
623 char*token2string(enum yytokentype nr, YYSTYPE v)
625 if(nr==T_STRING) return "<string>";
626 else if(nr==T_INT) return "<int>";
627 else if(nr==T_UINT) return "<uint>";
628 else if(nr==T_BYTE) return "<byte>";
629 else if(nr==T_FLOAT) return "<float>";
630 else if(nr==T_REGEXP) return "REGEXP";
631 else if(nr==T_EOF) return "***END***";
632 else if(nr==T_GE) return ">=";
633 else if(nr==T_LE) return "<=";
634 else if(nr==T_MINUSMINUS) return "--";
635 else if(nr==T_PLUSPLUS) return "++";
636 else if(nr==KW_IMPLEMENTS) return "implements";
637 else if(nr==KW_INTERFACE) return "interface";
638 else if(nr==KW_NAMESPACE) return "namespace";
639 else if(nr==KW_PROTECTED) return "protected";
640 else if(nr==KW_OVERRIDE) return "override";
641 else if(nr==KW_INTERNAL) return "internal";
642 else if(nr==KW_FUNCTION) return "function";
643 else if(nr==KW_PACKAGE) return "package";
644 else if(nr==KW_PRIVATE) return "private";
645 else if(nr==KW_BOOLEAN) return "Boolean";
646 else if(nr==KW_DYNAMIC) return "dynamic";
647 else if(nr==KW_EXTENDS) return "extends";
648 else if(nr==KW_PUBLIC) return "public";
649 else if(nr==KW_NATIVE) return "native";
650 else if(nr==KW_STATIC) return "static";
651 else if(nr==KW_IMPORT) return "import";
652 else if(nr==KW_NUMBER) return "number";
653 else if(nr==KW_CLASS) return "class";
654 else if(nr==KW_CONST) return "const";
655 else if(nr==KW_FINAL) return "final";
656 else if(nr==KW_FALSE) return "False";
657 else if(nr==KW_TRUE) return "True";
658 else if(nr==KW_UINT) return "uint";
659 else if(nr==KW_NULL) return "null";
660 else if(nr==KW_ELSE) return "else";
661 else if(nr==KW_USE) return "use";
662 else if(nr==KW_INT) return "int";
663 else if(nr==KW_NEW) return "new";
664 else if(nr==KW_GET) return "get";
665 else if(nr==KW_SET) return "set";
666 else if(nr==KW_VAR) return "var";
667 else if(nr==KW_IS) return "is";
668 else if(nr==KW_AS) return "as";
669 else if(nr==T_IDENTIFIER) return "ID";
671 sprintf(mbuf, "%d", nr);
676 void initialize_scanner()