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 */
32 #include "tokenizer.h"
35 unsigned int as3_tokencount = 0;
37 static void dbg(const char*format, ...)
44 va_start(arglist, format);
45 vsprintf(buf, format, arglist);
48 while(l && buf[l-1]=='\n') {
52 printf("(tokenizer) ");
57 #ifndef YY_CURRENT_BUFFER
58 #define YY_CURRENT_BUFFER yy_current_buffer
61 static void*as3_buffer=0;
62 static int as3_buffer_pos=0;
63 static int as3_buffer_len=0;
64 void as3_file_input(FILE*fi)
69 void as3_buffer_input(void*buffer, int len)
72 syntaxerror("trying to parse zero bytearray");
79 #define YY_INPUT(buf,result,max_size) { \
82 while((result = fread(buf, 1, max_size, as3_in))==0 && ferror(as3_in)) \
83 { if(errno != EINTR) {YY_FATAL_ERROR("input in flex scanner failed"); break;} \
84 errno=0; clearerr(as3_in); \
87 int to_read = max_size; \
88 if(to_read + as3_buffer_pos > as3_buffer_len) \
89 to_read = as3_buffer_len - as3_buffer_pos; \
90 memcpy(buf, as3_buffer+as3_buffer_pos, to_read); \
91 as3_buffer_pos += to_read; \
96 void handleInclude(char*text, int len, char quotes)
100 char*p1 = strchr(text, '"');
101 char*p2 = strrchr(text, '"');
102 if(!p1 || !p2 || p1==p2) {
103 syntaxerror("Invalid include in line %d\n", current_line);
106 filename = strdup(p1+1);
110 while(!strchr(" \n\r\t", text[i1])) i1++;
112 while(strchr(" \n\r\t", text[i1])) i1++;
113 while(strchr(" \n\r\t", text[i2-1])) i2--;
114 if(i2!=len) text[i2]=0;
115 filename = strdup(&text[i1]);
118 char*fullfilename = find_file(filename, 1);
119 enter_file2(filename, fullfilename, YY_CURRENT_BUFFER);
120 yyin = fopen(fullfilename, "rb");
122 syntaxerror("Couldn't open include file \"%s\"\n", fullfilename);
125 yy_switch_to_buffer(yy_create_buffer( yyin, YY_BUF_SIZE ) );
126 //BEGIN(INITIAL); keep context
129 static int do_unescape(const char*s, const char*end, char*n)
135 if(o) o[len] = *s;len++;
140 if(s==end) syntaxerror("invalid \\ at end of string");
142 /* handle the various line endings (mac, dos, unix) */
155 case '\\': if(o) o[len] = '\\';s++;len++; break;
156 case '"': if(o) o[len] = '"';s++;len++; break;
157 case '\'': if(o) o[len] = '\'';s++;len++; break;
158 case 'b': if(o) o[len] = '\b';s++;len++; break;
159 case 'f': if(o) o[len] = '\f';s++;len++; break;
160 case 'n': if(o) o[len] = '\n';s++;len++; break;
161 case 'r': if(o) o[len] = '\r';s++;len++; break;
162 case 't': if(o) o[len] = '\t';s++;len++; break;
163 case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': {
166 while(strchr("01234567", *s) && nr<3 && s<end) {
173 syntaxerror("octal number out of range (0-255): %d", num);
174 if(o) o[len] = num;len++;
177 case 'x': case 'u': {
186 if(s==end) syntaxerror("invalid \\u or \\x at end of string");
189 if(s==end) syntaxerror("invalid \\u{ at end of string");
194 while(strchr("0123456789abcdefABCDEF", *s) && (bracket || nr < max) && s<end) {
196 if(*s>='0' && *s<='9') num |= *s - '0';
197 if(*s>='a' && *s<='f') num |= *s - 'a' + 10;
198 if(*s>='A' && *s<='F') num |= *s - 'A' + 10;
203 if(*s=='}' && s<end) {
206 syntaxerror("missing terminating '}'");
210 char*utf8 = getUTF8(num);
212 if(o) o[len] = *utf8;utf8++;len++;
216 syntaxerror("byte out of range (0-255): %d", num);
217 if(o) o[len] = num;len++;
236 static string_t string_unescape(const char*in, int l)
239 const char*end = &in[l];
241 int len = do_unescape(s, end, 0);
242 char*n = (char*)malloc(len+1);
243 do_unescape(s, end, n);
244 string_t out = string_new(n, len);
248 static void handleCData(char*s, int len)
250 a3_lval.str.str = s+9; // <![CDATA[
251 a3_lval.str.len = len-9-3;// ]]>
252 a3_lval.str.str = strdup_n(a3_lval.str.str, a3_lval.str.len);
255 static void handleString(char*s, int len)
258 if(s[len-1]!='"') syntaxerror("String doesn't end with '\"'");
261 else if(s[0]=='\'') {
262 if(s[len-1]!='\'') syntaxerror("String doesn't end with '\"'");
265 else syntaxerror("String incorrectly terminated");
267 a3_lval.str = string_unescape(s, len);
271 char start_of_expression;
273 static inline int m(int type)
275 a3_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 a3_lval.number_int = v;
301 static inline int setuint(unsigned int v)
303 a3_lval.number_uint = v;
311 static inline int setfloat(double v)
313 a3_lval.number_float = v;
317 static inline int handlefloat()
320 a3_lval.number_float = atof(s);
324 static inline int handleint()
327 char l = (yytext[0]=='-');
329 char*max = l?"1073741824":"2147483647";
331 as3_softwarning("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_softwarning("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_softwarning("signed integer overflow: %s (converted to Number)", s);
409 if(!l && v>2147483647) {
411 as3_softwarning("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 a3_lval.regexp.pattern = s;
451 a3_lval.regexp.options = 0;
453 a3_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 trie_t*active_namespaces = 0;
479 /*void tokenizer_register_namespace(const char*id)
481 trie_put(namespaces, id, 0);
483 void tokenizer_unregister_namespace(const char*id)
485 trie_remove(namespaces, id);
487 static inline tokenizer_is_namespace(const char*id)
489 return trie_contains(active_namespaces, id);
492 static inline int handleIdentifier()
494 char*s = malloc(yyleng+1);
495 memcpy(s, yytext, yyleng);
498 if(tokenizer_is_namespace(s))
505 //Boolean {c();return m(KW_BOOLEAN);}
506 //int {c();return m(KW_INT);}
507 //uint {c();return m(KW_UINT);}
508 //Number {c();return m(KW_NUMBER);}
515 NAME [a-zA-Z_][a-zA-Z0-9_\\]*
518 HEXINT 0x[a-zA-Z0-9]+
519 HEXFLOAT 0x[a-zA-Z0-9]*\.[a-zA-Z0-9]*
521 FLOAT ([0-9]+(\.[0-9]*)?|\.[0-9]+)(e[0-9]+)?
523 HEXWITHSIGN [+-]?({HEXINT})
524 HEXFLOATWITHSIGN [+-]?({HEXFLOAT})
525 INTWITHSIGN [+-]?({INT})
526 FLOATWITHSIGN [+-]?({FLOAT})
528 CDATA <!\[CDATA\[([^]]|\][^]]|\]\][^>])*\]*\]\]\>
529 STRING ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*[']
531 MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[^*][/]|[\x00-\x1f])*[*]+[/]
532 SINGLELINE_COMMENT \/\/[^\n\r]*[\n\r]
533 REGEXP [/]([^/\n]|\\[/])*[/][a-zA-Z]*
537 {SINGLELINE_COMMENT} {l(); /* single line comment */}
538 {MULTILINE_COMMENT} {l(); /* multi line comment */}
539 [/][*] {syntaxerror("syntax error: unterminated comment", yytext);}
541 ^include{S}+{STRING}{S}*/\n {l();handleInclude(yytext, yyleng, 1);}
542 ^include{S}+[^" \t\r\n][\x20-\xff]*{S}*/\n {l();handleInclude(yytext, yyleng, 0);}
543 {STRING} {l(); BEGIN(INITIAL);handleString(yytext, yyleng);return T_STRING;}
544 {CDATA} {l(); BEGIN(INITIAL);handleCData(yytext, yyleng);return T_STRING;}
546 <BEGINNING,REGEXPOK>{
547 {REGEXP} {c(); BEGIN(INITIAL);return handleregexp();}
548 {HEXWITHSIGN}/{_} {c(); BEGIN(INITIAL);return handlehex();}
549 {HEXFLOATWITHSIGN}/{_} {c(); BEGIN(INITIAL);return handlehexfloat();}
550 {INTWITHSIGN}/{_} {c(); BEGIN(INITIAL);return handleint();}
551 {FLOATWITHSIGN}/{_} {c(); BEGIN(INITIAL);return handlefloat();}
554 <REGEXPOK>[\{] {c(); BEGIN(REGEXPOK);return m(T_DICTSTART);}
555 [\{] {c(); BEGIN(INITIAL); return m('{');}
557 \xef\xbb\xbf {/* utf 8 bom */}
560 {HEXINT}/{_} {c(); BEGIN(INITIAL);return handlehex();}
561 {HEXFLOAT}/{_} {c(); BEGIN(INITIAL);return handlehexfloat();}
562 {INT}/{_} {c(); BEGIN(INITIAL);return handleint();}
563 {FLOAT}/{_} {c(); BEGIN(INITIAL);return handlefloat();}
565 3rr0r {/* for debugging: generates a tokenizer-level error */
566 syntaxerror("3rr0r");}
568 {NAME}{S}*:{S}*for/{_} {l();BEGIN(INITIAL);handleLabel(yytext, yyleng-3);return T_FOR;}
569 {NAME}{S}*:{S}*do/{_} {l();BEGIN(INITIAL);handleLabel(yytext, yyleng-2);return T_DO;}
570 {NAME}{S}*:{S}*while/{_} {l();BEGIN(INITIAL);handleLabel(yytext, yyleng-5);return T_WHILE;}
571 {NAME}{S}*:{S}*switch/{_} {l();BEGIN(INITIAL);handleLabel(yytext, yyleng-6);return T_SWITCH;}
572 for {c();BEGIN(INITIAL);a3_lval.id="";return T_FOR;}
573 do {c();BEGIN(INITIAL);a3_lval.id="";return T_DO;}
574 while {c();BEGIN(INITIAL);a3_lval.id="";return T_WHILE;}
575 switch {c();BEGIN(INITIAL);a3_lval.id="";return T_SWITCH;}
577 [&][&] {c();BEGIN(REGEXPOK);return m(T_ANDAND);}
578 [|][|] {c();BEGIN(REGEXPOK);return m(T_OROR);}
579 [!][=] {c();BEGIN(REGEXPOK);return m(T_NE);}
580 [!][=][=] {c();BEGIN(REGEXPOK);return m(T_NEE);}
581 [=][=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQEQ);}
582 [=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQ);}
583 [>][=] {c();BEGIN(REGEXPOK);return m(T_GE);}
584 [<][=] {c();BEGIN(REGEXPOK);return m(T_LE);}
585 [-][-] {c();BEGIN(INITIAL);return m(T_MINUSMINUS);}
586 [+][+] {c();BEGIN(INITIAL);return m(T_PLUSPLUS);}
587 [+][=] {c();BEGIN(REGEXPOK);return m(T_PLUSBY);}
588 [\^][=] {c();BEGIN(REGEXPOK);return m(T_XORBY);}
589 [-][=] {c();BEGIN(REGEXPOK);return m(T_MINUSBY);}
590 [/][=] {c();BEGIN(REGEXPOK);return m(T_DIVBY);}
591 [%][=] {c();BEGIN(REGEXPOK);return m(T_MODBY);}
592 [*][=] {c();BEGIN(REGEXPOK);return m(T_MULBY);}
593 [|][=] {c();BEGIN(REGEXPOK);return m(T_ORBY);}
594 [>][>][=] {c();BEGIN(REGEXPOK);return m(T_SHRBY);}
595 [<][<][=] {c();BEGIN(REGEXPOK);return m(T_SHLBY);}
596 [>][>][>][=] {c();BEGIN(REGEXPOK);return m(T_USHRBY);}
597 [<][<] {c();BEGIN(REGEXPOK);return m(T_SHL);}
598 [>][>][>] {c();BEGIN(REGEXPOK);return m(T_USHR);}
599 [>][>] {c();BEGIN(REGEXPOK);return m(T_SHR);}
600 \.\.\. {c();BEGIN(REGEXPOK);return m(T_DOTDOTDOT);}
601 \.\. {c();BEGIN(REGEXPOK);return m(T_DOTDOT);}
602 \. {c();BEGIN(REGEXPOK);return m('.');}
603 :: {c();BEGIN(REGEXPOK);return m(T_COLONCOLON);}
604 : {c();BEGIN(REGEXPOK);return m(':');}
605 instanceof {c();BEGIN(REGEXPOK);return m(KW_INSTANCEOF);}
606 implements {c();BEGIN(REGEXPOK);return m(KW_IMPLEMENTS);}
607 interface {c();BEGIN(INITIAL);return m(KW_INTERFACE);}
608 namespace {c();BEGIN(INITIAL);return m(KW_NAMESPACE);}
609 protected {c();BEGIN(INITIAL);return m(KW_PROTECTED);}
610 undefined {c();BEGIN(INITIAL);return m(KW_UNDEFINED);}
611 continue {c();BEGIN(INITIAL);return m(KW_CONTINUE);}
612 override {c();BEGIN(INITIAL);return m(KW_OVERRIDE);}
613 internal {c();BEGIN(INITIAL);return m(KW_INTERNAL);}
614 function {c();BEGIN(INITIAL);return m(KW_FUNCTION);}
615 finally {c();BEGIN(INITIAL);return m(KW_FINALLY);}
616 default {c();BEGIN(INITIAL);return m(KW_DEFAULT);}
617 package {c();BEGIN(INITIAL);return m(KW_PACKAGE);}
618 private {c();BEGIN(INITIAL);return m(KW_PRIVATE);}
619 dynamic {c();BEGIN(INITIAL);return m(KW_DYNAMIC);}
620 extends {c();BEGIN(INITIAL);return m(KW_EXTENDS);}
621 delete {c();BEGIN(REGEXPOK);return m(KW_DELETE);}
622 return {c();BEGIN(REGEXPOK);return m(KW_RETURN);}
623 public {c();BEGIN(INITIAL);return m(KW_PUBLIC);}
624 native {c();BEGIN(INITIAL);return m(KW_NATIVE);}
625 static {c();BEGIN(INITIAL);return m(KW_STATIC);}
626 import {c();BEGIN(REGEXPOK);return m(KW_IMPORT);}
627 typeof {c();BEGIN(REGEXPOK);return m(KW_TYPEOF);}
628 throw {c();BEGIN(REGEXPOK);return m(KW_THROW);}
629 class {c();BEGIN(INITIAL);return m(KW_CLASS);}
630 const {c();BEGIN(INITIAL);return m(KW_CONST);}
631 catch {c();BEGIN(INITIAL);return m(KW_CATCH);}
632 final {c();BEGIN(INITIAL);return m(KW_FINAL);}
633 false {c();BEGIN(INITIAL);return m(KW_FALSE);}
634 break {c();BEGIN(INITIAL);return m(KW_BREAK);}
635 super {c();BEGIN(INITIAL);return m(KW_SUPER);}
636 each {c();BEGIN(INITIAL);return m(KW_EACH);}
637 void {c();BEGIN(INITIAL);return m(KW_VOID);}
638 true {c();BEGIN(INITIAL);return m(KW_TRUE);}
639 null {c();BEGIN(INITIAL);return m(KW_NULL);}
640 else {c();BEGIN(INITIAL);return m(KW_ELSE);}
641 case {c();BEGIN(REGEXPOK);return m(KW_CASE);}
642 with {c();BEGIN(REGEXPOK);return m(KW_WITH);}
643 use {c();BEGIN(REGEXPOK);return m(KW_USE);}
644 new {c();BEGIN(REGEXPOK);return m(KW_NEW);}
645 get {c();BEGIN(INITIAL);return m(KW_GET);}
646 set {c();BEGIN(INITIAL);return m(KW_SET);}
647 var {c();BEGIN(INITIAL);return m(KW_VAR);}
648 try {c();BEGIN(INITIAL);return m(KW_TRY);}
649 is {c();BEGIN(REGEXPOK);return m(KW_IS) ;}
650 in {c();BEGIN(REGEXPOK);return m(KW_IN) ;}
651 if {c();BEGIN(INITIAL);return m(KW_IF) ;}
652 as {c();BEGIN(REGEXPOK);return m(KW_AS);}
653 $?{NAME} {c();BEGIN(INITIAL);return handleIdentifier();}
655 [\]\}*] {c();BEGIN(INITIAL);return m(yytext[0]);}
656 [+-\/^~@$!%&\(=\[|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
657 [\)\]] {c();BEGIN(INITIAL);return m(yytext[0]);}
665 char c = buf[t]=input();
666 if(c=='\n' || c==EOF) {
671 if(c1>='0' && c1<='9')
672 syntaxerror("syntax error: %s (identifiers must not start with a digit)");
674 syntaxerror("syntax error: %s", buf);
680 void*b = leave_file();
683 yy_delete_buffer(YY_CURRENT_BUFFER);
686 yy_delete_buffer(YY_CURRENT_BUFFER);
687 yy_switch_to_buffer(b);
698 static char mbuf[256];
699 char*token2string(enum yytokentype nr, YYSTYPE v)
702 char*s = malloc(v.str.len+10);
703 strcpy(s, "<string>");
704 memcpy(s+8, v.str.str, v.str.len);
705 sprintf(s+8+v.str.len, " (%d bytes)", v.str.len);
708 else if(nr==T_REGEXP) {
709 char*s = malloc(strlen(v.regexp.pattern)+10);
710 sprintf(s, "<regexp>%s", v.regexp.pattern);
713 else if(nr==T_IDENTIFIER) {
714 char*s = malloc(strlen(v.id)+10);
715 sprintf(s, "<ID>%s", v.id);
718 else if(nr==T_INT) return "<int>";
719 else if(nr==T_UINT) return "<uint>";
720 else if(nr==T_BYTE) return "<byte>";
721 else if(nr==T_FLOAT) return "<float>";
722 else if(nr==T_EOF) return "***END***";
723 else if(nr==T_GE) return ">=";
724 else if(nr==T_LE) return "<=";
725 else if(nr==T_MINUSMINUS) return "--";
726 else if(nr==T_PLUSPLUS) return "++";
727 else if(nr==KW_IMPLEMENTS) return "implements";
728 else if(nr==KW_INTERFACE) return "interface";
729 else if(nr==KW_NAMESPACE) return "namespace";
730 else if(nr==KW_PROTECTED) return "protected";
731 else if(nr==KW_OVERRIDE) return "override";
732 else if(nr==KW_INTERNAL) return "internal";
733 else if(nr==KW_FUNCTION) return "function";
734 else if(nr==KW_PACKAGE) return "package";
735 else if(nr==KW_PRIVATE) return "private";
736 else if(nr==KW_BOOLEAN) return "Boolean";
737 else if(nr==KW_DYNAMIC) return "dynamic";
738 else if(nr==KW_EXTENDS) return "extends";
739 else if(nr==KW_PUBLIC) return "public";
740 else if(nr==KW_NATIVE) return "native";
741 else if(nr==KW_STATIC) return "static";
742 else if(nr==KW_IMPORT) return "import";
743 else if(nr==KW_NUMBER) return "number";
744 else if(nr==KW_CLASS) return "class";
745 else if(nr==KW_CONST) return "const";
746 else if(nr==KW_FINAL) return "final";
747 else if(nr==KW_FALSE) return "False";
748 else if(nr==KW_TRUE) return "True";
749 else if(nr==KW_UINT) return "uint";
750 else if(nr==KW_NULL) return "null";
751 else if(nr==KW_ELSE) return "else";
752 else if(nr==KW_USE) return "use";
753 else if(nr==KW_INT) return "int";
754 else if(nr==KW_NEW) return "new";
755 else if(nr==KW_GET) return "get";
756 else if(nr==KW_SET) return "set";
757 else if(nr==KW_VAR) return "var";
758 else if(nr==KW_IS) return "is";
759 else if(nr==KW_AS) return "as";
761 sprintf(mbuf, "%d", nr);
766 void initialize_scanner()