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"
35 int as3_verbosity = 1;
36 unsigned int as3_tokencount = 0;
38 void as3_error(const char*format, ...)
45 va_start(arglist, format);
46 vsprintf(buf, format, arglist);
48 fprintf(stderr, "%s:%d:%d: error: %s\n", current_filename_short, current_line, current_column, buf);
52 void as3_warning(const char*format, ...)
59 va_start(arglist, format);
60 vsprintf(buf, format, arglist);
62 fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
65 void as3_softwarning(const char*format, ...)
72 va_start(arglist, format);
73 vsprintf(buf, format, arglist);
75 fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
78 static void dbg(const char*format, ...)
85 va_start(arglist, format);
86 vsprintf(buf, format, arglist);
89 while(l && buf[l-1]=='\n') {
93 printf("(tokenizer) ");
98 #ifndef YY_CURRENT_BUFFER
99 #define YY_CURRENT_BUFFER yy_current_buffer
102 static void*as3_buffer=0;
103 static int as3_buffer_pos=0;
104 static int as3_buffer_len=0;
105 void as3_file_input(FILE*fi)
110 void as3_buffer_input(void*buffer, int len)
113 as3_buffer_len = len;
118 #define YY_INPUT(buf,result,max_size) { \
121 while((result = fread(buf, 1, max_size, as3_in))==0 && ferror(as3_in)) \
122 { if(errno != EINTR) {YY_FATAL_ERROR("input in flex scanner failed"); break;} \
123 errno=0; clearerr(as3_in); \
126 int to_read = max_size; \
127 if(to_read + as3_buffer_pos > as3_buffer_len) \
128 to_read = as3_buffer_len - as3_buffer_pos; \
129 memcpy(buf, as3_buffer+as3_buffer_pos, to_read); \
130 as3_buffer_pos += to_read; \
135 void handleInclude(char*text, int len, char quotes)
139 char*p1 = strchr(text, '"');
140 char*p2 = strrchr(text, '"');
141 if(!p1 || !p2 || p1==p2) {
142 syntaxerror("Invalid include in line %d\n", current_line);
145 filename = strdup(p1+1);
149 while(!strchr(" \n\r\t", text[i1])) i1++;
151 while(strchr(" \n\r\t", text[i1])) i1++;
152 while(strchr(" \n\r\t", text[i2-1])) i2--;
153 if(i2!=len) text[i2]=0;
154 filename = strdup(&text[i1]);
157 char*fullfilename = find_file(filename);
158 enter_file2(filename, fullfilename, YY_CURRENT_BUFFER);
159 yyin = fopen(fullfilename, "rb");
161 syntaxerror("Couldn't open include file \"%s\"\n", fullfilename);
164 yy_switch_to_buffer(yy_create_buffer( yyin, YY_BUF_SIZE ) );
165 //BEGIN(INITIAL); keep context
168 static int do_unescape(const char*s, const char*end, char*n)
174 if(o) o[len] = *s;len++;
179 if(s==end) syntaxerror("invalid \\ at end of string");
181 /* handle the various line endings (mac, dos, unix) */
194 case '\\': if(o) o[len] = '\\';s++;len++; break;
195 case '"': if(o) o[len] = '"';s++;len++; break;
196 case '\'': if(o) o[len] = '\'';s++;len++; break;
197 case 'b': if(o) o[len] = '\b';s++;len++; break;
198 case 'f': if(o) o[len] = '\f';s++;len++; break;
199 case 'n': if(o) o[len] = '\n';s++;len++; break;
200 case 'r': if(o) o[len] = '\r';s++;len++; break;
201 case 't': if(o) o[len] = '\t';s++;len++; break;
202 case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': {
205 while(strchr("01234567", *s) && nr<3 && s<end) {
212 syntaxerror("octal number out of range (0-255): %d", num);
213 if(o) o[len] = num;len++;
216 case 'x': case 'u': {
225 if(s==end) syntaxerror("invalid \\u or \\x at end of string");
228 if(s==end) syntaxerror("invalid \\u{ at end of string");
233 while(strchr("0123456789abcdefABCDEF", *s) && (bracket || nr < max) && s<end) {
235 if(*s>='0' && *s<='9') num |= *s - '0';
236 if(*s>='a' && *s<='f') num |= *s - 'a' + 10;
237 if(*s>='A' && *s<='F') num |= *s - 'A' + 10;
242 if(*s=='}' && s<end) {
245 syntaxerror("missing terminating '}'");
249 char*utf8 = getUTF8(num);
251 if(o) o[len] = *utf8;utf8++;len++;
255 syntaxerror("byte out of range (0-255): %d", num);
256 if(o) o[len] = num;len++;
261 syntaxerror("unknown escape sequence: \"\\%c\"", *s);
268 static string_t string_unescape(const char*in, int l)
271 const char*end = &in[l];
273 int len = do_unescape(s, end, 0);
274 char*n = (char*)malloc(len+1);
275 do_unescape(s, end, n);
276 string_t out = string_new(n, len);
280 static void handleCData(char*s, int len)
282 a3_lval.str.str = s+9; // <![CDATA[
283 a3_lval.str.len = len-9-3;// ]]>
284 a3_lval.str.str = strdup_n(a3_lval.str.str, a3_lval.str.len);
287 static void handleString(char*s, int len)
290 // don't bother decoding strings in pass 1
291 memset(&a3_lval, 0, sizeof(a3_lval));
296 if(s[len-1]!='"') syntaxerror("String doesn't end with '\"'");
299 else if(s[0]=='\'') {
300 if(s[len-1]!='\'') syntaxerror("String doesn't end with '\"'");
303 else syntaxerror("String incorrectly terminated");
306 a3_lval.str = string_unescape(s, len);
310 char start_of_expression;
312 static inline int handleIdentifier(int type)
314 char*s = malloc(yyleng+1);
315 memcpy(s, yytext, yyleng);
321 static inline int m(int type)
323 a3_lval.token = type;
328 static char numberbuf[64];
331 if(yyleng>sizeof(numberbuf)-1)
332 syntaxerror("decimal number overflow");
334 memcpy(s, yytext, yyleng);
339 static inline int setint(int v)
341 a3_lval.number_int = v;
349 static inline int setuint(unsigned int v)
351 a3_lval.number_uint = v;
359 static inline int setfloat(double v)
361 a3_lval.number_float = v;
365 static inline int handlefloat()
368 a3_lval.number_float = atof(s);
372 static inline int handleint()
375 char l = (yytext[0]=='-');
377 char*max = l?"1073741824":"2147483647";
379 as3_warning("integer overflow: %s (converted to Number)", s);
380 return handlefloat();
384 for(t=0;t<yyleng-l;t++) {
385 if(yytext[l+t]>max[t]) {
386 as3_warning("integer overflow: %s (converted to Number)", s);
387 return handlefloat();
389 else if(yytext[l+t]<max[t])
399 for(t=0;t<yyleng;t++) {
407 static inline int handlehexfloat()
409 char l = (yytext[0]=='-')+2;
414 for(t=l;t<yyleng;t++) {
427 else if((c>='a' && c<='f') || (c>='A' && c<='F'))
428 d+=((c&0x0f)+9)*base;
432 static inline int handlehex()
434 char l = (yytext[0]=='-')+2;
439 syntaxerror("integer overflow %s", s);
449 else if((c>='a' && c<='f') || (c>='A' && c<='F'))
452 if(l && v>1073741824) {
454 as3_warning("signed integer overflow: %s (converted to Number)", s);
457 if(!l && v>2147483647) {
459 as3_warning("unsigned integer overflow: %s (converted to Number)", s);
464 return setint(-(int)v);
470 void handleLabel(char*text, int len)
473 for(t=len-1;t>=0;--t) {
478 char*s = malloc(t+1);
479 memcpy(s, yytext, t);
484 static int handleregexp()
486 char*s = malloc(yyleng);
488 memcpy(s, yytext+1, len);
491 for(t=len;t>=0;--t) {
497 a3_lval.regexp.pattern = s;
499 a3_lval.regexp.options = 0;
501 a3_lval.regexp.options = s+t+1;
506 void initialize_scanner();
507 #define YY_USER_INIT initialize_scanner();
509 /* count the number of lines+columns consumed by this token */
510 static inline void l() {
512 for(t=0;t<yyleng;t++) {
513 if(yytext[t]=='\n') {
521 /* count the number of columns consumed by this token */
522 static inline void c() {
523 current_column+=yyleng;
526 //Boolean {c();return m(KW_BOOLEAN);}
527 //int {c();return m(KW_INT);}
528 //uint {c();return m(KW_UINT);}
529 //Number {c();return m(KW_NUMBER);}
536 NAME [a-zA-Z_][a-zA-Z0-9_\\]*
539 HEXINT 0x[a-zA-Z0-9]+
540 HEXFLOAT 0x[a-zA-Z0-9]*\.[a-zA-Z0-9]*
542 FLOAT [0-9]+(\.[0-9]*)?|\.[0-9]+
544 HEXWITHSIGN [+-]?({HEXINT})
545 HEXFLOATWITHSIGN [+-]?({HEXFLOAT})
546 INTWITHSIGN [+-]?({INT})
547 FLOATWITHSIGN [+-]?({FLOAT})
549 CDATA <!\[CDATA\[([^]]|\][^]]|\]\][^>])*\]*\]\]\>
550 STRING ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*[']
552 MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[^*][/]|[\x00-\x1f])*[*]+[/]
553 SINGLELINE_COMMENT \/\/[^\n]*\n
554 REGEXP [/]([^/\n]|\\[/])*[/][a-zA-Z]*
558 {SINGLELINE_COMMENT} {l(); /* single line comment */}
559 {MULTILINE_COMMENT} {l(); /* multi line comment */}
560 [/][*] {syntaxerror("syntax error: unterminated comment", yytext);}
562 ^include{S}+{STRING}{S}*/\n {l();handleInclude(yytext, yyleng, 1);}
563 ^include{S}+[^" \t\r\n][\x20-\xff]*{S}*/\n {l();handleInclude(yytext, yyleng, 0);}
564 {STRING} {l(); BEGIN(INITIAL);handleString(yytext, yyleng);return T_STRING;}
565 {CDATA} {l(); BEGIN(INITIAL);handleCData(yytext, yyleng);return T_STRING;}
567 <BEGINNING,REGEXPOK>{
568 {REGEXP} {c(); BEGIN(INITIAL);return handleregexp();}
569 {HEXWITHSIGN} {c(); BEGIN(INITIAL);return handlehex();}
570 {HEXFLOATWITHSIGN} {c(); BEGIN(INITIAL);return handlehexfloat();}
571 {INTWITHSIGN} {c(); BEGIN(INITIAL);return handleint();}
572 {FLOATWITHSIGN} {c(); BEGIN(INITIAL);return handlefloat();}
575 \xef\xbb\xbf {/* utf 8 bom */}
578 {HEXINT} {c(); BEGIN(INITIAL);return handlehex();}
579 {HEXFLOAT} {c(); BEGIN(INITIAL);return handlehexfloat();}
580 {INT} {c(); BEGIN(INITIAL);return handleint();}
581 {FLOAT} {c(); BEGIN(INITIAL);return handlefloat();}
583 3rr0r {/* for debugging: generates a tokenizer-level error */
584 syntaxerror("3rr0r");}
586 {NAME}{S}*:{S}*for/{_} {l();handleLabel(yytext, yyleng-3);return T_FOR;}
587 {NAME}{S}*:{S}*do/{_} {l();handleLabel(yytext, yyleng-2);return T_DO;}
588 {NAME}{S}*:{S}*while/{_} {l();handleLabel(yytext, yyleng-5);return T_WHILE;}
589 {NAME}{S}*:{S}*switch/{_} {l();handleLabel(yytext, yyleng-6);return T_SWITCH;}
590 for {c();a3_lval.id="";return T_FOR;}
591 do {c();a3_lval.id="";return T_DO;}
592 while {c();a3_lval.id="";return T_WHILE;}
593 switch {c();a3_lval.id="";return T_SWITCH;}
595 [&][&] {c();BEGIN(REGEXPOK);return m(T_ANDAND);}
596 [|][|] {c();BEGIN(REGEXPOK);return m(T_OROR);}
597 [!][=] {c();BEGIN(REGEXPOK);return m(T_NE);}
598 [!][=][=] {c();BEGIN(REGEXPOK);return m(T_NEE);}
599 [=][=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQEQ);}
600 [=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQ);}
601 [>][=] {c();return m(T_GE);}
602 [<][=] {c();return m(T_LE);}
603 [-][-] {c();BEGIN(INITIAL);return m(T_MINUSMINUS);}
604 [+][+] {c();BEGIN(INITIAL);return m(T_PLUSPLUS);}
605 [+][=] {c();return m(T_PLUSBY);}
606 [-][=] {c();return m(T_MINUSBY);}
607 [/][=] {c();return m(T_DIVBY);}
608 [%][=] {c();return m(T_MODBY);}
609 [*][=] {c();return m(T_MULBY);}
610 [|][=] {c();return m(T_ORBY);}
611 [>][>][=] {c();return m(T_SHRBY);}
612 [<][<][=] {c();return m(T_SHLBY);}
613 [>][>][>][=] {c();return m(T_USHRBY);}
614 [<][<] {c();return m(T_SHL);}
615 [>][>][>] {c();return m(T_USHR);}
616 [>][>] {c();return m(T_SHR);}
617 \.\.\. {c();return m(T_DOTDOTDOT);}
618 \.\. {c();return m(T_DOTDOT);}
619 \. {c();return m('.');}
620 :: {c();return m(T_COLONCOLON);}
621 : {c();return m(':');}
622 instanceof {c();return m(KW_INSTANCEOF);}
623 implements {c();return m(KW_IMPLEMENTS);}
624 interface {c();return m(KW_INTERFACE);}
625 namespace {c();return m(KW_NAMESPACE);}
626 protected {c();return m(KW_PROTECTED);}
627 undefined {c();return m(KW_UNDEFINED);}
628 continue {c();return m(KW_CONTINUE);}
629 override {c();return m(KW_OVERRIDE);}
630 internal {c();return m(KW_INTERNAL);}
631 function {c();return m(KW_FUNCTION);}
632 finally {c();return m(KW_FINALLY);}
633 default {c();return m(KW_DEFAULT);}
634 package {c();return m(KW_PACKAGE);}
635 private {c();return m(KW_PRIVATE);}
636 dynamic {c();return m(KW_DYNAMIC);}
637 extends {c();return m(KW_EXTENDS);}
638 delete {c();return m(KW_DELETE);}
639 return {c();return m(KW_RETURN);}
640 public {c();return m(KW_PUBLIC);}
641 native {c();return m(KW_NATIVE);}
642 static {c();return m(KW_STATIC);}
643 import {c();return m(KW_IMPORT);}
644 typeof {c();return m(KW_TYPEOF);}
645 throw {c();return m(KW_THROW);}
646 class {c();return m(KW_CLASS);}
647 const {c();return m(KW_CONST);}
648 catch {c();return m(KW_CATCH);}
649 final {c();return m(KW_FINAL);}
650 false {c();return m(KW_FALSE);}
651 break {c();return m(KW_BREAK);}
652 super {c();return m(KW_SUPER);}
653 each {c();return m(KW_EACH);}
654 void {c();return m(KW_VOID);}
655 true {c();return m(KW_TRUE);}
656 null {c();return m(KW_NULL);}
657 else {c();return m(KW_ELSE);}
658 case {c();return m(KW_CASE);}
659 with {c();return m(KW_WITH);}
660 use {c();return m(KW_USE);}
661 new {c();return m(KW_NEW);}
662 get {c();return m(KW_GET);}
663 set {c();return m(KW_SET);}
664 var {c();return m(KW_VAR);}
665 try {c();return m(KW_TRY);}
666 is {c();return m(KW_IS) ;}
667 in {c();return m(KW_IN) ;}
668 if {c();return m(KW_IF) ;}
669 as {c();return m(KW_AS);}
670 {NAME} {c();BEGIN(INITIAL);return handleIdentifier(T_IDENTIFIER);}
672 [+-\/*^~@$!%&\(=\[\]\{\}|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
673 [\)\]] {c();BEGIN(INITIAL);return m(yytext[0]);}
681 char c = buf[t]=input();
682 if(c=='\n' || c==EOF) {
687 if(c1>='0' && c1<='9')
688 syntaxerror("syntax error: %s (identifiers must not start with a digit)");
690 syntaxerror("syntax error: %s", buf);
696 void*b = leave_file();
699 yy_delete_buffer(YY_CURRENT_BUFFER);
702 yy_delete_buffer(YY_CURRENT_BUFFER);
703 yy_switch_to_buffer(b);
714 static char mbuf[256];
715 char*token2string(enum yytokentype nr, YYSTYPE v)
717 if(nr==T_STRING) return "<string>";
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_REGEXP) return "REGEXP";
723 else if(nr==T_EOF) return "***END***";
724 else if(nr==T_GE) return ">=";
725 else if(nr==T_LE) return "<=";
726 else if(nr==T_MINUSMINUS) return "--";
727 else if(nr==T_PLUSPLUS) return "++";
728 else if(nr==KW_IMPLEMENTS) return "implements";
729 else if(nr==KW_INTERFACE) return "interface";
730 else if(nr==KW_NAMESPACE) return "namespace";
731 else if(nr==KW_PROTECTED) return "protected";
732 else if(nr==KW_OVERRIDE) return "override";
733 else if(nr==KW_INTERNAL) return "internal";
734 else if(nr==KW_FUNCTION) return "function";
735 else if(nr==KW_PACKAGE) return "package";
736 else if(nr==KW_PRIVATE) return "private";
737 else if(nr==KW_BOOLEAN) return "Boolean";
738 else if(nr==KW_DYNAMIC) return "dynamic";
739 else if(nr==KW_EXTENDS) return "extends";
740 else if(nr==KW_PUBLIC) return "public";
741 else if(nr==KW_NATIVE) return "native";
742 else if(nr==KW_STATIC) return "static";
743 else if(nr==KW_IMPORT) return "import";
744 else if(nr==KW_NUMBER) return "number";
745 else if(nr==KW_CLASS) return "class";
746 else if(nr==KW_CONST) return "const";
747 else if(nr==KW_FINAL) return "final";
748 else if(nr==KW_FALSE) return "False";
749 else if(nr==KW_TRUE) return "True";
750 else if(nr==KW_UINT) return "uint";
751 else if(nr==KW_NULL) return "null";
752 else if(nr==KW_ELSE) return "else";
753 else if(nr==KW_USE) return "use";
754 else if(nr==KW_INT) return "int";
755 else if(nr==KW_NEW) return "new";
756 else if(nr==KW_GET) return "get";
757 else if(nr==KW_SET) return "set";
758 else if(nr==KW_VAR) return "var";
759 else if(nr==KW_IS) return "is";
760 else if(nr==KW_AS) return "as";
761 else if(nr==T_IDENTIFIER) return "ID";
763 sprintf(mbuf, "%d", nr);
768 void initialize_scanner()