lib/as3/tokenizer.lex

   1 /* tokenizer.lex
   2
   3    Routines for compiling Flash2 AVM2 ABC Actionscript
   4
   5    Extension module for the rfxswf library.
   6    Part of the swftools package.
   7
   8    Copyright (c) 2008 Matthias Kramm <kramm@quiss.org>
   9
  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.
  14
  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.
  19
  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 */
  23 %{
  24
  25
  26 #include <string.h>
  27 #include <stdlib.h>
  28 #include <stdio.h>
  29 #include <stdarg.h>
  30 #include "../utf8.h"
  31 #include "tokenizer.h"
  32 #include "files.h"
  33
  34 static void countlines(char*text, int len) {
  35     int t;
  36     for(t=0;t<len;t++) {
  37         if(text[t]=='\n') {
  38             current_line++;
  39             current_column=0;
  40         } else {
  41             current_column++;
  42         }
  43     }
  44 }
  45
  46 static int verbose = 1;
  47 static void dbg(const char*format, ...)
  48 {
  49     char buf[1024];
  50     int l;
  51     va_list arglist;
  52     if(!verbose)
  53         return;
  54     va_start(arglist, format);
  55     vsprintf(buf, format, arglist);
  56     va_end(arglist);
  57     l = strlen(buf);
  58     while(l && buf[l-1]=='\n') {
  59         buf[l-1] = 0;
  60         l--;
  61     }
  62     printf("(tokenizer) ");
  63     printf("%s\n", buf);
  64     fflush(stdout);
  65 }
  66
  67 void syntaxerror(const char*format, ...)
  68 {
  69     char buf[1024];
  70     int l;
  71     va_list arglist;
  72     if(!verbose)
  73         return;
  74     va_start(arglist, format);
  75     vsprintf(buf, format, arglist);
  76     va_end(arglist);
  77     fprintf(stderr, "%s:%d:%d: error: %s\n", current_filename, current_line, current_column, buf);
  78     fflush(stderr);
  79     exit(1);
  80 }
  81
  82
  83 #ifndef YY_CURRENT_BUFFER
  84 #define YY_CURRENT_BUFFER yy_current_buffer
  85 #endif
  86
  87 void handleInclude(char*text, int len, char quotes)
  88 {
  89     char*filename = 0;
  90     if(quotes) {
  91         char*p1 = strchr(text, '"');
  92         char*p2 = strrchr(text, '"');
  93         if(!p1 || !p2 || p1==p2) {
  94             syntaxerror("Invalid include in line %d\n", current_line);
  95         }
  96         *p2 = 0;
  97         filename = strdup(p1+1);
  98     } else {
  99         int i1=0,i2=len;
 100         // find start
 101         while(!strchr(" \n\r\t", text[i1])) i1++;
 102         // strip
 103         while(strchr(" \n\r\t", text[i1])) i1++;
 104         while(strchr(" \n\r\t", text[i2-1])) i2--;
 105         if(i2!=len) text[i2]=0;
 106         filename = strdup(&text[i1]);
 107     }
 108
 109     char*fullfilename = enter_file(filename, YY_CURRENT_BUFFER);
 110     yyin = fopen(fullfilename, "rb");
 111     if (!yyin) {
 112         syntaxerror("Couldn't open include file \"%s\"\n", fullfilename);
 113     }
 114
 115     yy_switch_to_buffer(yy_create_buffer( yyin, YY_BUF_SIZE ) );
 116     //BEGIN(INITIAL); keep context
 117 }
 118
 119 static void handleString(char*s, int len)
 120 {
 121     if(s[0]=='"') {
 122         if(s[len-1]!='"') syntaxerror("String doesn't end with '\"'");
 123         s++;len-=2;
 124     }
 125     else if(s[0]=='\'') {
 126         if(s[len-1]!='\'') syntaxerror("String doesn't end with '\"'");
 127         s++;len-=2;
 128     }
 129     else syntaxerror("String incorrectly terminated");
 130     s[len] = 0;
 131     avm2_lval.string = s;
 132 }
 133
 134
 135 char start_of_expression;
 136
 137 static inline int m(int type)
 138 {
 139     char*s = malloc(yyleng+1);
 140     memcpy(s, yytext, yyleng);
 141     s[yyleng]=0;
 142
 143     NEW(token_t,t);
 144     t->type = type;
 145     t->text = s;
 146     avm2_lval.token = t;
 147     return type;
 148 }
 149
 150 static char numberbuf[64];
 151 static inline int handlenumber()
 152 {
 153     if(yyleng>sizeof(numberbuf)-1)
 154         syntaxerror("decimal number overflow");
 155
 156     char*s = numberbuf;
 157     memcpy(s, yytext, yyleng);
 158     s[yyleng]=0;
 159
 160     int t;
 161     char is_float=0;
 162     for(t=0;t<yyleng;t++) {
 163         if(yytext[t]=='.') {
 164             if(is_float)
 165                 syntaxerror("Invalid number");
 166             is_float=1;
 167         } else if(!strchr("-0123456789", yytext[t])) {
 168             syntaxerror("Invalid number");
 169         }
 170     }
 171     if(is_float) {
 172         avm2_lval.number_float = atof(s);
 173         return T_FLOAT;
 174     }
 175     char l = (yytext[0]=='-');
 176
 177     char*max = l?"1073741824":"2147483647";
 178     if(yyleng-l>10)
 179         syntaxerror("integer overflow");
 180     if(yyleng-l==10) {
 181         int t;
 182         for(t=0;t<yyleng-l;t++) {
 183             if(yytext[l+t]>max[t])
 184                 syntaxerror("integer overflow %s > %s", s+l,max);
 185             else if(yytext[l+t]<max[t])
 186                 break;
 187         }
 188     }
 189     if(yytext[0]=='-') {
 190         int v = atoi(s);
 191         avm2_lval.number_int = v;
 192         if(v>-128)
 193             return T_BYTE;
 194         else if(v>=-32768)
 195             return T_SHORT;
 196         else
 197             return T_INT;
 198     } else {
 199         unsigned int v = 0;
 200         for(t=0;t<yyleng;t++) {
 201             v*=10;
 202             v+=yytext[t]-'0';
 203         }
 204         avm2_lval.number_uint = v;
 205         if(v<128)
 206             return T_BYTE;
 207         else if(v<32768)
 208             return T_SHORT;
 209         else
 210             return T_UINT;
 211     }
 212 }
 213
 214 void initialize_scanner();
 215 #define YY_USER_INIT initialize_scanner();
 216
 217 #define c() {countlines(yytext, yyleng);}
 218
 219 %}
 220
 221 %s REGEXPOK
 222 %s BEGINNING
 223
 224 NAME     [a-zA-Z_][a-zA-Z0-9_\\]*
 225
 226 NUMBER   -?[0-9]+(\.[0-9]*)?
 227
 228 STRING   ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*[']
 229 S        [ \n\r\t]
 230 MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[\x00-\x1f])*[*]+[/]
 231 SINGLELINE_COMMENT \/\/[^\n]*\n
 232 REGEXP   [/]([^/\n]|\\[/])*[/][a-zA-Z]*
 233 %%
 234
 235
 236 {SINGLELINE_COMMENT}         {c(); /* single line comment */}
 237 {MULTILINE_COMMENT}          {c(); /* multi line comment */}
 238 [/][*]                       {syntaxerror("syntax error: unterminated comment", yytext);}
 239
 240 ^include{S}+{STRING}{S}*/\n    {c();handleInclude(yytext, yyleng, 1);}
 241 ^include{S}+[^" \t\r\n][\x20-\xff]*{S}*/\n    {c();handleInclude(yytext, yyleng, 0);}
 242 {STRING}                     {c(); BEGIN(INITIAL);handleString(yytext, yyleng);return T_STRING;}
 243
 244 <BEGINNING,REGEXPOK>{
 245 {REGEXP}                     {c(); BEGIN(INITIAL);return m(T_REGEXP);}
 246 }
 247
 248 \xef\xbb\xbf                 {/* utf 8 bom */}
 249 {S}                          {c();}
 250
 251 {NUMBER}                     {c(); BEGIN(INITIAL);return handlenumber();}
 252
 253 3rr0r                        {/* for debugging: generates a tokenizer-level error */
 254                               syntaxerror("3rr0r");}
 255
 256 [!][=]                       {BEGIN(REGEXPOK);return m(T_NE);}
 257 [=][=][=]                    {BEGIN(REGEXPOK);return m(T_EQEQEQ);}
 258 [=][=]                       {BEGIN(REGEXPOK);return m(T_EQEQ);}
 259 [>][=]                       {return m(T_GE);}
 260 [<][=]                       {return m(T_LE);}
 261 [+][=]                       {return m(T_PLUSBY);}
 262 [-][=]                       {return m(T_MINUSBY);}
 263 [-][-]                       {BEGIN(INITIAL);return m(T_MINUSMINUS);}
 264 [+][+]                       {BEGIN(INITIAL);return m(T_PLUSPLUS);}
 265 \.\.                         {return m(T_DOTDOT);}
 266 \.                           {return m('.');}
 267 ::                           {return m(T_COLONCOLON);}
 268 :                            {return m(':');}
 269 implements                   {return m(KW_IMPLEMENTS);}
 270 interface                    {return m(KW_INTERFACE);}
 271 namespace                    {return m(KW_NAMESPACE);}
 272 protected                    {return m(KW_PROTECTED);}
 273 override                     {return m(KW_OVERRIDE);}
 274 internal                     {return m(KW_INTERNAL);}
 275 function                     {return m(KW_FUNCTION);}
 276 package                      {return m(KW_PACKAGE);}
 277 private                      {return m(KW_PRIVATE);}
 278 Boolean                      {return m(KW_BOOLEAN);}
 279 dynamic                      {return m(KW_DYNAMIC);}
 280 extends                      {return m(KW_EXTENDS);}
 281 public                       {return m(KW_PUBLIC);}
 282 native                       {return m(KW_NATIVE);}
 283 static                       {return m(KW_STATIC);}
 284 import                       {return m(KW_IMPORT);}
 285 Number                       {return m(KW_NUMBER);}
 286 while                        {return m(KW_WHILE);}
 287 class                        {return m(KW_CLASS);}
 288 const                        {return m(KW_CONST);}
 289 final                        {return m(KW_FINAL);}
 290 false                        {return m(KW_FALSE);}
 291 break                        {return m(KW_BREAK);}
 292 true                         {return m(KW_TRUE);}
 293 uint                         {return m(KW_UINT);}
 294 null                         {return m(KW_NULL);}
 295 else                         {return m(KW_ELSE);}
 296 use                          {return m(KW_USE);}
 297 int                          {return m(KW_INT);}
 298 new                          {return m(KW_NEW);}
 299 get                          {return m(KW_GET);}
 300 for                          {return m(KW_FOR);}
 301 set                          {return m(KW_SET);}
 302 var                          {return m(KW_VAR);}
 303 is                           {return m(KW_IS) ;}
 304 if                           {return m(KW_IF) ;}
 305 as                           {return m(KW_AS);}
 306 {NAME}                       {c();BEGIN(INITIAL);return m(T_IDENTIFIER);}
 307
 308 [+-\/*^~@$!%&\(=\[\]\{\}|?:;,.<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
 309 [\)\]]                            {c();BEGIN(INITIAL);return m(yytext[0]);}
 310
 311 .                            {char c1=yytext[0];
 312                               char buf[128];
 313                               buf[0] = yytext[0];
 314                               int t;
 315                               for(t=1;t<128;t++) {
 316                                   char c = buf[t]=input();
 317                                   if(c=='\n' || c==EOF)  {
 318                                       buf[t] = 0;
 319                                       break;
 320                                   }
 321                               }
 322                               if(c1>='0' && c1<='9')
 323                                   syntaxerror("syntax error: %s (identifiers must not start with a digit)");
 324                               else
 325                                   syntaxerror("syntax error: %s", buf);
 326                               printf("\n");
 327                               exit(1);
 328                               yyterminate();
 329                              }
 330 <<EOF>>                      {c();
 331                               void*b = leave_file();
 332                               if (!b) {
 333                                  yyterminate();
 334                                  yy_delete_buffer(YY_CURRENT_BUFFER);
 335                                  return m(T_EOF);
 336                               } else {
 337                                   yy_delete_buffer(YY_CURRENT_BUFFER);
 338                                   yy_switch_to_buffer(b);
 339                               }
 340                              }
 341
 342 %%
 343
 344 int yywrap()
 345 {
 346     return 1;
 347 }
 348
 349 static char mbuf[256];
 350 char*token2string(token_t*t)
 351 {
 352     int nr=t->type;
 353     if(nr==T_STRING)     return "<string>";
 354     else if(nr==T_INT)     return "<int>";
 355     else if(nr==T_UINT)     return "<uint>";
 356     else if(nr==T_FLOAT)     return "<float>";
 357     else if(nr==T_REGEXP)     return "REGEXP";
 358     else if(nr==T_EOF)        return "***END***";
 359     else if(nr==T_GE)         return ">=";
 360     else if(nr==T_LE)         return "<=";
 361     else if(nr==T_MINUSMINUS) return "--";
 362     else if(nr==T_PLUSPLUS)   return "++";
 363     else if(nr==KW_IMPLEMENTS) return "implements";
 364     else if(nr==KW_INTERFACE)  return "interface";
 365     else if(nr==KW_NAMESPACE)  return "namespace";
 366     else if(nr==KW_PROTECTED)  return "protected";
 367     else if(nr==KW_OVERRIDE)   return "override";
 368     else if(nr==KW_INTERNAL)   return "internal";
 369     else if(nr==KW_FUNCTION)   return "function";
 370     else if(nr==KW_PACKAGE)    return "package";
 371     else if(nr==KW_PRIVATE)    return "private";
 372     else if(nr==KW_BOOLEAN)    return "Boolean";
 373     else if(nr==KW_DYNAMIC)    return "dynamic";
 374     else if(nr==KW_EXTENDS)    return "extends";
 375     else if(nr==KW_PUBLIC)     return "public";
 376     else if(nr==KW_NATIVE)     return "native";
 377     else if(nr==KW_STATIC)     return "static";
 378     else if(nr==KW_IMPORT)     return "import";
 379     else if(nr==KW_NUMBER)     return "number";
 380     else if(nr==KW_CLASS)      return "class";
 381     else if(nr==KW_CONST)      return "const";
 382     else if(nr==KW_FINAL)      return "final";
 383     else if(nr==KW_FALSE)      return "False";
 384     else if(nr==KW_TRUE)       return "True";
 385     else if(nr==KW_UINT)       return "uint";
 386     else if(nr==KW_NULL)       return "null";
 387     else if(nr==KW_ELSE)       return "else";
 388     else if(nr==KW_USE)        return "use";
 389     else if(nr==KW_INT)        return "int";
 390     else if(nr==KW_NEW)        return "new";
 391     else if(nr==KW_GET)        return "get";
 392     else if(nr==KW_FOR)        return "for";
 393     else if(nr==KW_SET)        return "set";
 394     else if(nr==KW_VAR)        return "var";
 395     else if(nr==KW_IS)         return "is";
 396     else if(nr==KW_AS)         return "as";
 397     else if(nr==T_IDENTIFIER) {
 398         if(strlen(t->text)>sizeof(mbuf)-1)
 399             return "ID(...)";
 400         sprintf(mbuf, "ID(%s)", t->text);
 401         return mbuf;
 402     } else {
 403         sprintf(mbuf, "%d", nr);
 404         return mbuf;
 405     }
 406 }
 407
 408 void initialize_scanner()
 409 {
 410     BEGIN(BEGINNING);
 411 }
 412