tools/widl/parser.l

   1 /* -*-C-*-
   2  * IDL Compiler
   3  *
   4  * Copyright 2002 Ove Kaaven
   5  *
   6  * This library is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU Lesser General Public
   8  * License as published by the Free Software Foundation; either
   9  * version 2.1 of the License, or (at your option) any later version.
  10  *
  11  * This library is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * Lesser General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with this library; if not, write to the Free Software
  18  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
  19  */
  20
  21 %option stack
  22 %option noinput nounput noyy_top_state
  23 %option 8bit never-interactive prefix="parser_"
  24
  25 nl      \r?\n
  26 ws      [ \f\t\r]
  27 cident  [a-zA-Z_][0-9a-zA-Z_]*
  28 u_suffix        (u|U)
  29 l_suffix        (l|L)
  30 int     [0-9]+({l_suffix}?{u_suffix}?|{u_suffix}?{l_suffix}?)?
  31 hexd    [0-9a-fA-F]
  32 hex     0(x|X){hexd}+({l_suffix}?{u_suffix}?|{u_suffix}?{l_suffix}?)?
  33 uuid    {hexd}{8}-{hexd}{4}-{hexd}{4}-{hexd}{4}-{hexd}{12}
  34 double  [0-9]+\.[0-9]+([eE][+-]?[0-9]+)*
  35
  36 %x QUOTE
  37 %x WSTRQUOTE
  38 %x ATTR
  39 %x PP_LINE
  40 %x PP_PRAGMA
  41 %x SQUOTE
  42
  43 %{
  44
  45 #include "config.h"
  46 #include "wine/port.h"
  47
  48 #include <stdio.h>
  49 #include <stdlib.h>
  50 #include <string.h>
  51 #include <ctype.h>
  52 #include <assert.h>
  53 #include <errno.h>
  54 #include <limits.h>
  55
  56 #ifdef HAVE_UNISTD_H
  57 #include <unistd.h>
  58 #else
  59 #define YY_NO_UNISTD_H
  60 #endif
  61
  62 #include "widl.h"
  63 #include "utils.h"
  64 #include "parser.h"
  65 #include "wine/wpp.h"
  66
  67 #include "parser.tab.h"
  68
  69 static void addcchar(char c);
  70 static char *get_buffered_cstring(void);
  71
  72 static char *cbuffer;
  73 static int cbufidx;
  74 static int cbufalloc = 0;
  75
  76 static int kw_token(const char *kw);
  77 static int attr_token(const char *kw);
  78
  79 #define MAX_IMPORT_DEPTH 10
  80 struct {
  81   YY_BUFFER_STATE state;
  82   char *input_name;
  83   int   line_number;
  84   char *temp_name;
  85 } import_stack[MAX_IMPORT_DEPTH];
  86 int import_stack_ptr = 0;
  87
  88 /* converts an integer in string form to an unsigned long and prints an error
  89  * on overflow */
  90 static unsigned int xstrtoul(const char *nptr, char **endptr, int base)
  91 {
  92     unsigned long val;
  93
  94     errno = 0;
  95     val = strtoul(nptr, endptr, base);
  96     if ((val == ULONG_MAX && errno == ERANGE) || ((unsigned int)val != val))
  97         error_loc("integer constant %s is too large\n", nptr);
  98     return val;
  99 }
 100
 101 UUID *parse_uuid(const char *u)
 102 {
 103   UUID* uuid = xmalloc(sizeof(UUID));
 104   char b[3];
 105   /* it would be nice to use UuidFromStringA */
 106   uuid->Data1 = strtoul(u, NULL, 16);
 107   uuid->Data2 = strtoul(u+9, NULL, 16);
 108   uuid->Data3 = strtoul(u+14, NULL, 16);
 109   b[2] = 0;
 110   memcpy(b, u+19, 2); uuid->Data4[0] = strtoul(b, NULL, 16);
 111   memcpy(b, u+21, 2); uuid->Data4[1] = strtoul(b, NULL, 16);
 112   memcpy(b, u+24, 2); uuid->Data4[2] = strtoul(b, NULL, 16);
 113   memcpy(b, u+26, 2); uuid->Data4[3] = strtoul(b, NULL, 16);
 114   memcpy(b, u+28, 2); uuid->Data4[4] = strtoul(b, NULL, 16);
 115   memcpy(b, u+30, 2); uuid->Data4[5] = strtoul(b, NULL, 16);
 116   memcpy(b, u+32, 2); uuid->Data4[6] = strtoul(b, NULL, 16);
 117   memcpy(b, u+34, 2); uuid->Data4[7] = strtoul(b, NULL, 16);
 118   return uuid;
 119 }
 120
 121 %}
 122
 123 /*
 124  **************************************************************************
 125  * The flexer starts here
 126  **************************************************************************
 127  */
 128 %%
 129 <INITIAL>^{ws}*\#{ws}*pragma{ws}+ yy_push_state(PP_PRAGMA);
 130 <INITIAL,ATTR>^{ws}*\#{ws}*     yy_push_state(PP_LINE);
 131 <PP_LINE>[^\n]*         {
 132                             int lineno;
 133                             char *cptr, *fname;
 134                             yy_pop_state();
 135                             lineno = (int)strtol(yytext, &cptr, 10);
 136                             if(!lineno)
 137                                 error_loc("Malformed '#...' line-directive; invalid linenumber\n");
 138                             fname = strchr(cptr, '"');
 139                             if(!fname)
 140                                 error_loc("Malformed '#...' line-directive; missing filename\n");
 141                             fname++;
 142                             cptr = strchr(fname, '"');
 143                             if(!cptr)
 144                                 error_loc("Malformed '#...' line-directive; missing terminating \"\n");
 145                             *cptr = '\0';
 146                             line_number = lineno - 1;  /* We didn't read the newline */
 147                             input_name = xstrdup(fname);
 148                         }
 149 <PP_PRAGMA>midl_echo[^\n]*  yyless(9); yy_pop_state(); return tCPPQUOTE;
 150 <PP_PRAGMA>winrt[^\n]*  {
 151                             if(import_stack_ptr) {
 152                                 if(!winrt_mode)
 153                                     error_loc("winrt IDL file imported in non-winrt mode\n");
 154                             }else {
 155                                 const char *ptr = yytext+5;
 156
 157                                 winrt_mode = TRUE;
 158
 159                                 while(isspace(*ptr))
 160                                     ptr++;
 161                                 if(!strncmp(ptr, "ns_prefix", 9) && (!*(ptr += 9) || isspace(*ptr)))
 162                                     use_abi_namespace = TRUE;
 163                             }
 164                             yy_pop_state();
 165                         }
 166 <PP_PRAGMA>[^\n]*       parser_lval.str = xstrdup(yytext); yy_pop_state(); return aPRAGMA;
 167 <INITIAL,ATTR>\"        yy_push_state(QUOTE); cbufidx = 0;
 168 <QUOTE>\"               {
 169                                 yy_pop_state();
 170                                 parser_lval.str = get_buffered_cstring();
 171                                 return aSTRING;
 172                         }
 173 <INITIAL,ATTR>L\"       yy_push_state(WSTRQUOTE); cbufidx = 0;
 174 <WSTRQUOTE>\"           {
 175                                 yy_pop_state();
 176                                 parser_lval.str = get_buffered_cstring();
 177                                 return aWSTRING;
 178                         }
 179 <INITIAL,ATTR>\'        yy_push_state(SQUOTE); cbufidx = 0;
 180 <SQUOTE>\'              {
 181                                 yy_pop_state();
 182                                 parser_lval.str = get_buffered_cstring();
 183                                 return aSQSTRING;
 184                         }
 185 <QUOTE,WSTRQUOTE,SQUOTE>\\\\    |
 186 <QUOTE,WSTRQUOTE>\\\"   addcchar(yytext[1]);
 187 <SQUOTE>\\\'    addcchar(yytext[1]);
 188 <QUOTE,WSTRQUOTE,SQUOTE>\\.     addcchar('\\'); addcchar(yytext[1]);
 189 <QUOTE,WSTRQUOTE,SQUOTE>.       addcchar(yytext[0]);
 190 <INITIAL,ATTR>\[        yy_push_state(ATTR); return '[';
 191 <ATTR>\]                yy_pop_state(); return ']';
 192 <ATTR>{cident}          return attr_token(yytext);
 193 <ATTR>{uuid}                    {
 194                                 parser_lval.uuid = parse_uuid(yytext);
 195                                 return aUUID;
 196                         }
 197 <INITIAL,ATTR>{hex}     {
 198                                 parser_lval.num = xstrtoul(yytext, NULL, 0);
 199                                 return aHEXNUM;
 200                         }
 201 <INITIAL,ATTR>{int}     {
 202                                 parser_lval.num = xstrtoul(yytext, NULL, 0);
 203                                 return aNUM;
 204                         }
 205 <INITIAL>{double}       {
 206                                 parser_lval.dbl = strtod(yytext, NULL);
 207                                 return aDOUBLE;
 208                         }
 209 SAFEARRAY{ws}*/\(       return tSAFEARRAY;
 210 {cident}                return kw_token(yytext);
 211 <INITIAL,ATTR>\n        line_number++;
 212 <INITIAL,ATTR>{ws}
 213 <INITIAL,ATTR>\<\<      return SHL;
 214 <INITIAL,ATTR>\>\>      return SHR;
 215 <INITIAL,ATTR>\-\>      return MEMBERPTR;
 216 <INITIAL,ATTR>==        return EQUALITY;
 217 <INITIAL,ATTR>!=        return INEQUALITY;
 218 <INITIAL,ATTR>\>=       return GREATEREQUAL;
 219 <INITIAL,ATTR>\<=       return LESSEQUAL;
 220 <INITIAL,ATTR>\|\|      return LOGICALOR;
 221 <INITIAL,ATTR>&&        return LOGICALAND;
 222 <INITIAL,ATTR>\.\.\.    return ELLIPSIS;
 223 <INITIAL,ATTR>.         return yytext[0];
 224 <<EOF>>                 {
 225                                 if (import_stack_ptr)
 226                                         return aEOF;
 227                                 else yyterminate();
 228                         }
 229 %%
 230
 231 #ifndef parser_wrap
 232 int parser_wrap(void)
 233 {
 234         return 1;
 235 }
 236 #endif
 237
 238 struct keyword {
 239         const char *kw;
 240         int token;
 241 };
 242
 243 /* This table MUST be alphabetically sorted on the kw field */
 244 static const struct keyword keywords[] = {
 245         {"FALSE",                       tFALSE},
 246         {"NULL",                        tNULL},
 247         {"TRUE",                        tTRUE},
 248         {"__cdecl",                     tCDECL},
 249         {"__fastcall",                  tFASTCALL},
 250         {"__int3264",                   tINT3264},
 251         {"__int64",                     tINT64},
 252         {"__pascal",                    tPASCAL},
 253         {"__stdcall",                   tSTDCALL},
 254         {"_cdecl",                      tCDECL},
 255         {"_fastcall",                   tFASTCALL},
 256         {"_pascal",                     tPASCAL},
 257         {"_stdcall",                    tSTDCALL},
 258         {"boolean",                     tBOOLEAN},
 259         {"byte",                        tBYTE},
 260         {"case",                        tCASE},
 261         {"cdecl",                       tCDECL},
 262         {"char",                        tCHAR},
 263         {"coclass",                     tCOCLASS},
 264         {"const",                       tCONST},
 265         {"cpp_quote",                   tCPPQUOTE},
 266         {"default",                     tDEFAULT},
 267         {"dispinterface",               tDISPINTERFACE},
 268         {"double",                      tDOUBLE},
 269         {"enum",                        tENUM},
 270         {"error_status_t",              tERRORSTATUST},
 271         {"extern",                      tEXTERN},
 272         {"float",                       tFLOAT},
 273         {"handle_t",                    tHANDLET},
 274         {"hyper",                       tHYPER},
 275         {"import",                      tIMPORT},
 276         {"importlib",                   tIMPORTLIB},
 277         {"inline",                      tINLINE},
 278         {"int",                         tINT},
 279         {"interface",                   tINTERFACE},
 280         {"library",                     tLIBRARY},
 281         {"long",                        tLONG},
 282         {"methods",                     tMETHODS},
 283         {"module",                      tMODULE},
 284         {"namespace",                   tNAMESPACE},
 285         {"pascal",                      tPASCAL},
 286         {"properties",                  tPROPERTIES},
 287         {"register",                    tREGISTER},
 288         {"short",                       tSHORT},
 289         {"signed",                      tSIGNED},
 290         {"sizeof",                      tSIZEOF},
 291         {"small",                       tSMALL},
 292         {"static",                      tSTATIC},
 293         {"stdcall",                     tSTDCALL},
 294         {"struct",                      tSTRUCT},
 295         {"switch",                      tSWITCH},
 296         {"typedef",                     tTYPEDEF},
 297         {"union",                       tUNION},
 298         {"unsigned",                    tUNSIGNED},
 299         {"void",                        tVOID},
 300         {"wchar_t",                     tWCHAR},
 301 };
 302 #define NKEYWORDS (sizeof(keywords)/sizeof(keywords[0]))
 303
 304 /* keywords only recognized in attribute lists
 305  * This table MUST be alphabetically sorted on the kw field
 306  */
 307 static const struct keyword attr_keywords[] =
 308 {
 309         {"aggregatable",                tAGGREGATABLE},
 310         {"allocate",                    tALLOCATE},
 311         {"annotation",                  tANNOTATION},
 312         {"apartment",                   tAPARTMENT},
 313         {"appobject",                   tAPPOBJECT},
 314         {"async",                       tASYNC},
 315         {"async_uuid",                  tASYNCUUID},
 316         {"auto_handle",                 tAUTOHANDLE},
 317         {"bindable",                    tBINDABLE},
 318         {"both",                        tBOTH},
 319         {"broadcast",                   tBROADCAST},
 320         {"byte_count",                  tBYTECOUNT},
 321         {"call_as",                     tCALLAS},
 322         {"callback",                    tCALLBACK},
 323         {"code",                        tCODE},
 324         {"comm_status",                 tCOMMSTATUS},
 325         {"context_handle",              tCONTEXTHANDLE},
 326         {"context_handle_noserialize",  tCONTEXTHANDLENOSERIALIZE},
 327         {"context_handle_serialize",    tCONTEXTHANDLENOSERIALIZE},
 328         {"control",                     tCONTROL},
 329         {"decode",                      tDECODE},
 330         {"defaultbind",                 tDEFAULTBIND},
 331         {"defaultcollelem",             tDEFAULTCOLLELEM},
 332         {"defaultvalue",                tDEFAULTVALUE},
 333         {"defaultvtable",               tDEFAULTVTABLE},
 334         {"disable_consistency_check",   tDISABLECONSISTENCYCHECK},
 335         {"displaybind",                 tDISPLAYBIND},
 336         {"dllname",                     tDLLNAME},
 337         {"dual",                        tDUAL},
 338         {"enable_allocate",             tENABLEALLOCATE},
 339         {"encode",                      tENCODE},
 340         {"endpoint",                    tENDPOINT},
 341         {"entry",                       tENTRY},
 342         {"explicit_handle",             tEXPLICITHANDLE},
 343         {"fault_status",                tFAULTSTATUS},
 344         {"force_allocate",              tFORCEALLOCATE},
 345         {"free",                        tFREE},
 346         {"handle",                      tHANDLE},
 347         {"helpcontext",                 tHELPCONTEXT},
 348         {"helpfile",                    tHELPFILE},
 349         {"helpstring",                  tHELPSTRING},
 350         {"helpstringcontext",           tHELPSTRINGCONTEXT},
 351         {"helpstringdll",               tHELPSTRINGDLL},
 352         {"hidden",                      tHIDDEN},
 353         {"id",                          tID},
 354         {"idempotent",                  tIDEMPOTENT},
 355         {"ignore",                      tIGNORE},
 356         {"iid_is",                      tIIDIS},
 357         {"immediatebind",               tIMMEDIATEBIND},
 358         {"implicit_handle",             tIMPLICITHANDLE},
 359         {"in",                          tIN},
 360         {"in_line",                     tIN_LINE},
 361         {"input_sync",                  tINPUTSYNC},
 362         {"lcid",                        tLCID},
 363         {"length_is",                   tLENGTHIS},
 364         {"licensed",                    tLICENSED},
 365         {"local",                       tLOCAL},
 366         {"maybe",                       tMAYBE},
 367         {"message",                     tMESSAGE},
 368         {"neutral",                     tNEUTRAL},
 369         {"nocode",                      tNOCODE},
 370         {"nonbrowsable",                tNONBROWSABLE},
 371         {"noncreatable",                tNONCREATABLE},
 372         {"nonextensible",               tNONEXTENSIBLE},
 373         {"notify",                      tNOTIFY},
 374         {"notify_flag",                 tNOTIFYFLAG},
 375         {"object",                      tOBJECT},
 376         {"odl",                         tODL},
 377         {"oleautomation",               tOLEAUTOMATION},
 378         {"optimize",                    tOPTIMIZE},
 379         {"optional",                    tOPTIONAL},
 380         {"out",                         tOUT},
 381         {"partial_ignore",              tPARTIALIGNORE},
 382         {"pointer_default",             tPOINTERDEFAULT},
 383         {"progid",                      tPROGID},
 384         {"propget",                     tPROPGET},
 385         {"propput",                     tPROPPUT},
 386         {"propputref",                  tPROPPUTREF},
 387         {"proxy",                       tPROXY},
 388         {"ptr",                         tPTR},
 389         {"public",                      tPUBLIC},
 390         {"range",                       tRANGE},
 391         {"readonly",                    tREADONLY},
 392         {"ref",                         tREF},
 393         {"represent_as",                tREPRESENTAS},
 394         {"requestedit",                 tREQUESTEDIT},
 395         {"restricted",                  tRESTRICTED},
 396         {"retval",                      tRETVAL},
 397         {"single",                      tSINGLE},
 398         {"size_is",                     tSIZEIS},
 399         {"source",                      tSOURCE},
 400         {"strict_context_handle",       tSTRICTCONTEXTHANDLE},
 401         {"string",                      tSTRING},
 402         {"switch_is",                   tSWITCHIS},
 403         {"switch_type",                 tSWITCHTYPE},
 404         {"threading",                   tTHREADING},
 405         {"transmit_as",                 tTRANSMITAS},
 406         {"uidefault",                   tUIDEFAULT},
 407         {"unique",                      tUNIQUE},
 408         {"user_marshal",                tUSERMARSHAL},
 409         {"usesgetlasterror",            tUSESGETLASTERROR},
 410         {"uuid",                        tUUID},
 411         {"v1_enum",                     tV1ENUM},
 412         {"vararg",                      tVARARG},
 413         {"version",                     tVERSION},
 414         {"vi_progid",                   tVIPROGID},
 415         {"wire_marshal",                tWIREMARSHAL},
 416 };
 417
 418 /* attributes TODO:
 419     custom
 420     first_is
 421     last_is
 422     max_is
 423     min_is
 424 */
 425
 426 #define KWP(p) ((const struct keyword *)(p))
 427
 428 static int kw_cmp_func(const void *s1, const void *s2)
 429 {
 430         return strcmp(KWP(s1)->kw, KWP(s2)->kw);
 431 }
 432
 433 static int kw_token(const char *kw)
 434 {
 435         struct keyword key, *kwp;
 436         key.kw = kw;
 437         kwp = bsearch(&key, keywords, NKEYWORDS, sizeof(keywords[0]), kw_cmp_func);
 438         if (kwp && (winrt_mode || kwp->token != tNAMESPACE)) {
 439                 parser_lval.str = xstrdup(kwp->kw);
 440                 return kwp->token;
 441         }
 442         parser_lval.str = xstrdup(kw);
 443         return is_type(kw) ? aKNOWNTYPE : aIDENTIFIER;
 444 }
 445
 446 static int attr_token(const char *kw)
 447 {
 448         struct keyword key, *kwp;
 449         key.kw = kw;
 450         kwp = bsearch(&key, attr_keywords, sizeof(attr_keywords)/sizeof(attr_keywords[0]),
 451                       sizeof(attr_keywords[0]), kw_cmp_func);
 452         if (kwp) {
 453             parser_lval.str = xstrdup(kwp->kw);
 454             return kwp->token;
 455         }
 456         return kw_token(kw);
 457 }
 458
 459 static void addcchar(char c)
 460 {
 461         if(cbufidx >= cbufalloc)
 462         {
 463                 cbufalloc += 1024;
 464                 cbuffer = xrealloc(cbuffer, cbufalloc * sizeof(cbuffer[0]));
 465                 if(cbufalloc > 65536)
 466                         parser_warning("Reallocating string buffer larger than 64kB\n");
 467         }
 468         cbuffer[cbufidx++] = c;
 469 }
 470
 471 static char *get_buffered_cstring(void)
 472 {
 473         addcchar(0);
 474         return xstrdup(cbuffer);
 475 }
 476
 477 void pop_import(void)
 478 {
 479         int ptr = import_stack_ptr-1;
 480
 481         fclose(yyin);
 482         yy_delete_buffer( YY_CURRENT_BUFFER );
 483         yy_switch_to_buffer( import_stack[ptr].state );
 484         if (temp_name) {
 485                 unlink(temp_name);
 486                 free(temp_name);
 487         }
 488         temp_name = import_stack[ptr].temp_name;
 489         input_name = import_stack[ptr].input_name;
 490         line_number = import_stack[ptr].line_number;
 491         import_stack_ptr--;
 492 }
 493
 494 struct imports {
 495         char *name;
 496         struct imports *next;
 497 } *first_import;
 498
 499 int do_import(char *fname)
 500 {
 501         FILE *f;
 502         char *path, *name;
 503         struct imports *import;
 504         int ptr = import_stack_ptr;
 505         int ret, fd;
 506
 507         import = first_import;
 508         while (import && strcmp(import->name, fname))
 509                 import = import->next;
 510         if (import) return 0; /* already imported */
 511
 512         import = xmalloc(sizeof(struct imports));
 513         import->name = xstrdup(fname);
 514         import->next = first_import;
 515         first_import = import;
 516
 517         /* don't search for a file name with a path in the include directories,
 518          * for compatibility with MIDL */
 519         if (strchr( fname, '/' ) || strchr( fname, '\\' ))
 520             path = xstrdup( fname );
 521         else if (!(path = wpp_find_include( fname, input_name )))
 522             error_loc("Unable to open include file %s\n", fname);
 523
 524         import_stack[ptr].temp_name = temp_name;
 525         import_stack[ptr].input_name = input_name;
 526         import_stack[ptr].line_number = line_number;
 527         import_stack_ptr++;
 528         input_name = path;
 529         line_number = 1;
 530
 531         name = xstrdup( "widl.XXXXXX" );
 532         if((fd = mkstemps( name, 0 )) == -1)
 533             error("Could not generate a temp name from %s\n", name);
 534
 535         temp_name = name;
 536         if (!(f = fdopen(fd, "wt")))
 537             error("Could not open fd %s for writing\n", name);
 538
 539         ret = wpp_parse( path, f );
 540         fclose( f );
 541         if (ret) exit(1);
 542
 543         if((f = fopen(temp_name, "r")) == NULL)
 544                 error_loc("Unable to open %s\n", temp_name);
 545
 546         import_stack[ptr].state = YY_CURRENT_BUFFER;
 547         yy_switch_to_buffer(yy_create_buffer(f, YY_BUF_SIZE));
 548         return 1;
 549 }
 550
 551 void abort_import(void)
 552 {
 553         int ptr;
 554
 555         for (ptr=0; ptr<import_stack_ptr; ptr++)
 556                 unlink(import_stack[ptr].temp_name);
 557 }