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
  41 %{
  42
  43 #include "config.h"
  44
  45 #include <stdio.h>
  46 #include <stdlib.h>
  47 #include <string.h>
  48 #include <ctype.h>
  49 #include <assert.h>
  50 #include <errno.h>
  51 #include <limits.h>
  52
  53 #ifdef HAVE_UNISTD_H
  54 #include <unistd.h>
  55 #else
  56 #define YY_NO_UNISTD_H
  57 #endif
  58
  59 #include "widl.h"
  60 #include "utils.h"
  61 #include "parser.h"
  62 #include "wine/wpp.h"
  63
  64 #include "parser.tab.h"
  65
  66 extern char *temp_name;
  67
  68 static void addcchar(char c);
  69 static char *get_buffered_cstring(void);
  70
  71 static char *cbuffer;
  72 static int cbufidx;
  73 static int cbufalloc = 0;
  74
  75 static int kw_token(const char *kw);
  76 static int attr_token(const char *kw);
  77
  78 #define MAX_IMPORT_DEPTH 10
  79 struct {
  80   YY_BUFFER_STATE state;
  81   char *input_name;
  82   int   line_number;
  83   char *temp_name;
  84 } import_stack[MAX_IMPORT_DEPTH];
  85 int import_stack_ptr = 0;
  86
  87 /* converts an integer in string form to an unsigned long and prints an error
  88  * on overflow */
  89 static unsigned long xstrtoul(const char *nptr, char **endptr, int base)
  90 {
  91     unsigned long l;
  92
  93     errno = 0;
  94     l = strtoul(nptr, endptr, base);
  95     if (l == ULONG_MAX && errno == ERANGE)
  96         error_loc("integer constant %s is too large\n", nptr);
  97     return l;
  98 }
  99
 100 UUID *parse_uuid(const char *u)
 101 {
 102   UUID* uuid = xmalloc(sizeof(UUID));
 103   char b[3];
 104   /* it would be nice to use UuidFromStringA */
 105   uuid->Data1 = strtoul(u, NULL, 16);
 106   uuid->Data2 = strtoul(u+9, NULL, 16);
 107   uuid->Data3 = strtoul(u+14, NULL, 16);
 108   b[2] = 0;
 109   memcpy(b, u+19, 2); uuid->Data4[0] = strtoul(b, NULL, 16);
 110   memcpy(b, u+21, 2); uuid->Data4[1] = strtoul(b, NULL, 16);
 111   memcpy(b, u+24, 2); uuid->Data4[2] = strtoul(b, NULL, 16);
 112   memcpy(b, u+26, 2); uuid->Data4[3] = strtoul(b, NULL, 16);
 113   memcpy(b, u+28, 2); uuid->Data4[4] = strtoul(b, NULL, 16);
 114   memcpy(b, u+30, 2); uuid->Data4[5] = strtoul(b, NULL, 16);
 115   memcpy(b, u+32, 2); uuid->Data4[6] = strtoul(b, NULL, 16);
 116   memcpy(b, u+34, 2); uuid->Data4[7] = strtoul(b, NULL, 16);
 117   return uuid;
 118 }
 119
 120 %}
 121
 122 /*
 123  **************************************************************************
 124  * The flexer starts here
 125  **************************************************************************
 126  */
 127 %%
 128 <INITIAL,ATTR>^{ws}*\#{ws}*     yy_push_state(PP_LINE);
 129 <PP_LINE>[^\n]*         {
 130                             int lineno;
 131                             char *cptr, *fname;
 132                             yy_pop_state();
 133                             lineno = (int)strtol(yytext, &cptr, 10);
 134                             if(!lineno)
 135                                 error_loc("Malformed '#...' line-directive; invalid linenumber\n");
 136                             fname = strchr(cptr, '"');
 137                             if(!fname)
 138                                 error_loc("Malformed '#...' line-directive; missing filename\n");
 139                             fname++;
 140                             cptr = strchr(fname, '"');
 141                             if(!cptr)
 142                                 error_loc("Malformed '#...' line-directive; missing terminating \"\n");
 143                             *cptr = '\0';
 144                             line_number = lineno - 1;  /* We didn't read the newline */
 145                             free( input_name );
 146                             input_name = xstrdup(fname);
 147                         }
 148 <INITIAL,ATTR>\"        yy_push_state(QUOTE); cbufidx = 0;
 149 <QUOTE>\"               {
 150                                 yy_pop_state();
 151                                 parser_lval.str = get_buffered_cstring();
 152                                 return aSTRING;
 153                         }
 154 <INITIAL,ATTR>L\"       yy_push_state(WSTRQUOTE);
 155 <WSTRQUOTE>\"           {
 156                                 yy_pop_state();
 157                                 parser_lval.str = get_buffered_cstring();
 158                                 return aWSTRING;
 159                         }
 160 <QUOTE,WSTRQUOTE>\\\\   |
 161 <QUOTE,WSTRQUOTE>\\\"   addcchar(yytext[1]);
 162 <QUOTE,WSTRQUOTE>\\.    addcchar('\\'); addcchar(yytext[1]);
 163 <QUOTE,WSTRQUOTE>.      addcchar(yytext[0]);
 164 <INITIAL,ATTR>\[        yy_push_state(ATTR); return '[';
 165 <ATTR>\]                yy_pop_state(); return ']';
 166 <ATTR>{cident}          return attr_token(yytext);
 167 <ATTR>{uuid}                    {
 168                                 parser_lval.uuid = parse_uuid(yytext);
 169                                 return aUUID;
 170                         }
 171 <INITIAL,ATTR>{hex}     {
 172                                 parser_lval.num = xstrtoul(yytext, NULL, 0);
 173                                 return aHEXNUM;
 174                         }
 175 <INITIAL,ATTR>{int}     {
 176                                 parser_lval.num = xstrtoul(yytext, NULL, 0);
 177                                 return aNUM;
 178                         }
 179 <INITIAL>{double}       {
 180                                 parser_lval.dbl = strtod(yytext, NULL);
 181                                 return aDOUBLE;
 182                         }
 183 SAFEARRAY{ws}*/\(       return tSAFEARRAY;
 184 {cident}                return kw_token(yytext);
 185 <INITIAL,ATTR>\n        line_number++;
 186 <INITIAL,ATTR>{ws}
 187 <INITIAL,ATTR>\<\<      return SHL;
 188 <INITIAL,ATTR>\>\>      return SHR;
 189 <INITIAL,ATTR>\-\>      return MEMBERPTR;
 190 <INITIAL,ATTR>==        return EQUALITY;
 191 <INITIAL,ATTR>!=        return INEQUALITY;
 192 <INITIAL,ATTR>\>=       return GREATEREQUAL;
 193 <INITIAL,ATTR>\<=       return LESSEQUAL;
 194 <INITIAL,ATTR>\|\|      return LOGICALOR;
 195 <INITIAL,ATTR>&&        return LOGICALAND;
 196 <INITIAL,ATTR>.         return yytext[0];
 197 <<EOF>>                 {
 198                                 if (import_stack_ptr)
 199                                         return aEOF;
 200                                 else yyterminate();
 201                         }
 202 %%
 203
 204 #ifndef parser_wrap
 205 int parser_wrap(void)
 206 {
 207         return 1;
 208 }
 209 #endif
 210
 211 struct keyword {
 212         const char *kw;
 213         int token;
 214 };
 215
 216 /* This table MUST be alphabetically sorted on the kw field */
 217 static const struct keyword keywords[] = {
 218         {"FALSE",                       tFALSE},
 219         {"NULL",                        tNULL},
 220         {"TRUE",                        tTRUE},
 221         {"__cdecl",                     tCDECL},
 222         {"__fastcall",                  tFASTCALL},
 223         {"__int64",                     tINT64},
 224         {"__pascal",                    tPASCAL},
 225         {"__stdcall",                   tSTDCALL},
 226         {"_cdecl",                      tCDECL},
 227         {"_fastcall",                   tFASTCALL},
 228         {"_pascal",                     tPASCAL},
 229         {"_stdcall",                    tSTDCALL},
 230         {"boolean",                     tBOOLEAN},
 231         {"byte",                        tBYTE},
 232         {"case",                        tCASE},
 233         {"cdecl",                       tCDECL},
 234         {"char",                        tCHAR},
 235         {"coclass",                     tCOCLASS},
 236         {"const",                       tCONST},
 237         {"cpp_quote",                   tCPPQUOTE},
 238         {"default",                     tDEFAULT},
 239         {"dispinterface",               tDISPINTERFACE},
 240         {"double",                      tDOUBLE},
 241         {"enum",                        tENUM},
 242         {"error_status_t",              tERRORSTATUST},
 243         {"extern",                      tEXTERN},
 244         {"float",                       tFLOAT},
 245         {"handle_t",                    tHANDLET},
 246         {"hyper",                       tHYPER},
 247         {"import",                      tIMPORT},
 248         {"importlib",                   tIMPORTLIB},
 249         {"inline",                      tINLINE},
 250         {"int",                         tINT},
 251         {"interface",                   tINTERFACE},
 252         {"library",                     tLIBRARY},
 253         {"long",                        tLONG},
 254         {"methods",                     tMETHODS},
 255         {"module",                      tMODULE},
 256         {"pascal",                      tPASCAL},
 257         {"properties",                  tPROPERTIES},
 258         {"register",                    tREGISTER},
 259         {"short",                       tSHORT},
 260         {"signed",                      tSIGNED},
 261         {"sizeof",                      tSIZEOF},
 262         {"small",                       tSMALL},
 263         {"static",                      tSTATIC},
 264         {"stdcall",                     tSTDCALL},
 265         {"struct",                      tSTRUCT},
 266         {"switch",                      tSWITCH},
 267         {"typedef",                     tTYPEDEF},
 268         {"union",                       tUNION},
 269         {"unsigned",                    tUNSIGNED},
 270         {"void",                        tVOID},
 271         {"wchar_t",                     tWCHAR},
 272 };
 273 #define NKEYWORDS (sizeof(keywords)/sizeof(keywords[0]))
 274
 275 /* keywords only recognized in attribute lists
 276  * This table MUST be alphabetically sorted on the kw field
 277  */
 278 static const struct keyword attr_keywords[] =
 279 {
 280         {"aggregatable",                tAGGREGATABLE},
 281         {"allocate",                    tALLOCATE},
 282         {"appobject",                   tAPPOBJECT},
 283         {"async",                       tASYNC},
 284         {"async_uuid",                  tASYNCUUID},
 285         {"auto_handle",                 tAUTOHANDLE},
 286         {"bindable",                    tBINDABLE},
 287         {"broadcast",                   tBROADCAST},
 288         {"byte_count",                  tBYTECOUNT},
 289         {"call_as",                     tCALLAS},
 290         {"callback",                    tCALLBACK},
 291         {"code",                        tCODE},
 292         {"comm_status",                 tCOMMSTATUS},
 293         {"context_handle",              tCONTEXTHANDLE},
 294         {"context_handle_noserialize",  tCONTEXTHANDLENOSERIALIZE},
 295         {"context_handle_serialize",    tCONTEXTHANDLENOSERIALIZE},
 296         {"control",                     tCONTROL},
 297         {"defaultcollelem",             tDEFAULTCOLLELEM},
 298         {"defaultvalue",                tDEFAULTVALUE},
 299         {"defaultvtable",               tDEFAULTVTABLE},
 300         {"displaybind",                 tDISPLAYBIND},
 301         {"dllname",                     tDLLNAME},
 302         {"dual",                        tDUAL},
 303         {"endpoint",                    tENDPOINT},
 304         {"entry",                       tENTRY},
 305         {"explicit_handle",             tEXPLICITHANDLE},
 306         {"handle",                      tHANDLE},
 307         {"helpcontext",                 tHELPCONTEXT},
 308         {"helpfile",                    tHELPFILE},
 309         {"helpstring",                  tHELPSTRING},
 310         {"helpstringcontext",           tHELPSTRINGCONTEXT},
 311         {"helpstringdll",               tHELPSTRINGDLL},
 312         {"hidden",                      tHIDDEN},
 313         {"id",                          tID},
 314         {"idempotent",                  tIDEMPOTENT},
 315         {"iid_is",                      tIIDIS},
 316         {"immediatebind",               tIMMEDIATEBIND},
 317         {"implicit_handle",             tIMPLICITHANDLE},
 318         {"in",                          tIN},
 319         {"in_line",                     tIN_LINE},
 320         {"input_sync",                  tINPUTSYNC},
 321         {"lcid",                        tLCID},
 322         {"length_is",                   tLENGTHIS},
 323         {"local",                       tLOCAL},
 324         {"nonbrowsable",                tNONBROWSABLE},
 325         {"noncreatable",                tNONCREATABLE},
 326         {"nonextensible",               tNONEXTENSIBLE},
 327         {"object",                      tOBJECT},
 328         {"odl",                         tODL},
 329         {"oleautomation",               tOLEAUTOMATION},
 330         {"optional",                    tOPTIONAL},
 331         {"out",                         tOUT},
 332         {"pointer_default",             tPOINTERDEFAULT},
 333         {"propget",                     tPROPGET},
 334         {"propput",                     tPROPPUT},
 335         {"propputref",                  tPROPPUTREF},
 336         {"ptr",                         tPTR},
 337         {"public",                      tPUBLIC},
 338         {"range",                       tRANGE},
 339         {"readonly",                    tREADONLY},
 340         {"ref",                         tREF},
 341         {"requestedit",                 tREQUESTEDIT},
 342         {"restricted",                  tRESTRICTED},
 343         {"retval",                      tRETVAL},
 344         {"single",                      tSINGLE},
 345         {"size_is",                     tSIZEIS},
 346         {"source",                      tSOURCE},
 347         {"strict_context_handle",       tSTRICTCONTEXTHANDLE},
 348         {"string",                      tSTRING},
 349         {"switch_is",                   tSWITCHIS},
 350         {"switch_type",                 tSWITCHTYPE},
 351         {"transmit_as",                 tTRANSMITAS},
 352         {"unique",                      tUNIQUE},
 353         {"uuid",                        tUUID},
 354         {"v1_enum",                     tV1ENUM},
 355         {"vararg",                      tVARARG},
 356         {"version",                     tVERSION},
 357         {"wire_marshal",                tWIREMARSHAL},
 358 };
 359
 360
 361 #define KWP(p) ((const struct keyword *)(p))
 362
 363 static int kw_cmp_func(const void *s1, const void *s2)
 364 {
 365         return strcmp(KWP(s1)->kw, KWP(s2)->kw);
 366 }
 367
 368 static int kw_token(const char *kw)
 369 {
 370         struct keyword key, *kwp;
 371         key.kw = kw;
 372         kwp = bsearch(&key, keywords, NKEYWORDS, sizeof(keywords[0]), kw_cmp_func);
 373         if (kwp) {
 374                 parser_lval.str = xstrdup(kwp->kw);
 375                 return kwp->token;
 376         }
 377         parser_lval.str = xstrdup(kw);
 378         return is_type(kw) ? aKNOWNTYPE : aIDENTIFIER;
 379 }
 380
 381 static int attr_token(const char *kw)
 382 {
 383         struct keyword key, *kwp;
 384         key.kw = kw;
 385         kwp = bsearch(&key, attr_keywords, sizeof(attr_keywords)/sizeof(attr_keywords[0]),
 386                       sizeof(attr_keywords[0]), kw_cmp_func);
 387         if (kwp) {
 388             parser_lval.str = xstrdup(kwp->kw);
 389             return kwp->token;
 390         }
 391         return kw_token(kw);
 392 }
 393
 394 static void addcchar(char c)
 395 {
 396         if(cbufidx >= cbufalloc)
 397         {
 398                 cbufalloc += 1024;
 399                 cbuffer = xrealloc(cbuffer, cbufalloc * sizeof(cbuffer[0]));
 400                 if(cbufalloc > 65536)
 401                         parser_warning("Reallocating string buffer larger than 64kB\n");
 402         }
 403         cbuffer[cbufidx++] = c;
 404 }
 405
 406 static char *get_buffered_cstring(void)
 407 {
 408         addcchar(0);
 409         return xstrdup(cbuffer);
 410 }
 411
 412 void pop_import(void)
 413 {
 414         int ptr = import_stack_ptr-1;
 415
 416         fclose(yyin);
 417         yy_delete_buffer( YY_CURRENT_BUFFER );
 418         yy_switch_to_buffer( import_stack[ptr].state );
 419         if (temp_name) {
 420                 unlink(temp_name);
 421                 free(temp_name);
 422         }
 423         temp_name = import_stack[ptr].temp_name;
 424         input_name = import_stack[ptr].input_name;
 425         line_number = import_stack[ptr].line_number;
 426         import_stack_ptr--;
 427 }
 428
 429 struct imports {
 430         char *name;
 431         struct imports *next;
 432 } *first_import;
 433
 434 int do_import(char *fname)
 435 {
 436         FILE *f;
 437         char *path;
 438         struct imports *import;
 439         int ptr = import_stack_ptr;
 440         int ret;
 441
 442         import = first_import;
 443         while (import && strcmp(import->name, fname))
 444                 import = import->next;
 445         if (import) return 0; /* already imported */
 446
 447         import = xmalloc(sizeof(struct imports));
 448         import->name = xstrdup(fname);
 449         import->next = first_import;
 450         first_import = import;
 451
 452         /* don't search for a file name with a path in the include directories,
 453          * for compatibility with MIDL */
 454         if (strchr( fname, '/' ) || strchr( fname, '\\' ))
 455             path = strdup( fname );
 456         else if (!(path = wpp_find_include( fname, input_name )))
 457             error_loc("Unable to open include file %s\n", fname);
 458
 459         import_stack[ptr].temp_name = temp_name;
 460         import_stack[ptr].input_name = input_name;
 461         import_stack[ptr].line_number = line_number;
 462         import_stack_ptr++;
 463         input_name = path;
 464         line_number = 1;
 465
 466         ret = wpp_parse_temp( path, NULL, &temp_name );
 467         if (ret) exit(1);
 468
 469         if((f = fopen(temp_name, "r")) == NULL)
 470                 error_loc("Unable to open %s\n", temp_name);
 471
 472         import_stack[ptr].state = YY_CURRENT_BUFFER;
 473         yy_switch_to_buffer(yy_create_buffer(f, YY_BUF_SIZE));
 474         return 1;
 475 }
 476
 477 void abort_import(void)
 478 {
 479         int ptr;
 480
 481         for (ptr=0; ptr<import_stack_ptr; ptr++)
 482                 unlink(import_stack[ptr].temp_name);
 483 }