source/programparser.cpp

   1 #include <msp/strings/format.h>
   2 #include <msp/strings/regex.h>
   3 #include "programparser.h"
   4
   5 using namespace std;
   6
   7 namespace Msp {
   8 namespace GL {
   9
  10 using namespace ProgramSyntax;
  11
  12 ProgramParser::Operator ProgramParser::operators[] =
  13 {
  14         { "[", 2, BINARY, LEFT_TO_RIGHT },
  15         { "(", 2, BINARY, LEFT_TO_RIGHT },
  16         { ".", 2, BINARY, LEFT_TO_RIGHT },
  17         { "++", 2, POSTFIX, LEFT_TO_RIGHT },
  18         { "--", 2, POSTFIX, LEFT_TO_RIGHT },
  19         { "++", 3, PREFIX, RIGHT_TO_LEFT },
  20         { "--", 3, PREFIX, RIGHT_TO_LEFT },
  21         { "+", 3, PREFIX, RIGHT_TO_LEFT },
  22         { "-", 3, PREFIX, RIGHT_TO_LEFT },
  23         { "~", 3, PREFIX, RIGHT_TO_LEFT },
  24         { "!", 3, PREFIX, RIGHT_TO_LEFT },
  25         { "*", 4, BINARY, LEFT_TO_RIGHT },
  26         { "/", 4, BINARY, LEFT_TO_RIGHT },
  27         { "%", 4, BINARY, LEFT_TO_RIGHT },
  28         { "+", 5, BINARY, LEFT_TO_RIGHT },
  29         { "-", 5, BINARY, LEFT_TO_RIGHT },
  30         { "<<", 6, BINARY, LEFT_TO_RIGHT },
  31         { ">>", 6, BINARY, LEFT_TO_RIGHT },
  32         { "<", 7, BINARY, LEFT_TO_RIGHT },
  33         { ">", 7, BINARY, LEFT_TO_RIGHT },
  34         { "<=", 7, BINARY, LEFT_TO_RIGHT },
  35         { ">=", 7, BINARY, LEFT_TO_RIGHT },
  36         { "==", 8, BINARY, LEFT_TO_RIGHT },
  37         { "!=", 8, BINARY, LEFT_TO_RIGHT },
  38         { "&", 9, BINARY, LEFT_TO_RIGHT },
  39         { "^", 10, BINARY, LEFT_TO_RIGHT },
  40         { "|", 11, BINARY, LEFT_TO_RIGHT },
  41         { "&&", 12, BINARY, LEFT_TO_RIGHT },
  42         { "^^", 13, BINARY, LEFT_TO_RIGHT },
  43         { "||", 14, BINARY, LEFT_TO_RIGHT },
  44         { "?", 15, BINARY, RIGHT_TO_LEFT },
  45         { ":", 15, BINARY, RIGHT_TO_LEFT },
  46         { "=", 16, BINARY, RIGHT_TO_LEFT },
  47         { "+=", 16, BINARY, RIGHT_TO_LEFT },
  48         { "-=", 16, BINARY, RIGHT_TO_LEFT },
  49         { "*=", 16, BINARY, RIGHT_TO_LEFT },
  50         { "/=", 16, BINARY, RIGHT_TO_LEFT },
  51         { "%=", 16, BINARY, RIGHT_TO_LEFT },
  52         { "<<=", 16, BINARY, RIGHT_TO_LEFT },
  53         { ">>=", 16, BINARY, RIGHT_TO_LEFT },
  54         { "&=", 16, BINARY, RIGHT_TO_LEFT },
  55         { "^=", 16, BINARY, RIGHT_TO_LEFT },
  56         { "|=", 16, BINARY, RIGHT_TO_LEFT },
  57         { ",", 17, BINARY, LEFT_TO_RIGHT },
  58         { { 0 }, 18, NO_OPERATOR, LEFT_TO_RIGHT }
  59 };
  60
  61 ProgramParser::ProgramParser():
  62         module(0)
  63 { }
  64
  65 ProgramParser::~ProgramParser()
  66 {
  67         delete module;
  68 }
  69
  70 Module &ProgramParser::parse(const string &s)
  71 {
  72         source = s;
  73         parse_source();
  74         return *module;
  75 }
  76
  77 Module &ProgramParser::parse(IO::Base &io)
  78 {
  79         source = string();
  80         while(!io.eof())
  81         {
  82                 char buffer[4096];
  83                 unsigned len = io.read(buffer, sizeof(buffer));
  84                 source.append(buffer, len);
  85         }
  86         parse_source();
  87         return *module;
  88 }
  89
  90 void ProgramParser::parse_source()
  91 {
  92         delete module;
  93         module = new Module;
  94         cur_stage = &module->shared;
  95         iter = source.begin();
  96         while(1)
  97         {
  98                 while(Node *statement = parse_global_declaration())
  99                         cur_stage->content.body.push_back(statement);
 100
 101                 parse_token();
 102                 string token = parse_token();
 103                 if(token.empty())
 104                         break;
 105                 else if(token=="vertex")
 106                         module->stages.push_back(VERTEX);
 107                 else if(token=="geometry")
 108                         module->stages.push_back(GEOMETRY);
 109                 else if(token=="fragment")
 110                         module->stages.push_back(FRAGMENT);
 111                 else
 112                         throw runtime_error(format("Parse error at '%s': expected stage identifier", token));
 113
 114                 if(cur_stage->type!=SHARED)
 115                         module->stages.back().previous = cur_stage;
 116                 cur_stage = &module->stages.back();
 117
 118                 for(; (iter!=source.end() && *iter!='\n'); ++iter) ;
 119         }
 120 }
 121
 122 const string &ProgramParser::peek_token(unsigned index)
 123 {
 124         while(next_tokens.size()<=index)
 125                 next_tokens.push_back(parse_token_());
 126         return next_tokens[index];
 127 }
 128
 129 string ProgramParser::parse_token()
 130 {
 131         if(!next_tokens.empty())
 132         {
 133                 string token = next_tokens.front();
 134                 next_tokens.pop_front();
 135                 return token;
 136         }
 137
 138         return parse_token_();
 139 }
 140
 141 string ProgramParser::parse_token_()
 142 {
 143         if(!skip_comment_and_whitespace())
 144                 return string();
 145
 146         if(isalpha(*iter) || *iter=='_')
 147                 return parse_identifier();
 148         else if(isdigit(*iter))
 149                 return parse_number();
 150         else
 151                 return parse_other();
 152 }
 153
 154 string ProgramParser::parse_identifier()
 155 {
 156         string ident;
 157         while(iter!=source.end())
 158         {
 159                 if(isalnum(*iter) || *iter=='_')
 160                         ident += *iter++;
 161                 else
 162                         break;
 163         }
 164
 165         return ident;
 166 }
 167
 168 string ProgramParser::parse_number()
 169 {
 170         bool accept_sign = false;
 171         string number;
 172         while(iter!=source.end())
 173         {
 174                 if(isdigit(*iter) || *iter=='.')
 175                         number += *iter++;
 176                 else if(*iter=='e' || *iter=='E')
 177                 {
 178                         number += *iter++;
 179                         accept_sign = true;
 180                 }
 181                 else if(accept_sign && (*iter=='+' || *iter=='-'))
 182                         number += *iter++;
 183                 else
 184                         break;
 185         }
 186
 187         return number;
 188 }
 189
 190 string ProgramParser::parse_other()
 191 {
 192         if(iter==source.end())
 193                 return string();
 194
 195         string token(1, *iter++);
 196         for(unsigned i=1; (i<3 && iter!=source.end()); ++i)
 197         {
 198                 bool matched = false;
 199                 for(const Operator *j=operators; (!matched && j->type); ++j)
 200                 {
 201                         matched = (j->token[i]==*iter);
 202                         for(unsigned k=0; (matched && k<i && j->token[k]); ++k)
 203                                 matched = (j->token[k]==token[k]);
 204                 }
 205
 206                 if(!matched)
 207                         break;
 208
 209                 token += *iter++;
 210         }
 211
 212         return token;
 213 }
 214
 215 bool ProgramParser::skip_comment_and_whitespace()
 216 {
 217         unsigned comment = 0;
 218         unsigned slashes = 0;
 219         while(iter!=source.end())
 220         {
 221                 if(comment==0)
 222                 {
 223                         if(*iter=='/')
 224                                 comment = 1;
 225                         else if(!isspace(*iter))
 226                                 break;
 227                 }
 228                 else if(comment==1)
 229                 {
 230                         if(*iter=='/')
 231                         {
 232                                 comment = 2;
 233                                 slashes = 2;
 234                         }
 235                         else if(*iter=='*')
 236                                 comment = 3;
 237                         else
 238                         {
 239                                 comment = 0;
 240                                 --iter;
 241                                 break;
 242                         }
 243                 }
 244                 else if(comment==2)
 245                 {
 246                         if(*iter=='\n')
 247                                 comment = 0;
 248                         else if(*iter=='/')
 249                                 ++slashes;
 250                         else if(!isspace(*iter) && slashes>=6)
 251                                 return false;
 252                 }
 253                 else if(comment==3 && *iter=='*')
 254                         comment = 4;
 255                 else if(comment==4)
 256                 {
 257                         if(*iter=='/')
 258                                 comment = 0;
 259                         else
 260                                 comment = 3;
 261                 }
 262
 263                 ++iter;
 264         }
 265
 266         return iter!=source.end();
 267 }
 268
 269 void ProgramParser::expect(const string &token)
 270 {
 271         string parsed = parse_token();
 272         if(parsed!=token)
 273                 throw runtime_error(format("Parse error at '%s': expected '%s'", parsed, token));
 274 }
 275
 276 string ProgramParser::expect_type()
 277 {
 278         string token = parse_token();
 279         if(!is_type(token))
 280                 throw runtime_error(format("Parse error at '%s': expected a type", token));
 281         return token;
 282 }
 283
 284 string ProgramParser::expect_identifier()
 285 {
 286         string token = parse_token();
 287         if(!is_identifier(token))
 288                 throw runtime_error(format("Parse error at '%s': expected an identifier", token));
 289         return token;
 290 }
 291
 292 bool ProgramParser::check(const string &token)
 293 {
 294         bool result = (peek_token()==token);
 295         if(result)
 296                 parse_token();
 297         return result;
 298 }
 299
 300 bool ProgramParser::is_interface_qualifier(const string &token)
 301 {
 302         return (token=="uniform" || token=="in" || token=="out");
 303 }
 304
 305 bool ProgramParser::is_sampling_qualifier(const string &token)
 306 {
 307         return token=="centroid";
 308 }
 309
 310 bool ProgramParser::is_qualifier(const string &token)
 311 {
 312         return (token=="const" || is_interface_qualifier(token) || is_sampling_qualifier(token));
 313 }
 314
 315 bool ProgramParser::is_builtin_type(const string &token)
 316 {
 317         static Regex re("^(void|float|int|bool|[ib]?vec[234]|mat[234](x[234])?|sampler((1D|2D)(Array)?(Shadow)?|Cube(Shadow)?|3D))$");
 318         return re.match(token);
 319 }
 320
 321 bool ProgramParser::is_type(const string &token)
 322 {
 323         return is_builtin_type(token) || declared_types.count(token);
 324 }
 325
 326 bool ProgramParser::is_identifier(const string &token)
 327 {
 328         static Regex re("^[a-zA-Z_][a-zA-Z0-9_]*$");
 329         return re.match(token);
 330 }
 331
 332 Node *ProgramParser::parse_global_declaration()
 333 {
 334         string token = peek_token();
 335         if(token=="import")
 336                 return parse_import();
 337         else if(token=="layout")
 338                 return parse_layout();
 339         else if(token=="struct")
 340                 return parse_struct_declaration();
 341         else if(is_sampling_qualifier(token) || token=="const")
 342                 return parse_variable_declaration();
 343         else if(is_interface_qualifier(token))
 344         {
 345                 if(is_type(peek_token(1)))
 346                         return parse_variable_declaration();
 347                 else
 348                         return parse_interface_block();
 349         }
 350         else if(is_type(token))
 351         {
 352                 if(peek_token(2)=="(")
 353                         return parse_function_declaration();
 354                 else
 355                         return parse_variable_declaration();
 356         }
 357         else if(token.empty())
 358                 return 0;
 359         else
 360                 throw runtime_error(format("Syntax error at '%s': expected a global declaration", token));
 361 }
 362
 363 Node *ProgramParser::parse_statement()
 364 {
 365         string token = peek_token();
 366         if(token=="if")
 367                 return parse_conditional();
 368         else if(token=="for")
 369                 return parse_iteration();
 370         else if(token=="passthrough")
 371                 return parse_passthrough();
 372         else if(token=="return")
 373                 return parse_return();
 374         else if(is_qualifier(token) || is_type(token))
 375                 return parse_variable_declaration();
 376         else if(!token.empty())
 377         {
 378                 RefPtr<ExpressionStatement> expr = new ExpressionStatement;
 379                 expr->expression = parse_expression();
 380                 expect(";");
 381
 382                 return expr.release();
 383         }
 384         else
 385                 throw runtime_error(format("Syntax error at '%s': expected a statement", token));
 386 }
 387
 388 Import *ProgramParser::parse_import()
 389 {
 390         if(cur_stage->type!=SHARED)
 391                 throw runtime_error("Imports are only allowed in the shared section");
 392
 393         expect("import");
 394         RefPtr<Import> import = new Import;
 395         import->module = parse_token();
 396         expect(";");
 397         return import.release();
 398 }
 399
 400 Layout *ProgramParser::parse_layout()
 401 {
 402         expect("layout");
 403         expect("(");
 404         RefPtr<Layout> layout = new Layout;
 405         while(1)
 406         {
 407                 string token = parse_token();
 408                 if(token==")")
 409                         throw runtime_error(format("Parse error at '%s': expected layout qualifier id", token));
 410
 411                 layout->qualifiers.push_back(Layout::Qualifier());
 412                 Layout::Qualifier &qual = layout->qualifiers.back();
 413                 qual.identifier = token;
 414
 415                 if(check("="))
 416                         qual.value = parse_token();
 417
 418                 if(peek_token()==")")
 419                         break;
 420
 421                 expect(",");
 422         }
 423         expect(")");
 424         layout->interface = parse_token();
 425         expect(";");
 426
 427         return layout.release();
 428 }
 429
 430 void ProgramParser::parse_block(Block &block, bool require_braces)
 431 {
 432         bool have_braces = (require_braces || peek_token()=="{");
 433         if(have_braces)
 434                 expect("{");
 435
 436         if(have_braces)
 437         {
 438                 while(peek_token()!="}")
 439                         block.body.push_back(parse_statement());
 440         }
 441         else
 442                 block.body.push_back(parse_statement());
 443
 444         block.use_braces = (require_braces || block.body.size()!=1);
 445
 446         if(have_braces)
 447                 expect("}");
 448 }
 449
 450 Expression *ProgramParser::parse_expression(unsigned precedence)
 451 {
 452         RefPtr<Expression> left;
 453         VariableReference *left_var = 0;
 454         while(1)
 455         {
 456                 string token = peek_token();
 457
 458                 const Operator *oper = 0;
 459                 for(Operator *i=operators; (!oper && i->type); ++i)
 460                         if(token==i->token && (!left || i->type!=PREFIX) && (left || i->type!=POSTFIX))
 461                                 oper = i;
 462
 463                 if(token==";" || token==")" || token=="]" || token=="," || (oper && precedence && oper->precedence>=precedence))
 464                 {
 465                         if(left)
 466                                 return left.release();
 467                         else
 468                                 throw runtime_error(format("Parse error at '%s': expected an expression", token));
 469                 }
 470                 else if(left)
 471                 {
 472                         if(token=="(")
 473                         {
 474                                 if(!left_var)
 475                                         throw runtime_error(format("Parse error at '%s': function name must be an identifier", token));
 476                                 left = parse_function_call(left_var);
 477                         }
 478                         else if(token==".")
 479                         {
 480                                 RefPtr<MemberAccess> memacc = new MemberAccess;
 481                                 memacc->left = left.release();
 482                                 parse_token();
 483                                 memacc->member = expect_identifier();
 484                                 left = memacc;
 485                         }
 486                         else if(oper && oper->type==POSTFIX)
 487                         {
 488                                 RefPtr<UnaryExpression> unary = new UnaryExpression;
 489                                 unary->oper = parse_token();
 490                                 unary->prefix = false;
 491                                 unary->expression = left.release();
 492                                 left = unary;
 493                         }
 494                         else if(oper && oper->type==BINARY)
 495                                 left = parse_binary(left.release(), oper);
 496                         else
 497                                 throw runtime_error(format("Parse error at '%s': expected an operator", token));
 498                         left_var = 0;
 499                 }
 500                 else
 501                 {
 502                         if(token=="(")
 503                         {
 504                                 parse_token();
 505                                 RefPtr<ParenthesizedExpression> parexpr = new ParenthesizedExpression;
 506                                 parexpr->expression = parse_expression();
 507                                 expect(")");
 508                                 left = parexpr;
 509                         }
 510                         else if(isdigit(token[0]) || token=="true" || token=="false")
 511                         {
 512                                 RefPtr<Literal> literal = new Literal;
 513                                 literal->token = parse_token();
 514                                 left = literal;
 515                         }
 516                         else if(is_identifier(token))
 517                         {
 518                                 RefPtr<VariableReference> var = new VariableReference;
 519                                 var->name = expect_identifier();
 520                                 left = var;
 521                                 left_var = var.get();
 522                         }
 523                         else if(oper && oper->type==PREFIX)
 524                         {
 525                                 RefPtr<UnaryExpression> unary = new UnaryExpression;
 526                                 unary->oper = parse_token();
 527                                 unary->prefix = true;
 528                                 unary->expression = parse_expression(oper->precedence);
 529                                 left = unary;
 530                         }
 531                         else
 532                                 throw runtime_error(format("Parse error at '%s': expected an expression", token));
 533                 }
 534         }
 535 }
 536
 537 BinaryExpression *ProgramParser::parse_binary(Expression *left, const Operator *oper)
 538 {
 539         RefPtr<BinaryExpression> binary = (oper->precedence==16 ? new Assignment : new BinaryExpression);
 540         binary->left = left;
 541         binary->oper = parse_token();
 542         if(binary->oper=="[")
 543         {
 544                 binary->right = parse_expression();
 545                 expect("]");
 546                 binary->after = "]";
 547         }
 548         else
 549                 binary->right = parse_expression(oper->precedence+(oper->assoc==RIGHT_TO_LEFT));
 550         return binary.release();
 551 }
 552
 553 FunctionCall *ProgramParser::parse_function_call(VariableReference *var)
 554 {
 555         RefPtr<FunctionCall> call = new FunctionCall;
 556         call->name = var->name;
 557         call->constructor = is_type(call->name);
 558         expect("(");
 559         while(peek_token()!=")")
 560         {
 561                 if(!call->arguments.empty())
 562                         expect(",");
 563                 call->arguments.push_back(parse_expression());
 564         }
 565         expect(")");
 566         return call.release();
 567 }
 568
 569 StructDeclaration *ProgramParser::parse_struct_declaration()
 570 {
 571         expect("struct");
 572         RefPtr<StructDeclaration> strct = new StructDeclaration;
 573
 574         strct->name = expect_identifier();
 575         parse_block(strct->members, true);
 576         expect(";");
 577
 578         declared_types.insert(strct->name);
 579         return strct.release();
 580 }
 581
 582 VariableDeclaration *ProgramParser::parse_variable_declaration()
 583 {
 584         RefPtr<VariableDeclaration> var = new VariableDeclaration;
 585
 586         string token = peek_token();
 587         if(is_sampling_qualifier(token))
 588         {
 589                 var->sampling = parse_token();
 590                 token = peek_token();
 591                 if(!is_interface_qualifier(token))
 592                         throw runtime_error(format("Parse error at '%s': expected an interface qualifier", token));
 593         }
 594
 595         if(is_interface_qualifier(token))
 596                 var->interface = parse_token();
 597         else if(token=="const")
 598         {
 599                 var->constant = true;
 600                 parse_token();
 601         }
 602
 603         var->type = expect_type();
 604         var->name = expect_identifier();
 605
 606         if(check("["))
 607         {
 608                 var->array = true;
 609                 if(!check("]"))
 610                 {
 611                         var->array_size = parse_expression();
 612                         expect("]");
 613                 }
 614         }
 615
 616         if(check("="))
 617                 var->init_expression = parse_expression();
 618
 619         expect(";");
 620         return var.release();
 621 }
 622
 623 FunctionDeclaration *ProgramParser::parse_function_declaration()
 624 {
 625         RefPtr<FunctionDeclaration> func = new FunctionDeclaration;
 626
 627         func->return_type = expect_type();
 628         func->name = expect_identifier();
 629         expect("(");
 630         while(peek_token()!=")")
 631         {
 632                 if(!func->parameters.empty())
 633                         expect(",");
 634
 635                 RefPtr<VariableDeclaration> var = new VariableDeclaration;
 636                 var->type = expect_type();
 637                 var->name = expect_identifier();
 638                 func->parameters.push_back(var.release());
 639         }
 640         expect(")");
 641
 642         string token = peek_token();
 643         if(token=="{")
 644         {
 645                 func->definition = func.get();
 646                 parse_block(func->body, true);
 647         }
 648         else if(token==";")
 649                 parse_token();
 650         else
 651                 throw runtime_error(format("Parse error at '%s': expected '{' or ';'", token));
 652
 653         return func.release();
 654 }
 655
 656 InterfaceBlock *ProgramParser::parse_interface_block()
 657 {
 658         RefPtr<InterfaceBlock> iface = new InterfaceBlock;
 659
 660         iface->interface = parse_token();
 661         if(!is_interface_qualifier(iface->interface))
 662                 throw runtime_error(format("Parse error at '%s': expected an interface qualifier", iface->interface));
 663
 664         iface->name = expect_identifier();
 665         parse_block(iface->members, true);
 666         if(!check(";"))
 667         {
 668                 iface->instance_name = expect_identifier();
 669                 if(check("["))
 670                 {
 671                         iface->array = true;
 672                         expect("]");
 673                 }
 674                 expect(";");
 675         }
 676
 677         return iface.release();
 678 }
 679
 680 Conditional *ProgramParser::parse_conditional()
 681 {
 682         expect("if");
 683         expect("(");
 684         RefPtr<Conditional> cond = new Conditional;
 685         cond->condition = parse_expression();
 686         expect(")");
 687
 688         parse_block(cond->body, false);
 689
 690         string token = peek_token();
 691         if(token=="else")
 692         {
 693                 parse_token();
 694                 parse_block(cond->else_body, false);
 695         }
 696
 697         return cond.release();
 698 }
 699
 700 Iteration *ProgramParser::parse_iteration()
 701 {
 702         expect("for");
 703         expect("(");
 704         RefPtr<Iteration> loop = new Iteration;
 705         string token = peek_token();
 706         if(is_type(token))
 707                 loop->init_statement = parse_statement();
 708         else
 709         {
 710                 RefPtr<ExpressionStatement> expr = new ExpressionStatement;
 711                 expr->expression = parse_expression();
 712                 expect(";");
 713                 loop->init_statement = expr.release();
 714         }
 715         loop->condition = parse_expression();
 716         expect(";");
 717         loop->loop_expression = parse_expression();
 718         expect(")");
 719
 720         parse_block(loop->body, false);
 721
 722         return loop.release();
 723 }
 724
 725 Passthrough *ProgramParser::parse_passthrough()
 726 {
 727         expect("passthrough");
 728         RefPtr<Passthrough> pass = new Passthrough;
 729         if(cur_stage->type==GEOMETRY)
 730         {
 731                 expect("[");
 732                 pass->subscript = parse_expression();
 733                 expect("]");
 734         }
 735         expect(";");
 736         return pass.release();
 737 }
 738
 739 Return *ProgramParser::parse_return()
 740 {
 741         expect("return");
 742         RefPtr<Return> ret = new Return;
 743         if(peek_token()!=";")
 744                 ret->expression = parse_expression();
 745         expect(";");
 746         return ret.release();
 747 }
 748
 749 } // namespace GL
 750 } // namespace Msp