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(RefPtr<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 RefPtr<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         {
 339                 RefPtr<Layout> layout = parse_layout();
 340                 token = peek_token();
 341                 if(is_interface_qualifier(token) && peek_token(1)==";")
 342                 {
 343                         RefPtr<InterfaceLayout> iface_lo = new InterfaceLayout;
 344                         iface_lo->layout.qualifiers = layout->qualifiers;
 345                         iface_lo->interface = parse_token();
 346                         expect(";");
 347                         return iface_lo;
 348                 }
 349                 else
 350                 {
 351                         RefPtr<VariableDeclaration> var = parse_variable_declaration();
 352                         var->layout = layout;
 353                         return var;
 354                 }
 355         }
 356         else if(token=="struct")
 357                 return parse_struct_declaration();
 358         else if(is_sampling_qualifier(token) || token=="const")
 359                 return parse_variable_declaration();
 360         else if(is_interface_qualifier(token))
 361         {
 362                 if(is_type(peek_token(1)))
 363                         return parse_variable_declaration();
 364                 else
 365                         return parse_interface_block();
 366         }
 367         else if(is_type(token))
 368         {
 369                 if(peek_token(2)=="(")
 370                         return parse_function_declaration();
 371                 else
 372                         return parse_variable_declaration();
 373         }
 374         else if(token.empty())
 375                 return 0;
 376         else
 377                 throw runtime_error(format("Syntax error at '%s': expected a global declaration", token));
 378 }
 379
 380 RefPtr<Node> ProgramParser::parse_statement()
 381 {
 382         string token = peek_token();
 383         if(token=="if")
 384                 return parse_conditional();
 385         else if(token=="for")
 386                 return parse_iteration();
 387         else if(token=="passthrough")
 388                 return parse_passthrough();
 389         else if(token=="return")
 390                 return parse_return();
 391         else if(is_qualifier(token) || is_type(token))
 392                 return parse_variable_declaration();
 393         else if(!token.empty())
 394         {
 395                 RefPtr<ExpressionStatement> expr = new ExpressionStatement;
 396                 expr->expression = parse_expression();
 397                 expect(";");
 398
 399                 return expr;
 400         }
 401         else
 402                 throw runtime_error(format("Syntax error at '%s': expected a statement", token));
 403 }
 404
 405 RefPtr<Import> ProgramParser::parse_import()
 406 {
 407         if(cur_stage->type!=SHARED)
 408                 throw runtime_error("Imports are only allowed in the shared section");
 409
 410         expect("import");
 411         RefPtr<Import> import = new Import;
 412         import->module = parse_token();
 413         expect(";");
 414         return import;
 415 }
 416
 417 RefPtr<Layout> ProgramParser::parse_layout()
 418 {
 419         expect("layout");
 420         expect("(");
 421         RefPtr<Layout> layout = new Layout;
 422         while(1)
 423         {
 424                 string token = parse_token();
 425                 if(token==")")
 426                         throw runtime_error(format("Parse error at '%s': expected layout qualifier id", token));
 427
 428                 layout->qualifiers.push_back(Layout::Qualifier());
 429                 Layout::Qualifier &qual = layout->qualifiers.back();
 430                 qual.identifier = token;
 431
 432                 if(check("="))
 433                         qual.value = parse_token();
 434
 435                 if(peek_token()==")")
 436                         break;
 437
 438                 expect(",");
 439         }
 440         expect(")");
 441
 442         return layout;
 443 }
 444
 445 void ProgramParser::parse_block(Block &block, bool require_braces)
 446 {
 447         bool have_braces = (require_braces || peek_token()=="{");
 448         if(have_braces)
 449                 expect("{");
 450
 451         if(have_braces)
 452         {
 453                 while(peek_token()!="}")
 454                         block.body.push_back(parse_statement());
 455         }
 456         else
 457                 block.body.push_back(parse_statement());
 458
 459         block.use_braces = (require_braces || block.body.size()!=1);
 460
 461         if(have_braces)
 462                 expect("}");
 463 }
 464
 465 RefPtr<Expression> ProgramParser::parse_expression(unsigned precedence)
 466 {
 467         RefPtr<Expression> left;
 468         VariableReference *left_var = 0;
 469         while(1)
 470         {
 471                 string token = peek_token();
 472
 473                 const Operator *oper = 0;
 474                 for(Operator *i=operators; (!oper && i->type); ++i)
 475                         if(token==i->token && (!left || i->type!=PREFIX) && (left || i->type!=POSTFIX))
 476                                 oper = i;
 477
 478                 if(token==";" || token==")" || token=="]" || token=="," || (oper && precedence && oper->precedence>=precedence))
 479                 {
 480                         if(left)
 481                                 return left;
 482                         else
 483                                 throw runtime_error(format("Parse error at '%s': expected an expression", token));
 484                 }
 485                 else if(left)
 486                 {
 487                         if(token=="(")
 488                         {
 489                                 if(!left_var)
 490                                         throw runtime_error(format("Parse error at '%s': function name must be an identifier", token));
 491                                 left = parse_function_call(*left_var);
 492                         }
 493                         else if(token==".")
 494                         {
 495                                 RefPtr<MemberAccess> memacc = new MemberAccess;
 496                                 memacc->left = left;
 497                                 parse_token();
 498                                 memacc->member = expect_identifier();
 499                                 left = memacc;
 500                         }
 501                         else if(oper && oper->type==POSTFIX)
 502                         {
 503                                 RefPtr<UnaryExpression> unary = new UnaryExpression;
 504                                 unary->oper = parse_token();
 505                                 unary->prefix = false;
 506                                 unary->expression = left;
 507                                 left = unary;
 508                         }
 509                         else if(oper && oper->type==BINARY)
 510                                 left = parse_binary(left, oper);
 511                         else
 512                                 throw runtime_error(format("Parse error at '%s': expected an operator", token));
 513                         left_var = 0;
 514                 }
 515                 else
 516                 {
 517                         if(token=="(")
 518                         {
 519                                 parse_token();
 520                                 RefPtr<ParenthesizedExpression> parexpr = new ParenthesizedExpression;
 521                                 parexpr->expression = parse_expression();
 522                                 expect(")");
 523                                 left = parexpr;
 524                         }
 525                         else if(isdigit(token[0]) || token=="true" || token=="false")
 526                         {
 527                                 RefPtr<Literal> literal = new Literal;
 528                                 literal->token = parse_token();
 529                                 left = literal;
 530                         }
 531                         else if(is_identifier(token))
 532                         {
 533                                 RefPtr<VariableReference> var = new VariableReference;
 534                                 var->name = expect_identifier();
 535                                 left = var;
 536                                 left_var = var.get();
 537                         }
 538                         else if(oper && oper->type==PREFIX)
 539                         {
 540                                 RefPtr<UnaryExpression> unary = new UnaryExpression;
 541                                 unary->oper = parse_token();
 542                                 unary->prefix = true;
 543                                 unary->expression = parse_expression(oper->precedence);
 544                                 left = unary;
 545                         }
 546                         else
 547                                 throw runtime_error(format("Parse error at '%s': expected an expression", token));
 548                 }
 549         }
 550 }
 551
 552 RefPtr<BinaryExpression> ProgramParser::parse_binary(const RefPtr<Expression> &left, const Operator *oper)
 553 {
 554         RefPtr<BinaryExpression> binary = (oper->precedence==16 ? new Assignment : new BinaryExpression);
 555         binary->left = left;
 556         binary->oper = parse_token();
 557         if(binary->oper=="[")
 558         {
 559                 binary->right = parse_expression();
 560                 expect("]");
 561                 binary->after = "]";
 562         }
 563         else
 564                 binary->right = parse_expression(oper->precedence+(oper->assoc==RIGHT_TO_LEFT));
 565         return binary;
 566 }
 567
 568 RefPtr<FunctionCall> ProgramParser::parse_function_call(const VariableReference &var)
 569 {
 570         RefPtr<FunctionCall> call = new FunctionCall;
 571         call->name = var.name;
 572         call->constructor = is_type(call->name);
 573         expect("(");
 574         while(peek_token()!=")")
 575         {
 576                 if(!call->arguments.empty())
 577                         expect(",");
 578                 call->arguments.push_back(parse_expression());
 579         }
 580         expect(")");
 581         return call;
 582 }
 583
 584 RefPtr<StructDeclaration> ProgramParser::parse_struct_declaration()
 585 {
 586         expect("struct");
 587         RefPtr<StructDeclaration> strct = new StructDeclaration;
 588
 589         strct->name = expect_identifier();
 590         parse_block(strct->members, true);
 591         expect(";");
 592
 593         declared_types.insert(strct->name);
 594         return strct;
 595 }
 596
 597 RefPtr<VariableDeclaration> ProgramParser::parse_variable_declaration()
 598 {
 599         RefPtr<VariableDeclaration> var = new VariableDeclaration;
 600
 601         string token = peek_token();
 602         if(is_sampling_qualifier(token))
 603         {
 604                 var->sampling = parse_token();
 605                 token = peek_token();
 606                 if(!is_interface_qualifier(token))
 607                         throw runtime_error(format("Parse error at '%s': expected an interface qualifier", token));
 608         }
 609
 610         if(is_interface_qualifier(token))
 611                 var->interface = parse_token();
 612         else if(token=="const")
 613         {
 614                 var->constant = true;
 615                 parse_token();
 616         }
 617
 618         var->type = expect_type();
 619         var->name = expect_identifier();
 620
 621         if(check("["))
 622         {
 623                 var->array = true;
 624                 if(!check("]"))
 625                 {
 626                         var->array_size = parse_expression();
 627                         expect("]");
 628                 }
 629         }
 630
 631         if(check("="))
 632                 var->init_expression = parse_expression();
 633
 634         expect(";");
 635         return var;
 636 }
 637
 638 RefPtr<FunctionDeclaration> ProgramParser::parse_function_declaration()
 639 {
 640         RefPtr<FunctionDeclaration> func = new FunctionDeclaration;
 641
 642         func->return_type = expect_type();
 643         func->name = expect_identifier();
 644         expect("(");
 645         while(peek_token()!=")")
 646         {
 647                 if(!func->parameters.empty())
 648                         expect(",");
 649
 650                 RefPtr<VariableDeclaration> var = new VariableDeclaration;
 651                 var->type = expect_type();
 652                 var->name = expect_identifier();
 653                 func->parameters.push_back(var);
 654         }
 655         expect(")");
 656
 657         string token = peek_token();
 658         if(token=="{")
 659         {
 660                 func->definition = func.get();
 661                 parse_block(func->body, true);
 662         }
 663         else if(token==";")
 664                 parse_token();
 665         else
 666                 throw runtime_error(format("Parse error at '%s': expected '{' or ';'", token));
 667
 668         return func;
 669 }
 670
 671 RefPtr<InterfaceBlock> ProgramParser::parse_interface_block()
 672 {
 673         RefPtr<InterfaceBlock> iface = new InterfaceBlock;
 674
 675         iface->interface = parse_token();
 676         if(!is_interface_qualifier(iface->interface))
 677                 throw runtime_error(format("Parse error at '%s': expected an interface qualifier", iface->interface));
 678
 679         iface->name = expect_identifier();
 680         parse_block(iface->members, true);
 681         if(!check(";"))
 682         {
 683                 iface->instance_name = expect_identifier();
 684                 if(check("["))
 685                 {
 686                         iface->array = true;
 687                         expect("]");
 688                 }
 689                 expect(";");
 690         }
 691
 692         return iface;
 693 }
 694
 695 RefPtr<Conditional> ProgramParser::parse_conditional()
 696 {
 697         expect("if");
 698         expect("(");
 699         RefPtr<Conditional> cond = new Conditional;
 700         cond->condition = parse_expression();
 701         expect(")");
 702
 703         parse_block(cond->body, false);
 704
 705         string token = peek_token();
 706         if(token=="else")
 707         {
 708                 parse_token();
 709                 parse_block(cond->else_body, false);
 710         }
 711
 712         return cond;
 713 }
 714
 715 RefPtr<Iteration> ProgramParser::parse_iteration()
 716 {
 717         expect("for");
 718         expect("(");
 719         RefPtr<Iteration> loop = new Iteration;
 720         string token = peek_token();
 721         if(is_type(token))
 722                 loop->init_statement = parse_statement();
 723         else
 724         {
 725                 RefPtr<ExpressionStatement> expr = new ExpressionStatement;
 726                 expr->expression = parse_expression();
 727                 expect(";");
 728                 loop->init_statement = expr;
 729         }
 730         loop->condition = parse_expression();
 731         expect(";");
 732         loop->loop_expression = parse_expression();
 733         expect(")");
 734
 735         parse_block(loop->body, false);
 736
 737         return loop;
 738 }
 739
 740 RefPtr<Passthrough> ProgramParser::parse_passthrough()
 741 {
 742         expect("passthrough");
 743         RefPtr<Passthrough> pass = new Passthrough;
 744         if(cur_stage->type==GEOMETRY)
 745         {
 746                 expect("[");
 747                 pass->subscript = parse_expression();
 748                 expect("]");
 749         }
 750         expect(";");
 751         return pass;
 752 }
 753
 754 RefPtr<Return> ProgramParser::parse_return()
 755 {
 756         expect("return");
 757         RefPtr<Return> ret = new Return;
 758         if(peek_token()!=";")
 759                 ret->expression = parse_expression();
 760         expect(";");
 761         return ret;
 762 }
 763
 764 } // namespace GL
 765 } // namespace Msp