/* * Cppcheck - A tool for static C/C++ code analysis * Copyright (C) 2007-2009 Daniel Marjamäki and Cppcheck team. * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation, either version 3 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program. If not, see #include #include #include #include #include #include #include #include //--------------------------------------------------------------------------- Tokenizer::Tokenizer() { _tokens = 0; _tokensBack = 0; _errorLogger = 0; } Tokenizer::Tokenizer(const Settings &settings, ErrorLogger *errorLogger) { _tokens = 0; _tokensBack = 0; _settings = settings; _errorLogger = errorLogger; } Tokenizer::~Tokenizer() { DeallocateTokens(); } //--------------------------------------------------------------------------- // Helper functions.. //--------------------------------------------------------------------------- const Token *Tokenizer::tokens() const { return _tokens; } const std::vector *Tokenizer::getFiles() const { return &_files; } //--------------------------------------------------------------------------- //--------------------------------------------------------------------------- // addtoken // add a token. Used by 'Tokenizer' //--------------------------------------------------------------------------- void Tokenizer::addtoken(const char str[], const unsigned int lineno, const unsigned int fileno) { if (str[0] == 0) return; // Replace hexadecimal value with decimal std::ostringstream str2; if (strncmp(str, "0x", 2) == 0) { str2 << std::strtoul(str + 2, NULL, 16); } else { str2 << str; } if (_tokensBack) { _tokensBack->insertToken(str2.str().c_str()); _tokensBack = _tokensBack->next(); } else { _tokens = new Token; _tokensBack = _tokens; _tokensBack->str(str2.str()); } _tokensBack->linenr(lineno); _tokensBack->fileIndex(fileno); } //--------------------------------------------------------------------------- //--------------------------------------------------------------------------- // SizeOfType - gives the size of a type //--------------------------------------------------------------------------- int Tokenizer::SizeOfType(const char type[]) const { if (!type) return 0; std::map::const_iterator it = _typeSize.find(type); if (it == _typeSize.end()) return 0; return it->second; } //--------------------------------------------------------------------------- //--------------------------------------------------------------------------- // InsertTokens - Copy and insert tokens //--------------------------------------------------------------------------- void Tokenizer::InsertTokens(Token *dest, Token *src, unsigned int n) { while (n > 0) { dest->insertToken(src->str().c_str()); dest = dest->next(); dest->fileIndex(src->fileIndex()); dest->linenr(src->linenr()); dest->varId(src->varId()); src = src->next(); --n; } } //--------------------------------------------------------------------------- //--------------------------------------------------------------------------- // Tokenize - tokenizes a given file. //--------------------------------------------------------------------------- void Tokenizer::createTokens(std::istream &code) { // line number in parsed code unsigned int lineno = 1; // The current token being parsed std::string CurrentToken; // lineNumbers holds line numbers for files in fileIndexes // every time an include file is complitely parsed, last item in the vector // is removed and lineno is set to point to that value. std::vector lineNumbers; // fileIndexes holds index for _files vector about currently parsed files // every time an include file is complitely parsed, last item in the vector // is removed and FileIndex is set to point to that value. std::vector fileIndexes; // FileIndex. What file in the _files vector is read now? unsigned int FileIndex = 0; // Read one byte at a time from code and create tokens for (char ch = (char)code.get(); code.good(); ch = (char)code.get()) { // We are not handling UTF and stuff like that. Code is supposed to plain simple text. if (ch < 0) continue; // char/string.. if (ch == '\'' || ch == '\"') { std::string line; // read char bool special = false; char c = ch; do { // Append token.. line += c; if (c == '\n') ++lineno; // Special sequence '\.' if (special) special = false; else special = (c == '\\'); // Get next character c = (char)code.get(); } while (code.good() && (special || c != ch)); line += ch; // Handle #file "file.h" if (CurrentToken == "#file") { // Extract the filename line = line.substr(1, line.length() - 2); // Has this file been tokenized already? ++lineno; bool foundOurfile = false; fileIndexes.push_back(FileIndex); for (unsigned int i = 0; i < _files.size(); i++) { if (FileLister::SameFileName(_files[i].c_str(), line.c_str())) { // Use this index foundOurfile = true; FileIndex = i; } } if (!foundOurfile) { // The "_files" vector remembers what files have been tokenized.. _files.push_back(FileLister::simplifyPath(line.c_str())); FileIndex = static_cast(_files.size() - 1); } lineNumbers.push_back(lineno); lineno = 0; } else { // Add previous token addtoken(CurrentToken.c_str(), lineno, FileIndex); // Add content of the string addtoken(line.c_str(), lineno, FileIndex); } CurrentToken.clear(); continue; } if (strchr("+-*/%&|^?!=<>[](){};:,.~\n ", ch)) { if (ch == '.' && CurrentToken.length() > 0 && std::isdigit(CurrentToken[0])) { // Don't separate doubles "5.4" } else if (strchr("+-", ch) && CurrentToken.length() > 0 && std::isdigit(CurrentToken[0]) && CurrentToken[CurrentToken.length()-1] == 'e') { // Don't separate doubles "4.2e+10" } else { if (CurrentToken == "#file") { // Handle this where strings are handled continue; } else if (CurrentToken == "#endfile") { if (lineNumbers.empty() || fileIndexes.empty()) { std::cerr << "####### Preprocessor bug! #######\n"; std::exit(0); } lineno = lineNumbers.back(); lineNumbers.pop_back(); FileIndex = fileIndexes.back(); fileIndexes.pop_back(); CurrentToken.clear(); continue; } // If token contains # characters, split it up std::string temp; for (std::string::size_type i = 0; i < CurrentToken.length(); ++i) { if (CurrentToken[i] == '#' && CurrentToken.length() + 1 > i && CurrentToken[i+1] == '#') { addtoken(temp.c_str(), lineno, FileIndex); temp.clear(); addtoken("##", lineno, FileIndex); ++i; } else temp += CurrentToken[i]; } addtoken(temp.c_str(), lineno, FileIndex); CurrentToken.clear(); if (ch == '\n') { ++lineno; continue; } else if (ch == ' ') { continue; } CurrentToken += ch; // Add "++", "--" or ">>" token if ((ch == '+' || ch == '-' || ch == '>') && (code.peek() == ch)) CurrentToken += (char)code.get(); addtoken(CurrentToken.c_str(), lineno, FileIndex); CurrentToken.clear(); continue; } } CurrentToken += ch; } addtoken(CurrentToken.c_str(), lineno, FileIndex); } bool Tokenizer::tokenize(std::istream &code, const char FileName[]) { // The "_files" vector remembers what files have been tokenized.. _files.push_back(FileLister::simplifyPath(FileName)); createTokens(code); if (!createLinks()) { // Source has syntax errors, can't proceed return false; } // Combine "- %num%" .. for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok, "[(+-*/=,] - %num%") && tok->strAt(2)[0] != '-') { tok->next()->str((std::string("-") + tok->strAt(2)).c_str()); tok->next()->deleteNext(); } if (Token::Match(tok, "return - %num%") && tok->strAt(2)[0] != '-') { tok->next()->str((std::string("-") + tok->strAt(2)).c_str()); tok->next()->deleteNext(); } } // Combine tokens.. for (Token *tok = _tokens; tok && tok->next(); tok = tok->next()) { static const char * const combineWithNext[][3] = { { "<", "<", "<<" }, { "&", "&", "&&" }, { "|", "|", "||" }, { "+", "=", "+=" }, { "-", "=", "-=" }, { "*", "=", "*=" }, { "/", "=", "/=" }, { "&", "=", "&=" }, { "|", "=", "|=" }, { "=", "=", "==" }, { "!", "=", "!=" }, { "<", "=", "<=" }, { ">", "=", ">=" }, { ":", ":", "::" }, { "-", ">", "." }, // Replace "->" with "." { "private", ":", "private:" }, { "protected", ":", "protected:" }, { "public", ":", "public:" } }; for (unsigned ui = 0; ui < sizeof(combineWithNext) / sizeof(combineWithNext[0]); ui++) { if (tok->str() == combineWithNext[ui][0] && tok->next()->str() == combineWithNext[ui][1]) { tok->str(combineWithNext[ui][2]); tok->deleteNext(); } } } // typedef.. for (Token *tok = _tokens; tok; tok = tok->next()) { if (tok->str() != "typedef") continue; if (Token::Match(tok->next(), "%type% %type% ;")) { const char *type1 = tok->strAt(1); const char *type2 = tok->strAt(2); tok = const_cast(tok->tokAt(3)); for (Token *tok2 = tok; tok2; tok2 = tok2->next()) { if (tok2->str() == type2) tok2->str(type1); } } else if (Token::Match(tok->next(), "%type% %type% %type% ;")) { const char *type1 = tok->strAt(1); const char *type2 = tok->strAt(2); const char *type3 = tok->strAt(3); tok = const_cast(tok->tokAt(4)); for (Token *tok2 = tok; tok2; tok2 = tok2->next()) { if (tok2->str() == type3) { tok2->str(type1); tok2->insertToken(type2); tok2 = tok2->next(); } } } } // Remove __asm.. for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok->next(), "__asm|_asm|asm {")) { while (tok->next()) { bool last = Token::simpleMatch(tok->next(), "}"); // Unlink and delete tok->next() tok->deleteNext(); // break if this was the last token to delete.. if (last) break; } } } // Remove "volatile" and "mutable" while (_tokens && (_tokens->str() == "volatile" || _tokens->str() == "mutable")) { _tokens->deleteThis(); } for (Token *tok = _tokens; tok; tok = tok->next()) { while (tok->next() && (tok->next()->str() == "volatile" || tok->next()->str() == "mutable")) { tok->deleteNext(); } } // replace "unsigned i" with "unsigned int i" unsignedint(); // Split up variable declarations. simplifyVarDecl(); // Handle templates.. simplifyTemplates(); // change array to pointer.. for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok, "%type% %var% [ ] [,;=]")) { tok->next()->deleteNext(); tok->next()->deleteNext(); tok->insertToken("*"); } } return true; } //--------------------------------------------------------------------------- void Tokenizer::simplifyTemplates() { // Locate templates.. std::list templates; for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::simpleMatch(tok, "template <")) { for (const Token *tok2 = tok; tok2; tok2 = tok2->next()) { // Just a declaration => ignore this if (tok2->str() == ";") break; // Implementation => add to "templates" if (tok2->str() == "{") { templates.push_back(tok); break; } } } } if (templates.empty()) return; // Locate possible instantiations of templates.. std::list used; for (Token *tok = _tokens; tok; tok = tok->next()) { // template definition.. skip it if (Token::simpleMatch(tok, "template <")) { unsigned int indentlevel = 0; for (; tok; tok = tok->next()) { if (tok->str() == "{") { ++indentlevel; } else if (tok->str() == "}") { if (indentlevel <= 1) break; --indentlevel; } else if (indentlevel == 0 && tok->str() == ";") { break; } } if (!tok) break; } else if (Token::Match(tok->previous(), "[{};=] %var% <")) { used.push_back(tok); } } if (used.empty()) return; // expand templates for (std::list::iterator iter1 = templates.begin(); iter1 != templates.end(); ++iter1) { Token *tok = *iter1; std::vector type; for (tok = tok->tokAt(2); tok && tok->str() != ">"; tok = tok->next()) { if (Token::Match(tok, "%var% ,|>")) type.push_back(tok->str()); } // bail out if the end of the file was reached if (!tok) break; // if this is a template function, get the position of the function name unsigned int pos = 0; if (Token::Match(tok, "> %type% *| %var% (")) pos = 2; else if (Token::Match(tok, "> %type% %type% *| %var% (")) pos = 3; if (pos > 0 && tok->tokAt(pos)->str() == "*") ++pos; // name of template function/class.. const std::string name(tok->strAt(pos > 0 ? pos : 2)); const bool isfunc(pos > 0); // locate template usage.. std::string s(name + " <"); for (unsigned int i = 0; i < type.size(); ++i) { if (i > 0) s += ","; s += " %any% "; } const std::string pattern(s + "> "); for (std::list::iterator iter2 = used.begin(); iter2 != used.end(); ++iter2) { Token *tok2 = *iter2; if (tok2->str() != name) continue; if (!Token::Match(tok2, (pattern + (isfunc ? "(" : "%var%")).c_str())) continue; // New type.. std::vector types2; s = ""; for (const Token *tok3 = tok2->tokAt(2); tok3->str() != ">"; tok3 = tok3->next()) { if (tok3->str() != ",") types2.push_back(tok3->str()); s += tok3->str(); } const std::string type2(s); // New classname/funcname.. const std::string name2(name + "<" + type2 + ">"); // Copy template.. int _indentlevel = 0; for (const Token *tok3 = _tokens; tok3; tok3 = tok3->next()) { if (tok3->str() == "{") ++_indentlevel; else if (tok3->str() == "}") --_indentlevel; // Start of template.. if (tok3 == tok) { tok3 = tok3->next(); } // member function implemented outside class definition else if (_indentlevel == 0 && Token::Match(tok3, (pattern + " :: %var% (").c_str())) { addtoken(name2.c_str(), tok3->linenr(), tok3->fileIndex()); while (tok3->str() != "::") tok3 = tok3->next(); } // not part of template.. go on to next token else continue; int indentlevel = 0; for (; tok3; tok3 = tok3->next()) { if (tok3->str() == "{") ++indentlevel; else if (tok3->str() == "}") { if (indentlevel <= 1) { // there is a bug if indentlevel is 0 // the "}" token should only be added if indentlevel is 1 but I add it always intentionally // if indentlevel ever becomes 0, cppcheck will write: // ### Error: Invalid number of character { addtoken("}", tok3->linenr(), tok3->fileIndex()); break; } --indentlevel; } { // search for this token in the type vector unsigned int itype = 0; while (itype < type.size() && type[itype] != tok3->str()) ++itype; // replace type with given type.. if (itype < type.size()) addtoken(types2[itype].c_str(), tok3->linenr(), tok3->fileIndex()); // replace name.. else if (tok3->str() == name) addtoken(name2.c_str(), tok3->linenr(), tok3->fileIndex()); // copy else addtoken(tok3->str().c_str(), tok3->linenr(), tok3->fileIndex()); } } } // Replace all these template usages.. s = name + " < " + type2 + " >"; for (std::string::size_type pos = s.find(","); pos != std::string::npos; pos = s.find(",", pos + 2)) { s.insert(pos + 1, " "); s.insert(pos, " "); } for (Token *tok4 = tok2; tok4; tok4 = tok4->next()) { if (Token::simpleMatch(tok4, s.c_str())) { tok4->str(name2.c_str()); while (tok4->next()->str() != ">") tok4->deleteNext(); tok4->deleteNext(); } } } } } //--------------------------------------------------------------------------- void Tokenizer::setVarId() { // Clear all variable ids for (Token *tok = _tokens; tok; tok = tok->next()) tok->varId(0); // Set variable ids.. unsigned int _varId = 0; for (Token *tok = _tokens; tok; tok = tok->next()) { if (tok != _tokens && !Token::Match(tok, "[,;{}(] %type%")) continue; if (Token::Match(tok, "[,;{}(] %type%")) tok = tok->next(); if (Token::Match(tok, "else|return|typedef|delete")) continue; if (Token::simpleMatch(tok, "const")) tok = tok->next(); if (Token::simpleMatch(tok, "std ::")) tok = tok->tokAt(2); // Skip template arguments.. if (Token::Match(tok, "%type% <")) { Token *tok2 = tok->tokAt(2); while (tok2 && (tok2->isName() || tok2->str() == "*")) tok2 = tok2->next(); if (Token::Match(tok2, "> %var%")) tok = tok2; else if (Token::Match(tok2, "> :: %var%")) tok = tok2->next(); else continue; // Not code that I understand / not a variable declaration } // Determine name of declared variable.. const char *varname = 0; Token *tok2 = tok->tokAt(1); while (tok2) { if (tok2->isName()) varname = tok2->strAt(0); else if (tok2->str() != "*" && tok2->str() != "&") break; tok2 = tok2->next(); } // End of tokens reached.. if (!tok2) break; // Is it a function? if (tok2->str() == "(") { // Search for function declaration, e.g. void f(); if (Token::simpleMatch(tok2->next(), ") ;")) continue; // Search for function declaration, e.g. void f( int c ); if (Token::Match(tok2->next(), "%num%") || Token::Match(tok2->next(), "%bool%") || tok2->next()->str()[0] == '"' || tok2->next()->varId() != 0) { // This is not a function } else { continue; } } // Variable declaration found => Set variable ids if (Token::Match(tok2, "[,();[=]") && varname) { ++_varId; int indentlevel = 0; int parlevel = 0; bool dot = false; for (tok2 = tok->next(); tok2; tok2 = tok2->next()) { if (!dot && tok2->str() == varname && !Token::Match(tok2->previous(), "struct|union")) tok2->varId(_varId); else if (tok2->str() == "{") ++indentlevel; else if (tok2->str() == "}") { --indentlevel; if (indentlevel < 0) break; } else if (tok2->str() == "(") ++parlevel; else if (tok2->str() == ")") { // Is this a function parameter or a variable declared in for example a for loop? if (parlevel == 0 && indentlevel == 0 && Token::Match(tok2, ") const| {")) ; else --parlevel; } else if (parlevel < 0 && tok2->str() == ";") break; dot = bool(tok2->str() == "."); } } } // Struct/Class members for (Token *tok = _tokens; tok; tok = tok->next()) { // str.clear is a variable // str.clear() is a member function if (tok->varId() != 0 && Token::Match(tok->next(), ". %var%") && !Token::Match(tok->tokAt(3), "(") && tok->tokAt(2)->varId() == 0) { ++_varId; const std::string pattern(std::string(". ") + tok->strAt(2)); for (Token *tok2 = tok; tok2; tok2 = tok2->next()) { if (tok2->varId() == tok->varId() && Token::simpleMatch(tok2->next(), pattern.c_str())) tok2->tokAt(2)->varId(_varId); } } } // class members.. for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok, "class %var% {")) { const std::string &classname(tok->next()->str()); // What member variables are there in this class? std::list varlist; { unsigned int indentlevel = 0; for (const Token *tok2 = tok; tok2; tok2 = tok2->next()) { // Indentation.. if (tok2->str() == "{") ++indentlevel; else if (tok2->str() == "}") { if (indentlevel <= 1) break; --indentlevel; } // Found a member variable.. else if (indentlevel == 1 && tok2->varId() > 0) varlist.push_back(tok2); } } // Are there any member variables in this class? if (varlist.empty()) continue; // Member functions for this class.. std::list funclist; { const std::string funcpattern(classname + " :: %var% ("); for (Token *tok2 = tok; tok2; tok2 = tok2->next()) { // Found a class function.. if (Token::Match(tok2, funcpattern.c_str())) { // Goto the end paranthesis.. while (tok2 && tok2->str() != ")") tok2 = tok2->next(); if (!tok2) break; // If this is a function implementation.. add it to funclist if (Token::Match(tok2, ") const|volatile| {")) funclist.push_back(tok2); } } } // Are there any member functions for this class? if (funclist.empty()) continue; // Update the variable ids.. for (std::list::const_iterator var = varlist.begin(); var != varlist.end(); ++var) { const unsigned int varid((*var)->varId()); const std::string &varname((*var)->str()); // Parse each function.. for (std::list::iterator func = funclist.begin(); func != funclist.end(); ++func) { unsigned int indentlevel = 0; for (Token *tok2 = *func; tok2; tok2 = tok2->next()) { if (tok2->str() == "{") ++indentlevel; else if (tok2->str() == "}") { if (indentlevel <= 1) break; --indentlevel; } else if (indentlevel > 0 && tok2->str() == varname && tok2->varId() == 0) tok2->varId(varid); } } } } } } //--------------------------------------------------------------------------- // Simplify token list //--------------------------------------------------------------------------- void Tokenizer::simplifyNamespaces() { for (Token *token = _tokens; token; token = token->next()) { while (token && token->str() == "namespace" && (!token->previous() || token->previous()->str() != "using")) { // Token is namespace and there is no "using" before it. Token *start = token; Token *tok = token->tokAt(2); if (!tok) return; tok = tok->link(); if (tok && tok->str() == "}") { tok = tok->previous(); tok->deleteNext(); start->deleteNext(); start->deleteNext(); if (start->previous()) { token = start->next(); start = start->previous(); start->deleteNext(); } else { // First token in the list, don't delete // as _tokens is attached to it. start->deleteThis(); } } else { return; } } if (!token) break; } } bool Tokenizer::createLinks() { std::list links; std::list links2; for (Token *token = _tokens; token; token = token->next()) { if (token->link()) { token->link(0); } if (token->str() == "{") { links.push_back(token); } else if (token->str() == "}") { if (links.size() == 0) { // Error, { and } don't match. syntaxError(token, '{'); return false; } token->link(links.back()); links.back()->link(token); links.pop_back(); } else if (token->str() == "(") { links2.push_back(token); } else if (token->str() == ")") { if (links2.size() == 0) { // Error, ( and ) don't match. syntaxError(token, '('); return false; } token->link(links2.back()); links2.back()->link(token); links2.pop_back(); } } if (links.size() > 0) { // Error, { and } don't match. syntaxError(_tokens, '{'); return false; } if (links2.size() > 0) { // Error, ( and ) don't match. syntaxError(_tokens, '('); return false; } return true; } void Tokenizer::simplifyTokenList() { createLinks(); simplifyNamespaces(); // Combine wide strings for (Token *tok = _tokens; tok; tok = tok->next()) { while (tok->str() == "L" && tok->next() && tok->next()->str()[0] == '"') { // Combine 'L "string"' tok->str(tok->next()->str().c_str()); tok->deleteNext(); } } // Combine strings for (Token *tok = _tokens; tok; tok = tok->next()) { while (tok->str()[0] == '"' && tok->next() && tok->next()->str()[0] == '"') { // Two strings after each other, combine them tok->concatStr(tok->next()->str()); tok->deleteNext(); } } // Remove unwanted keywords static const char * const unwantedWords[] = { "unsigned", "unlikely", "likely" }; for (Token *tok = _tokens; tok; tok = tok->next()) { for (unsigned ui = 0; ui < sizeof(unwantedWords) / sizeof(unwantedWords[0]) && tok->next(); ui++) { if (tok->next()->str() == unwantedWords[ui]) { tok->deleteNext(); break; } } if (Token::simpleMatch(tok->next(), "__builtin_expect (")) { unsigned int parlevel = 0; for (Token *tok2 = tok->next(); tok2; tok2 = tok2->next()) { if (tok2->str() == "(") ++parlevel; else if (tok2->str() == ")") { if (parlevel <= 1) break; --parlevel; } if (parlevel == 1 && tok2->str() == ",") { if (Token::Match(tok2, ", %num% )")) { tok->deleteNext(); tok2->deleteThis(); tok2->deleteThis(); } break; } } } } // Convert + + into + and + - into - for (Token *tok = _tokens; tok; tok = tok->next()) { while (tok->next()) { if (tok->str() == "+") { if (tok->next()->str() == "+") { tok->deleteNext(); continue; } else if (tok->next()->str() == "-") { tok->str("-"); tok->deleteNext(); continue; } } else if (tok->str() == "-") { if (tok->next()->str() == "-") { tok->str("+"); tok->deleteNext(); continue; } else if (tok->next()->str() == "+") { tok->deleteNext(); continue; } } break; } } // Fill the map _typeSize.. _typeSize.clear(); _typeSize["char"] = sizeof(char); _typeSize["short"] = sizeof(short); _typeSize["int"] = sizeof(int); _typeSize["long"] = sizeof(long); _typeSize["float"] = sizeof(float); _typeSize["double"] = sizeof(double); _typeSize["*"] = sizeof(void *); for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok, "class|struct %var%")) { _typeSize[tok->strAt(1)] = 100; } } // Replace 'sizeof(var)' with 'sizeof(type)' for (Token *tok = _tokens; tok; tok = tok->next()) { if (! Token::Match(tok, "[;{}] %type% *| %var% ;")) continue; const int type_tok = (tok->tokAt(2)->str() == "*" ? 2 : 1); const int varname_tok = type_tok + 1; const unsigned int varid = tok->tokAt(varname_tok)->varId(); if (varid <= 0) continue; int indentlevel = 0; for (Token *tok2 = tok; tok2; tok2 = tok2->next()) { if (tok2->str() == "{") ++indentlevel; else if (tok2->str() == "}") { --indentlevel; if (indentlevel < 0) break; } else if (Token::Match(tok2, "sizeof ( %varid% )", varid)) { tok2 = tok2->tokAt(2); tok2->str(tok->strAt(type_tok)); } } } // Replace 'sizeof(type)'.. for (Token *tok = _tokens; tok; tok = tok->next()) { if (tok->str() != "sizeof") continue; if (tok->strAt(1) != std::string("(")) { // Add parenthesis around the sizeof for (Token *tempToken = tok->next(); tempToken; tempToken = tempToken->next()) { if (Token::Match(tempToken, "%var%")) { if (Token::simpleMatch(tempToken->next(), ".")) { // We are checking a class or struct, search next varname tempToken = tempToken->tokAt(1); continue; } else if (Token::simpleMatch(tempToken->next(), "- >")) { // We are checking a class or struct, search next varname tempToken = tempToken->tokAt(2); continue; } else if (Token::Match(tempToken->next(), "++|--")) { // We have variable++ or variable--, there should be // nothing after this tempToken = tempToken->tokAt(2); } else if (Token::simpleMatch(tempToken->next(), "[")) { // TODO: We need to find closing ], then check for // dots and arrows "var[some[0]]->other" // But for now, just bail out break; } // Ok, we should be clean. Add ) after tempToken tok->insertToken("("); tempToken->insertToken(")"); break; } } } // sizeof(type *) => sizeof(*) if (Token::Match(tok, "sizeof ( %type% *)")) { tok->next()->deleteNext(); } if (Token::Match(tok, "sizeof ( * )")) { std::ostringstream str; str << SizeOfType(tok->strAt(2)); tok->str(str.str()); for (int i = 0; i < 3; i++) { tok->deleteNext(); } } else if (Token::Match(tok, "sizeof ( %var% )") && tok->tokAt(2)->varId() > 0) { // don't try to replace size of variable if variable has // similar name with type (#329) } else if (Token::Match(tok, "sizeof ( %type% )")) { const char *type = tok->strAt(2); int size = SizeOfType(type); if (size > 0) { std::ostringstream str; str << size; tok->str(str.str()); for (int i = 0; i < 3; i++) { tok->deleteNext(); } } } else if (Token::Match(tok, "sizeof ( * %var% )") || Token::Match(tok, "sizeof ( %var% [ %num% ] )")) { // Some default value.. int sz = 100; unsigned int varid = tok->tokAt((tok->tokAt(2)->str() == "*") ? 3 : 2)->varId(); if (varid != 0) { // Try to locate variable declaration.. const Token *decltok = Token::findmatch(_tokens, "%type% %varid% [", varid); if (decltok) { sz = SizeOfType(decltok->strAt(0)); } } if (sz > 0) { std::ostringstream ostr; ostr << sz; tok->str(ostr.str()); while (tok->next()->str() != ")") tok->deleteNext(); tok->deleteNext(); } } } // Replace 'sizeof(var)' for (Token *tok = _tokens; tok; tok = tok->next()) { if (! Token::Match(tok, "%type% *| %var% [ %num% ] ;")) continue; const int type_tok = ((tok->next()->str() == "*") ? 1 : 0); int size = SizeOfType(tok->tokAt(type_tok)->str().c_str()); if (size <= 0) continue; const int varname_tok = type_tok + 1; const unsigned int varid = tok->tokAt(varname_tok)->varId(); if (varid == 0) continue; const int num_tok = varname_tok + 2; int total_size = size * MathLib::toLongNumber(tok->strAt(num_tok)); // Replace 'sizeof(var)' with number int indentlevel = 0; const int next_tok = num_tok + 3; for (Token *tok2 = tok->tokAt(next_tok); tok2; tok2 = tok2->next()) { if (tok2->str() == "{") { ++indentlevel; } else if (tok2->str() == "}") { --indentlevel; if (indentlevel < 0) break; } else if (Token::Match(tok2, "sizeof ( %varid% )", varid)) { std::ostringstream str; str << total_size; tok2->str(str.str()); // Delete the other tokens.. for (int i = 0; i < 3; i++) { tok2->deleteNext(); } } } } // Replace constants.. for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok, "const %type% %var% = %num% ;")) { unsigned int varId = tok->tokAt(2)->varId(); const char *num = tok->strAt(4); int indent = 1; for (Token *tok2 = tok->tokAt(6); tok2; tok2 = tok2->next()) { if (tok2->str() == "{") { ++indent; } else if (tok2->str() == "}") { --indent; if (indent == 0) break; } // Compare constants, but don't touch members of other structures else if (tok2->varId() == varId) { tok2->str(num); } } } } simplifyCasts(); // Simplify simple calculations.. while (simplifyCalculations()) ; // Replace "*(str + num)" => "str[num]" for (Token *tok = _tokens; tok; tok = tok->next()) { if (! strchr(";{}(=<>", tok->str()[0])) continue; Token *next = tok->next(); if (! next) break; if (Token::Match(next, "* ( %var% + %num% )")) { const char *str[4] = {"var", "[", "num", "]"}; str[0] = tok->strAt(3); str[2] = tok->strAt(5); for (int i = 0; i < 4; i++) { tok = tok->next(); tok->str(str[i]); } tok->deleteNext(); tok->deleteNext(); } } // Simplify variable declarations simplifyVarDecl(); // Replace NULL with 0.. for (Token *tok = _tokens; tok; tok = tok->next()) { if (tok->str() == "NULL") tok->str("0"); } // Replace pointer casts of 0.. "(char *)0" => "0" for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok->next(), "( %type% * ) 0") || Token::Match(tok->next(), "( %type% %type% * ) 0")) { while (!Token::simpleMatch(tok->next(), "0")) tok->deleteNext(); } } simplifyIfAddBraces(); simplifyFunctionParameters(); elseif(); simplifyIfNot(); simplifyNot(); simplifyIfAssign(); for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok, "case %any% : %var%")) tok->tokAt(2)->insertToken(";"); if (Token::Match(tok, "default : %var%")) tok->next()->insertToken(";"); } // In case variable declarations have been updated... setVarId(); bool modified = true; while (modified) { modified = false; modified |= simplifyConditions(); modified |= simplifyFunctionReturn(); modified |= simplifyKnownVariables(); modified |= removeReduntantConditions(); modified |= simplifyRedundantParanthesis(); modified |= simplifyCalculations(); modified |= simplifyQuestionMark(); } simplifyCommaNearKeyWords(); createLinks(); if (_settings._debug) { _tokens->printOut(); } } //--------------------------------------------------------------------------- const Token *Tokenizer::findClosing(const Token *tok, const char *start, const char *end) { if (!tok) return 0; // Find the closing "}" int indentLevel = 0; for (const Token *closing = tok->next(); closing; closing = closing->next()) { if (closing->str() == start) { ++indentLevel; continue; } if (closing->str() == end) --indentLevel; if (indentLevel >= 0) continue; // Closing } is found. return closing; } return 0; } bool Tokenizer::removeReduntantConditions() { bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (!Token::simpleMatch(tok, "if")) continue; if (!Token::Match(tok->tokAt(1), "( %bool% ) {")) continue; // Find matching else const Token *elseTag = 0; // Find the closing "}" elseTag = Tokenizer::findClosing(tok->tokAt(4), "{", "}"); if (elseTag) elseTag = elseTag->next(); bool boolValue = false; if (tok->tokAt(2)->str() == "true") boolValue = true; // Handle if with else if (elseTag && elseTag->str() == "else") { if (Token::simpleMatch(elseTag->next(), "if")) { // Handle "else if" if (boolValue == false) { // Convert "if( false ) {aaa;} else if() {bbb;}" => "if() {bbb;}" Token::eraseTokens(tok, elseTag->tokAt(2)); ret = true; } else { // Keep first if, remove every else if and else after it const Token *lastTagInIf = elseTag->tokAt(2); while (lastTagInIf) { if (lastTagInIf->str() == "(") { lastTagInIf = Tokenizer::findClosing(lastTagInIf, "(", ")"); lastTagInIf = lastTagInIf->next(); } lastTagInIf = Tokenizer::findClosing(lastTagInIf, "{", "}"); lastTagInIf = lastTagInIf->next(); if (!Token::simpleMatch(lastTagInIf, "else")) break; lastTagInIf = lastTagInIf->next(); if (Token::simpleMatch(lastTagInIf, "if")) lastTagInIf = lastTagInIf->next(); } Token::eraseTokens(elseTag->previous(), lastTagInIf); ret = true; } } else { // Handle else if (boolValue == false) { // Convert "if( false ) {aaa;} else {bbb;}" => "{bbb;}" or ";{bbb;}" if (tok->previous()) tok = tok->previous(); else tok->str(";"); Token::eraseTokens(tok, elseTag->tokAt(1)); } else { if (Token::simpleMatch(elseTag->tokAt(1), "{")) { // Convert "if( true ) {aaa;} else {bbb;}" => "{aaa;}" const Token *end = Tokenizer::findClosing(elseTag->tokAt(1), "{", "}"); if (!end) { // Possibly syntax error in code return false; } // Remove the "else { aaa; }" Token::eraseTokens(elseTag->previous(), end->tokAt(1)); } // Remove "if( true )" if (tok->previous()) tok = tok->previous(); else tok->str(";"); Token::eraseTokens(tok, tok->tokAt(5)); } ret = true; } } // Handle if without else else { if (boolValue == false) { // Remove if and its content if (tok->previous()) tok = tok->previous(); else tok->str(";"); Token::eraseTokens(tok, elseTag); } else { // convert "if( true ) {aaa;}" => "{aaa;}" if (tok->previous()) tok = tok->previous(); else tok->str(";"); Token::eraseTokens(tok, tok->tokAt(5)); } ret = true; } } return ret; } bool Tokenizer::simplifyIfAddBraces() { bool ret = false; for (Token *tok = _tokens; tok; tok = tok ? tok->next() : NULL) { if (Token::Match(tok, "if|for|while (")) { // Goto the ending ')' int parlevel = 1; tok = tok->next(); while (parlevel >= 1 && (tok = tok->next())) { if (tok->str() == "(") ++parlevel; else if (tok->str() == ")") --parlevel; } // ')' should be followed by '{' if (!tok || Token::simpleMatch(tok, ") {")) continue; } else if (tok->str() == "else") { // An else followed by an if or brace don't need to be processed further if (Token::Match(tok, "else if|{")) continue; } else { continue; } // If there is no code after he if(), abort if (!tok->next()) return false; // insert open brace.. tok->insertToken("{"); tok = tok->next(); Token *tempToken = tok; // insert close brace.. // In most cases it would work to just search for the next ';' and insert a closing brace after it. // But here are special cases.. // * if (cond) for (;;) break; // * if (cond1) if (cond2) { } int parlevel = 0; int indentlevel = 0; while ((tempToken = tempToken->next()) != NULL) { if (tempToken->str() == "{") ++indentlevel; else if (tempToken->str() == "}") { --indentlevel; if (indentlevel == 0) break; } else if (tempToken->str() == "(") ++parlevel; else if (tempToken->str() == ")") --parlevel; else if (indentlevel == 0 && parlevel == 0 && tempToken->str() == ";") break; } if (tempToken) { tempToken->insertToken("}"); ret = true; } } return ret; } bool Tokenizer::simplifyConditions() { bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok, "! %num%") || Token::Match(tok, "! %bool%")) { if (tok->next()->str() == "0" || tok->next()->str() == "false") tok->str("true"); else tok->str("false"); tok->deleteNext(); ret = true; } if (Token::simpleMatch(tok, "( true &&") || Token::simpleMatch(tok, "&& true &&") || Token::simpleMatch(tok->next(), "&& true )")) { tok->deleteNext(); tok->deleteNext(); ret = true; } else if (Token::simpleMatch(tok, "( false ||") || Token::simpleMatch(tok, "|| false ||") || Token::simpleMatch(tok->next(), "|| false )")) { tok->deleteNext(); tok->deleteNext(); ret = true; } // Change numeric constant in condition to "true" or "false" if (Token::Match(tok, "if|while ( %num%") && (tok->tokAt(3)->str() == ")" || tok->tokAt(3)->str() == "||" || tok->tokAt(3)->str() == "&&")) { tok->tokAt(2)->str((tok->tokAt(2)->str() != "0") ? "true" : "false"); ret = true; } Token *tok2 = tok->tokAt(2); if (tok2 && (tok->str() == "&&" || tok->str() == "||") && Token::Match(tok->next(), "%num%") && (tok2->str() == ")" || tok2->str() == "&&" || tok2->str() == "||")) { tok->next()->str((tok->next()->str() != "0") ? "true" : "false"); ret = true; } // Reduce "(%num% == %num%)" => "(true)"/"(false)" const Token *tok4 = tok->tokAt(4); if (! tok4) break; if ((tok->str() == "&&" || tok->str() == "||" || tok->str() == "(") && (Token::Match(tok->tokAt(1), "%num% %any% %num%") || Token::Match(tok->tokAt(1), "%bool% %any% %bool%")) && (tok4->str() == "&&" || tok4->str() == "||" || tok4->str() == ")")) { std::string cmp = tok->strAt(2); bool result = false; if (Token::Match(tok->tokAt(1), "%num%")) { // Compare numbers double op1 = (strstr(tok->strAt(1), "0x")) ? std::strtol(tok->strAt(1), 0, 16) : std::atof(tok->strAt(1)); double op2 = (strstr(tok->strAt(3), "0x")) ? std::strtol(tok->strAt(3), 0, 16) : std::atof(tok->strAt(3)); if (cmp == "==") result = (op1 == op2); else if (cmp == "!=") result = (op1 != op2); else if (cmp == ">=") result = (op1 >= op2); else if (cmp == ">") result = (op1 > op2); else if (cmp == "<=") result = (op1 <= op2); else if (cmp == "<") result = (op1 < op2); else cmp = ""; } else { // Compare boolean bool op1 = (tok->strAt(1) == std::string("true")); bool op2 = (tok->strAt(3) == std::string("true")); if (cmp == "==") result = (op1 == op2); else if (cmp == "!=") result = (op1 != op2); else if (cmp == ">=") result = (op1 >= op2); else if (cmp == ">") result = (op1 > op2); else if (cmp == "<=") result = (op1 <= op2); else if (cmp == "<") result = (op1 < op2); else cmp = ""; } if (! cmp.empty()) { tok = tok->next(); tok->deleteNext(); tok->deleteNext(); tok->str(result ? "true" : "false"); ret = true; } } } return ret; } bool Tokenizer::simplifyQuestionMark() { bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (tok->str() != "?") continue; if (!tok->previous() || !tok->tokAt(-2)) continue; if (!Token::Match(tok->tokAt(-2), "[=,(]")) continue; if (!Token::Match(tok->previous(), "%bool%") && !Token::Match(tok->previous(), "%num%")) continue; if (tok->previous()->str() == "false" || tok->previous()->str() == "0") { // Use code after semicolon, remove code before it. const Token *end = Token::findmatch(tok, ":"); if (!end || !end->next()) continue; end = end->next(); tok = tok->tokAt(-2); while (tok->next() != end) { tok->deleteNext(); } tok = tok->next(); ret = true; } else { // Use code before semicolon } } return ret; } bool Tokenizer::simplifyCasts() { bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok->next(), "( %type% *| )") || Token::Match(tok->next(), "( %type% %type% *| )")) { if (tok->isName() && tok->str() != "return") continue; // Is it a cast of some variable? const Token *tok2 = tok->tokAt(3); while (tok2 && tok2->str() != ")") tok2 = tok2->next(); if (!Token::Match(tok2, ") %var%")) continue; // Remove cast.. while (tok->next()->str() != ")") tok->deleteNext(); tok->deleteNext(); ret = true; } else if (Token::Match(tok->next(), "dynamic_cast|reinterpret_cast|const_cast|static_cast <")) { Token *tok2 = tok->next(); unsigned int level = 0; while (tok2) { if (tok2->str() == "<") ++level; else if (tok2->str() == ">") { --level; if (level == 0) break; } tok2 = tok2->next(); } if (Token::simpleMatch(tok2, "> (")) { Token::eraseTokens(tok, tok2->tokAt(2)); tok2 = tok; int parlevel = 0; while (tok2->next() && parlevel >= 0) { tok2 = tok2->next(); if (Token::simpleMatch(tok2->next(), "(")) ++parlevel; else if (Token::simpleMatch(tok2->next(), ")")) --parlevel; } if (tok2->next()) tok2->deleteNext(); ret = true; } } } return ret; } bool Tokenizer::simplifyFunctionParameters() { bool ret = false; int indentlevel = 0; for (Token *tok = _tokens; tok; tok = tok->next()) { if (tok->str() == "{") ++indentlevel; else if (tok->str() == "}") --indentlevel; // Find the function e.g. foo( x ) or foo( x, y ) else if (indentlevel == 0 && Token::Match(tok, "%var% ( %var% [,)]")) { // We have found old style function, now we need to change it // Get list of argument names std::map argumentNames; bool bailOut = false; for (tok = tok->tokAt(2); tok; tok = tok->tokAt(2)) { if (!Token::Match(tok, "%var% [,)]")) { bailOut = true; break; } argumentNames[tok->str()] = tok; if (tok->next()->str() == ")") { tok = tok->tokAt(2); break; } } if (bailOut) { continue; } Token *start = tok; while (tok && tok->str() != "{") { if (tok->str() == ";") { tok = tok->previous(); // Move tokens from start to tok into the place of // argumentNames[tok->str()] and remove the ";" if (argumentNames.find(tok->str()) == argumentNames.end()) { bailOut = true; break; } // Remove the following ";" Token *temp = tok->tokAt(2); tok->deleteNext(); // Replace "x" with "int x" or similar Token::replace(argumentNames[tok->str()], start, tok); ret = true; tok = temp; start = tok; } else { tok = tok->next(); } } if (tok == NULL) { break; } if (bailOut) { continue; } ++indentlevel; } } return ret; } bool Tokenizer::simplifyFunctionReturn() { bool ret = false; int indentlevel = 0; for (const Token *tok = tokens(); tok; tok = tok->next()) { if (tok->str() == "{") ++indentlevel; else if (tok->str() == "}") --indentlevel; else if (indentlevel == 0 && Token::Match(tok, "%var% ( ) { return %num% ; }")) { std::ostringstream pattern; pattern << "[(=+-*/] " << tok->str() << " ( ) [;)+-*/]"; for (Token *tok2 = _tokens; tok2; tok2 = tok2->next()) { if (Token::Match(tok2, pattern.str().c_str())) { tok2 = tok2->next(); tok2->str(tok->strAt(5)); tok2->deleteNext(); tok2->deleteNext(); ret = true; } } } } return ret; } static void incdec(std::string &value, const std::string &op) { int ivalue = 0; std::istringstream istr(value.c_str()); istr >> ivalue; if (op == "++") ++ivalue; else if (op == "--") --ivalue; std::ostringstream ostr; ostr << ivalue; value = ostr.str(); } bool Tokenizer::simplifyVarDecl() { // Split up variable declarations.. // "int a=4;" => "int a; a=4;" bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (tok->previous() && !Token::Match(tok->previous(), "[{};)]")) continue; Token *type0 = tok; if (!Token::Match(type0, "%type%")) continue; if (Token::Match(type0, "else|return")) continue; bool isconst = false; Token *tok2 = type0; unsigned int typelen = 1; while (Token::Match(tok2, "%type% %type% *| %var%")) { if (tok2->str() == "const") isconst = true; tok2 = tok2->next(); ++typelen; } // Don't split up const declaration.. if (isconst && Token::Match(tok2, "%type% %var% =")) continue; if (Token::Match(tok2, "%type% %var% ,|=")) { if (tok2->next()->str() != "operator") tok2 = tok2->tokAt(2); // The ',' or '=' token else tok2 = NULL; } else if (Token::Match(tok2, "%type% * %var% ,|=")) { if (tok2->tokAt(2)->str() != "operator") tok2 = tok2->tokAt(3); // The ',' token else tok2 = NULL; } else if (Token::Match(tok2, "%type% * const %var% ,|=")) { if (tok2->tokAt(3)->str() != "operator") { tok2 = tok2->tokAt(4); // The ',' token } else { tok2 = NULL; } } else if (Token::Match(tok2, "%type% %var% [ %num% ] ,")) { tok2 = tok2->tokAt(5); // The ',' token } else if (Token::Match(tok2, "%type% * %var% [ %num% ] ,")) { tok2 = tok2->tokAt(6); // The ',' token } else { tok2 = NULL; typelen = 0; } if (tok2) { ret = true; if (tok2->str() == ",") { tok2->str(";"); InsertTokens(tok2, type0, typelen); } else { Token *eq = tok2; int parlevel = 0; while (tok2) { if (strchr("{(<", tok2->str()[0])) { ++parlevel; } else if (strchr("})>", tok2->str()[0])) { if (parlevel <= 0) break; --parlevel; } else if (parlevel == 0 && strchr(";,", tok2->str()[0])) { // "type var =" => "type var; var =" Token *VarTok = type0->tokAt(typelen); while (Token::Match(VarTok, "*|const")) VarTok = VarTok->next(); InsertTokens(eq, VarTok, 2); eq->str(";"); // "= x, " => "= x; type " if (tok2->str() == ",") { tok2->str(";"); InsertTokens(tok2, type0, typelen); } break; } tok2 = tok2->next(); } } } } return ret; } void Tokenizer::unsignedint() { for (Token *tok = _tokens; tok; tok = tok->next()) { // A variable declaration where the "int" is left out? if (!Token::Match(tok, "unsigned %var% [;,=]")) continue; // Previous token should either be a symbol or one of "{};" if (tok->previous() && !tok->previous()->isName() && !Token::Match(tok->previous(), "[{};]")) continue; // next token should not be a standard type? if (tok->next()->isStandardType()) continue; // The "int" is missing.. add it tok->insertToken("int"); } } bool Tokenizer::simplifyIfAssign() { bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok->next(), "if|while ( (| %var% =") || Token::Match(tok->next(), "if|while ( ! ( %var% =")) { ret = true; // simplifying a "while" condition ? const bool iswhile(tok->next()->str() == "while"); // delete the "if" tok->deleteNext(); // Remember if there is a "!" or not. And delete it if there are. bool isNot = false; if (Token::simpleMatch(tok->tokAt(2), "!")) { isNot = true; tok->next()->deleteNext(); } // Delete paranthesis.. and remember how many there are. int numpar = 0; while (tok->next()->str() == "(") { ++numpar; tok->deleteNext(); } // Skip the "%var% = ..." Token *tok2 = tok; int indentlevel = 0; for (tok2 = tok; tok2; tok2 = tok2->next()) { if (tok2->str() == "(") ++indentlevel; else if (tok2->str() == ")") { if (indentlevel <= 0) break; --indentlevel; } } // Insert "; if|while ( .." if (tok2) { tok2 = tok2->previous(); tok2->insertToken(tok->strAt(1)); for (int p = 0; p < numpar; ++p) tok2->insertToken("("); if (isNot) tok2->next()->insertToken("!"); tok2->insertToken(iswhile ? "while" : "if"); tok2->insertToken(";"); // If it's a while loop.. insert the assignment in the loop if (iswhile) { indentlevel = 0; Token *tok3 = tok2; for (tok3 = tok2; tok3; tok3 = tok3->next()) { if (tok3->str() == "{") ++indentlevel; else if (tok3->str() == "}") { if (indentlevel <= 1) break; --indentlevel; } } if (tok3 && indentlevel == 1) { tok3 = tok3->previous(); for (tok2 = tok2->next(); tok2 && tok2 != tok; tok2 = tok2->previous()) { tok3->insertToken(tok2->str().c_str()); tok3->next()->fileIndex(tok2->fileIndex()); tok3->next()->linenr(tok2->linenr()); } } } } } } return ret; } bool Tokenizer::simplifyIfNot() { // Make sure we have working links createLinks(); bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::simpleMatch(tok, "0 == (") || Token::Match(tok, "0 == %var%")) { tok->deleteNext(); tok->str("!"); ret = true; } if (Token::Match(tok, "%var% == 0")) { tok->deleteNext(); tok->next()->str(tok->str()); tok->str("!"); ret = true; } if (tok->link() && Token::simpleMatch(tok, ") == 0")) { tok->deleteNext(); tok->deleteNext(); tok->link()->insertToken("("); tok->link()->str("!"); ret = true; } } return ret; } bool Tokenizer::simplifyNot() { // "if (not p)" => "if (!p)" bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::Match(tok, "if|while ( not %var%")) tok->tokAt(2)->str("!"); if (Token::Match(tok, "&& not %var%")) tok->next()->str("!"); if (Token::Match(tok, "|| not %var%")) tok->next()->str("!"); } return ret; } bool Tokenizer::simplifyKnownVariables() { createLinks(); bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { // Search for a block of code if (! Token::Match(tok, ") const| {")) continue; // parse the block of code.. int indentlevel = 0; for (Token *tok2 = tok; tok2; tok2 = tok2->next()) { if (tok2->str() == "{") ++indentlevel; else if (tok2->str() == "}") { --indentlevel; if (indentlevel <= 0) break; } else if (tok2->previous()->str() != "*" && (Token::Match(tok2, "%var% = %num% ;") || Token::Match(tok2, "%var% = %bool% ;"))) { unsigned int varid = tok2->varId(); if (varid == 0) continue; std::string value(tok2->strAt(2)); Token* bailOutFromLoop = 0; int indentlevel3 = indentlevel; // indentlevel for tok3 for (Token *tok3 = tok2->next(); tok3; tok3 = tok3->next()) { if (tok3->str() == "{") { ++indentlevel3; } else if (tok3->str() == "}") { --indentlevel3; if (indentlevel3 < indentlevel) break; } if (bailOutFromLoop) { // This could be a loop, skip it, but only if it doesn't contain // the variable we are checking for. If it contains the variable // we will bail out. if (tok3->varId() == varid) { // Continue tok2 = bailOutFromLoop; break; } else if (tok3 == bailOutFromLoop) { // We have skipped the loop bailOutFromLoop = 0; continue; } continue; } else if (tok3->str() == "{" && Token::simpleMatch(tok3->previous(), ")")) { // There is a possible loop after the assignment. Try to skip it. bailOutFromLoop = tok3->link(); continue; } else if (tok3->str() == "}" && Token::simpleMatch(tok3->link()->previous(), ")")) { // Assignment was in the middle of possible loop, bail out. break; } // Variable is used somehow in a non-defined pattern => bail out if (tok3->varId() == varid) break; // Using the variable in condition.. if (Token::Match(tok3, "(|!|==|!=|<|<=|>|>= %varid% )|==|!=|<|<=|>|>=", varid)) { tok3 = tok3->next(); tok3->str(value.c_str()); ret = true; } // Variable is used in calculation.. if (Token::Match(tok3, "[=+-*/[] %varid% [+-*/;]]", varid)) { tok3 = tok3->next(); tok3->str(value.c_str()); ret = true; } if (Token::Match(tok3->next(), "%varid% ++|--", varid)) { const std::string op(tok3->strAt(2)); if (Token::Match(tok3, "[{};] %any% %any% ;")) { tok3->deleteNext(); tok3->deleteNext(); } else { tok3 = tok3->next(); tok3->str(value.c_str()); tok3->deleteNext(); } incdec(value, op); tok2->tokAt(2)->str(value.c_str()); ret = true; } if (Token::Match(tok3->next(), "++|-- %varid%", varid)) { incdec(value, tok3->strAt(1)); tok2->tokAt(2)->str(value.c_str()); if (Token::Match(tok3, "[;{}] %any% %any% ;")) { tok3->deleteNext(); tok3->deleteNext(); } else { tok3->deleteNext(); tok3->next()->str(value.c_str()); } tok3 = tok3->next(); ret = true; } } } } } return ret; } bool Tokenizer::elseif() { bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (!Token::simpleMatch(tok, "else if")) continue; int indent = 0; for (Token *tok2 = tok; indent >= 0 && tok2; tok2 = tok2->next()) { if (Token::Match(tok2, "(|{")) ++indent; else if (Token::Match(tok2, ")|}")) --indent; if (indent == 0 && Token::Match(tok2, "}|;")) { if (!Token::simpleMatch(tok2->next(), "else")) { tok->insertToken("{"); tok2->insertToken("}"); ret = true; break; } } } } return ret; } bool Tokenizer::simplifyRedundantParanthesis() { if (!createLinks()) return false; bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (tok->str() != "(") continue; while (Token::simpleMatch(tok, "( (") && tok->link()->previous() == tok->next()->link()) { // We have "(( *something* ))", remove the inner // paranthesis tok->deleteNext(); tok->link()->tokAt(-2)->deleteNext(); ret = true; } while (Token::Match(tok->previous(), "[;{] ( %var% (") && tok->link()->previous() == tok->tokAt(2)->link()) { // We have "( func ( *something* ))", remove the outer // paranthesis tok->link()->deleteThis(); tok->deleteThis(); ret = true; } while (Token::Match(tok->previous(), "[;{] ( delete %var% ) ;")) { // We have "( delete var )", remove the outer // paranthesis tok->tokAt(3)->deleteThis(); tok->deleteThis(); ret = true; } while (Token::Match(tok->previous(), "[;{] ( delete [ ] %var% ) ;")) { // We have "( delete [] var )", remove the outer // paranthesis tok->tokAt(5)->deleteThis(); tok->deleteThis(); ret = true; } if (Token::Match(tok, "( ( %bool% )") || Token::Match(tok, "( ( %num% )")) { tok->tokAt(2)->deleteNext(); tok->deleteNext(); ret = true; } } return ret; } bool Tokenizer::simplifyCalculations() { bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (Token::simpleMatch(tok->next(), "* 1") || Token::simpleMatch(tok->next(), "1 *")) { for (int i = 0; i < 2; i++) tok->deleteNext(); ret = true; } // (1-2) if (Token::Match(tok, "[[,(=<>] %num% [+-*/] %num% [],);=<>]")) { tok = tok->next(); // Don't simplify "%num% / 0" if (Token::simpleMatch(tok->next(), "/ 0")) continue; switch (*(tok->strAt(1))) { case '+': tok->str(MathLib::add(tok->str(), tok->strAt(2)).c_str()); break; case '-': tok->str(MathLib::subtract(tok->str(), tok->strAt(2)).c_str()); break; case '*': tok->str(MathLib::multiply(tok->str(), tok->strAt(2)).c_str()); break; case '/': tok->str(MathLib::divide(tok->str(), tok->strAt(2)).c_str()); break; } tok->deleteNext(); tok->deleteNext(); ret = true; } // Remove parantheses around number.. if (!tok->isName() && Token::Match(tok->next(), "( %num% )")) { tok->deleteNext(); tok = tok->next(); tok->deleteNext(); ret = true; } // Remove parantheses around variable.. // keep parantheses here: dynamic_cast(p); if (!tok->isName() && tok->str() != ">" && Token::Match(tok->next(), "( %var% ) [;),+-*/><]]")) { tok->deleteNext(); tok = tok->next(); tok->deleteNext(); ret = true; } } return ret; } //--------------------------------------------------------------------------- // Helper functions for handling the tokens list //--------------------------------------------------------------------------- //--------------------------------------------------------------------------- const Token *Tokenizer::GetFunctionTokenByName(const char funcname[]) const { for (unsigned int i = 0; i < _functionList.size(); ++i) { if (_functionList[i]->str() == funcname) { return _functionList[i]; } } return NULL; } void Tokenizer::fillFunctionList() { _functionList.clear(); int indentlevel = 0; for (const Token *tok = _tokens; tok; tok = tok->next()) { if (tok->str() == "{") ++indentlevel; else if (tok->str() == "}") --indentlevel; if (indentlevel > 0) { continue; } if (Token::Match(tok, "%var% (")) { // Check if this is the first token of a function implementation.. for (const Token *tok2 = tok->tokAt(2); tok2; tok2 = tok2->next()) { if (tok2->str() == ";") { tok = tok2; break; } else if (tok2->str() == "{") { break; } else if (tok2->str() == ")") { if (Token::Match(tok2, ") const| {")) { _functionList.push_back(tok); tok = tok2; } else { tok = tok2; while (tok->next() && !strchr(";{", tok->strAt(1)[0])) tok = tok->next(); } break; } } } } // If the _functionList functions with duplicate names, remove them // TODO this will need some better handling for (unsigned int func1 = 0; func1 < _functionList.size();) { bool hasDuplicates = false; for (unsigned int func2 = func1 + 1; func2 < _functionList.size();) { if (_functionList[func1]->str() == _functionList[func2]->str()) { hasDuplicates = true; _functionList.erase(_functionList.begin() + func2); } else { ++func2; } } if (! hasDuplicates) { ++func1; } else { _functionList.erase(_functionList.begin() + func1); } } } //--------------------------------------------------------------------------- // Deallocate lists.. void Tokenizer::DeallocateTokens() { deleteTokens(_tokens); _tokens = 0; _tokensBack = 0; _files.clear(); } void Tokenizer::deleteTokens(Token *tok) { while (tok) { Token *next = tok->next(); delete tok; tok = next; } } //--------------------------------------------------------------------------- const char *Tokenizer::getParameterName(const Token *ftok, int par) { int _par = 1; for (; ftok; ftok = ftok->next()) { if (ftok->str() == ",") ++_par; if (par == _par && Token::Match(ftok, "%var% [,)]")) return ftok->str().c_str(); } return NULL; } //--------------------------------------------------------------------------- std::string Tokenizer::fileLine(const Token *tok) const { std::ostringstream ostr; ostr << "[" << _files.at(tok->fileIndex()) << ":" << tok->linenr() << "]"; return ostr.str(); } std::string Tokenizer::file(const Token *tok) const { return _files.at(tok->fileIndex()); } //--------------------------------------------------------------------------- const Token * Tokenizer::FindClassFunction(const Token *tok, const char classname[], const char funcname[], int &indentlevel) { if (indentlevel < 0 || tok == NULL) return NULL; std::ostringstream classPattern; classPattern << "class " << classname << " :|{"; std::ostringstream internalPattern; internalPattern << funcname << " ("; std::ostringstream externalPattern; externalPattern << classname << " :: " << funcname << " ("; for (;tok; tok = tok->next()) { if (indentlevel == 0 && Token::Match(tok, classPattern.str().c_str())) { while (tok && tok->str() != "{") tok = tok->next(); if (tok) tok = tok->next(); if (! tok) break; indentlevel = 1; } if (tok->str() == "{") { // If indentlevel==0 don't go to indentlevel 1. Skip the block. if (indentlevel > 0) ++indentlevel; else { for (; tok; tok = tok->next()) { if (tok->str() == "{") ++indentlevel; else if (tok->str() == "}") { --indentlevel; if (indentlevel <= 0) break; } } if (tok == NULL) return NULL; continue; } } if (tok->str() == "}") { --indentlevel; if (indentlevel < 0) return NULL; } if (indentlevel == 1) { // Member function implemented in the class declaration? if (tok->str() != "~" && Token::Match(tok->next(), internalPattern.str().c_str())) { const Token *tok2 = tok->next(); while (tok2 && tok2->str() != "{" && tok2->str() != ";") tok2 = tok2->next(); if (tok2 && tok2->str() == "{") return tok->next(); } } else if (indentlevel == 0 && Token::Match(tok, externalPattern.str().c_str())) { return tok; } } // Not found return NULL; } //--------------------------------------------------------------------------- // Error message for bad iterator usage.. void Tokenizer::syntaxError(const Token *tok, char c) { if (_settings._debug) { _tokens->printOut(); } if (!_errorLogger) { std::cout << "### Unlogged error at Tokenizer::syntaxError" << std::endl; return; } std::list locationList; ErrorLogger::ErrorMessage::FileLocation loc; loc.line = tok->linenr(); loc.file = file(tok); locationList.push_back(loc); _errorLogger->reportErr( ErrorLogger::ErrorMessage(locationList, "error", std::string("Invalid number of character (") + c + "). Can't process file.", "syntaxError")); } bool Tokenizer::simplifyCommaNearKeyWords() { bool ret = false; for (Token *tok = _tokens; tok; tok = tok->next()) { if (tok->str() != ",") continue; // We must not accept just any keyword, e.g. accepting int // would cause function parameters to corrupt. if (Token::Match(tok->next(), "delete")) { // Handle "delete a, delete b;" tok->str(";"); ret = true; } else if (tok->previous() && Token::Match(tok->previous()->previous(), "delete") && tok->next()->varId() != 0) { // Handle "delete a, b;" tok->str(";"); tok->insertToken("delete"); ret = true; } } return ret; }