cppcheck/lib/token.h

1283 lines
41 KiB
C
Raw Normal View History

2008-12-18 22:28:57 +01:00
/*
* Cppcheck - A tool for static C/C++ code analysis
2020-05-10 11:16:32 +02:00
* Copyright (C) 2007-2020 Cppcheck team.
2008-12-18 22:28:57 +01:00
*
* 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 <http://www.gnu.org/licenses/>.
2008-12-18 22:28:57 +01:00
*/
//---------------------------------------------------------------------------
#ifndef tokenH
#define tokenH
//---------------------------------------------------------------------------
2008-12-18 22:28:57 +01:00
2017-05-27 04:33:47 +02:00
#include "config.h"
#include "mathlib.h"
#include "valueflow.h"
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
#include "templatesimplifier.h"
#include "utils.h"
2017-05-27 04:33:47 +02:00
#include <cstddef>
#include <functional>
#include <list>
#include <memory>
2017-05-27 04:33:47 +02:00
#include <ostream>
2008-12-18 22:28:57 +01:00
#include <string>
#include <vector>
2008-12-18 22:28:57 +01:00
2017-05-27 04:33:47 +02:00
class Enumerator;
class Function;
class Scope;
2017-05-27 04:33:47 +02:00
class Settings;
class Type;
class ValueType;
2017-05-27 04:33:47 +02:00
class Variable;
class TokenList;
Fix crash bug #8579 (#1238) * Added declaration for deletePrevious function * Added definition for deletePrevious function * Fixed crash from deleteThis invalidating pointers The crash was caused by deleteThis() invalidating the pointer to a constant variable usage. This happened when a usage followed an assignment. This fixes bug #8579. * Added tokensFront to match tokensBack This means deletePrevious can set the list's front if necessary. * Initialised tokensFront in appropriate places * Switched to using default Token constructor * Switched to using Token default constructor * Switched to using default constructor for Token * Added missing argument to Token constructor * Changed to use default constructor for Tokens * Switched to using default constructor for Tokens * Switched to using default constructor for Token * Added new test for deleting front Token Also made sure to use the correct constructor for Token in other tests. * Syntax error * Replaced tokensFront and tokensBack with a struct This decreases the size of the Token class for performance purposes. * Replaced tokensFront and tokensBack with a struct * Added tokensFrontBack to destructor * Reworked to use TokensBackFront struct Also ran astyle. * Reworked to use TokenList's TokensFrontBack member * Reworked to use TokensFrontBack struct * Reworked to use TokensFrontBack struct * Reworked to work with TokensFrontBack struct * Removed unnecessary scope operator * Added missing parentheses * Fixed syntax error * Removed unnecessary constructor * Default constructor now 0-initialises everything This is safer for not using a temporary TokensFrontBack object, and doesn't use delegating constructors which aren't supported yet. * Fixed unsafe null check * Added missing explicit keyword * Fixing stylistic nits Removed default constructor as it has been superseded by the single-argument constructor with a default argument value. Renamed listEnds to tokensFrontBack. Fixed if statement that was supposed to be adding safety but would actually cause a crash if tokensFrontBack was null. * Fixing stylistic nits Removed default constructor and replaced it with a single-argument constructor with a default value. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack.
2018-05-25 07:15:05 +02:00
/**
* @brief This struct stores pointers to the front and back tokens of the list this token is in.
*/
struct TokensFrontBack {
Token *front;
Token *back;
const TokenList* list;
Fix crash bug #8579 (#1238) * Added declaration for deletePrevious function * Added definition for deletePrevious function * Fixed crash from deleteThis invalidating pointers The crash was caused by deleteThis() invalidating the pointer to a constant variable usage. This happened when a usage followed an assignment. This fixes bug #8579. * Added tokensFront to match tokensBack This means deletePrevious can set the list's front if necessary. * Initialised tokensFront in appropriate places * Switched to using default Token constructor * Switched to using Token default constructor * Switched to using default constructor for Token * Added missing argument to Token constructor * Changed to use default constructor for Tokens * Switched to using default constructor for Tokens * Switched to using default constructor for Token * Added new test for deleting front Token Also made sure to use the correct constructor for Token in other tests. * Syntax error * Replaced tokensFront and tokensBack with a struct This decreases the size of the Token class for performance purposes. * Replaced tokensFront and tokensBack with a struct * Added tokensFrontBack to destructor * Reworked to use TokensBackFront struct Also ran astyle. * Reworked to use TokenList's TokensFrontBack member * Reworked to use TokensFrontBack struct * Reworked to use TokensFrontBack struct * Reworked to work with TokensFrontBack struct * Removed unnecessary scope operator * Added missing parentheses * Fixed syntax error * Removed unnecessary constructor * Default constructor now 0-initialises everything This is safer for not using a temporary TokensFrontBack object, and doesn't use delegating constructors which aren't supported yet. * Fixed unsafe null check * Added missing explicit keyword * Fixing stylistic nits Removed default constructor as it has been superseded by the single-argument constructor with a default argument value. Renamed listEnds to tokensFrontBack. Fixed if statement that was supposed to be adding safety but would actually cause a crash if tokensFrontBack was null. * Fixing stylistic nits Removed default constructor and replaced it with a single-argument constructor with a default value. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack.
2018-05-25 07:15:05 +02:00
};
struct ScopeInfo2 {
ScopeInfo2(const std::string &name_, const Token *bodyEnd_, const std::set<std::string> &usingNamespaces_ = std::set<std::string>()) : name(name_), bodyEnd(bodyEnd_), usingNamespaces(usingNamespaces_) {}
std::string name;
const Token * const bodyEnd;
std::set<std::string> usingNamespaces;
};
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
struct TokenImpl {
nonneg int mVarId;
nonneg int mFileIndex;
nonneg int mLineNumber;
nonneg int mColumn;
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
// AST..
Token *mAstOperand1;
Token *mAstOperand2;
Token *mAstParent;
// symbol database information
const Scope *mScope;
union {
const Function *mFunction;
const Variable *mVariable;
const ::Type* mType;
const Enumerator *mEnumerator;
};
/**
* A value from 0-100 that provides a rough idea about where in the token
* list this token is located.
*/
nonneg int mProgressValue;
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
/**
* Token index. Position in token list
*/
nonneg int mIndex;
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
// original name like size_t
std::string* mOriginalName;
// ValueType
ValueType *mValueType;
// ValueFlow
std::list<ValueFlow::Value>* mValues;
static const std::list<ValueFlow::Value> mEmptyValueList;
// Pointer to a template in the template simplifier
std::set<TemplateSimplifier::TokenAndName*>* mTemplateSimplifierPointers;
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
// Pointer to the object representing this token's scope
std::shared_ptr<ScopeInfo2> mScopeInfo;
// __cppcheck_in_range__
struct CppcheckAttributes {
enum Type {LOW,HIGH} type;
MathLib::bigint value;
struct CppcheckAttributes *next;
};
struct CppcheckAttributes *mCppcheckAttributes;
// For memoization, to speed up parsing of huge arrays #8897
enum class Cpp11init {UNKNOWN, CPP11INIT, NOINIT} mCpp11init;
/** Bitfield bit count. */
unsigned char mBits;
void setCppcheckAttribute(CppcheckAttributes::Type type, MathLib::bigint value);
bool getCppcheckAttribute(CppcheckAttributes::Type type, MathLib::bigint *value) const;
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
TokenImpl()
: mVarId(0)
, mFileIndex(0)
, mLineNumber(0)
, mColumn(0)
, mAstOperand1(nullptr)
, mAstOperand2(nullptr)
, mAstParent(nullptr)
, mScope(nullptr)
, mFunction(nullptr) // Initialize whole union
, mProgressValue(0)
, mIndex(0)
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
, mOriginalName(nullptr)
, mValueType(nullptr)
, mValues(nullptr)
, mTemplateSimplifierPointers(nullptr)
, mScopeInfo(nullptr)
, mCppcheckAttributes(nullptr)
, mCpp11init(Cpp11init::UNKNOWN)
, mBits(0)
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
{}
~TokenImpl();
};
/// @addtogroup Core
/// @{
2009-07-13 13:35:33 +02:00
/**
* @brief The token list that the TokenList generates is a linked-list of this class.
2009-07-13 13:35:33 +02:00
*
* Tokens are stored as strings. The "if", "while", etc are stored in plain text.
* The reason the Token class is needed (instead of using the string class) is that some extra functionality is also needed for tokens:
* - location of the token is stored (fileIndex, linenr, column)
2009-07-13 13:35:33 +02:00
* - functions for classifying the token (isName, isNumber, isBoolean, isStandardType)
*
* The Token class also has other functions for management of token list, matching tokens, etc.
*/
class CPPCHECKLIB Token {
private:
TokensFrontBack* mTokensFrontBack;
// Not implemented..
Token(const Token &);
Token operator=(const Token &);
public:
enum Type {
eVariable, eType, eFunction, eKeyword, eName, // Names: Variable (varId), Type (typeId, later), Function (FuncId, later), Language keyword, Name (unknown identifier)
eNumber, eString, eChar, eBoolean, eLiteral, eEnumerator, // Literals: Number, String, Character, Boolean, User defined literal (C++11), Enumerator
eArithmeticalOp, eComparisonOp, eAssignmentOp, eLogicalOp, eBitOp, eIncDecOp, eExtendedOp, // Operators: Arithmetical, Comparison, Assignment, Logical, Bitwise, ++/--, Extended
eBracket, // {, }, <, >: < and > only if link() is set. Otherwise they are comparison operators.
eLambda, // A function without a name
eEllipsis, // "..."
eOther,
eNone
};
Fix crash bug #8579 (#1238) * Added declaration for deletePrevious function * Added definition for deletePrevious function * Fixed crash from deleteThis invalidating pointers The crash was caused by deleteThis() invalidating the pointer to a constant variable usage. This happened when a usage followed an assignment. This fixes bug #8579. * Added tokensFront to match tokensBack This means deletePrevious can set the list's front if necessary. * Initialised tokensFront in appropriate places * Switched to using default Token constructor * Switched to using Token default constructor * Switched to using default constructor for Token * Added missing argument to Token constructor * Changed to use default constructor for Tokens * Switched to using default constructor for Tokens * Switched to using default constructor for Token * Added new test for deleting front Token Also made sure to use the correct constructor for Token in other tests. * Syntax error * Replaced tokensFront and tokensBack with a struct This decreases the size of the Token class for performance purposes. * Replaced tokensFront and tokensBack with a struct * Added tokensFrontBack to destructor * Reworked to use TokensBackFront struct Also ran astyle. * Reworked to use TokenList's TokensFrontBack member * Reworked to use TokensFrontBack struct * Reworked to use TokensFrontBack struct * Reworked to work with TokensFrontBack struct * Removed unnecessary scope operator * Added missing parentheses * Fixed syntax error * Removed unnecessary constructor * Default constructor now 0-initialises everything This is safer for not using a temporary TokensFrontBack object, and doesn't use delegating constructors which aren't supported yet. * Fixed unsafe null check * Added missing explicit keyword * Fixing stylistic nits Removed default constructor as it has been superseded by the single-argument constructor with a default argument value. Renamed listEnds to tokensFrontBack. Fixed if statement that was supposed to be adding safety but would actually cause a crash if tokensFrontBack was null. * Fixing stylistic nits Removed default constructor and replaced it with a single-argument constructor with a default value. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack.
2018-05-25 07:15:05 +02:00
explicit Token(TokensFrontBack *tokensFrontBack = nullptr);
~Token();
template<typename T>
2014-11-20 14:20:09 +01:00
void str(T&& s) {
2018-06-16 23:03:15 +02:00
mStr = s;
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mVarId = 0;
update_property_info();
}
2008-12-18 22:28:57 +01:00
2011-10-23 21:21:42 +02:00
/**
* Concatenate two (quoted) strings. Automatically cuts of the last/first character.
* Example: "hello ""world" -> "hello world". Used by the token simplifier.
2011-10-23 21:21:42 +02:00
*/
void concatStr(std::string const& b);
2014-11-20 14:20:09 +01:00
const std::string &str() const {
2018-06-16 23:03:15 +02:00
return mStr;
}
2008-12-18 22:28:57 +01:00
/**
* Unlink and delete the next 'count' tokens.
2008-12-18 22:28:57 +01:00
*/
void deleteNext(nonneg int count = 1);
2008-12-18 22:28:57 +01:00
Fix crash bug #8579 (#1238) * Added declaration for deletePrevious function * Added definition for deletePrevious function * Fixed crash from deleteThis invalidating pointers The crash was caused by deleteThis() invalidating the pointer to a constant variable usage. This happened when a usage followed an assignment. This fixes bug #8579. * Added tokensFront to match tokensBack This means deletePrevious can set the list's front if necessary. * Initialised tokensFront in appropriate places * Switched to using default Token constructor * Switched to using Token default constructor * Switched to using default constructor for Token * Added missing argument to Token constructor * Changed to use default constructor for Tokens * Switched to using default constructor for Tokens * Switched to using default constructor for Token * Added new test for deleting front Token Also made sure to use the correct constructor for Token in other tests. * Syntax error * Replaced tokensFront and tokensBack with a struct This decreases the size of the Token class for performance purposes. * Replaced tokensFront and tokensBack with a struct * Added tokensFrontBack to destructor * Reworked to use TokensBackFront struct Also ran astyle. * Reworked to use TokenList's TokensFrontBack member * Reworked to use TokensFrontBack struct * Reworked to use TokensFrontBack struct * Reworked to work with TokensFrontBack struct * Removed unnecessary scope operator * Added missing parentheses * Fixed syntax error * Removed unnecessary constructor * Default constructor now 0-initialises everything This is safer for not using a temporary TokensFrontBack object, and doesn't use delegating constructors which aren't supported yet. * Fixed unsafe null check * Added missing explicit keyword * Fixing stylistic nits Removed default constructor as it has been superseded by the single-argument constructor with a default argument value. Renamed listEnds to tokensFrontBack. Fixed if statement that was supposed to be adding safety but would actually cause a crash if tokensFrontBack was null. * Fixing stylistic nits Removed default constructor and replaced it with a single-argument constructor with a default value. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack.
2018-05-25 07:15:05 +02:00
/**
* Unlink and delete the previous 'count' tokens.
Fix crash bug #8579 (#1238) * Added declaration for deletePrevious function * Added definition for deletePrevious function * Fixed crash from deleteThis invalidating pointers The crash was caused by deleteThis() invalidating the pointer to a constant variable usage. This happened when a usage followed an assignment. This fixes bug #8579. * Added tokensFront to match tokensBack This means deletePrevious can set the list's front if necessary. * Initialised tokensFront in appropriate places * Switched to using default Token constructor * Switched to using Token default constructor * Switched to using default constructor for Token * Added missing argument to Token constructor * Changed to use default constructor for Tokens * Switched to using default constructor for Tokens * Switched to using default constructor for Token * Added new test for deleting front Token Also made sure to use the correct constructor for Token in other tests. * Syntax error * Replaced tokensFront and tokensBack with a struct This decreases the size of the Token class for performance purposes. * Replaced tokensFront and tokensBack with a struct * Added tokensFrontBack to destructor * Reworked to use TokensBackFront struct Also ran astyle. * Reworked to use TokenList's TokensFrontBack member * Reworked to use TokensFrontBack struct * Reworked to use TokensFrontBack struct * Reworked to work with TokensFrontBack struct * Removed unnecessary scope operator * Added missing parentheses * Fixed syntax error * Removed unnecessary constructor * Default constructor now 0-initialises everything This is safer for not using a temporary TokensFrontBack object, and doesn't use delegating constructors which aren't supported yet. * Fixed unsafe null check * Added missing explicit keyword * Fixing stylistic nits Removed default constructor as it has been superseded by the single-argument constructor with a default argument value. Renamed listEnds to tokensFrontBack. Fixed if statement that was supposed to be adding safety but would actually cause a crash if tokensFrontBack was null. * Fixing stylistic nits Removed default constructor and replaced it with a single-argument constructor with a default value. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack.
2018-05-25 07:15:05 +02:00
*/
void deletePrevious(nonneg int count = 1);
Fix crash bug #8579 (#1238) * Added declaration for deletePrevious function * Added definition for deletePrevious function * Fixed crash from deleteThis invalidating pointers The crash was caused by deleteThis() invalidating the pointer to a constant variable usage. This happened when a usage followed an assignment. This fixes bug #8579. * Added tokensFront to match tokensBack This means deletePrevious can set the list's front if necessary. * Initialised tokensFront in appropriate places * Switched to using default Token constructor * Switched to using Token default constructor * Switched to using default constructor for Token * Added missing argument to Token constructor * Changed to use default constructor for Tokens * Switched to using default constructor for Tokens * Switched to using default constructor for Token * Added new test for deleting front Token Also made sure to use the correct constructor for Token in other tests. * Syntax error * Replaced tokensFront and tokensBack with a struct This decreases the size of the Token class for performance purposes. * Replaced tokensFront and tokensBack with a struct * Added tokensFrontBack to destructor * Reworked to use TokensBackFront struct Also ran astyle. * Reworked to use TokenList's TokensFrontBack member * Reworked to use TokensFrontBack struct * Reworked to use TokensFrontBack struct * Reworked to work with TokensFrontBack struct * Removed unnecessary scope operator * Added missing parentheses * Fixed syntax error * Removed unnecessary constructor * Default constructor now 0-initialises everything This is safer for not using a temporary TokensFrontBack object, and doesn't use delegating constructors which aren't supported yet. * Fixed unsafe null check * Added missing explicit keyword * Fixing stylistic nits Removed default constructor as it has been superseded by the single-argument constructor with a default argument value. Renamed listEnds to tokensFrontBack. Fixed if statement that was supposed to be adding safety but would actually cause a crash if tokensFrontBack was null. * Fixing stylistic nits Removed default constructor and replaced it with a single-argument constructor with a default value. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack. * Fixing stylistic nits Renamed _listEnds to _tokensFrontBack.
2018-05-25 07:15:05 +02:00
/**
* Swap the contents of this token with the next token.
*/
void swapWithNext();
2008-12-18 22:28:57 +01:00
/**
* @return token in given index, related to this token.
2008-12-18 22:28:57 +01:00
* For example index 1 would return next token, and 2
* would return next from that one.
*/
const Token *tokAt(int index) const;
2014-11-20 14:20:09 +01:00
Token *tokAt(int index) {
return const_cast<Token *>(const_cast<const Token *>(this)->tokAt(index));
}
2008-12-18 22:28:57 +01:00
/**
* @return the link to the token in given index, related to this token.
* For example index 1 would return the link to next token.
*/
const Token *linkAt(int index) const;
2014-11-20 14:20:09 +01:00
Token *linkAt(int index) {
return const_cast<Token *>(const_cast<const Token *>(this)->linkAt(index));
}
/**
* @return String of the token in given index, related to this token.
* If that token does not exist, an empty string is being returned.
*/
const std::string &strAt(int index) const;
2008-12-18 22:28:57 +01:00
2008-12-23 22:51:54 +01:00
/**
* Match given token (or list of tokens) to a pattern list.
*
* Possible patterns
* "someRandomText" If token contains "someRandomText".
* @note Use Match() if you want to use flags in patterns
*
* The patterns can be also combined to compare to multiple tokens at once
* by separating tokens with a space, e.g.
* ") void {" will return true if first token is ')' next token
* is "void" and token after that is '{'. If even one of the tokens does
* not match its pattern, false is returned.
*
* @param tok List of tokens to be compared to the pattern
* @param pattern The pattern against which the tokens are compared,
* e.g. "const" or ") void {".
* @return true if given token matches with given pattern
* false if given token does not match with given pattern
*/
template<size_t count>
static bool simpleMatch(const Token *tok, const char (&pattern)[count]) {
return simpleMatch(tok, pattern, count-1);
}
static bool simpleMatch(const Token *tok, const char pattern[], size_t pattern_len);
2008-12-18 22:28:57 +01:00
/**
* Match given token (or list of tokens) to a pattern list.
*
* Possible patterns
2010-03-13 22:16:06 +01:00
* - "%any%" any token
2015-12-31 01:25:36 +01:00
* - "%assign%" a assignment operand
2010-03-13 22:16:06 +01:00
* - "%bool%" true or false
* - "%char%" Any token enclosed in &apos;-character.
* - "%comp%" Any token such that isComparisonOp() returns true.
* - "%cop%" Any token such that isConstOp() returns true.
2015-12-31 01:25:36 +01:00
* - "%name%" any token which is a name, variable or type e.g. "hello" or "int"
* - "%num%" Any numeric token, e.g. "23"
* - "%op%" Any token such that isOp() returns true.
* - "%or%" A bitwise-or operator '|'
* - "%oror%" A logical-or operator '||'
2015-12-31 01:25:36 +01:00
* - "%type%" Anything that can be a variable type, e.g. "int", but not "delete".
* - "%str%" Any token starting with &quot;-character (C-string).
* - "%var%" Match with token with varId > 0
* - "%varid%" Match with parameter varid
2010-03-13 22:16:06 +01:00
* - "[abc]" Any of the characters 'a' or 'b' or 'c'
* - "int|void|char" Any of the strings, int, void or char
* - "int|void|char|" Any of the strings, int, void or char or empty string
* - "!!else" No tokens or any token that is not "else".
* - "someRandomText" If token contains "someRandomText".
2008-12-18 22:28:57 +01:00
*
2014-01-01 20:46:00 +01:00
* multi-compare patterns such as "int|void|char" can contain %%or%, %%oror% and %%op%
2015-12-31 01:25:36 +01:00
* it is recommended to put such an %%cmd% as the first pattern.
* For example: "%var%|%num%|)" means yes to a variable, a number or ')'.
*
2008-12-18 22:28:57 +01:00
* The patterns can be also combined to compare to multiple tokens at once
* by separating tokens with a space, e.g.
* ") const|void {" will return true if first token is ')' next token is either
* "const" or "void" and token after that is '{'. If even one of the tokens does not
* match its pattern, false is returned.
*
* @param tok List of tokens to be compared to the pattern
2008-12-23 22:51:54 +01:00
* @param pattern The pattern against which the tokens are compared,
* e.g. "const" or ") const|volatile| {".
2014-01-01 20:46:00 +01:00
* @param varid if %%varid% is given in the pattern the Token::varId
* will be matched against this argument
2008-12-18 22:28:57 +01:00
* @return true if given token matches with given pattern
* false if given token does not match with given pattern
*/
static bool Match(const Token *tok, const char pattern[], nonneg int varid = 0);
2008-12-18 22:28:57 +01:00
/**
* @return length of C-string.
*
2014-01-01 20:46:00 +01:00
* Should be called for %%str%% tokens only.
*
* @param tok token with C-string
**/
static nonneg int getStrLength(const Token *tok);
/**
* @return array length of C-string.
*
* Should be called for %%str%% tokens only.
*
* @param tok token with C-string
**/
static nonneg int getStrArraySize(const Token *tok);
/**
* @return sizeof of C-string.
*
* Should be called for %%str%% tokens only.
*
* @param tok token with C-string
* @param settings Settings
**/
static nonneg int getStrSize(const Token *tok, const Settings *const);
2013-01-13 20:52:38 +01:00
/**
* @return char of C-string at index (possible escaped "\\n")
2013-01-13 20:52:38 +01:00
*
2014-01-01 20:46:00 +01:00
* Should be called for %%str%% tokens only.
2013-01-13 20:52:38 +01:00
*
* @param tok token with C-string
* @param index position of character
**/
2019-07-13 16:06:24 +02:00
static std::string getCharAt(const Token *tok, MathLib::bigint index);
2013-01-13 20:52:38 +01:00
const ValueType *valueType() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mValueType;
}
void setValueType(ValueType *vt);
const ValueType *argumentType() const {
const Token *top = this;
while (top && !Token::Match(top->astParent(), ",|("))
top = top->astParent();
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return top ? top->mImpl->mValueType : nullptr;
}
Token::Type tokType() const {
2018-06-16 16:40:02 +02:00
return mTokType;
2009-08-23 17:17:57 +02:00
}
void tokType(Token::Type t) {
2018-06-16 16:40:02 +02:00
mTokType = t;
2017-10-15 01:27:47 +02:00
2018-06-16 16:40:02 +02:00
const bool memoizedIsName = (mTokType == eName || mTokType == eType || mTokType == eVariable ||
mTokType == eFunction || mTokType == eKeyword || mTokType == eBoolean ||
mTokType == eEnumerator); // TODO: "true"/"false" aren't really a name...
2017-10-15 01:27:47 +02:00
setFlag(fIsName, memoizedIsName);
2018-06-16 16:40:02 +02:00
const bool memoizedIsLiteral = (mTokType == eNumber || mTokType == eString || mTokType == eChar ||
mTokType == eBoolean || mTokType == eLiteral || mTokType == eEnumerator);
2017-10-15 01:27:47 +02:00
setFlag(fIsLiteral, memoizedIsLiteral);
2011-03-08 02:04:25 +01:00
}
void isKeyword(const bool kwd) {
if (kwd)
2017-10-15 01:27:47 +02:00
tokType(eKeyword);
2018-06-16 16:40:02 +02:00
else if (mTokType == eKeyword)
2017-10-15 01:27:47 +02:00
tokType(eName);
}
2014-11-20 14:20:09 +01:00
bool isKeyword() const {
2018-06-16 16:40:02 +02:00
return mTokType == eKeyword;
}
2014-11-20 14:20:09 +01:00
bool isName() const {
2017-10-15 01:27:47 +02:00
return getFlag(fIsName);
2009-08-23 17:17:57 +02:00
}
bool isNameOnly() const {
return mFlags == fIsName && mTokType == eName;
}
bool isUpperCaseName() const;
2014-11-20 14:20:09 +01:00
bool isLiteral() const {
2017-10-15 01:27:47 +02:00
return getFlag(fIsLiteral);
}
2014-11-20 14:20:09 +01:00
bool isNumber() const {
2018-06-16 16:40:02 +02:00
return mTokType == eNumber;
2011-03-08 02:04:25 +01:00
}
bool isEnumerator() const {
2018-06-16 16:40:02 +02:00
return mTokType == eEnumerator;
}
2014-11-20 14:20:09 +01:00
bool isOp() const {
return (isConstOp() ||
isAssignmentOp() ||
2018-06-16 16:40:02 +02:00
mTokType == eIncDecOp);
}
2014-11-20 14:20:09 +01:00
bool isConstOp() const {
2011-04-09 15:54:36 +02:00
return (isArithmeticalOp() ||
2018-06-16 16:40:02 +02:00
mTokType == eLogicalOp ||
mTokType == eComparisonOp ||
mTokType == eBitOp);
}
2014-11-20 14:20:09 +01:00
bool isExtendedOp() const {
return isConstOp() ||
2018-06-16 16:40:02 +02:00
mTokType == eExtendedOp;
}
2014-11-20 14:20:09 +01:00
bool isArithmeticalOp() const {
2018-06-16 16:40:02 +02:00
return mTokType == eArithmeticalOp;
}
2014-11-20 14:20:09 +01:00
bool isComparisonOp() const {
2018-06-16 16:40:02 +02:00
return mTokType == eComparisonOp;
}
2014-11-20 14:20:09 +01:00
bool isAssignmentOp() const {
2018-06-16 16:40:02 +02:00
return mTokType == eAssignmentOp;
}
2014-11-20 14:20:09 +01:00
bool isBoolean() const {
2018-06-16 16:40:02 +02:00
return mTokType == eBoolean;
2011-03-08 02:04:25 +01:00
}
bool isBinaryOp() const {
return astOperand1() != nullptr && astOperand2() != nullptr;
}
bool isUnaryOp(const std::string &s) const {
return s == mStr && astOperand1() != nullptr && astOperand2() == nullptr;
2018-07-13 16:40:32 +02:00
}
bool isUnaryPreOp() const;
2014-11-20 14:20:09 +01:00
unsigned int flags() const {
2018-06-16 16:14:34 +02:00
return mFlags;
}
void flags(const unsigned int flags_) {
2018-06-16 16:14:34 +02:00
mFlags = flags_;
}
2014-11-20 14:20:09 +01:00
bool isUnsigned() const {
return getFlag(fIsUnsigned);
}
void isUnsigned(const bool sign) {
setFlag(fIsUnsigned, sign);
}
2014-11-20 14:20:09 +01:00
bool isSigned() const {
return getFlag(fIsSigned);
}
void isSigned(const bool sign) {
setFlag(fIsSigned, sign);
}
2014-11-20 14:20:09 +01:00
bool isPointerCompare() const {
return getFlag(fIsPointerCompare);
}
void isPointerCompare(const bool b) {
setFlag(fIsPointerCompare, b);
}
2014-11-20 14:20:09 +01:00
bool isLong() const {
return getFlag(fIsLong);
}
2014-11-20 14:20:09 +01:00
void isLong(bool size) {
setFlag(fIsLong, size);
}
2014-11-20 14:20:09 +01:00
bool isStandardType() const {
return getFlag(fIsStandardType);
}
void isStandardType(const bool b) {
setFlag(fIsStandardType, b);
}
2014-11-20 14:20:09 +01:00
bool isExpandedMacro() const {
return getFlag(fIsExpandedMacro);
}
void isExpandedMacro(const bool m) {
setFlag(fIsExpandedMacro, m);
}
2014-12-24 10:35:40 +01:00
bool isCast() const {
return getFlag(fIsCast);
}
2014-12-24 10:35:40 +01:00
void isCast(bool c) {
setFlag(fIsCast, c);
}
2014-11-20 14:20:09 +01:00
bool isAttributeConstructor() const {
return getFlag(fIsAttributeConstructor);
}
void isAttributeConstructor(const bool ac) {
setFlag(fIsAttributeConstructor, ac);
}
2014-11-20 14:20:09 +01:00
bool isAttributeDestructor() const {
return getFlag(fIsAttributeDestructor);
}
void isAttributeDestructor(const bool value) {
setFlag(fIsAttributeDestructor, value);
}
2014-11-20 14:20:09 +01:00
bool isAttributeUnused() const {
return getFlag(fIsAttributeUnused);
}
2014-11-20 14:20:09 +01:00
void isAttributeUnused(bool unused) {
setFlag(fIsAttributeUnused, unused);
}
2014-11-20 14:20:09 +01:00
bool isAttributeUsed() const {
return getFlag(fIsAttributeUsed);
}
void isAttributeUsed(const bool unused) {
setFlag(fIsAttributeUsed, unused);
}
2014-11-20 14:20:09 +01:00
bool isAttributePure() const {
return getFlag(fIsAttributePure);
}
void isAttributePure(const bool value) {
setFlag(fIsAttributePure, value);
}
2014-11-20 14:20:09 +01:00
bool isAttributeConst() const {
return getFlag(fIsAttributeConst);
}
2014-11-20 14:20:09 +01:00
void isAttributeConst(bool value) {
setFlag(fIsAttributeConst, value);
}
bool isAttributeNoreturn() const {
return getFlag(fIsAttributeNoreturn);
}
void isAttributeNoreturn(const bool value) {
setFlag(fIsAttributeNoreturn, value);
}
2014-11-20 14:20:09 +01:00
bool isAttributeNothrow() const {
return getFlag(fIsAttributeNothrow);
}
void isAttributeNothrow(const bool value) {
setFlag(fIsAttributeNothrow, value);
}
bool isAttributePacked() const {
return getFlag(fIsAttributePacked);
}
void isAttributePacked(const bool value) {
setFlag(fIsAttributePacked, value);
}
bool isAttributeNodiscard() const {
return getFlag(fIsAttributeNodiscard);
}
void isAttributeNodiscard(const bool value) {
setFlag(fIsAttributeNodiscard, value);
}
bool isMaybeUnused() const {
return getFlag(fIsMaybeUnused);
}
void isMaybeUnused(const bool value) {
setFlag(fIsMaybeUnused, value);
}
void setCppcheckAttribute(TokenImpl::CppcheckAttributes::Type type, MathLib::bigint value) {
mImpl->setCppcheckAttribute(type, value);
}
bool getCppcheckAttribute(TokenImpl::CppcheckAttributes::Type type, MathLib::bigint *value) const {
return mImpl->getCppcheckAttribute(type, value);
}
bool isControlFlowKeyword() const {
return getFlag(fIsControlFlowKeyword);
}
bool isOperatorKeyword() const {
return getFlag(fIsOperatorKeyword);
2015-07-22 13:48:30 +02:00
}
void isOperatorKeyword(const bool value) {
setFlag(fIsOperatorKeyword, value);
2015-07-22 13:48:30 +02:00
}
bool isComplex() const {
return getFlag(fIsComplex);
}
void isComplex(const bool value) {
setFlag(fIsComplex, value);
}
bool isEnumType() const {
return getFlag(fIsEnumType);
}
void isEnumType(const bool value) {
setFlag(fIsEnumType, value);
}
bool isAtAddress() const {
return getFlag(fAtAddress);
}
void isAtAddress(bool b) {
setFlag(fAtAddress, b);
}
bool isIncompleteVar() const {
return getFlag(fIncompleteVar);
}
void isIncompleteVar(bool b) {
setFlag(fIncompleteVar, b);
}
bool isConstexpr() const {
return getFlag(fConstexpr);
}
void isConstexpr(bool b) {
setFlag(fConstexpr, b);
}
bool isExternC() const {
return getFlag(fExternC);
}
void isExternC(bool b) {
setFlag(fExternC, b);
}
2018-03-12 12:49:27 +01:00
2018-05-02 20:55:11 +02:00
bool isBitfield() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mBits > 0;
2018-05-02 20:55:11 +02:00
}
unsigned char bits() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mBits;
2018-05-02 20:55:11 +02:00
}
std::set<TemplateSimplifier::TokenAndName*>* templateSimplifierPointers() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mTemplateSimplifierPointers;
}
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
void templateSimplifierPointer(TemplateSimplifier::TokenAndName* tokenAndName) {
if (!mImpl->mTemplateSimplifierPointers)
mImpl->mTemplateSimplifierPointers = new std::set<TemplateSimplifier::TokenAndName*>;
mImpl->mTemplateSimplifierPointers->insert(tokenAndName);
}
void setBits(const unsigned char b) {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mBits = b;
2018-05-02 20:55:11 +02:00
}
bool isUtf8() const {
return (((mTokType == eString) && isPrefixStringCharLiteral(mStr, '"', "u8")) ||
((mTokType == eChar) && isPrefixStringCharLiteral(mStr, '\'', "u8")));
}
bool isUtf16() const {
return (((mTokType == eString) && isPrefixStringCharLiteral(mStr, '"', "u")) ||
((mTokType == eChar) && isPrefixStringCharLiteral(mStr, '\'', "u")));
}
bool isUtf32() const {
return (((mTokType == eString) && isPrefixStringCharLiteral(mStr, '"', "U")) ||
((mTokType == eChar) && isPrefixStringCharLiteral(mStr, '\'', "U")));
}
bool isCChar() const {
return (((mTokType == eString) && isPrefixStringCharLiteral(mStr, '"', "")) ||
((mTokType == eChar) && isPrefixStringCharLiteral(mStr, '\'', "") && mStr.length() == 3));
}
bool isCMultiChar() const {
return (((mTokType == eChar) && isPrefixStringCharLiteral(mStr, '\'', "")) &&
(mStr.length() > 3));
}
2018-03-12 12:49:27 +01:00
/**
* @brief Is current token a template argument?
*
* Original code:
*
* template<class C> struct S {
* C x;
* };
* S<int> s;
*
* Resulting code:
*
* struct S<int> {
* int x ; // <- "int" is a template argument
* }
* S<int> s;
*/
bool isTemplateArg() const {
return getFlag(fIsTemplateArg);
}
void isTemplateArg(const bool value) {
setFlag(fIsTemplateArg, value);
}
template<size_t count>
static const Token *findsimplematch(const Token * const startTok, const char (&pattern)[count]) {
return findsimplematch(startTok, pattern, count-1);
}
static const Token *findsimplematch(const Token * const startTok, const char pattern[], size_t pattern_len);
template<size_t count>
static const Token *findsimplematch(const Token * const startTok, const char (&pattern)[count], const Token * const end) {
return findsimplematch(startTok, pattern, count-1, end);
}
static const Token *findsimplematch(const Token * const startTok, const char pattern[], size_t pattern_len, const Token * const end);
static const Token *findmatch(const Token * const startTok, const char pattern[], const nonneg int varId = 0);
static const Token *findmatch(const Token * const startTok, const char pattern[], const Token * const end, const nonneg int varId = 0);
template<size_t count>
static Token *findsimplematch(Token * const startTok, const char (&pattern)[count]) {
return findsimplematch(startTok, pattern, count-1);
}
static Token *findsimplematch(Token * const startTok, const char pattern[], size_t pattern_len) {
return const_cast<Token *>(findsimplematch(const_cast<const Token *>(startTok), pattern, pattern_len));
}
template<size_t count>
static Token *findsimplematch(Token * const startTok, const char (&pattern)[count], const Token * const end) {
return findsimplematch(startTok, pattern, count-1, end);
}
static Token *findsimplematch(Token * const startTok, const char pattern[], size_t pattern_len, const Token * const end) {
return const_cast<Token *>(findsimplematch(const_cast<const Token *>(startTok), pattern, pattern_len, end));
}
static Token *findmatch(Token * const startTok, const char pattern[], const nonneg int varId = 0) {
return const_cast<Token *>(findmatch(const_cast<const Token *>(startTok), pattern, varId));
}
static Token *findmatch(Token * const startTok, const char pattern[], const Token * const end, const nonneg int varId = 0) {
return const_cast<Token *>(findmatch(const_cast<const Token *>(startTok), pattern, end, varId));
}
2008-12-18 22:28:57 +01:00
/**
* Needle is build from multiple alternatives. If one of
* them is equal to haystack, return value is 1. If there
* are no matches, but one alternative to needle is empty
* string, return value is 0. If needle was not found, return
* value is -1.
*
2017-05-06 11:57:02 +02:00
* @param tok Current token (needle)
* @param haystack e.g. "one|two" or "|one|two"
2014-11-16 19:40:04 +01:00
* @param varid optional varid of token
2008-12-18 22:28:57 +01:00
* @return 1 if needle is found from the haystack
* 0 if needle was empty string
* -1 if needle was not found
*/
2019-07-14 17:31:26 +02:00
static int multiCompare(const Token *tok, const char *haystack, nonneg int varid);
2008-12-18 22:28:57 +01:00
nonneg int fileIndex() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mFileIndex;
}
void fileIndex(nonneg int indexOfFile) {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mFileIndex = indexOfFile;
}
nonneg int linenr() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mLineNumber;
2009-08-23 17:17:57 +02:00
}
void linenr(nonneg int lineNumber) {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mLineNumber = lineNumber;
2009-08-23 17:17:57 +02:00
}
2008-12-18 22:28:57 +01:00
nonneg int column() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mColumn;
2009-08-23 17:17:57 +02:00
}
void column(nonneg int c) {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mColumn = c;
2009-08-23 17:17:57 +02:00
}
2008-12-18 22:28:57 +01:00
2014-11-20 14:20:09 +01:00
Token *next() const {
2018-06-16 16:16:55 +02:00
return mNext;
2009-08-23 17:17:57 +02:00
}
2008-12-18 22:28:57 +01:00
/**
* Delete tokens between begin and end. E.g. if begin = 1
* and end = 5, tokens 2,3 and 4 would be erased.
*
* @param begin Tokens after this will be erased.
* @param end Tokens before this will be erased.
*/
static void eraseTokens(Token *begin, const Token *end);
2008-12-18 22:28:57 +01:00
/**
* Insert new token after this token. This function will handle
* relations between next and previous token also.
2010-04-09 21:40:37 +02:00
* @param tokenStr String for the new token.
* @param originalNameStr String used for Token::originalName().
* @param prepend Insert the new token before this token when it's not
* the first one on the tokens list.
2008-12-18 22:28:57 +01:00
*/
void insertToken(const std::string &tokenStr, const std::string &originalNameStr=emptyString, bool prepend=false);
2014-11-20 14:20:09 +01:00
Token *previous() const {
2018-06-16 16:16:55 +02:00
return mPrevious;
2009-08-23 17:17:57 +02:00
}
2008-12-18 22:28:57 +01:00
nonneg int varId() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mVarId;
2009-08-23 17:17:57 +02:00
}
void varId(nonneg int id) {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mVarId = id;
if (id != 0) {
2017-10-15 01:27:47 +02:00
tokType(eVariable);
isStandardType(false);
} else {
update_property_info();
}
2009-08-23 17:17:57 +02:00
}
2008-12-18 22:28:57 +01:00
/**
* For debugging purposes, prints token and all tokens
* followed by it.
* @param title Title for the printout or use default parameter or 0
* for no title.
*/
void printOut(const char *title = nullptr) const;
2008-12-18 22:28:57 +01:00
/**
* For debugging purposes, prints token and all tokens
* followed by it.
* @param title Title for the printout or use default parameter or 0
* for no title.
* @param fileNames Prints out file name instead of file index.
* File index should match the index of the string in this vector.
*/
void printOut(const char *title, const std::vector<std::string> &fileNames) const;
2020-05-23 11:30:54 +02:00
/**
* print out tokens
*/
void printLines(int lines=5) const;
/**
* Replace token replaceThis with tokens between start and end,
* including start and end. The replaceThis token is deleted.
* @param replaceThis This token will be deleted.
* @param start This will be in the place of replaceThis
* @param end This is also in the place of replaceThis
*/
static void replace(Token *replaceThis, Token *start, Token *end);
/**
* Stringify a token
* @param os The result is shifted into that output stream
* @param varid Print varids. (Style: "varname@id")
* @param attributes Print attributes of tokens like "unsigned" in front of it.
* @param macro Prints $ in front of the token if it was expanded from a macro.
*/
void stringify(std::ostream& os, bool varid, bool attributes, bool macro) const;
/**
* Stringify a list of token, from current instance on.
* @param varid Print varids. (Style: "varname@id")
* @param attributes Print attributes of tokens like "unsigned" in front of it.
* @param linenumbers Print line number in front of each line
* @param linebreaks Insert "\\n" into string when line number changes
* @param files print Files as numbers or as names (if fileNames is given)
* @param fileNames Vector of filenames. Used (if given) to print filenames as strings instead of numbers.
* @param end Stringification ends before this token is reached. 0 to stringify until end of list.
* @return Stringified token list as a string
*/
std::string stringifyList(bool varid, bool attributes, bool linenumbers, bool linebreaks, bool files, const std::vector<std::string>* fileNames = nullptr, const Token* end = nullptr) const;
std::string stringifyList(const Token* end, bool attributes = true) const;
std::string stringifyList(bool varid = false) const;
/**
2011-10-16 08:25:11 +02:00
* Remove the contents for this token from the token list.
*
* The contents are replaced with the contents of the next token and
* the next token is unlinked and deleted from the token list.
*
* So this token will still be valid after the 'deleteThis()'.
*/
void deleteThis();
/**
* Create link to given token
* @param linkToToken The token where this token should link
* to.
*/
2014-11-20 14:20:09 +01:00
void link(Token *linkToToken) {
2018-06-16 20:30:09 +02:00
mLink = linkToToken;
2018-06-16 23:03:15 +02:00
if (mStr == "<" || mStr == ">")
update_property_info();
2009-08-23 17:17:57 +02:00
}
/**
* Return token where this token links to.
* Supported links are:
* "{" <-> "}"
2011-10-16 08:25:11 +02:00
* "(" <-> ")"
* "[" <-> "]"
*
* @return The token where this token links to.
*/
2014-11-20 14:20:09 +01:00
Token *link() const {
2018-06-16 20:30:09 +02:00
return mLink;
2009-08-23 17:17:57 +02:00
}
/**
* Associate this token with given scope
* @param s Scope to be associated
*/
2014-11-20 14:20:09 +01:00
void scope(const Scope *s) {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mScope = s;
}
/**
* @return a pointer to the scope containing this token.
*/
2014-11-20 14:20:09 +01:00
const Scope *scope() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mScope;
}
/**
* Associate this token with given function
* @param f Function to be associated
*/
2019-07-05 14:00:59 +02:00
void function(const Function *f);
/**
* @return a pointer to the Function associated with this token.
*/
2014-11-20 14:20:09 +01:00
const Function *function() const {
return mTokType == eFunction || mTokType == eLambda ? mImpl->mFunction : nullptr;
}
/**
* Associate this token with given variable
* @param v Variable to be associated
*/
2014-11-20 14:20:09 +01:00
void variable(const Variable *v) {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mVariable = v;
if (v || mImpl->mVarId)
2017-10-15 01:27:47 +02:00
tokType(eVariable);
2018-06-16 16:40:02 +02:00
else if (mTokType == eVariable)
2017-10-15 01:27:47 +02:00
tokType(eName);
}
/**
* @return a pointer to the variable associated with this token.
*/
2014-11-20 14:20:09 +01:00
const Variable *variable() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mTokType == eVariable ? mImpl->mVariable : nullptr;
}
/**
* Associate this token with given type
* @param t Type to be associated
*/
void type(const ::Type *t);
/**
* @return a pointer to the type associated with this token.
*/
const ::Type *type() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mTokType == eType ? mImpl->mType : nullptr;
}
static const ::Type *typeOf(const Token *tok);
static std::pair<const Token*, const Token*> typeDecl(const Token * tok);
static std::string typeStr(const Token* tok);
/**
* @return a pointer to the Enumerator associated with this token.
*/
const Enumerator *enumerator() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mTokType == eEnumerator ? mImpl->mEnumerator : nullptr;
}
/**
* Associate this token with given enumerator
* @param e Enumerator to be associated
*/
void enumerator(const Enumerator *e) {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mEnumerator = e;
if (e)
2017-10-15 01:27:47 +02:00
tokType(eEnumerator);
2018-06-16 16:40:02 +02:00
else if (mTokType == eEnumerator)
2017-10-15 01:27:47 +02:00
tokType(eName);
}
/**
* Links two elements against each other.
**/
static void createMutualLinks(Token *begin, Token *end);
/**
* This can be called only for tokens that are strings, else
* the assert() is called. If Token is e.g. '"hello"', this will return
* 'hello' (removing the double quotes).
* @return String value
*/
std::string strValue() const;
/**
* Move srcStart and srcEnd tokens and all tokens between them
* into new a location. Only links between tokens are changed.
* @param srcStart This is the first token to be moved
* @param srcEnd The last token to be moved
* @param newLocation srcStart will be placed after this token.
*/
static void move(Token *srcStart, Token *srcEnd, Token *newLocation);
/** Get progressValue (0 - 100) */
nonneg int progressValue() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mProgressValue;
}
/** Calculate progress values for all tokens */
static void assignProgressValues(Token *tok);
2011-10-23 11:23:48 +02:00
/**
* @return the first token of the next argument. Does only work on argument
* lists. Requires that Tokenizer::createLinks2() has been called before.
* Returns 0, if there is no next argument.
2011-10-23 11:23:48 +02:00
*/
Token* nextArgument() const;
2011-10-23 11:23:48 +02:00
/**
* @return the first token of the next argument. Does only work on argument
* lists. Should be used only before Tokenizer::createLinks2() was called.
* Returns 0, if there is no next argument.
*/
Token* nextArgumentBeforeCreateLinks2() const;
/**
* @return the first token of the next template argument. Does only work on template argument
* lists. Requires that Tokenizer::createLinks2() has been called before.
* Returns 0, if there is no next argument.
*/
Token* nextTemplateArgument() const;
/**
* Returns the closing bracket of opening '<'. Should only be used if link()
* is unavailable.
2013-07-31 10:30:20 +02:00
* @return closing '>', ')', ']' or '}'. if no closing bracket is found, NULL is returned
*/
2013-07-31 10:30:20 +02:00
const Token* findClosingBracket() const;
Token* findClosingBracket();
const Token* findOpeningBracket() const;
Token* findOpeningBracket();
/**
* @return the original name.
*/
2014-11-20 14:20:09 +01:00
const std::string & originalName() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mOriginalName ? *mImpl->mOriginalName : emptyString;
}
const std::list<ValueFlow::Value>& values() const {
return mImpl->mValues ? *mImpl->mValues : TokenImpl::mEmptyValueList;
}
/**
* Sets the original name.
*/
template<typename T>
2014-11-20 14:20:09 +01:00
void originalName(T&& name) {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
if (!mImpl->mOriginalName)
mImpl->mOriginalName = new std::string(name);
else
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
*mImpl->mOriginalName = name;
}
bool hasKnownIntValue() const;
bool hasKnownValue() const;
2018-03-24 07:58:37 +01:00
MathLib::bigint getKnownIntValue() const {
return mImpl->mValues->front().intvalue;
}
bool isImpossibleIntValue(const MathLib::bigint val) const;
const ValueFlow::Value* getValue(const MathLib::bigint val) const;
const ValueFlow::Value* getMaxValue(bool condition) const;
const ValueFlow::Value* getMovedValue() const;
const ValueFlow::Value * getValueLE(const MathLib::bigint val, const Settings *settings) const;
const ValueFlow::Value * getValueGE(const MathLib::bigint val, const Settings *settings) const;
const ValueFlow::Value * getInvalidValue(const Token *ftok, nonneg int argnr, const Settings *settings) const;
const ValueFlow::Value* getContainerSizeValue(const MathLib::bigint val) const;
const Token *getValueTokenMaxStrLength() const;
const Token *getValueTokenMinStrSize(const Settings *settings) const;
const Token *getValueTokenDeadPointer() const;
/** Add token value. Return true if value is added. */
bool addValue(const ValueFlow::Value &value);
2020-02-01 08:20:08 +01:00
void removeValues(std::function<bool(const ValueFlow::Value &)> pred) {
if (mImpl->mValues)
mImpl->mValues->remove_if(pred);
}
nonneg int index() const {
return mImpl->mIndex;
}
void assignIndexes();
2008-12-18 22:28:57 +01:00
private:
2014-11-20 14:20:09 +01:00
void next(Token *nextToken) {
2018-06-16 16:16:55 +02:00
mNext = nextToken;
2009-08-23 17:17:57 +02:00
}
2014-11-20 14:20:09 +01:00
void previous(Token *previousToken) {
2018-06-16 16:16:55 +02:00
mPrevious = previousToken;
2009-08-23 17:17:57 +02:00
}
2008-12-18 22:28:57 +01:00
2017-08-25 23:30:04 +02:00
/** used by deleteThis() to take data from token to delete */
void takeData(Token *fromToken);
/**
* Works almost like strcmp() except returns only true or false and
2010-03-13 22:16:06 +01:00
* if str has empty space &apos; &apos; character, that character is handled
* as if it were &apos;\\0&apos;
*/
static bool firstWordEquals(const char *str, const char *word);
/**
* Works almost like strchr() except
2010-03-13 22:16:06 +01:00
* if str has empty space &apos; &apos; character, that character is handled
* as if it were &apos;\\0&apos;
*/
static const char *chrInFirstWord(const char *str, char c);
2018-06-16 23:03:15 +02:00
std::string mStr;
2018-06-16 16:16:55 +02:00
Token *mNext;
Token *mPrevious;
2018-06-16 20:30:09 +02:00
Token *mLink;
enum {
fIsUnsigned = (1 << 0),
fIsSigned = (1 << 1),
fIsPointerCompare = (1 << 2),
fIsLong = (1 << 3),
fIsStandardType = (1 << 4),
fIsExpandedMacro = (1 << 5),
2014-12-24 10:35:40 +01:00
fIsCast = (1 << 6),
fIsAttributeConstructor = (1 << 7), // __attribute__((constructor)) __attribute__((constructor(priority)))
fIsAttributeDestructor = (1 << 8), // __attribute__((destructor)) __attribute__((destructor(priority)))
fIsAttributeUnused = (1 << 9), // __attribute__((unused))
fIsAttributePure = (1 << 10), // __attribute__((pure))
fIsAttributeConst = (1 << 11), // __attribute__((const))
fIsAttributeNoreturn = (1 << 12), // __attribute__((noreturn)), __declspec(noreturn)
fIsAttributeNothrow = (1 << 13), // __attribute__((nothrow)), __declspec(nothrow)
2015-07-22 13:48:30 +02:00
fIsAttributeUsed = (1 << 14), // __attribute__((used))
fIsAttributePacked = (1 << 15), // __attribute__((packed))
fIsControlFlowKeyword = (1 << 16), // if/switch/while/...
fIsOperatorKeyword = (1 << 17), // operator=, etc
fIsComplex = (1 << 18), // complex/_Complex type
fIsEnumType = (1 << 19), // enumeration type
fIsName = (1 << 20),
fIsLiteral = (1 << 21),
fIsTemplateArg = (1 << 22),
fIsAttributeNodiscard = (1 << 23), // __attribute__ ((warn_unused_result)), [[nodiscard]]
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
fAtAddress = (1 << 24), // @ 0x4000
fIncompleteVar = (1 << 25),
fConstexpr = (1 << 26),
fExternC = (1 << 27),
fIsMaybeUnused = (1 << 28), // [[maybe_unsed]]
};
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
Token::Type mTokType;
unsigned int mFlags;
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
TokenImpl *mImpl;
/**
* Get specified flag state.
* @param flag_ flag to get state of
* @return true if flag set or false in flag not set
*/
2014-11-20 14:20:09 +01:00
bool getFlag(unsigned int flag_) const {
2018-06-16 16:14:34 +02:00
return ((mFlags & flag_) != 0);
}
/**
* Set specified flag state.
* @param flag_ flag to set state
* @param state_ new state of flag
*/
2014-11-20 14:20:09 +01:00
void setFlag(unsigned int flag_, bool state_) {
2018-06-16 16:14:34 +02:00
mFlags = state_ ? mFlags | flag_ : mFlags & ~flag_;
}
/** Updates internal property cache like _isName or _isBoolean.
2018-06-16 23:03:15 +02:00
Called after any mStr() modification. */
void update_property_info();
/** Update internal property cache about isStandardType() */
void update_property_isStandardType();
/** Update internal property cache about string and char literals */
void update_property_char_string_literal();
Optimize astStringVerbose() for large arrays (#1815) Change the astStringVerbose() recursion to extend a string instead of returning one. This has the benefit that for tokens where the recursion runs deep (typically large arrays), the time savings can be substantial (see comments on benchmarks further down). The reason is that previously, for each token, the astString of its operands was constructed, and then appended to this tokens astString. This led to a lot of unnecessary string copying (and with that allocations). Instead, by passing the string by reference, the number of temporary strings is greatly reduced. Another way of seeing it is that previously, the string was constructed from end to beginning, but now it is constructed from the beginning to end. There was no notable speedup by preallocating the entire string using string::reserve() (at least not on Linux). To benchmark, the changes and master were tested on Linux using the commands: make time cppcheck --debug --verbose $file >/dev/null i.e., the cppcheck binary was compiled with the settings in the Makefile. Printing the output to screen or file will of course take longer time. In Trac ticket #8355 which triggered this change, an example file from the Wine repository was attached. Running the above cppcheck on master took 24 minutes and with the changes in this commmit, took 22 seconds. Another test made was on lib/tokenlist.cpp in the cppcheck repo, which is more "normal" file. On that file there was no measurable time difference. A synthetic benchmark was generated to illustrate the effects on dumping the ast for arrays of different sizes. The generate code looked as follows: const int array[] = {...}; with different number of elements. The results are as follows (times are in seconds): N master optimized 10 0.1 0.1 100 0.1 0.1 1000 2.8 0.7 2000 19 1.8 3000 53 3.8 5000 350 10 10000 3215 38 As we can see, for small arrays, there is no time difference, but for large arrays the time savings are substantial.
2019-04-30 13:35:48 +02:00
/** Internal helper function to avoid excessive string allocations */
void astStringVerboseRecursive(std::string& ret, const nonneg int indent1 = 0, const nonneg int indent2 = 0) const;
Optimize astStringVerbose() for large arrays (#1815) Change the astStringVerbose() recursion to extend a string instead of returning one. This has the benefit that for tokens where the recursion runs deep (typically large arrays), the time savings can be substantial (see comments on benchmarks further down). The reason is that previously, for each token, the astString of its operands was constructed, and then appended to this tokens astString. This led to a lot of unnecessary string copying (and with that allocations). Instead, by passing the string by reference, the number of temporary strings is greatly reduced. Another way of seeing it is that previously, the string was constructed from end to beginning, but now it is constructed from the beginning to end. There was no notable speedup by preallocating the entire string using string::reserve() (at least not on Linux). To benchmark, the changes and master were tested on Linux using the commands: make time cppcheck --debug --verbose $file >/dev/null i.e., the cppcheck binary was compiled with the settings in the Makefile. Printing the output to screen or file will of course take longer time. In Trac ticket #8355 which triggered this change, an example file from the Wine repository was attached. Running the above cppcheck on master took 24 minutes and with the changes in this commmit, took 22 seconds. Another test made was on lib/tokenlist.cpp in the cppcheck repo, which is more "normal" file. On that file there was no measurable time difference. A synthetic benchmark was generated to illustrate the effects on dumping the ast for arrays of different sizes. The generate code looked as follows: const int array[] = {...}; with different number of elements. The results are as follows (times are in seconds): N master optimized 10 0.1 0.1 100 0.1 0.1 1000 2.8 0.7 2000 19 1.8 3000 53 3.8 5000 350 10 10000 3215 38 As we can see, for small arrays, there is no time difference, but for large arrays the time savings are substantial.
2019-04-30 13:35:48 +02:00
public:
void astOperand1(Token *tok);
void astOperand2(Token *tok);
Token * astOperand1() {
return mImpl->mAstOperand1;
}
2014-11-20 14:20:09 +01:00
const Token * astOperand1() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mAstOperand1;
}
Token * astOperand2() {
return mImpl->mAstOperand2;
}
2014-11-20 14:20:09 +01:00
const Token * astOperand2() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mAstOperand2;
}
Token * astParent() {
return mImpl->mAstParent;
}
2014-11-20 14:20:09 +01:00
const Token * astParent() const {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
return mImpl->mAstParent;
}
Token *astTop() {
Token *ret = this;
while (ret->mImpl->mAstParent)
ret = ret->mImpl->mAstParent;
return ret;
}
2014-11-20 14:20:09 +01:00
const Token *astTop() const {
const Token *ret = this;
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
while (ret->mImpl->mAstParent)
ret = ret->mImpl->mAstParent;
return ret;
}
std::pair<const Token *, const Token *> findExpressionStartEndTokens() const;
/**
* Is current token a calculation? Only true for operands.
* For '*' and '&' tokens it is looked up if this is a
* dereference or address-of. A dereference or address-of is not
* counted as a calculation.
* @return returns true if current token is a calculation
*/
bool isCalculation() const;
2014-11-20 14:20:09 +01:00
void clearAst() {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
mImpl->mAstOperand1 = mImpl->mAstOperand2 = mImpl->mAstParent = nullptr;
}
void clearValueFlow() {
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
delete mImpl->mValues;
mImpl->mValues = nullptr;
}
2014-11-20 14:20:09 +01:00
std::string astString(const char *sep = "") const {
std::string ret;
Several fairly significant optimisations (#1518) * Code changes for Token::mImpl optimisation * Added new TokenImpl optimisation Moving members to the TokenImpl struct reduces the size of the Token class, which is a fairly significant optimisation. In my testing on Windows with 32-bit Release-PCRE, this change reduced the size of the Token class from 108 bits to 52 bits and reduced run-time of my test case by around 20%. * Several optimisations Deleted some code that ran very slowly and did nothing, as there is no need to change a Token's string to null if you are about to delete it. Added a frontToken to simplifyCalculations to reduce the amount of work it has to do on already-simplified calculations. Moved template removal to the end of the list as this reduces redundant iteration and saves time. * Added tok argument to simplifyCalculations This means callers can avoid unnecessary work if they know which tokens have already been simplified. Passing nullptr indicates the original behaviour (starting from the front of the list). * Removed mention of member from another change * Re-added and optimised some code deleted in error Changing mTemplateInstantiations to a vector avoids the high cost of doing repeated linear searches. Changing how the code iterates through the array was necessary because the vector can be resized at several points during the loop, which breaks existing references and iterators. * Changed mTemplateInstantiations to a vector This is an optimisation that makes repeated linear searches of this collection significantly faster. Also added a copy constructor to TokenAndName so code can make copies of these objects to keep a reference if a vector gets resized. * A cleaner optimisation to removing template tokens This reverts the previous change to made mInstantiatedTemplates a vector and the iterator changes to support this, and makes mTypesUsedInTemplateInstantiation so the eraseTokens logic can be unified. * Reverted vector to list Also made mTypesUsedInTemplateInstantiation a vector of TokenAndName objects so it can share the same logic as the other members. * Added member for template simplifier pointer This can be used more efficiently than marking Tokens with a flag and then searching through all templates to find the one that matches. * Turned loop inside out This means we only have to iterate through the std::list once. std::list is very expensive to iterate through. * Latest code from danmar and fixed optimisations In particular I have optimised simplifying template instantiation names as this was incredibly slow because of the number of times it had to iterate through the template instantiation list. Previous optimisations to this weren't very effective and broke some edge cases. * Added changes from danmar Made mExplicitInstantiationsToDelete a vector of TokenAndName to be consistent with the rest of the members, which are cleaned up very efficiently. * Tokens can have many templateSimplifierPointers * templateSimplifierPointers must be kept in sync
2018-12-21 13:51:45 +01:00
if (mImpl->mAstOperand1)
ret = mImpl->mAstOperand1->astString(sep);
if (mImpl->mAstOperand2)
ret += mImpl->mAstOperand2->astString(sep);
2018-06-16 23:03:15 +02:00
return ret + sep + mStr;
}
Optimize astStringVerbose() for large arrays (#1815) Change the astStringVerbose() recursion to extend a string instead of returning one. This has the benefit that for tokens where the recursion runs deep (typically large arrays), the time savings can be substantial (see comments on benchmarks further down). The reason is that previously, for each token, the astString of its operands was constructed, and then appended to this tokens astString. This led to a lot of unnecessary string copying (and with that allocations). Instead, by passing the string by reference, the number of temporary strings is greatly reduced. Another way of seeing it is that previously, the string was constructed from end to beginning, but now it is constructed from the beginning to end. There was no notable speedup by preallocating the entire string using string::reserve() (at least not on Linux). To benchmark, the changes and master were tested on Linux using the commands: make time cppcheck --debug --verbose $file >/dev/null i.e., the cppcheck binary was compiled with the settings in the Makefile. Printing the output to screen or file will of course take longer time. In Trac ticket #8355 which triggered this change, an example file from the Wine repository was attached. Running the above cppcheck on master took 24 minutes and with the changes in this commmit, took 22 seconds. Another test made was on lib/tokenlist.cpp in the cppcheck repo, which is more "normal" file. On that file there was no measurable time difference. A synthetic benchmark was generated to illustrate the effects on dumping the ast for arrays of different sizes. The generate code looked as follows: const int array[] = {...}; with different number of elements. The results are as follows (times are in seconds): N master optimized 10 0.1 0.1 100 0.1 0.1 1000 2.8 0.7 2000 19 1.8 3000 53 3.8 5000 350 10 10000 3215 38 As we can see, for small arrays, there is no time difference, but for large arrays the time savings are substantial.
2019-04-30 13:35:48 +02:00
std::string astStringVerbose() const;
std::string astStringZ3() const;
std::string expressionString() const;
2014-07-14 15:51:45 +02:00
void printAst(bool verbose, bool xml, std::ostream &out) const;
2014-01-18 09:58:32 +01:00
2014-07-14 15:51:45 +02:00
void printValueFlow(bool xml, std::ostream &out) const;
void scopeInfo(std::shared_ptr<ScopeInfo2> newScopeInfo);
std::shared_ptr<ScopeInfo2> scopeInfo() const;
2019-09-03 06:43:08 +02:00
void setCpp11init(bool cpp11init) const {
mImpl->mCpp11init=cpp11init ? TokenImpl::Cpp11init::CPP11INIT : TokenImpl::Cpp11init::NOINIT;
}
TokenImpl::Cpp11init isCpp11init() const {
return mImpl->mCpp11init;
}
2008-12-18 22:28:57 +01:00
};
/// @}
//---------------------------------------------------------------------------
#endif // tokenH