#include "comparison.h" #include #include "shared/globalFunctions.h" #include #include #include #include #include #include "algorithm.h" #include "ui/util.h" #include #include #include "shared/stringConv.h" #include "library/statusHandler.h" #include "shared/fileHandling.h" #include "shared/systemFunctions.h" #include "shared/fileTraverser.h" #include "library/filter.h" #include #include "fileHierarchy.h" #include #include using namespace FreeFileSync; std::vector FreeFileSync::extractCompareCfg(const MainConfiguration& mainCfg) { //merge first and additional pairs std::vector allPairs; allPairs.push_back(mainCfg.firstPair); allPairs.insert(allPairs.end(), mainCfg.additionalPairs.begin(), //add additional pairs mainCfg.additionalPairs.end()); const FilterProcess::FilterRef globalFilter(new NameFilter(mainCfg.includeFilter, mainCfg.excludeFilter)); std::vector output; for (std::vector::const_iterator i = allPairs.begin(); i != allPairs.end(); ++i) output.push_back( FolderPairCfg(i->leftDirectory, i->rightDirectory, mainCfg.filterIsActive ? combineFilters(globalFilter, FilterProcess::FilterRef( new NameFilter( i->localFilter.includeFilter, i->localFilter.excludeFilter))) : FilterProcess::FilterRef(new NullFilter), i->altSyncConfig.get() ? i->altSyncConfig->syncConfiguration : mainCfg.syncConfiguration)); return output; } class BaseDirCallback; class DirCallback : public FreeFileSync::TraverseCallback { public: DirCallback(BaseDirCallback* baseCallback, const Zstring& relNameParentPf, //postfixed with FILE_NAME_SEPARATOR! DirContainer& output, StatusHandler* handler) : baseCallback_(baseCallback), relNameParentPf_(relNameParentPf), output_(output), statusHandler(handler) {} virtual ~DirCallback() {} virtual ReturnValue onFile(const DefaultChar* shortName, const Zstring& fullName, const FileInfo& details); virtual ReturnValDir onDir(const DefaultChar* shortName, const Zstring& fullName); virtual ReturnValue onError(const wxString& errorText); private: BaseDirCallback* const baseCallback_; const Zstring relNameParentPf_; DirContainer& output_; StatusHandler* const statusHandler; }; class BaseDirCallback : public DirCallback { friend class DirCallback; public: BaseDirCallback(DirContainer& output, const FilterProcess::FilterRef& filter, StatusHandler* handler) : DirCallback(this, Zstring(), output, handler), textScanning(wxToZ(wxString(_("Scanning:")) + wxT(" \n"))), filterInstance(filter) {} virtual TraverseCallback::ReturnValue onFile(const DefaultChar* shortName, const Zstring& fullName, const TraverseCallback::FileInfo& details); private: typedef boost::shared_ptr CallbackPointer; const Zstring textScanning; std::vector callBackBox; //collection of callback pointers to handle ownership const FilterProcess::FilterRef filterInstance; //always bound! }; TraverseCallback::ReturnValue DirCallback::onFile(const DefaultChar* shortName, const Zstring& fullName, const FileInfo& details) { //assemble status message (performance optimized) = textScanning + wxT("\"") + fullName + wxT("\"") Zstring statusText = baseCallback_->textScanning; statusText.reserve(statusText.length() + fullName.length() + 2); statusText += DefaultChar('\"'); statusText += fullName; statusText += DefaultChar('\"'); //update UI/commandline status information statusHandler->updateStatusText(statusText); //------------------------------------------------------------------------------------ //apply filter before processing (use relative name!) if (!baseCallback_->filterInstance->passFileFilter(relNameParentPf_ + shortName)) { statusHandler->requestUiRefresh(); return TRAVERSING_CONTINUE; } output_.addSubFile(shortName, FileDescriptor(details.lastWriteTimeRaw, details.fileSize)); //add 1 element to the progress indicator statusHandler->updateProcessedData(1, 0); //NO performance issue at all //trigger display refresh statusHandler->requestUiRefresh(); return TRAVERSING_CONTINUE; } TraverseCallback::ReturnValDir DirCallback::onDir(const DefaultChar* shortName, const Zstring& fullName) { using globalFunctions::FILE_NAME_SEPARATOR; //assemble status message (performance optimized) = textScanning + wxT("\"") + fullName + wxT("\"") Zstring statusText = baseCallback_->textScanning; statusText.reserve(statusText.length() + fullName.length() + 2); statusText += DefaultChar('\"'); statusText += fullName; statusText += DefaultChar('\"'); //update UI/commandline status information statusHandler->updateStatusText(statusText); //------------------------------------------------------------------------------------ Zstring relName = relNameParentPf_; relName += shortName; //apply filter before processing (use relative name!) bool subObjMightMatch = true; if (!baseCallback_->filterInstance->passDirFilter(relName, &subObjMightMatch)) { statusHandler->requestUiRefresh(); if (subObjMightMatch) { DirContainer& subDir = output_.addSubDir(shortName); DirCallback* subDirCallback = new DirCallback(baseCallback_, relName += FILE_NAME_SEPARATOR, subDir, statusHandler); baseCallback_->callBackBox.push_back(BaseDirCallback::CallbackPointer(subDirCallback)); //handle ownership //attention: ensure directory filtering is applied later to exclude actually filtered directories return ReturnValDir(ReturnValDir::Continue(), subDirCallback); } else return ReturnValDir::Ignore(); //do NOT traverse subdirs } DirContainer& subDir = output_.addSubDir(shortName); //add 1 element to the progress indicator statusHandler->updateProcessedData(1, 0); //NO performance issue at all //trigger display refresh statusHandler->requestUiRefresh(); DirCallback* subDirCallback = new DirCallback(baseCallback_, relName += FILE_NAME_SEPARATOR, subDir, statusHandler); baseCallback_->callBackBox.push_back(BaseDirCallback::CallbackPointer(subDirCallback)); //handle ownership return ReturnValDir(ReturnValDir::Continue(), subDirCallback); } TraverseCallback::ReturnValue DirCallback::onError(const wxString& errorText) { while (true) { switch (statusHandler->reportError(errorText)) { case ErrorHandler::IGNORE_ERROR: return TRAVERSING_CONTINUE; case ErrorHandler::RETRY: break; //I have to admit "retry" is a bit of a fake here... at least the user has opportunity to abort! } } return TRAVERSING_CONTINUE; //dummy value } TraverseCallback::ReturnValue BaseDirCallback::onFile( const DefaultChar* shortName, const Zstring& fullName, const TraverseCallback::FileInfo& details) { //do not scan the database file #ifdef FFS_WIN if (getSyncDBFilename().CmpNoCase(shortName) == 0) #elif defined FFS_LINUX if (getSyncDBFilename().Cmp(shortName) == 0) #endif return TraverseCallback::TRAVERSING_CONTINUE; return DirCallback::onFile(shortName, fullName, details); } //------------------------------------------------------------------------------------------ struct DirBufferKey { DirBufferKey(const Zstring& dirname, const FilterProcess::FilterRef& filterIn) : //filter interface: always bound by design! directoryName(dirname), filter(filterIn->isNull() ? //some optimization of "Null" filter FilterProcess::FilterRef(new NullFilter) : filterIn) {} const Zstring directoryName; const FilterProcess::FilterRef filter; //buffering has to consider filtering! bool operator < (const DirBufferKey& b) const { #ifdef FFS_WIN //Windows does NOT distinguish between upper/lower-case const int rv = directoryName.CmpNoCase(b.directoryName); #elif defined FFS_LINUX //Linux DOES distinguish between upper/lower-case const int rv = directoryName.Cmp(b.directoryName); #endif if (rv != 0) return rv < 0; return *filter < *b.filter; } }; //------------------------------------------------------------------------------------------ class CompareProcess::DirectoryBuffer //buffer multiple scans of the same directories { public: DirectoryBuffer(const bool traverseDirectorySymlinks, StatusHandler* statusUpdater) : m_traverseDirectorySymlinks(traverseDirectorySymlinks), m_statusUpdater(statusUpdater) {} const DirContainer& getDirectoryDescription(const Zstring& directoryPostfixed, const FilterProcess::FilterRef& filter); private: typedef boost::shared_ptr DirBufferValue; //exception safety: avoid memory leak typedef std::map BufferType; DirContainer& insertIntoBuffer(const DirBufferKey& newKey); BufferType buffer; const bool m_traverseDirectorySymlinks; StatusHandler* m_statusUpdater; }; //------------------------------------------------------------------------------------------ DirContainer& CompareProcess::DirectoryBuffer::insertIntoBuffer(const DirBufferKey& newKey) { DirBufferValue baseContainer(new DirContainer); buffer.insert(std::make_pair(newKey, baseContainer)); if (FreeFileSync::dirExists(newKey.directoryName.c_str())) //folder existence already checked in startCompareProcess(): do not treat as error when arriving here! { std::auto_ptr traverser(new BaseDirCallback(*baseContainer, newKey.filter, m_statusUpdater)); //get all files and folders from directoryPostfixed (and subdirectories) traverseFolder(newKey.directoryName, m_traverseDirectorySymlinks, traverser.get()); //exceptions may be thrown! } return *baseContainer.get(); } const DirContainer& CompareProcess::DirectoryBuffer::getDirectoryDescription( const Zstring& directoryPostfixed, const FilterProcess::FilterRef& filter) { const DirBufferKey searchKey(directoryPostfixed, filter); BufferType::const_iterator entryFound = buffer.find(searchKey); if (entryFound != buffer.end()) return *entryFound->second.get(); //entry found in buffer; return else return insertIntoBuffer(searchKey); //entry not found; create new one } //------------------------------------------------------------------------------------------ void foldersAreValidForComparison(const std::vector& folderPairsForm, StatusHandler* statusUpdater) { bool checkEmptyDirnameActive = true; //check for empty dirs just once const wxString additionalInfo = _("You can ignore the error to consider not existing directories as empty."); for (std::vector::const_iterator i = folderPairsForm.begin(); i != folderPairsForm.end(); ++i) { //check if folder name is empty if (i->leftDirectory.empty() || i->rightDirectory.empty()) { if (checkEmptyDirnameActive) { checkEmptyDirnameActive = false; while (true) { const ErrorHandler::Response rv = statusUpdater->reportError(wxString(_("Please fill all empty directory fields.")) + wxT(" \n\n") + + wxT("(") + additionalInfo + wxT(")")); if (rv == ErrorHandler::IGNORE_ERROR) break; else if (rv == ErrorHandler::RETRY) ; //continue with loop else throw std::logic_error("Programming Error: Unknown return value!"); } } } //check if folders exist if (!i->leftDirectory.empty()) while (!FreeFileSync::dirExists(i->leftDirectory.c_str())) { ErrorHandler::Response rv = statusUpdater->reportError(wxString(_("Directory does not exist:")) + wxT(" \n") + wxT("\"") + zToWx(i->leftDirectory) + wxT("\"") + wxT("\n\n") + FreeFileSync::getLastErrorFormatted() + wxT(" ") + additionalInfo); if (rv == ErrorHandler::IGNORE_ERROR) break; else if (rv == ErrorHandler::RETRY) ; //continue with loop else throw std::logic_error("Programming Error: Unknown return value!"); } if (!i->rightDirectory.empty()) while (!FreeFileSync::dirExists(i->rightDirectory.c_str())) { ErrorHandler::Response rv = statusUpdater->reportError(wxString(_("Directory does not exist:")) + wxT("\n") + wxT("\"") + zToWx(i->rightDirectory) + wxT("\"") + wxT("\n\n") + FreeFileSync::getLastErrorFormatted() + wxT(" ") + additionalInfo); if (rv == ErrorHandler::IGNORE_ERROR) break; else if (rv == ErrorHandler::RETRY) ; //continue with loop else throw std::logic_error("Programming Error: Unknown return value!"); } } } bool dependencyExists(const std::set& folders, const Zstring& newFolder, wxString& warningMessage) { for (std::set::const_iterator i = folders.begin(); i != folders.end(); ++i) { Zstring newFolderFmt = newFolder; Zstring refFolderFmt = *i; #ifdef FFS_WIN //Windows does NOT distinguish between upper/lower-case newFolderFmt.MakeLower(); refFolderFmt.MakeLower(); #elif defined FFS_LINUX //Linux DOES distinguish between upper/lower-case //nothing to do here #endif if (newFolderFmt.StartsWith(refFolderFmt) || refFolderFmt.StartsWith(newFolderFmt)) { warningMessage = wxString(_("Directories are dependent! Be careful when setting up synchronization rules:")) + wxT("\n") + wxT("\"") + zToWx(*i) + wxT("\"\n") + wxT("\"") + zToWx(newFolder) + wxT("\""); return true; } } return false; } bool foldersHaveDependencies(const std::vector& folderPairsFrom, wxString& warningMessage) { warningMessage.Clear(); std::set folders; for (std::vector::const_iterator i = folderPairsFrom.begin(); i != folderPairsFrom.end(); ++i) { if (!i->leftDirectory.empty()) //empty folders names might be accepted by user { if (dependencyExists(folders, i->leftDirectory, warningMessage)) return true; folders.insert(i->leftDirectory); } if (!i->rightDirectory.empty()) //empty folders names might be accepted by user { if (dependencyExists(folders, i->rightDirectory, warningMessage)) return true; folders.insert(i->rightDirectory); } } return false; } CompareProcess::CompareProcess(const bool traverseSymLinks, const unsigned int fileTimeTol, const bool ignoreOneHourDiff, xmlAccess::OptionalDialogs& warnings, StatusHandler* handler) : fileTimeTolerance(fileTimeTol), ignoreOneHourDifference(ignoreOneHourDiff), m_warnings(warnings), statusUpdater(handler), txtComparingContentOfFiles(wxToZ(_("Comparing content of files %x")).Replace(DefaultStr("%x"), DefaultStr("\n\"%x\""), false)) { directoryBuffer.reset(new DirectoryBuffer(traverseSymLinks, handler)); } //callback functionality for status updates while comparing class CompareCallback { public: virtual ~CompareCallback() {} virtual void updateCompareStatus(const wxLongLong& totalBytesTransferred) = 0; }; bool filesHaveSameContent(const Zstring& filename1, const Zstring& filename2, CompareCallback* callback) { const unsigned int BUFFER_SIZE = 512 * 1024; //512 kb seems to be the perfect buffer size static boost::scoped_array memory1(new unsigned char[BUFFER_SIZE]); static boost::scoped_array memory2(new unsigned char[BUFFER_SIZE]); #ifdef FFS_WIN wxFFile file1(filename1.c_str(), DefaultStr("rb")); #elif defined FFS_LINUX wxFFile file1(::fopen(filename1.c_str(), DefaultStr("rb"))); //utilize UTF-8 filename #endif if (!file1.IsOpened()) throw FileError(wxString(_("Error opening file:")) + wxT(" \"") + zToWx(filename1) + wxT("\"")); #ifdef FFS_WIN wxFFile file2(filename2.c_str(), DefaultStr("rb")); #elif defined FFS_LINUX wxFFile file2(::fopen(filename2.c_str(), DefaultStr("rb"))); //utilize UTF-8 filename #endif if (!file2.IsOpened()) //NO cleanup necessary for (wxFFile) file1 throw FileError(wxString(_("Error opening file:")) + wxT(" \"") + zToWx(filename2) + wxT("\"")); wxLongLong bytesCompared; do { const size_t length1 = file1.Read(memory1.get(), BUFFER_SIZE); if (file1.Error()) throw FileError(wxString(_("Error reading file:")) + wxT(" \"") + zToWx(filename1) + wxT("\"")); const size_t length2 = file2.Read(memory2.get(), BUFFER_SIZE); if (file2.Error()) throw FileError(wxString(_("Error reading file:")) + wxT(" \"") + zToWx(filename2) + wxT("\"")); if (length1 != length2 || ::memcmp(memory1.get(), memory2.get(), length1) != 0) return false; bytesCompared += length1 * 2; //send progress updates callback->updateCompareStatus(bytesCompared); } while (!file1.Eof()); if (!file2.Eof()) return false; return true; } //callback implementation class CmpCallbackImpl : public CompareCallback { public: CmpCallbackImpl(StatusHandler* handler, wxLongLong& bytesComparedLast) : m_handler(handler), m_bytesComparedLast(bytesComparedLast) {} virtual void updateCompareStatus(const wxLongLong& totalBytesTransferred) { //called every 512 kB //inform about the (differential) processed amount of data m_handler->updateProcessedData(0, totalBytesTransferred - m_bytesComparedLast); m_bytesComparedLast = totalBytesTransferred; m_handler->requestUiRefresh(); //exceptions may be thrown here! } private: StatusHandler* m_handler; wxLongLong& m_bytesComparedLast; }; bool filesHaveSameContentUpdating(const Zstring& filename1, const Zstring& filename2, const wxULongLong& totalBytesToCmp, StatusHandler* handler) { wxLongLong bytesComparedLast; //amount of bytes that have been compared and communicated to status handler CmpCallbackImpl callback(handler, bytesComparedLast); bool sameContent = true; try { sameContent = filesHaveSameContent(filename1, filename2, &callback); } catch (...) { //error situation: undo communication of processed amount of data handler->updateProcessedData(0, bytesComparedLast * -1); throw; } //inform about the (remaining) processed amount of data handler->updateProcessedData(0, globalFunctions::convertToSigned(totalBytesToCmp) - bytesComparedLast); return sameContent; } struct ToBeRemoved { bool operator()(const DirMapping& dirObj) const { return !dirObj.isActive() && dirObj.subDirs.size() == 0 && dirObj.subFiles.size() == 0; } }; class RemoveFilteredDirs { public: RemoveFilteredDirs(const FilterProcess& filterProc) : filterProc_(filterProc) {} void execute(HierarchyObject& hierObj) { //process subdirs recursively std::for_each(hierObj.subDirs.begin(), hierObj.subDirs.end(), *this); //remove superfluous directories hierObj.subDirs.erase(std::remove_if(hierObj.subDirs.begin(), hierObj.subDirs.end(), ::ToBeRemoved()), hierObj.subDirs.end()); } private: template friend Function std::for_each(Iterator, Iterator, Function); void operator()(DirMapping& dirObj) { dirObj.setActive(filterProc_.passDirFilter(dirObj.getObjRelativeName().c_str(), NULL)); //subObjMightMatch is always true in this context! execute(dirObj); } const FilterProcess& filterProc_; }; inline void formatPair(FolderPairCfg& input) { //ensure they end with globalFunctions::FILE_NAME_SEPARATOR and replace macros input.leftDirectory = FreeFileSync::getFormattedDirectoryName(input.leftDirectory); input.rightDirectory = FreeFileSync::getFormattedDirectoryName(input.rightDirectory); } //############################################################################################################################# void CompareProcess::startCompareProcess(const std::vector& directoryPairs, const CompareVariant cmpVar, FolderComparison& output) { #ifndef __WXDEBUG__ wxLogNull noWxLogs; //hide wxWidgets log messages in release build #endif //PERF_START; //init process: keep at beginning so that all gui elements are initialized properly statusUpdater->initNewProcess(-1, 0, StatusHandler::PROCESS_SCANNING); //it's not known how many files will be scanned => -1 objects //format directory pairs: ensure they end with globalFunctions::FILE_NAME_SEPARATOR and replace macros! std::vector directoryPairsFormatted = directoryPairs; std::for_each(directoryPairsFormatted.begin(), directoryPairsFormatted.end(), formatPair); //-------------------some basic checks:------------------------------------------ //check if folders are valid foldersAreValidForComparison(directoryPairsFormatted, statusUpdater); //check if folders have dependencies { wxString warningMessage; if (foldersHaveDependencies(directoryPairsFormatted, warningMessage)) statusUpdater->reportWarning(warningMessage.c_str(), m_warnings.warningDependentFolders); } //-------------------end of basic checks------------------------------------------ try { FolderComparison output_tmp; //write to output not before END of process! switch (cmpVar) { case CMP_BY_TIME_SIZE: compareByTimeSize(directoryPairsFormatted, output_tmp); break; case CMP_BY_CONTENT: compareByContent(directoryPairsFormatted, output_tmp); break; } assert (output_tmp.size() == directoryPairsFormatted.size()); for (FolderComparison::iterator j = output_tmp.begin(); j != output_tmp.end(); ++j) { const FolderPairCfg& fpCfg = directoryPairsFormatted[j - output_tmp.begin()]; //attention: some filtered directories are still in the comparison result! (see include filter handling!) if (!fpCfg.filter->isNull()) //let's filter them now... (and remove those that contain excluded elements only) { //filters and removes all excluded directories (but keeps those serving as parent folders) RemoveFilteredDirs(*fpCfg.filter).execute(*j); } //set initial sync-direction class RedetermineCallback : public DeterminationProblem { public: RedetermineCallback(bool& warningSyncDatabase, StatusHandler& statusUpdater) : warningSyncDatabase_(warningSyncDatabase), statusUpdater_(statusUpdater) {} virtual void reportWarning(const wxString& text) { statusUpdater_.reportWarning(text, warningSyncDatabase_); } private: bool& warningSyncDatabase_; StatusHandler& statusUpdater_; } redetCallback(m_warnings.warningSyncDatabase, *statusUpdater); FreeFileSync::redetermineSyncDirection(fpCfg.syncConfiguration, *j, &redetCallback); } //only if everything was processed correctly output is written to! //note: output mustn't change during this process to be in sync with GUI grid view!!! output_tmp.swap(output); } catch (const std::exception& e) { if (dynamic_cast(&e) != NULL) statusUpdater->reportFatalError(wxString(_("System out of memory!")) + wxT(" ") + wxString::FromAscii(e.what())); else statusUpdater->reportFatalError(wxString::FromAscii(e.what())); return; //should be obsolete! } } //--------------------assemble conflict descriptions--------------------------- //check for very old dates or dates in the future wxString getConflictInvalidDate(const Zstring& fileNameFull, const wxLongLong& utcTime) { wxString msg = _("File %x has an invalid date!"); msg.Replace(wxT("%x"), wxString(wxT("\"")) + zToWx(fileNameFull) + wxT("\"")); msg += wxString(wxT("\n\n")) + _("Date") + wxT(": ") + utcTimeToLocalString(utcTime, fileNameFull); return wxString(_("Conflict detected:")) + wxT("\n") + msg; } //check for changed files with same modification date wxString getConflictSameDateDiffSize(const FileMapping& fileObj) { //some beautification... wxString left = wxString(_("Left")) + wxT(": "); wxString right = wxString(_("Right")) + wxT(": "); const size_t maxPref = std::max(left.length(), right.length()); left.Pad(maxPref - left.length(), wxT(' '), true); right.Pad(maxPref - right.length(), wxT(' '), true); wxString msg = _("Files %x have the same date but a different size!"); msg.Replace(wxT("%x"), wxString(wxT("\"")) + zToWx(fileObj.getRelativeName()) + wxT("\"")); msg += wxT("\n\n"); msg += left + wxT("\t") + _("Date") + wxT(": ") + utcTimeToLocalString(fileObj.getLastWriteTime(), fileObj.getFullName()) + wxT(" \t") + _("Size") + wxT(": ") + fileObj.getFileSize().ToString() + wxT("\n"); msg += right + wxT("\t") + _("Date") + wxT(": ") + utcTimeToLocalString(fileObj.getLastWriteTime(), fileObj.getFullName()) + wxT(" \t") + _("Size") + wxT(": ") + fileObj.getFileSize().ToString(); return wxString(_("Conflict detected:")) + wxT("\n") + msg; } //check for files that have a difference in file modification date below 1 hour when DST check is active wxString getConflictChangeWithinHour(const FileMapping& fileObj) { //some beautification... wxString left = wxString(_("Left")) + wxT(": "); wxString right = wxString(_("Right")) + wxT(": "); const size_t maxPref = std::max(left.length(), right.length()); left.Pad(maxPref - left.length(), wxT(' '), true); right.Pad(maxPref - right.length(), wxT(' '), true); wxString msg = _("Files %x have a file time difference of less than 1 hour!\n\nIt's not safe to decide which one is newer due to Daylight Saving Time issues."); msg += wxString(wxT("\n")) + _("(Note that only FAT/FAT32 drives are affected by this problem!\nIn all other cases you can disable the setting \"ignore 1-hour difference\".)"); msg.Replace(wxT("%x"), wxString(wxT("\"")) + zToWx(fileObj.getRelativeName()) + wxT("\"")); msg += wxT("\n\n"); msg += left + wxT("\t") + _("Date") + wxT(": ") + utcTimeToLocalString(fileObj.getLastWriteTime(), fileObj.getFullName()) + wxT("\n"); msg += right + wxT("\t") + _("Date") + wxT(": ") + utcTimeToLocalString(fileObj.getLastWriteTime(), fileObj.getFullName()); return wxString(_("Conflict detected:")) + wxT("\n") + msg; } //----------------------------------------------------------------------------- inline bool sameFileTime(const wxLongLong& a, const wxLongLong& b, const unsigned int tolerance) { if (a < b) return b - a <= tolerance; else return a - b <= tolerance; } void CompareProcess::compareByTimeSize(const std::vector& directoryPairsFormatted, FolderComparison& output) { output.reserve(output.size() + directoryPairsFormatted.size()); //process one folder pair after each other for (std::vector::const_iterator pair = directoryPairsFormatted.begin(); pair != directoryPairsFormatted.end(); ++pair) { BaseDirMapping newEntry(pair->leftDirectory, pair->rightDirectory, pair->filter); output.push_back(newEntry); //attention: push_back() copies by value!!! performance: append BEFORE writing values into fileCmp! //do basis scan and retrieve files existing on both sides as "compareCandidates" std::vector compareCandidates; performBaseComparison(output.back(), compareCandidates); //PERF_START; //categorize files that exist on both sides for (std::vector::iterator i = compareCandidates.begin(); i != compareCandidates.end(); ++i) { FileMapping* const line = *i; if (line->getLastWriteTime() != line->getLastWriteTime()) { //number of seconds since Jan 1st 1970 + 1 year (needn't be too precise) static const long oneYearFromNow = wxGetUTCTime() + 365 * 24 * 3600; //check for erroneous dates (but only if dates are not (EXACTLY) the same) if ( line->getLastWriteTime() < 0 || //earlier than Jan 1st 1970 line->getLastWriteTime() < 0 || //earlier than Jan 1st 1970 line->getLastWriteTime() > oneYearFromNow || //dated more than one year in future line->getLastWriteTime() > oneYearFromNow) //dated more than one year in future { if (line->getLastWriteTime() < 0 || line->getLastWriteTime() > oneYearFromNow) line->setCategoryConflict(getConflictInvalidDate(line->getFullName(), line->getLastWriteTime())); else line->setCategoryConflict(getConflictInvalidDate(line->getFullName(), line->getLastWriteTime())); } else //from this block on all dates are at least "valid" { //last write time may differ by up to 2 seconds (NTFS vs FAT32) if (sameFileTime(line->getLastWriteTime(), line->getLastWriteTime(), fileTimeTolerance)) { if (line->getFileSize() == line->getFileSize()) line->setCategory(); else line->setCategoryConflict(getConflictSameDateDiffSize(*line)); //same date, different filesize } else { //finally: DST +/- 1-hour check: test if time diff is exactly +/- 1-hour (respecting 2 second FAT precision) if (ignoreOneHourDifference && sameFileTime(line->getLastWriteTime(), line->getLastWriteTime(), 3600 + 2)) { //date diff < 1 hour is a conflict: it's not safe to determine which file is newer if (sameFileTime(line->getLastWriteTime(), line->getLastWriteTime(), 3600 - 2 - 1)) line->setCategoryConflict(getConflictChangeWithinHour(*line)); else //exact +/- 1-hour detected: treat as equal { if (line->getFileSize() == line->getFileSize()) line->setCategory(); else line->setCategoryConflict(getConflictSameDateDiffSize(*line)); //same date, different filesize } } else { if (line->getLastWriteTime() < line->getLastWriteTime()) line->setCategory(); else line->setCategory(); } } } } else //same write time { if (line->getFileSize() == line->getFileSize()) line->setCategory(); else line->setCategoryConflict(getConflictSameDateDiffSize(*line)); //same date, different filesize } } } } wxULongLong getBytesToCompare(const std::vector& rowsToCompare) { wxULongLong dataTotal; for (std::vector::const_iterator j = rowsToCompare.begin(); j != rowsToCompare.end(); ++j) dataTotal += (*j)->getFileSize(); //left and right filesizes should be the same return dataTotal * 2; } void CompareProcess::compareByContent(const std::vector& directoryPairsFormatted, FolderComparison& output) { //PERF_START; std::vector compareCandidates; //attention: make sure pointers in "compareCandidates" remain valid!!! output.reserve(output.size() + directoryPairsFormatted.size()); //process one folder pair after each other for (std::vector::const_iterator pair = directoryPairsFormatted.begin(); pair != directoryPairsFormatted.end(); ++pair) { BaseDirMapping newEntry(pair->leftDirectory, pair->rightDirectory, pair->filter); output.push_back(newEntry); //attention: push_back() copies by value!!! performance: append BEFORE writing values into fileCmp! //do basis scan and retrieve candidates for binary comparison (files existing on both sides) performBaseComparison(output.back(), compareCandidates); } //finish categorization... std::vector filesToCompareBytewise; //content comparison of file content happens AFTER finding corresponding files //in order to separate into two processes (scanning and comparing) for (std::vector::iterator i = compareCandidates.begin(); i != compareCandidates.end(); ++i) { //pre-check: files have different content if they have a different filesize if ((*i)->getFileSize() != (*i)->getFileSize()) (*i)->setCategory(); else filesToCompareBytewise.push_back(*i); } const size_t objectsTotal = filesToCompareBytewise.size() * 2; const wxULongLong bytesTotal = getBytesToCompare(filesToCompareBytewise); statusUpdater->initNewProcess(objectsTotal, globalFunctions::convertToSigned(bytesTotal), StatusHandler::PROCESS_COMPARING_CONTENT); //compare files (that have same size) bytewise... for (std::vector::const_iterator j = filesToCompareBytewise.begin(); j != filesToCompareBytewise.end(); ++j) { FileMapping* const gridline = *j; Zstring statusText = txtComparingContentOfFiles; statusText.Replace(DefaultStr("%x"), gridline->getRelativeName(), false); statusUpdater->updateStatusText(statusText); //check files that exist in left and right model but have different content while (true) { //trigger display refresh statusUpdater->requestUiRefresh(); try { if (filesHaveSameContentUpdating(gridline->getFullName(), gridline->getFullName(), gridline->getFileSize() * 2, statusUpdater)) gridline->setCategory(); else gridline->setCategory(); statusUpdater->updateProcessedData(2, 0); //processed data is communicated in subfunctions! break; } catch (FileError& error) { ErrorHandler::Response rv = statusUpdater->reportError(error.show()); if (rv == ErrorHandler::IGNORE_ERROR) { gridline->setCategoryConflict(wxString(_("Conflict detected:")) + wxT("\n") + _("Comparing files by content failed.")); break; } else if (rv == ErrorHandler::RETRY) ; //continue with loop else throw std::logic_error("Programming Error: Unknown return value!"); } } } } class MergeSides { public: MergeSides(std::vector& appendUndefinedOut) : appendUndefined(appendUndefinedOut) {} void execute(const DirContainer& leftSide, const DirContainer& rightSide, HierarchyObject& output) { //ATTENTION: HierarchyObject::retrieveById() can only work correctly if the following conditions are fulfilled: //1. on each level, files are added first, then directories (=> file id < dir id) //2. when a directory is added, all subdirectories must be added immediately (recursion) before the next dir on this level is added //3. entries may be deleted but NEVER new ones inserted!!! //=> this allows for a quasi-binary search by id! //reserve() fulfills two task here: 1. massive performance improvement! 2. ensure references in appendUndefined remain valid! output.subFiles.reserve(leftSide.getSubFiles().size() + rightSide.getSubFiles().size()); //assume worst case! output.subDirs.reserve( leftSide.getSubDirs().size() + rightSide.getSubDirs().size()); // for (DirContainer::SubFileList::const_iterator i = leftSide.getSubFiles().begin(); i != leftSide.getSubFiles().end(); ++i) { DirContainer::SubFileList::const_iterator j = rightSide.getSubFiles().find(i->first); //find files that exist on left but not on right if (j == rightSide.getSubFiles().end()) output.addSubFile(i->second.getData(), i->first); //find files that exist on left and right else { appendUndefined.push_back( &output.addSubFile(i->second.getData(), i->first, FILE_EQUAL, j->second.getData())); //FILE_EQUAL is just a dummy-value here } } //find files that exist on right but not on left for (DirContainer::SubFileList::const_iterator j = rightSide.getSubFiles().begin(); j != rightSide.getSubFiles().end(); ++j) { if (leftSide.getSubFiles().find(j->first) == leftSide.getSubFiles().end()) output.addSubFile(j->first, j->second.getData()); } //----------------------------------------------------------------------------------------------- for (DirContainer::SubDirList::const_iterator i = leftSide.getSubDirs().begin(); i != leftSide.getSubDirs().end(); ++i) { DirContainer::SubDirList::const_iterator j = rightSide.getSubDirs().find(i->first); //find directories that exist on left but not on right if (j == rightSide.getSubDirs().end()) { DirMapping& newDirMap = output.addSubDir(true, i->first, false); fillOneSide(i->second, newDirMap); //recurse into subdirectories } else //directories that exist on both sides { DirMapping& newDirMap = output.addSubDir(true, i->first, true); execute(i->second, j->second, newDirMap); //recurse into subdirectories } } //find directories that exist on right but not on left for (DirContainer::SubDirList::const_iterator j = rightSide.getSubDirs().begin(); j != rightSide.getSubDirs().end(); ++j) { if (leftSide.getSubDirs().find(j->first) == leftSide.getSubDirs().end()) { DirMapping& newDirMap = output.addSubDir(false, j->first, true); fillOneSide(j->second, newDirMap); //recurse into subdirectories } } } private: template void fillOneSide(const DirContainer& dirCont, HierarchyObject& output) { //reserve() fulfills two task here: 1. massive performance improvement! 2. ensure references in appendUndefined remain valid! output.subFiles.reserve(dirCont.getSubFiles().size()); output.subDirs.reserve( dirCont.getSubDirs(). size()); for (DirContainer::SubFileList::const_iterator i = dirCont.getSubFiles().begin(); i != dirCont.getSubFiles().end(); ++i) { if (leftSide) output.addSubFile(i->second.getData(), i->first); else output.addSubFile(i->first, i->second.getData()); } for (DirContainer::SubDirList::const_iterator i = dirCont.getSubDirs().begin(); i != dirCont.getSubDirs().end(); ++i) { DirMapping& newDirMap = leftSide ? output.addSubDir(true, i->first, false) : output.addSubDir(false, i->first, true); fillOneSide(i->second, newDirMap); //recurse into subdirectories } } std::vector& appendUndefined; }; void CompareProcess::performBaseComparison(BaseDirMapping& output, std::vector& appendUndefined) { assert(output.subDirs.empty()); assert(output.subFiles.empty()); //PERF_START; //scan directories const DirContainer& directoryLeft = directoryBuffer->getDirectoryDescription( output.getBaseDir(), output.getFilter()); const DirContainer& directoryRight = directoryBuffer->getDirectoryDescription( output.getBaseDir(), output.getFilter()); statusUpdater->updateStatusText(wxToZ(_("Generating file list..."))); statusUpdater->forceUiRefresh(); //keep total number of scanned files up to date //PERF_STOP; MergeSides(appendUndefined).execute(directoryLeft, directoryRight, output); }