X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=src%2FgdcmDocument.cxx;h=5ebc0cb127b4806418b0090e40274eda6d11d6f0;hb=01940747ce53f8c87bc883c47fd152e787e73b07;hp=23a0051201fb0329e673e4d04884f410a1faf48d;hpb=9871e41fed63a2c08545218af2f09c717c64470f;p=gdcm.git diff --git a/src/gdcmDocument.cxx b/src/gdcmDocument.cxx index 23a00512..5ebc0cb1 100644 --- a/src/gdcmDocument.cxx +++ b/src/gdcmDocument.cxx @@ -3,8 +3,8 @@ Program: gdcm Module: $RCSfile: gdcmDocument.cxx,v $ Language: C++ - Date: $Date: 2005/01/13 09:53:42 $ - Version: $Revision: 1.188 $ + Date: $Date: 2005/09/21 09:42:19 $ + Version: $Revision: 1.282 $ Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de l'Image). All rights reserved. See Doc/License.txt or @@ -24,88 +24,205 @@ #include "gdcmUtil.h" #include "gdcmDebug.h" #include "gdcmTS.h" -#include "gdcmException.h" #include "gdcmDictSet.h" -#include "gdcmRLEFramesInfo.h" -#include "gdcmJPEGFragmentsInfo.h" #include "gdcmDocEntrySet.h" #include "gdcmSQItem.h" #include #include - -// For nthos: -#if defined(_MSC_VER) || defined(__BORLANDC__) - #include -#else - #include -#endif +#include +#include // for isdigit +#include // for atoi namespace gdcm { - //----------------------------------------------------------------------------- -// Refer to Document::CheckSwap() -//const unsigned int Document::HEADER_LENGTH_TO_READ = 256; // Refer to Document::SetMaxSizeLoadEntry() const unsigned int Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096 -const unsigned int Document::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff; //----------------------------------------------------------------------------- // Constructor / Destructor +// Constructors and destructors are protected to avoid user to invoke directly + +/** + * \brief This default constructor neither loads nor parses the file. + * You should then invoke \ref Document::Load. + * + */ +Document::Document() + :ElementSet(-1) +{ + Fp = 0; + + SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); + Initialize(); + SwapCode = 1234; + Filetype = ExplicitVR; + // Load will set it to true if sucessfull + Group0002Parsed = false; + IsDocumentAlreadyLoaded = false; + IsDocumentModified = true; + LoadMode = LD_ALL; // default : load everything, later + SetFileName(""); +} + +/** + * \brief Constructor (DEPRECATED : not to break the API) + * @param fileName 'Document' (File or DicomDir) to be open for parsing + */ +Document::Document( std::string const &fileName ) + :ElementSet(-1) +{ + Fp = 0; + + SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); + Initialize(); + SwapCode = 1234; + Filetype = ExplicitVR; + Group0002Parsed = false; + LoadMode = LD_ALL; // Load everything, later + + // Load will set it to true if sucessfull + IsDocumentAlreadyLoaded = false; + IsDocumentModified = true; + + SetFileName(fileName); + Load( ); +} +/** + * \brief Canonical destructor. + */ +Document::~Document () +{ + CloseFile(); +} + +//----------------------------------------------------------------------------- +// Public + +/** + * \brief Loader. use SetLoadMode(), SetFileName() before ! + * @return false if file cannot be open or no swap info was found, + * or no tag was found. + */ +bool Document::Load( ) +{ + if ( GetFileName() == "" ) + { + gdcmWarningMacro( "Use SetFileName, before !" ); + return false; + } + return DoTheLoadingDocumentJob( ); +} +/** + * \brief Loader. (DEPRECATED : not to break the API) + * @param fileName 'Document' (File or DicomDir) to be open for parsing + * @return false if file cannot be open or no swap info was found, + * or no tag was found. + */ +bool Document::Load( std::string const &fileName ) +{ + Filename = fileName; + return DoTheLoadingDocumentJob( ); +} /** - * \brief constructor - * @param filename file to be opened for parsing + * \brief Performs the Loading Job (internal use only) + * @return false if file cannot be open or no swap info was found, + * or no tag was found. */ -Document::Document( std::string const &filename ) : ElementSet(-1) +bool Document::DoTheLoadingDocumentJob( ) { - SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); - Filename = filename; - Initialise(); + if ( ! IsDocumentModified ) // Nothing to do ! + return true; + + // if ( Filename == fileName ) + // { + // gdcmWarningMacro( "The file was already parsed inside this " + // << "gdcm::Document (its name is: " + // << Filename.c_str() ); + // return true; + // } + + //gdcmWarningMacro( "A file was already parsed inside this " + // << "gdcm::Document (previous name was: " + // << Filename.c_str() << ". New name is :" + // << fileName ); + // clean out the Entries, if already parsed + // (probabely a mistake from the user) + + ClearEntry(); Fp = 0; if ( !OpenFile() ) { - return; + // warning already performed in OpenFile() + //gdcmWarningMacro( "Unable to open as an ACR/DICOM file: " + // << Filename.c_str() ); + Filetype = Unknown; + return false; } Group0002Parsed = false; - gdcmVerboseMacro( "Starting parsing of file: " << Filename.c_str()); - // Fp->seekg( 0, std::ios::beg); - - Fp->seekg(0, std::ios::end); - long lgt = Fp->tellg(); - - Fp->seekg( 0, std::ios::beg); + gdcmWarningMacro( "Starting parsing of file: " << Filename.c_str()); - CheckSwap(); - long beg = Fp->tellg(); - lgt -= beg; - - ParseDES( this, beg, lgt, false); // le Load sera fait a la volee + Fp->seekg(0, std::ios::end); + long lgt = Fp->tellg(); // total length of the file + + Fp->seekg(0, std::ios::beg); + + // CheckSwap returns a boolean + // (false if no swap info of any kind was found) + if (! CheckSwap() ) + { + gdcmWarningMacro( "Neither a DICOM V3 nor an ACR-NEMA file: " + << Filename.c_str()); + CloseFile(); + return false; + } + + long beg = Fp->tellg(); // just after DICOM preamble (if any) + + lgt -= beg; // remaining length to parse + + // Recursive call. + // Loading is done during parsing + ParseDES( this, beg, lgt, false); // delim_mode is first defaulted to false + + if ( IsEmpty() ) + { + gdcmWarningMacro( "No tag in internal hash table for: " + << Filename.c_str()); + CloseFile(); + return false; + } + IsDocumentAlreadyLoaded = true; - Fp->seekg( 0, std::ios::beg); + Fp->seekg( 0, std::ios::beg); // Load 'non string' values - std::string PhotometricInterpretation = GetEntry(0x0028,0x0004); - if( PhotometricInterpretation == "PALETTE COLOR " ) + std::string PhotometricInterpretation = GetEntryValue(0x0028,0x0004); + if ( PhotometricInterpretation == "PALETTE COLOR " ) { - LoadEntryBinArea(0x0028,0x1200); // gray LUT - /// FIXME FIXME FIXME + // FIXME + // Probabely this line should be outside the 'if' + // Try to find an image sample holding a 'gray LUT' + LoadEntryBinArea(0x0028,0x1200); // gray LUT + + /// FIXME /// The tags refered by the three following lines used to be CORRECTLY /// defined as having an US Value Representation in the public - /// dictionnary. BUT the semantics implied by the three following + /// dictionary. BUT the semantics implied by the three following /// lines state that the corresponding tag contents are in fact /// the ones of a BinEntry. - /// In order to fix things "Quick and Dirty" the dictionnary was + /// In order to fix things "Quick and Dirty" the dictionary was /// altered on PURPOSE but now contains a WRONG value. /// In order to fix things and restore the dictionary to its /// correct value, one needs to decided of the semantics by deciding - /// wether the following tags are either: + /// whether the following tags are either : /// - multivaluated US, and hence loaded as ValEntry, but afterwards /// also used as BinEntry, which requires the proper conversion, /// - OW, and hence loaded as BinEntry, but afterwards also used @@ -120,89 +237,109 @@ Document::Document( std::string const &filename ) : ElementSet(-1) LoadEntryBinArea(0x0028,0x1222); // Segmented Blue Palette Color LUT Data LoadEntryBinArea(0x0028,0x1223); - } + } + //FIXME later : how to use it? - LoadEntryBinArea(0x0028,0x3006); //LUT Data (CTX dependent) + SeqEntry *modLutSeq = GetSeqEntry(0x0028,0x3000); + if ( modLutSeq !=0 ) + { + SQItem *sqi= modLutSeq->GetFirstSQItem(); + if ( sqi != 0 ) + { + BinEntry *b = sqi->GetBinEntry(0x0028,0x3006); + if ( b != 0 ) + { + if ( b->GetLength() != 0 ) + { + LoadEntryBinArea(b); //LUT Data (CTX dependent) + } + } + } + } + + // Force Loading some more elements if user asked to. + + gdcm::DocEntry *d; + for (ListElements::iterator it = UserForceLoadList.begin(); + it != UserForceLoadList.end(); + ++it) + { + gdcmWarningMacro( "Force Load " << std::hex + << (*it).Group << "|" <<(*it).Elem ); + + d = GetDocEntry( (*it).Group, (*it).Elem); + + if ( d == NULL) + { + gdcmWarningMacro( "You asked toForce Load " << std::hex + << (*it).Group <<"|"<< (*it).Elem + << " that doesn't exist" ); + continue; + } + + if ( dynamic_cast(d) ) + { + LoadDocEntry(d, true); + continue; + } + + BinEntry *b = dynamic_cast(d); + if ( b ) + { + LoadEntryBinArea(b); + b->SetValue(GDCM_BINLOADED); + continue; + } + + if ( dynamic_cast(d) ) + { + gdcmWarningMacro( "You cannot 'ForceLoad' a SeqEntry :" << std::hex + << (*it).Group <<"|"<< (*it).Elem ); + continue; + } + } CloseFile(); - // -------------------------------------------------------------- + // ---------------------------- // Specific code to allow gdcm to read ACR-LibIDO formated images // Note: ACR-LibIDO is an extension of the ACR standard that was - // used at CREATIS. For the time being (say a couple years) - // we keep this kludge to allow a smooth move to gdcm for - // CREATIS developpers (sorry folks). + // used at CREATIS. For the time being (say a couple of years) + // we keep this kludge to allow CREATIS users + // reading their old images. // // if recognition code tells us we deal with a LibIDO image // we switch lineNumber and columnNumber // std::string RecCode; - RecCode = GetEntry(0x0008, 0x0010); // recognition code + RecCode = GetEntryValue(0x0008, 0x0010); // recognition code (RET) if (RecCode == "ACRNEMA_LIBIDO_1.1" || RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares // with "little-endian strings" { Filetype = ACR_LIBIDO; - std::string rows = GetEntry(0x0028, 0x0010); - std::string columns = GetEntry(0x0028, 0x0011); - SetEntry(columns, 0x0028, 0x0010); - SetEntry(rows , 0x0028, 0x0011); + std::string rows = GetEntryValue(0x0028, 0x0010); + std::string columns = GetEntryValue(0x0028, 0x0011); + SetValEntry(columns, 0x0028, 0x0010); + SetValEntry(rows , 0x0028, 0x0011); } - // ----------------- End of ACR-LibIDO kludge ------------------ + // --- End of ACR-LibIDO kludge --- + return true; } -/** - * \brief This default constructor doesn't parse the file. You should - * then invoke \ref Document::SetFileName and then the parsing. - */ -Document::Document() : ElementSet(-1) -{ - Fp = 0; - - SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); - Initialise(); - SwapCode = 1234; - Filetype = ExplicitVR; - Group0002Parsed = false; -} /** - * \brief Canonical destructor. + * \brief Adds a new element we want to load anyway + * @param group Group number of the target tag. + * @param elem Element number of the target tag. */ -Document::~Document () -{ - RefPubDict = NULL; - RefShaDict = NULL; - - delete RLEInfo; - delete JPEGInfo; -} - -//----------------------------------------------------------------------------- -// Print - -/** - * \brief Prints The Dict Entries of THE public Dicom Dictionary - * @return - */ -void Document::PrintPubDict(std::ostream &os) -{ - RefPubDict->SetPrintLevel(PrintLevel); - RefPubDict->Print(os); -} - -/** - * \brief Prints The Dict Entries of THE shadow Dicom Dictionary - * @return - */ -void Document::PrintShaDict(std::ostream &os) -{ - RefShaDict->SetPrintLevel(PrintLevel); - RefShaDict->Print(os); +void Document::AddForceLoadElement (uint16_t group, uint16_t elem) +{ + Element el; + el.Group = group; + el.Elem = elem; + UserForceLoadList.push_back(el); } - -//----------------------------------------------------------------------------- -// Public /** * \brief Get the public dictionary used */ @@ -240,30 +377,69 @@ bool Document::SetShaDict(DictKey const &dictName) } /** - * \brief This predicate, based on hopefully reasonable heuristics, - * decides whether or not the current Document was properly parsed - * and contains the mandatory information for being considered as - * a well formed and usable Dicom/Acr File. - * @return true when Document is the one of a reasonable Dicom/Acr file, - * false otherwise. + * \brief This predicate tells us whether or not the current Document + * was properly parsed and contains at least *one* Dicom Element + * (and nothing more, sorry). + * @return false when we're 150 % sure it's NOT a Dicom/Acr file, + * true otherwise. */ bool Document::IsReadable() { - if( Filetype == Unknown) + if ( Filetype == Unknown ) { - gdcmVerboseMacro( "Wrong filetype"); + gdcmWarningMacro( "Wrong filetype"); return false; } - if( TagHT.empty() ) - { - gdcmVerboseMacro( "No tags in internal hash table."); + if ( IsEmpty() ) + { + gdcmWarningMacro( "No tag in internal hash table."); return false; } return true; } +/** + * \brief Predicate for dicom version 3 file. + * @return True when the file is a dicom version 3. + */ +bool Document::IsDicomV3() +{ + // Checking if Transfer Syntax exists is enough + // Anyway, it's too late check if the 'Preamble' was found ... + // And ... would it be a rich idea to check ? + // (some 'no Preamble' DICOM images exist !) + return GetDocEntry(0x0002, 0x0010) != NULL; +} + +/** + * \brief Predicate for Papyrus file + * Dedicated to whomsoever it may concern + * @return True when the file is a Papyrus file. + */ +bool Document::IsPapyrus() +{ + // check for Papyrus private Sequence + DocEntry *e = GetDocEntry(0x0041, 0x1050); + if ( !e ) + return false; + // check if it's actually a Sequence + if ( !dynamic_cast(e) ) + return false; + return true; +} + +/** + * \brief returns the File Type + * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown) + * @return the FileType code + */ +FileType Document::GetFileType() +{ + return Filetype; +} + /** * \brief Accessor to the Transfer Syntax (when present) of the * current document (it internally handles reading the @@ -302,28 +478,83 @@ std::string Document::GetTransferSyntax() } /** - * \brief Predicate for dicom version 3 file. - * @return True when the file is a dicom version 3. + * \brief Accesses the info from 0002,0010 : Transfer Syntax and TS + * @return The full Transfer Syntax Name (as opposed to Transfer Syntax UID) */ -bool Document::IsDicomV3() +std::string Document::GetTransferSyntaxName() { - // Checking if Transfer Syntax exists is enough - // Anyway, it's to late check if the 'Preamble' was found ... - // And ... would it be a rich idea to check ? - // (some 'no Preamble' DICOM images exist !) - return GetDocEntry(0x0002, 0x0010) != NULL; -} + // use the TS (TS : Transfer Syntax) + std::string transferSyntax = GetEntryValue(0x0002,0x0010); + + if ( (transferSyntax.find(GDCM_NOTLOADED) < transferSyntax.length()) ) + { + gdcmErrorMacro( "Transfer Syntax not loaded. " << std::endl + << "Better you increase MAX_SIZE_LOAD_ELEMENT_VALUE" ); + return "Uncompressed ACR-NEMA"; + } + if ( transferSyntax == GDCM_UNFOUND ) + { + gdcmWarningMacro( "Unfound Transfer Syntax (0002,0010)"); + return "Uncompressed ACR-NEMA"; + } + + // we do it only when we need it + const TSKey &tsName = Global::GetTS()->GetValue( transferSyntax ); + // Global::GetTS() is a global static you shall never try to delete it! + return tsName; +} +// +// --------------- Swap Code ------------------ /** - * \brief returns the File Type - * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown) - * @return the FileType code + * \brief Swaps the bytes so they agree with the processor order + * @return The properly swaped 16 bits integer. */ -FileType Document::GetFileType() +uint16_t Document::SwapShort(uint16_t a) { - return Filetype; + if ( SwapCode == 4321 || SwapCode == 2143 ) + { + //a = ((( a << 8 ) & 0xff00 ) | (( a >> 8 ) & 0x00ff ) ); + // Save CPU time + a = ( a << 8 ) | ( a >> 8 ); + } + return a; } +/** + * \brief Swaps back the bytes of 4-byte long integer accordingly to + * processor order. + * @return The properly swaped 32 bits integer. + */ +uint32_t Document::SwapLong(uint32_t a) +{ + switch (SwapCode) + { + case 1234 : + break; + case 4321 : +// a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) | +// ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) ); +// save CPU time + a=( ( a<<24) | ((a<<8) & 0x00ff0000) | + ((a>>8) & 0x0000ff00) | (a>>24) ); + break; + case 3412 : +// a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) ); + a=( (a<<16) | (a>>16) ); + break; + case 2143 : + a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) ); + break; + default : + gdcmErrorMacro( "Unset swap code:" << SwapCode ); + a = 0; + } + return a; +} + +// +// -----------------File I/O --------------- /** * \brief Tries to open the file \ref Document::Filename and * checks the preamble when existing. @@ -331,66 +562,73 @@ FileType Document::GetFileType() */ std::ifstream *Document::OpenFile() { - HasDCMPreamble = false; if (Filename.length() == 0) { return 0; } - if(Fp) + if ( Fp ) { - gdcmVerboseMacro( "File already open: " << Filename.c_str()); + gdcmWarningMacro( "File already open: " << Filename.c_str()); + CloseFile(); } Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary); - if( ! *Fp ) + if ( ! *Fp ) { - gdcmDebugMacro( "Cannot open file: " << Filename.c_str()); + // Don't user gdcmErrorMacro : + // a spurious message will appear when you use, for instance + // gdcm::FileHelper *fh = new gdcm::FileHelper( outputFileName ); + // to create outputFileName. + gdcmWarningMacro( "Cannot open file: " << Filename.c_str()); delete Fp; Fp = 0; return 0; + //exit(1); // No function is allowed to leave the application instead + // of warning the caller } - uint16_t zero; + uint16_t zero = 0; Fp->read((char*)&zero, (size_t)2); - if( Fp->eof() ) + if ( Fp->eof() ) { CloseFile(); return 0; } - //ACR -- or DICOM with no Preamble; may start with a Shadow Group -- - if( + //-- ACR or DICOM with no Preamble; may start with a Shadow Group -- + if ( zero == 0x0001 || zero == 0x0100 || zero == 0x0002 || zero == 0x0200 || zero == 0x0003 || zero == 0x0300 || zero == 0x0004 || zero == 0x0400 || zero == 0x0005 || zero == 0x0500 || zero == 0x0006 || zero == 0x0600 || zero == 0x0007 || zero == 0x0700 || zero == 0x0008 || zero == 0x0800 ) { - std::string msg - = Util::Format("ACR/DICOM with no preamble: (%04x)\n", zero); - gdcmVerboseMacro( msg.c_str() ); + std::string msg = Util::Format( + "ACR/DICOM starting at the beginning of the file:(%04x)\n", zero); + gdcmWarningMacro( msg.c_str() ); return Fp; } - //DICOM + //-- DICOM -- Fp->seekg(126L, std::ios::cur); - char dicm[4]; + char dicm[4]; // = {' ',' ',' ',' '}; Fp->read(dicm, (size_t)4); - if( Fp->eof() ) + if ( Fp->eof() ) { CloseFile(); return 0; } - if( memcmp(dicm, "DICM", 4) == 0 ) + if ( memcmp(dicm, "DICM", 4) == 0 ) { HasDCMPreamble = true; return Fp; } - + + // -- Neither ACR/No Preamble Dicom nor DICOMV3 file CloseFile(); - gdcmVerboseMacro( "Not DICOM/ACR (missing preamble)" << Filename.c_str()); - + gdcmWarningMacro( "Neither ACR/No Preamble Dicom nor DICOMV3 file: " + << Filename.c_str()); return 0; } @@ -400,29 +638,24 @@ std::ifstream *Document::OpenFile() */ bool Document::CloseFile() { - if( Fp ) + if ( Fp ) { Fp->close(); delete Fp; Fp = 0; } - - return true; //FIXME how do we detect a non-close ifstream ? + return true; } /** - * \brief Writes in a file all the Header Entries (Dicom Elements) - * @param fp file pointer on an already open file + * \brief Writes in a file all the Entries (Dicom Elements) + * @param fp file pointer on an already open file (actually: Output File Stream) * @param filetype Type of the File to be written * (ACR-NEMA, ExplicitVR, ImplicitVR) - * \return Always true. */ void Document::WriteContent(std::ofstream *fp, FileType filetype) { - /// \todo move the following lines (and a lot of others, to be written) - /// to a future function CheckAndCorrectHeader - /// (necessary if user wants to write a DICOM V3 file - /// starting from an ACR-NEMA (V2) Header + // Skip if user wants to write an ACR-NEMA file if ( filetype == ImplicitVR || filetype == ExplicitVR ) { @@ -433,652 +666,307 @@ void Document::WriteContent(std::ofstream *fp, FileType filetype) fp->write("DICM", 4); } -/** - * \todo rewrite later, if really usefull - * - 'Group Length' element is optional in DICOM - * - but un-updated odd groups lengthes can causes pb - * (xmedcon breaker) - * - * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) ) - * UpdateGroupLength(false,filetype); - * if ( filetype == ACR) - * UpdateGroupLength(true,ACR); - */ - + /* + * \todo rewrite later, if really usefull + * - 'Group Length' element is optional in DICOM + * - but un-updated odd groups lengthes can causes pb + * (xmedcon breaker) + * + * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) ) + * UpdateGroupLength(false,filetype); + * if ( filetype == ACR) + * UpdateGroupLength(true,ACR); + */ + ElementSet::WriteContent(fp, filetype); // This one is recursive } +// ----------------------------------------- +// Content entries /** - * \brief Modifies the value of a given Doc Entry (Dicom Element) - * when it exists. Create it with the given value when unexistant. - * @param value (string) Value to be set - * @param group Group number of the Entry - * @param elem Element number of the Entry - * @param vr V(alue) R(epresentation) of the Entry -if private Entry- - * \return pointer to the modified/created Header Entry (NULL when creation - * failed). - */ -ValEntry *Document::ReplaceOrCreate(std::string const &value, - uint16_t group, - uint16_t elem, - TagName const &vr ) + * \brief Loads (from disk) the element content + * when a string is not suitable + * @param group group number of the Entry + * @param elem element number of the Entry + */ +void Document::LoadEntryBinArea(uint16_t group, uint16_t elem) { - ValEntry *valEntry = 0; - DocEntry *currentEntry = GetDocEntry( group, elem); - - if (currentEntry) + // Search the corresponding DocEntry + DocEntry *docElement = GetDocEntry(group, elem); + if ( !docElement ) { - valEntry = dynamic_cast< ValEntry* >(currentEntry); - - // Verify the VR - if( valEntry ) - if( valEntry->GetVR()!=vr ) - valEntry=NULL; - - // if currentEntry doesn't correspond to the requested valEntry - if( !valEntry) - { - if (!RemoveEntry(currentEntry)) - { - gdcmVerboseMacro( "Removal of previous DocEntry failed."); - - return NULL; - } - } + gdcmWarningMacro(std::hex << group << "|" << elem + << "doesn't exist" ); + return; } - - // Create a new valEntry if necessary - if (!valEntry) + BinEntry *binElement = dynamic_cast(docElement); + if ( !binElement ) { - valEntry = NewValEntry(group, elem, vr); - - if ( !AddEntry(valEntry)) - { - gdcmVerboseMacro("AddEntry failed although this is a creation."); - - delete valEntry; - return NULL; - } - } - - // Set the binEntry value - SetEntry(value, valEntry); - return valEntry; -} - -/* - * \brief Modifies the value of a given Header Entry (Dicom Element) - * when it exists. Create it with the given value when unexistant. - * A copy of the binArea is made to be kept in the Document. - * @param binArea (binary) value to be set - * @param Group Group number of the Entry - * @param Elem Element number of the Entry - * @param vr V(alue) R(epresentation) of the Entry -if private Entry- - * \return pointer to the modified/created Header Entry (NULL when creation - * failed). - */ -BinEntry *Document::ReplaceOrCreate(uint8_t *binArea, - int lgth, - uint16_t group, - uint16_t elem, - TagName const &vr ) -{ - BinEntry *binEntry = 0; - DocEntry *currentEntry = GetDocEntry( group, elem); - - // Verify the currentEntry - if (currentEntry) - { - binEntry = dynamic_cast< BinEntry* >(currentEntry); - - // Verify the VR - if( binEntry ) - if( binEntry->GetVR()!=vr ) - binEntry=NULL; - - // if currentEntry doesn't correspond to the requested valEntry - if( !binEntry) - { - if (!RemoveEntry(currentEntry)) - { - gdcmVerboseMacro( "Removal of previous DocEntry failed."); - - return NULL; - } - } - } - - // Create a new binEntry if necessary - if (!binEntry) - { - binEntry = NewBinEntry(group, elem, vr); - - if ( !AddEntry(binEntry)) - { - gdcmVerboseMacro( "AddEntry failed allthough this is a creation."); - - delete binEntry; - return NULL; - } - } - - // Set the binEntry value - uint8_t *tmpArea; - if (lgth>0 && binArea) - { - tmpArea = new uint8_t[lgth]; - memcpy(tmpArea,binArea,lgth); - } - else - { - tmpArea = 0; - } - if (!SetEntry(tmpArea,lgth,binEntry)) - { - if (tmpArea) - { - delete[] tmpArea; - } - } - - return binEntry; -} - -/* - * \brief Modifies the value of a given Header Entry (Dicom Element) - * when it exists. Create it when unexistant. - * @param Group Group number of the Entry - * @param Elem Element number of the Entry - * \return pointer to the modified/created SeqEntry (NULL when creation - * failed). - */ -SeqEntry *Document::ReplaceOrCreate( uint16_t group, uint16_t elem) -{ - SeqEntry *seqEntry = 0; - DocEntry *currentEntry = GetDocEntry( group, elem); - - // Verify the currentEntry - if (currentEntry) - { - seqEntry = dynamic_cast< SeqEntry* >(currentEntry); - - // Verify the VR - if( seqEntry ) - if( seqEntry->GetVR()!="SQ" ) - seqEntry=NULL; - - // if currentEntry doesn't correspond to the requested valEntry - if( !seqEntry) - { - if (!RemoveEntry(currentEntry)) - { - gdcmVerboseMacro( "Removal of previous DocEntry failed."); - - return NULL; - } - } - } - - // Create a new seqEntry if necessary - if (!seqEntry) - { - seqEntry = NewSeqEntry(group, elem); - - if ( !AddEntry(seqEntry)) - { - gdcmVerboseMacro( "AddEntry failed allthough this is a creation."); - - delete seqEntry; - return NULL; - } - } - - return seqEntry; -} - -/** - * \brief Set a new value if the invoked element exists - * Seems to be useless !!! - * @param value new element value - * @param group group number of the Entry - * @param elem element number of the Entry - * \return boolean - */ -bool Document::ReplaceIfExist(std::string const &value, - uint16_t group, uint16_t elem ) -{ - SetEntry(value, group, elem); - - return true; -} - -//----------------------------------------------------------------------------- -// Protected - -/** - * \brief Checks if a given Dicom Element exists within the H table - * @param group Group number of the searched Dicom Element - * @param element Element number of the searched Dicom Element - * @return true is found - */ -bool Document::CheckIfEntryExist(uint16_t group, uint16_t element ) -{ - const std::string &key = DictEntry::TranslateToKey(group, element ); - return TagHT.count(key) != 0; -} - - -/** - * \brief Searches within Header Entries (Dicom Elements) parsed with - * the public and private dictionaries - * for the element value representation of a given tag. - * @param group Group number of the searched tag. - * @param element Element number of the searched tag. - * @return Corresponding element value representation when it exists, - * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. - */ -std::string Document::GetEntry(uint16_t group, uint16_t element) -{ - TagKey key = DictEntry::TranslateToKey(group, element); - if ( !TagHT.count(key)) - { - return GDCM_UNFOUND; - } - - return ((ValEntry *)TagHT.find(key)->second)->GetValue(); -} - -/** - * \brief Searches within Header Entries (Dicom Elements) parsed with - * the public and private dictionaries - * for the element value representation of a given tag.. - * - * Obtaining the VR (Value Representation) might be needed by caller - * to convert the string typed content to caller's native type - * (think of C++ vs Python). The VR is actually of a higher level - * of semantics than just the native C++ type. - * @param group Group number of the searched tag. - * @param element Element number of the searched tag. - * @return Corresponding element value representation when it exists, - * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. - */ -std::string Document::GetEntryVR(uint16_t group, uint16_t element) -{ - DocEntry *elem = GetDocEntry(group, element); - if ( !elem ) - { - return GDCM_UNFOUND; - } - return elem->GetVR(); -} - -/** - * \brief Searches within Header Entries (Dicom Elements) parsed with - * the public and private dictionaries - * for the value length of a given tag.. - * @param group Group number of the searched tag. - * @param element Element number of the searched tag. - * @return Corresponding element length; -2 if not found - */ -int Document::GetEntryLength(uint16_t group, uint16_t element) -{ - DocEntry *elem = GetDocEntry(group, element); - if ( !elem ) - { - return -2; //magic number - } - return elem->GetLength(); -} - -/** - * \brief Accesses an existing DocEntry (i.e. a Dicom Element) - * through it's (group, element) and modifies it's content with - * the given value. - * @param content new value (string) to substitute with - * @param group group number of the Dicom Element to modify - * @param element element number of the Dicom Element to modify - */ -bool Document::SetEntry(std::string const& content, - uint16_t group, uint16_t element) -{ - ValEntry *entry = GetValEntry(group, element); - if (!entry ) - { - gdcmVerboseMacro( "No corresponding ValEntry (try promotion first)."); - return false; - } - return SetEntry(content,entry); -} - -/** - * \brief Accesses an existing DocEntry (i.e. a Dicom Element) - * through it's (group, element) and modifies it's content with - * the given value. - * @param content new value (void* -> uint8_t*) to substitute with - * @param lgth new value length - * @param group group number of the Dicom Element to modify - * @param element element number of the Dicom Element to modify - */ -bool Document::SetEntry(uint8_t*content, int lgth, - uint16_t group, uint16_t element) -{ - BinEntry *entry = GetBinEntry(group, element); - if (!entry ) - { - gdcmVerboseMacro( "No corresponding ValEntry (try promotion first)."); - return false; - } - - return SetEntry(content,lgth,entry); -} - -/** - * \brief Accesses an existing DocEntry (i.e. a Dicom Element) - * and modifies it's content with the given value. - * @param content new value (string) to substitute with - * @param entry Entry to be modified - */ -bool Document::SetEntry(std::string const &content,ValEntry *entry) -{ - if(entry) - { - entry->SetValue(content); - return true; - } - return false; -} - -/** - * \brief Accesses an existing BinEntry (i.e. a Dicom Element) - * and modifies it's content with the given value. - * @param content new value (void* -> uint8_t*) to substitute with - * @param entry Entry to be modified - * @param lgth new value length - */ -bool Document::SetEntry(uint8_t *content, int lgth, BinEntry *entry) -{ - if(entry) - { - // Hope Binary field length is *never* wrong - /*if(lgth%2) // Non even length are padded with a space (020H). - { - lgth++; - //content = content + '\0'; // fing a trick to enlarge a binary field? - }*/ - - entry->SetBinArea(content); - entry->SetLength(lgth); - entry->SetValue(GDCM_BINLOADED); - return true; - } - return false; -} - -/** - * \brief Gets (from Header) a 'non string' element value - * (LoadElementValues has already be executed) - * @param group group number of the Entry - * @param elem element number of the Entry - * @return Pointer to the 'non string' area - */ -void *Document::GetEntryBinArea(uint16_t group, uint16_t elem) -{ - DocEntry *entry = GetDocEntry(group, elem); - if (!entry) - { - gdcmVerboseMacro( "No entry"); - return 0; - } - if ( BinEntry *binEntry = dynamic_cast(entry) ) - { - return binEntry->GetBinArea(); - } - - return 0; -} - -/** - * \brief Loads (from disk) the element content - * when a string is not suitable - * @param group group number of the Entry - * @param elem element number of the Entry - */ -void Document::LoadEntryBinArea(uint16_t group, uint16_t elem) -{ - // Search the corresponding DocEntry - DocEntry *docElement = GetDocEntry(group, elem); - if ( !docElement ) + gdcmWarningMacro(std::hex << group << "|" << elem + << "is NOT a BinEntry"); return; - - BinEntry *binElement = dynamic_cast(docElement); - if( !binElement ) - return; - + } LoadEntryBinArea(binElement); } /** - * \brief Loads (from disk) the element content - * when a string is not suitable - * @param element Entry whose binArea is going to be loaded + * \brief Loads (from disk) the element content + * when a string is not suitable + * @param elem Entry whose binArea is going to be loaded */ -void Document::LoadEntryBinArea(BinEntry *element) +void Document::LoadEntryBinArea(BinEntry *elem) { - if(element->GetBinArea()) + if (elem->GetBinArea() ) return; bool openFile = !Fp; - if(openFile) + if ( openFile ) OpenFile(); - size_t o =(size_t)element->GetOffset(); + size_t o =(size_t)elem->GetOffset(); Fp->seekg(o, std::ios::beg); - size_t l = element->GetLength(); + size_t l = elem->GetLength(); uint8_t *a = new uint8_t[l]; - if( !a ) + if ( !a ) { - gdcmVerboseMacro( "Cannot allocate a"); + gdcmWarningMacro( "Cannot allocate BinEntry content for : " + << std::hex << elem->GetGroup() + << "|" << elem->GetElement() ); return; } - /// \todo check the result Fp->read((char*)a, l); - if( Fp->fail() || Fp->eof()) + if ( Fp->fail() || Fp->eof() ) { delete[] a; return; } - element->SetBinArea(a); + elem->SetBinArea(a); - if(openFile) + if ( openFile ) CloseFile(); } /** - * \brief Sets a 'non string' value to a given Dicom Element - * @param area area containing the 'non string' value - * @param group Group number of the searched Dicom Element - * @param element Element number of the searched Dicom Element - * @return + * \brief Loads the element while preserving the current + * underlying file position indicator as opposed to + * LoadDocEntry that modifies it. + * @param entry DocEntry whose value will be loaded. */ -/*bool Document::SetEntryBinArea(uint8_t *area, - uint16_t group, uint16_t element) +void Document::LoadDocEntrySafe(DocEntry *entry) { - DocEntry *currentEntry = GetDocEntry(group, element); - if ( !currentEntry ) + if ( Fp ) { - return false; - } - - if ( BinEntry *binEntry = dynamic_cast(currentEntry) ) - { - binEntry->SetBinArea( area ); - return true; + long PositionOnEntry = Fp->tellg(); + LoadDocEntry(entry); + Fp->seekg(PositionOnEntry, std::ios::beg); } - - return false; -}*/ +} /** - * \brief retrieves a Dicom Element (the first one) using (group, element) - * \warning (group, element) IS NOT an identifier inside the Dicom Header - * if you think it's NOT UNIQUE, check the count number - * and use iterators to retrieve ALL the Dicoms Elements within - * a given couple (group, element) - * @param group Group number of the searched Dicom Element - * @param element Element number of the searched Dicom Element - * @return + * \brief Compares two documents, according to \ref DicomDir rules + * \warning Does NOT work with ACR-NEMA files + * \todo Find a trick to solve the pb (use RET fields ?) + * @param document to compare with current one + * @return true if 'smaller' */ -DocEntry *Document::GetDocEntry(uint16_t group, uint16_t element) +bool Document::operator<(Document &document) { - TagKey key = DictEntry::TranslateToKey(group, element); - if ( !TagHT.count(key)) + // Patient Name + std::string s1 = GetEntryValue(0x0010,0x0010); + std::string s2 = document.GetEntryValue(0x0010,0x0010); + if (s1 < s2) { - return NULL; + return true; } - return TagHT.find(key)->second; -} - -/** - * \brief Same as \ref Document::GetDocEntry except it only - * returns a result when the corresponding entry is of type - * ValEntry. - * @return When present, the corresponding ValEntry. - */ -ValEntry *Document::GetValEntry(uint16_t group, uint16_t element) -{ - DocEntry *currentEntry = GetDocEntry(group, element); - if ( !currentEntry ) + else if ( s1 > s2 ) { - return 0; + return false; } - if ( ValEntry *entry = dynamic_cast(currentEntry) ) + else { - return entry; + // Patient ID + s1 = GetEntryValue(0x0010,0x0020); + s2 = document.GetEntryValue(0x0010,0x0020); + if ( s1 < s2 ) + { + return true; + } + else if ( s1 > s2 ) + { + return false; + } + else + { + // Study Instance UID + s1 = GetEntryValue(0x0020,0x000d); + s2 = document.GetEntryValue(0x0020,0x000d); + if ( s1 < s2 ) + { + return true; + } + else if ( s1 > s2 ) + { + return false; + } + else + { + // Serie Instance UID + s1 = GetEntryValue(0x0020,0x000e); + s2 = document.GetEntryValue(0x0020,0x000e); + if ( s1 < s2 ) + { + return true; + } + else if ( s1 > s2 ) + { + return false; + } + } + } } - gdcmVerboseMacro( "Unfound ValEntry."); - - return 0; + return false; } +//----------------------------------------------------------------------------- +// Protected /** - * \brief Same as \ref Document::GetDocEntry except it only - * returns a result when the corresponding entry is of type - * BinEntry. - * @return When present, the corresponding BinEntry. + * \brief Reads a supposed to be 16 Bits integer + * (swaps it depending on processor endianness) + * @return read value */ -BinEntry *Document::GetBinEntry(uint16_t group, uint16_t element) +uint16_t Document::ReadInt16() + throw( FormatError ) { - DocEntry *currentEntry = GetDocEntry(group, element); - if ( !currentEntry ) + uint16_t g; + Fp->read ((char*)&g, (size_t)2); + if ( Fp->fail() ) { - return 0; + throw FormatError( "Document::ReadInt16()", " file error." ); } - if ( BinEntry *entry = dynamic_cast(currentEntry) ) + if ( Fp->eof() ) { - return entry; + throw FormatError( "Document::ReadInt16()", "EOF." ); } - gdcmVerboseMacro( "Unfound BinEntry."); - - return 0; + g = SwapShort(g); + return g; } /** - * \brief Loads the element while preserving the current - * underlying file position indicator as opposed to - * to LoadDocEntry that modifies it. - * @param entry Header Entry whose value will be loaded. - * @return + * \brief Reads a supposed to be 32 Bits integer + * (swaps it depending on processor endianness) + * @return read value */ -void Document::LoadDocEntrySafe(DocEntry *entry) +uint32_t Document::ReadInt32() + throw( FormatError ) { - if(Fp) + uint32_t g; + Fp->read ((char*)&g, (size_t)4); + if ( Fp->fail() ) { - long PositionOnEntry = Fp->tellg(); - LoadDocEntry(entry); - Fp->seekg(PositionOnEntry, std::ios::beg); + throw FormatError( "Document::ReadInt32()", " file error." ); } -} - -/** - * \brief Swaps back the bytes of 4-byte long integer accordingly to - * processor order. - * @return The properly swaped 32 bits integer. - */ -uint32_t Document::SwapLong(uint32_t a) -{ - switch (SwapCode) + if ( Fp->eof() ) { - case 1234 : - break; - case 4321 : - a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) | - ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) ); - break; - case 3412 : - a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) ); - break; - case 2143 : - a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) ); - break; - default : - gdcmErrorMacro( "Unset swap code:" << SwapCode ); - a = 0; + throw FormatError( "Document::ReadInt32()", "EOF." ); } - return a; -} + g = SwapLong(g); + return g; +} /** - * \brief Unswaps back the bytes of 4-byte long integer accordingly to - * processor order. - * @return The properly unswaped 32 bits integer. + * \brief skips bytes inside the source file + * \warning NOT end user intended method ! + * @return */ -uint32_t Document::UnswapLong(uint32_t a) +void Document::SkipBytes(uint32_t nBytes) { - return SwapLong(a); + //FIXME don't dump the returned value + Fp->seekg((long)nBytes, std::ios::cur); } /** - * \brief Swaps the bytes so they agree with the processor order - * @return The properly swaped 16 bits integer. + * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader */ -uint16_t Document::SwapShort(uint16_t a) +int Document::ComputeGroup0002Length( /*FileType filetype*/ ) { - if ( SwapCode == 4321 || SwapCode == 2143 ) + uint16_t gr; + std::string vr; + + int groupLength = 0; + bool found0002 = false; + + // for each zero-level Tag in the DCM Header + DocEntry *entry = GetFirstEntry(); + while( entry ) { - a = ((( a << 8 ) & 0x0ff00 ) | (( a >> 8 ) & 0x00ff ) ); + gr = entry->GetGroup(); + + if ( gr == 0x0002 ) + { + found0002 = true; + + if ( entry->GetElement() != 0x0000 ) + { + vr = entry->GetVR(); + + // FIXME : group 0x0002 is *always* Explicit VR! + + //if ( filetype == ExplicitVR ) + //{ + // if ( (vr == "OB") || (vr == "OW") || (vr == "UT") || (vr == "SQ") ) + // (no SQ, OW, UT in group 0x0002;) + if ( vr == "OB" ) + { + // explicit VR AND OB, OW, SQ, UT : 4 more bytes + groupLength += 4; + } + //} + groupLength += 2 + 2 + 4 + entry->GetLength(); + } + } + else if (found0002 ) + break; + + entry = GetNextEntry(); } - return a; + return groupLength; } +//----------------------------------------------------------------------------- +// Private /** - * \brief Unswaps the bytes so they agree with the processor order - * @return The properly unswaped 16 bits integer. + * \brief Loads all the needed Dictionaries + * \warning NOT end user intended method ! */ -uint16_t Document::UnswapShort(uint16_t a) +void Document::Initialize() { - return SwapShort(a); + RefPubDict = Global::GetDicts()->GetDefaultPubDict(); + RefShaDict = NULL; + Filetype = Unknown; } -//----------------------------------------------------------------------------- -// Private - /** * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries) - * @return length of the parsed set. + * @param set DocEntrySet we are going to parse ('zero level' or a SQItem) + * @param offset start of parsing + * @param l_max length to parse (meaningless when we are in 'delimitor mode') + * @param delim_mode : whether we are in 'delimitor mode' (l=0xffffff) or not */ void Document::ParseDES(DocEntrySet *set, long offset, long l_max, bool delim_mode) { - DocEntry *newDocEntry = 0; + DocEntry *newDocEntry; ValEntry *newValEntry; BinEntry *newBinEntry; SeqEntry *newSeqEntry; VRKey vr; - bool used=false; - + bool used; // will be set to false when something wrong happens to an Entry. + // (Entry will then be deleted) + bool delim_mode_intern = delim_mode; + bool first = true; while (true) { if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) @@ -1086,175 +974,257 @@ void Document::ParseDES(DocEntrySet *set, long offset, break; } - used=true; newDocEntry = ReadNextDocEntry( ); + + // FIXME : + // Private tag, in IMplicit VR are defaulted as a BinEntry, + // Very often they are only composed of Printable characters, + // and could be defaulted as a ValEntry. + // It's too late to do the Job + // (we should check the value, but we know it after LoadDocEntry ...) + // --> in next gdcm major release let's unify ValEntry and BinEntry ! + + // Uncoment this printf line to be able to 'follow' the DocEntries + // when something *very* strange happens + + //printf( "%04x|%04x %s\n",newDocEntry->GetGroup(), + // newDocEntry->GetElement(), + // newDocEntry->GetVR().c_str() ); + if ( !newDocEntry ) { break; } - vr = newDocEntry->GetVR(); + // an Item Starter found elsewhere but the first position + // of a SeqEntry means previous entry was a Sequence + // but we didn't get it (private Sequence + Implicit VR) + // we have to backtrack. + if ( !first && newDocEntry->IsItemStarter() ) + { + newDocEntry = Backtrack(newDocEntry); + } + else + { + PreviousDocEntry = newDocEntry; + } + + used = true; newValEntry = dynamic_cast(newDocEntry); newBinEntry = dynamic_cast(newDocEntry); - newSeqEntry = dynamic_cast(newDocEntry); - if ( newValEntry || newBinEntry ) + if ( newValEntry || newBinEntry ) { + //////////////////////////// ContentEntry if ( newBinEntry ) { - if ( ! Global::GetVR()->IsVROfBinaryRepresentable(vr) ) + vr = newDocEntry->GetVR(); + if ( Filetype == ExplicitVR && + !Global::GetVR()->IsVROfBinaryRepresentable(vr) ) { ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR - gdcmVerboseMacro( std::hex << newDocEntry->GetGroup() + gdcmWarningMacro( std::hex << newDocEntry->GetGroup() << "|" << newDocEntry->GetElement() << " : Neither Valentry, nor BinEntry." "Probably unknown VR."); } //////////////////// BinEntry or UNKOWN VR: + // When "this" is a Document the Key is simply of the // form ( group, elem )... - if (Document *dummy = dynamic_cast< Document* > ( set ) ) - { - (void)dummy; - newBinEntry->SetKey( newBinEntry->GetKey() ); - } + //if ( set == this ) // ( dynamic_cast< Document* > ( set ) ) + //{ + // newBinEntry->SetKey( newBinEntry->GetKey() ); + //} // but when "this" is a SQItem, we are inserting this new - // valEntry in a sequence item, and the kay has the + // valEntry in a sequence item, and the key has the // generalized form (refer to \ref BaseTagKey): - if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + + // time waste hunting + //if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + //{ + // newBinEntry->SetKey( parentSQItem->GetBaseTagKey() + // + newBinEntry->GetKey() ); + //} + + if ( !set->AddEntry( newBinEntry ) ) { - newBinEntry->SetKey( parentSQItem->GetBaseTagKey() - + newBinEntry->GetKey() ); + gdcmWarningMacro( "in ParseDES : cannot add a BinEntry " + << newBinEntry->GetKey() + << " (at offset : " + << newBinEntry->GetOffset() << " )" ); + used=false; } - - LoadDocEntry( newBinEntry ); - if( !set->AddEntry( newBinEntry ) ) + else { - //Expect big troubles if here - //delete newBinEntry; - used=false; + // Load only if we can add (not a duplicate key) + LoadDocEntry( newBinEntry ); } - } + } // end BinEntry else { /////////////////////// ValEntry + // When "set" is a Document, then we are at the top of the // hierarchy and the Key is simply of the form ( group, elem )... - if (Document *dummy = dynamic_cast< Document* > ( set ) ) - { - (void)dummy; - newValEntry->SetKey( newValEntry->GetKey() ); - } + //if ( set == this ) // ( dynamic_cast< Document* > ( set ) ) + //{ + // newValEntry->SetKey( newValEntry->GetKey() ); + //} // ...but when "set" is a SQItem, we are inserting this new // valEntry in a sequence item. Hence the key has the // generalized form (refer to \ref BaseTagKey): - if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + + // time waste hunting + //if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + //{ + // newValEntry->SetKey( parentSQItem->GetBaseTagKey() + // + newValEntry->GetKey() ); + //} + + if ( !set->AddEntry( newValEntry ) ) { - newValEntry->SetKey( parentSQItem->GetBaseTagKey() - + newValEntry->GetKey() ); + gdcmWarningMacro( "in ParseDES : cannot add a ValEntry " + << newValEntry->GetKey() + << " (at offset : " + << newValEntry->GetOffset() << " )" ); + used=false; } - - LoadDocEntry( newValEntry ); - bool delimitor=newValEntry->IsItemDelimitor(); - if( !set->AddEntry( newValEntry ) ) + else { - // If here expect big troubles - //delete newValEntry; //otherwise mem leak - used=false; + // Load only if we can add (not a duplicate key) + LoadDocEntry( newValEntry ); } - if (delimitor) + if ( newValEntry->GetElement() == 0x0000 ) // if on group length { - if(!used) - delete newDocEntry; - break; + if ( newValEntry->GetGroup()%2 != 0 ) // if Shadow Group + { + if ( LoadMode & LD_NOSHADOW ) // if user asked to skip shad.gr + { + std::string strLgrGroup = newValEntry->GetValue(); + int lgrGroup; + if ( strLgrGroup != GDCM_UNFOUND) + { + lgrGroup = atoi(strLgrGroup.c_str()); + Fp->seekg(lgrGroup, std::ios::cur); + used = false; + RemoveEntry( newDocEntry ); + // bcc 5.5 is right "assigned a value that's never used" + // newDocEntry = 0; + continue; + } + } + } } - if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) + + bool delimitor = newValEntry->IsItemDelimitor(); + + if ( (delimitor) || + (!delim_mode && ((long)(Fp->tellg())-offset) >= l_max) ) { - if(!used) + if ( !used ) delete newDocEntry; break; } } - if ( ( newDocEntry->GetGroup() == 0x7fe0 ) - && ( newDocEntry->GetElement() == 0x0010 ) ) - { - std::string ts = GetTransferSyntax(); - if ( Global::GetTS()->IsRLELossless(ts) ) - { - long positionOnEntry = Fp->tellg(); - Fp->seekg( newDocEntry->GetOffset(), std::ios::beg ); - ComputeRLEInfo(); - Fp->seekg( positionOnEntry, std::ios::beg ); - } - else if ( Global::GetTS()->IsJPEG(ts) ) - { - long positionOnEntry = Fp->tellg(); - Fp->seekg( newDocEntry->GetOffset(), std::ios::beg ); - ComputeJPEGFragmentInfo(); - Fp->seekg( positionOnEntry, std::ios::beg ); - } - } - // Just to make sure we are at the beginning of next entry. SkipToNextDocEntry(newDocEntry); } else { - // VR = "SQ" - unsigned long l = newDocEntry->GetReadLength(); - if ( l != 0 ) // don't mess the delim_mode for zero-length sequence + /////////////////////// SeqEntry : VR = "SQ" + + unsigned long l = newDocEntry->GetReadLength(); + if ( l != 0 ) // don't mess the delim_mode for 'zero-length sequence' { if ( l == 0xffffffff ) { - delim_mode = true; + delim_mode_intern = true; } else { - delim_mode = false; + delim_mode_intern = false; } } - // no other way to create it ... - newSeqEntry->SetDelimitorMode( delim_mode ); + + if ( (LoadMode & LD_NOSHADOWSEQ) && ! delim_mode_intern ) + { + // User asked to skip SeQuences *only* if they belong to Shadow Group + if ( newDocEntry->GetGroup()%2 != 0 ) + { + Fp->seekg( l, std::ios::cur); + used = false; + continue; + } + } + if ( (LoadMode & LD_NOSEQ) && ! delim_mode_intern ) + { + // User asked to skip *any* SeQuence + Fp->seekg( l, std::ios::cur); + used = false; + continue; + } + // delay the dynamic cast as late as possible + newSeqEntry = dynamic_cast(newDocEntry); + + // no other way to create the Delimitor ... + newSeqEntry->SetDelimitorMode( delim_mode_intern ); // At the top of the hierarchy, stands a Document. When "set" // is a Document, then we are building the first depth level. // Hence the SeqEntry we are building simply has a depth // level of one: - if (Document *dummy = dynamic_cast< Document* > ( set ) ) +// SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ); + if ( set == this ) // ( dynamic_cast< Document* > ( set ) ) { - (void)dummy; newSeqEntry->SetDepthLevel( 1 ); - newSeqEntry->SetKey( newSeqEntry->GetKey() ); + // newSeqEntry->SetKey( newSeqEntry->GetKey() ); } // But when "set" is already a SQItem, we are building a nested // sequence, and hence the depth level of the new SeqEntry // we are building, is one level deeper: - if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + + // time waste hunting + else if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) { newSeqEntry->SetDepthLevel( parentSQItem->GetDepthLevel() + 1 ); - newSeqEntry->SetKey( parentSQItem->GetBaseTagKey() - + newSeqEntry->GetKey() ); + + // newSeqEntry->SetKey( parentSQItem->GetBaseTagKey() + // + newSeqEntry->GetKey() ); } if ( l != 0 ) { // Don't try to parse zero-length sequences ParseSQ( newSeqEntry, newDocEntry->GetOffset(), - l, delim_mode); + l, delim_mode_intern); } - set->AddEntry( newSeqEntry ); - if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) + if ( !set->AddEntry( newSeqEntry ) ) { - break; + gdcmWarningMacro( "in ParseDES : cannot add a SeqEntry " + << newSeqEntry->GetKey() + << " (at offset : " + << newSeqEntry->GetOffset() << " )" ); + used = false; } - } + + if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) + { + if ( !used ) + delete newDocEntry; + break; + } + } // end SeqEntry : VR = "SQ" - if(!used) + if ( !used ) + { delete newDocEntry; - } + } + first = false; + } // end While } /** @@ -1266,20 +1236,24 @@ void Document::ParseSQ( SeqEntry *seqEntry, { int SQItemNumber = 0; bool dlm_mod; + long offsetStartCurrentSQItem = offset; while (true) { - DocEntry *newDocEntry = ReadNextDocEntry(); + // the first time, we read the fff0,e000 of the first SQItem + DocEntry *newDocEntry = ReadNextDocEntry(); + if ( !newDocEntry ) { // FIXME Should warn user + gdcmWarningMacro("in ParseSQ : should never get here!"); break; } - if( delim_mode ) + if ( delim_mode ) { if ( newDocEntry->IsSequenceDelimitor() ) { - seqEntry->SetSequenceDelimitationItem( newDocEntry ); + seqEntry->SetDelimitationItem( newDocEntry ); break; } } @@ -1288,14 +1262,16 @@ void Document::ParseSQ( SeqEntry *seqEntry, delete newDocEntry; break; } - + // create the current SQItem SQItem *itemSQ = new SQItem( seqEntry->GetDepthLevel() ); +/* std::ostringstream newBase; newBase << seqEntry->GetKey() << "/" << SQItemNumber << "#"; itemSQ->SetBaseTagKey( newBase.str() ); +*/ unsigned int l = newDocEntry->GetReadLength(); if ( l == 0xffffffff ) @@ -1306,11 +1282,19 @@ void Document::ParseSQ( SeqEntry *seqEntry, { dlm_mod = false; } - - ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod); + + // Let's try :------------ + // remove fff0,e000, created out of the SQItem delete newDocEntry; - - seqEntry->AddEntry( itemSQ, SQItemNumber ); + Fp->seekg(offsetStartCurrentSQItem, std::ios::beg); + // fill up the current SQItem, starting at the beginning of fff0,e000 + + ParseDES(itemSQ, offsetStartCurrentSQItem, l+8, dlm_mod); + + offsetStartCurrentSQItem = Fp->tellg(); + // end try ----------------- + + seqEntry->AddSQItem( itemSQ, SQItemNumber ); SQItemNumber++; if ( !delim_mode && ((long)(Fp->tellg())-offset ) >= l_max ) { @@ -1320,11 +1304,46 @@ void Document::ParseSQ( SeqEntry *seqEntry, } /** - * \brief Loads the element content if its length doesn't exceed - * the value specified with Document::SetMaxSizeLoadEntry() - * @param entry Header Entry (Dicom Element) to be dealt with + * \brief When a private Sequence + Implicit VR is encountered + * we cannot guess it's a Sequence till we find the first + * Item Starter. We then backtrack to do the job. + * @param docEntry Item Starter that warned us + */ +DocEntry *Document::Backtrack(DocEntry *docEntry) +{ + // delete the Item Starter, built erroneously out of any Sequence + // it's not yet in the HTable/chained list + delete docEntry; + + // Get all info we can from PreviousDocEntry + uint16_t group = PreviousDocEntry->GetGroup(); + uint16_t elem = PreviousDocEntry->GetElement(); + uint32_t lgt = PreviousDocEntry->GetLength(); + long offset = PreviousDocEntry->GetOffset(); + + gdcmWarningMacro( "Backtrack :" << std::hex << group + << "|" << elem + << " at offset " << offset ); + RemoveEntry( PreviousDocEntry ); + + // forge the Seq Entry + DocEntry *newEntry = NewSeqEntry(group, elem); + newEntry->SetLength(lgt); + newEntry->SetOffset(offset); + + // Move back to the beginning of the Sequence + Fp->seekg( 0, std::ios::beg); + Fp->seekg(offset, std::ios::cur); + +return newEntry; +} + +/** + * \brief Loads (or not) the element content depending if its length exceeds + * or not the value specified with Document::SetMaxSizeLoadEntry() + * @param entry Header Entry (Dicom Element) to be dealt with */ -void Document::LoadDocEntry(DocEntry *entry) +void Document::LoadDocEntry(DocEntry *entry, bool forceLoad) { uint16_t group = entry->GetGroup(); std::string vr = entry->GetVR(); @@ -1336,7 +1355,7 @@ void Document::LoadDocEntry(DocEntry *entry) // (fffe e000) tells us an Element is beginning // (fffe e00d) tells us an Element just ended // (fffe e0dd) tells us the current SeQuence just ended - if( group == 0xfffe ) + if ( group == 0xfffe ) { // NO more value field for SQ ! return; @@ -1350,42 +1369,45 @@ void Document::LoadDocEntry(DocEntry *entry) } // The elements whose length is bigger than the specified upper bound - // are not loaded. Instead we leave a short notice of the offset of + // are not loaded. Instead we leave a short notice on the offset of // the element content and it's length. std::ostringstream s; - if (length > MaxSizeLoadEntry) + + if (!forceLoad) { - if (BinEntry *binEntryPtr = dynamic_cast< BinEntry* >(entry) ) - { - //s << "gdcm::NotLoaded (BinEntry)"; - s << GDCM_NOTLOADED; - s << " Address:" << (long)entry->GetOffset(); - s << " Length:" << entry->GetLength(); - s << " x(" << std::hex << entry->GetLength() << ")"; - binEntryPtr->SetValue(s.str()); - } - // Be carefull : a BinEntry IS_A ValEntry ... - else if (ValEntry *valEntryPtr = dynamic_cast< ValEntry* >(entry) ) - { - // s << "gdcm::NotLoaded. (ValEntry)"; - s << GDCM_NOTLOADED; - s << " Address:" << (long)entry->GetOffset(); - s << " Length:" << entry->GetLength(); - s << " x(" << std::hex << entry->GetLength() << ")"; - valEntryPtr->SetValue(s.str()); - } - else + if (length > MaxSizeLoadEntry) { - // fusible - gdcmErrorMacro( "MaxSizeLoadEntry exceeded, neither a BinEntry " - << "nor a ValEntry ?! Should never print that !" ); - } + if (BinEntry *binEntryPtr = dynamic_cast< BinEntry* >(entry) ) + { + s << GDCM_NOTLOADED; + s << " Ad.:" << (long)entry->GetOffset(); + s << " x(" << std::hex << entry->GetOffset() << ")"; + s << std::dec; + s << " Lgt:" << entry->GetLength(); + s << " x(" << std::hex << entry->GetLength() << ")"; + binEntryPtr->SetValue(s.str()); + } + else if (ValEntry *valEntryPtr = dynamic_cast< ValEntry* >(entry) ) + { + s << GDCM_NOTLOADED; + s << " Address:" << (long)entry->GetOffset(); + s << " Length:" << entry->GetLength(); + s << " x(" << std::hex << entry->GetLength() << ")"; + valEntryPtr->SetValue(s.str()); + } + else + { + // fusible + gdcmErrorMacro( "MaxSizeLoadEntry exceeded, neither a BinEntry " + << "nor a ValEntry ?! Should never print that !" ); + } - // to be sure we are at the end of the value ... - Fp->seekg((long)entry->GetOffset()+(long)entry->GetLength(), - std::ios::beg); - return; + // to be sure we are at the end of the value ... + Fp->seekg((long)entry->GetOffset()+(long)entry->GetLength(), + std::ios::beg); + return; + } } // When we find a BinEntry not very much can be done : @@ -1397,13 +1419,12 @@ void Document::LoadDocEntry(DocEntry *entry) return; } - /// \todo Any compacter code suggested (?) if ( IsDocEntryAnInteger(entry) ) { uint32_t NewInt; int nbInt; // When short integer(s) are expected, read and convert the following - // n *two characters properly i.e. consider them as short integers as + // (n * 2) characters properly i.e. consider them as short integers as // opposed to strings. // Elements with Value Multiplicity > 1 // contain a set of integers (not a single one) @@ -1452,11 +1473,11 @@ void Document::LoadDocEntry(DocEntry *entry) str[length] = '\0'; //this is only useful when length is odd // Special DicomString call to properly handle \0 and even length std::string newValue; - if( length % 2 ) + if ( length % 2 ) { newValue = Util::DicomString(str, length+1); - gdcmVerboseMacro("Warning: bad length: " << length << - ",For string :" << newValue.c_str()); + gdcmWarningMacro("Warning: bad length: " << length << + " For string :" << newValue.c_str()); // Since we change the length of string update it length //entry->SetReadLength(length+1); } @@ -1470,12 +1491,17 @@ void Document::LoadDocEntry(DocEntry *entry) { if ( Fp->fail() || Fp->eof()) { - gdcmVerboseMacro("Unread element value"); + if ( Fp->fail() ) + gdcmWarningMacro("--> fail"); + + gdcmWarningMacro("Unread element value " << valEntry->GetKey() + << " lgt : " << valEntry->GetReadLength() + << " at " << std::hex << valEntry->GetOffset()); valEntry->SetValue(GDCM_UNREAD); return; } - if( vr == "UI" ) + if ( vr == "UI" ) { // Because of correspondance with the VR dic valEntry->SetValue(newValue); @@ -1487,13 +1513,14 @@ void Document::LoadDocEntry(DocEntry *entry) } else { - gdcmErrorMacro( "Should have a ValEntry, here !"); + gdcmWarningMacro("Should have a ValEntry, here ! " << valEntry->GetKey() + << " lgt : " << valEntry->GetReadLength() + << " at " << std::hex << valEntry->GetOffset()); } } - /** - * \brief Find the value Length of the passed Header Entry + * \brief Find the value Length of the passed Doc Entry * @param entry Header Entry whose length of the value shall be loaded. */ void Document::FindDocEntryLength( DocEntry *entry ) @@ -1504,7 +1531,8 @@ void Document::FindDocEntryLength( DocEntry *entry ) if ( Filetype == ExplicitVR && !entry->IsImplicitVR() ) { - if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UN" ) + if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UT" + /*|| vr == "UN"*/ ) { // The following reserved two bytes (see PS 3.5-2003, section // "7.1.2 Data element structure with explicit vr", p 27) must be @@ -1526,6 +1554,9 @@ void Document::FindDocEntryLength( DocEntry *entry ) // chance to get the pixels by deciding the element goes // until the end of the file. Hence we artificially fix the // the length and proceed. + gdcmWarningMacro( " Computing the length failed for " << + entry->GetKey() <<" in " <tellg(); Fp->seekg(0L,std::ios::end); @@ -1546,73 +1577,6 @@ void Document::FindDocEntryLength( DocEntry *entry ) // Length is encoded on 2 bytes. length16 = ReadInt16(); - - // FIXME : This heuristic supposes that the first group following - // group 0002 *has* and element 0000. - // BUT ... Element 0000 is optionnal :-( - - - // Fixed using : HandleOutOfGroup0002() - // (first hereafter strategy ...) - - // We can tell the current file is encoded in big endian (like - // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag - // and it's value is the one of the encoding of a big endian file. - // In order to deal with such big endian encoded files, we have - // (at least) two strategies: - // * when we load the "Transfer Syntax" tag with value of big endian - // encoding, we raise the proper flags. Then we wait for the end - // of the META group (0x0002) among which is "Transfer Syntax", - // before switching the swap code to big endian. We have to postpone - // the switching of the swap code since the META group is fully encoded - // in little endian, and big endian coding only starts at the next - // group. The corresponding code can be hard to analyse and adds - // many additional unnecessary tests for regular tags. - // * the second strategy consists in waiting for trouble, that shall - // appear when we find the first group with big endian encoding. This - // is easy to detect since the length of a "Group Length" tag (the - // ones with zero as element number) has to be of 4 (0x0004). When we - // encounter 1024 (0x0400) chances are the encoding changed and we - // found a group with big endian encoding. - //---> Unfortunately, element 0000 is optional. - //---> This will not work when missing! - // We shall use this second strategy. In order to make sure that we - // can interpret the presence of an apparently big endian encoded - // length of a "Group Length" without committing a big mistake, we - // add an additional check: we look in the already parsed elements - // for the presence of a "Transfer Syntax" whose value has to be "big - // endian encoding". When this is the case, chances are we have got our - // hands on a big endian encoded file: we switch the swap code to - // big endian and proceed... - - // - // if ( element == 0x0000 && length16 == 0x0400 ) - // { - // std::string ts = GetTransferSyntax(); - // if ( Global::GetTS()->GetSpecialTransferSyntax(ts) - // != TS::ExplicitVRBigEndian ) - // { - // throw FormatError( "Document::FindDocEntryLength()", - // " not explicit VR." ); - // return; - // } - // length16 = 4; - // SwitchByteSwapCode(); - - // Restore the unproperly loaded values i.e. the group, the element - // and the dictionary entry depending on them. -// uint16_t correctGroup = SwapShort( entry->GetGroup() ); -// uint16_t correctElem = SwapShort( entry->GetElement() ); -// DictEntry *newTag = GetDictEntry( correctGroup, correctElem ); if ( !newTag ) -// { - // This correct tag is not in the dictionary. Create a new one. -// newTag = NewVirtualDictEntry(correctGroup, correctElem); -// } - // FIXME this can create a memory leaks on the old entry that be - // left unreferenced. -// entry->SetDictEntry( newTag ); -// } - // 0xffff means that we deal with 'No Length' Sequence // or 'No Length' SQItem @@ -1631,12 +1595,74 @@ void Document::FindDocEntryLength( DocEntry *entry ) // on Data elements "Implicit and Explicit VR Data Elements shall // not coexist in a Data Set and Data Sets nested within it".] // Length is on 4 bytes. + + // Well ... group 0002 is always coded in 'Explicit VR Litle Endian' + // even if Transfer Syntax is 'Implicit VR ...' FixDocEntryFoundLength( entry, ReadInt32() ); return; } } +/** + * \brief Find the Length till the next sequence delimiter + * \warning NOT end user intended method ! + * @return + */ +uint32_t Document::FindDocEntryLengthOBOrOW() + throw( FormatUnexpected ) +{ + // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data. + long positionOnEntry = Fp->tellg(); + bool foundSequenceDelimiter = false; + uint32_t totalLength = 0; + + while ( !foundSequenceDelimiter ) + { + uint16_t group; + uint16_t elem; + try + { + group = ReadInt16(); + elem = ReadInt16(); + } + catch ( FormatError ) + { + throw FormatError("Unexpected end of file encountered during ", + "Document::FindDocEntryLengthOBOrOW()"); + } + // We have to decount the group and element we just read + totalLength += 4; + if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) ) + { + long filePosition = Fp->tellg(); + gdcmWarningMacro( + "Neither an Item tag nor a Sequence delimiter tag on :" + << std::hex << group << " , " << elem + << ") -before- position x(" << filePosition << ")" ); + + Fp->seekg(positionOnEntry, std::ios::beg); + throw FormatUnexpected( + "Neither an Item tag nor a Sequence delimiter tag."); + } + if ( elem == 0xe0dd ) + { + foundSequenceDelimiter = true; + } + uint32_t itemLength = ReadInt32(); + // We add 4 bytes since we just read the ItemLength with ReadInt32 + totalLength += itemLength + 4; + SkipBytes(itemLength); + + if ( foundSequenceDelimiter ) + { + break; + } + } + Fp->seekg( positionOnEntry, std::ios::beg); + return totalLength; +} + /** * \brief Find the Value Representation of the current Dicom Element. * @return Value Representation of the current Entry @@ -1659,7 +1685,7 @@ std::string Document::FindDocEntryVR() Fp->read (vr, (size_t)2); vr[2] = 0; - if( !CheckDocEntryVR(vr) ) + if ( !CheckDocEntryVR(vr) ) { Fp->seekg(positionOnEntry, std::ios::beg); return GDCM_UNKNOWN; @@ -1672,14 +1698,11 @@ std::string Document::FindDocEntryVR() * and the taken VR. If they are different, the header entry is * updated with the new VR. * @param vr Dicom Value Representation - * @return false if the VR is incorrect of if the VR isn't referenced + * @return false if the VR is incorrect or if the VR isn't referenced * otherwise, it returns true */ bool Document::CheckDocEntryVR(VRKey vr) { - // CLEANME searching the dicom_vr at each occurence is expensive. - // PostPone this test in an optional integrity check at the end - // of parsing or only in debug mode. if ( !Global::GetVR()->IsValidVR(vr) ) return false; @@ -1710,14 +1733,14 @@ std::string Document::GetDocEntryValue(DocEntry *entry) // Elements with Value Multiplicity > 1 // contain a set of short integers (not a single one) - if( vr == "US" || vr == "SS" ) + if ( vr == "US" || vr == "SS" ) { uint16_t newInt16; nbInt = length / 2; for (int i=0; i < nbInt; i++) { - if( i != 0 ) + if ( i != 0 ) { s << '\\'; } @@ -1733,14 +1756,14 @@ std::string Document::GetDocEntryValue(DocEntry *entry) // as usual for standard multivaluated filels // Elements with Value Multiplicity > 1 // contain a set of integers (not a single one) - else if( vr == "UL" || vr == "SL" ) + else if ( vr == "UL" || vr == "SL" ) { uint32_t newInt32; nbInt = length / 4; for (int i=0; i < nbInt; i++) { - if( i != 0) + if ( i != 0) { s << '\\'; } @@ -1757,7 +1780,6 @@ std::string Document::GetDocEntryValue(DocEntry *entry) #endif //GDCM_NO_ANSI_STRING_STREAM return s.str(); } - return ((ValEntry *)entry)->GetValue(); } @@ -1828,14 +1850,21 @@ void Document::SkipDocEntry(DocEntry *entry) } /** - * \brief Skips to the begining of the next Header Entry + * \brief Skips to the beginning of the next Header Entry * \warning NOT end user intended method ! * @param currentDocEntry entry to skip */ void Document::SkipToNextDocEntry(DocEntry *currentDocEntry) { - Fp->seekg((long)(currentDocEntry->GetOffset()), std::ios::beg); - Fp->seekg( (long)(currentDocEntry->GetReadLength()),std::ios::cur); + int l = currentDocEntry->GetReadLength(); + if ( l == -1 ) // length = 0xffff shouldn't appear here ... + // ... but PMS imagers happen ! + return; + Fp->seekg((long)(currentDocEntry->GetOffset()), std::ios::beg); + if (currentDocEntry->GetGroup() != 0xfffe) // for fffe pb + { + Fp->seekg( (long)(currentDocEntry->GetReadLength()),std::ios::cur); + } } /** @@ -1849,242 +1878,111 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, uint32_t foundLength) { entry->SetReadLength( foundLength ); // will be updated only if a bug is found - if ( foundLength == 0xffffffff) - { - foundLength = 0; - } - - uint16_t gr = entry->GetGroup(); - uint16_t elem = entry->GetElement(); - - if ( foundLength % 2) - { - gdcmVerboseMacro( "Warning : Tag with uneven length " << foundLength - << " in x(" << std::hex << gr << "," << elem <<")"); - } - - //////// Fix for some naughty General Electric images. - // Allthough not recent many such GE corrupted images are still present - // on Creatis hard disks. Hence this fix shall remain when such images - // are no longer in use (we are talking a few years, here)... - // Note: XMedCom probably uses such a trick since it is able to read - // those pesky GE images ... - if ( foundLength == 13) - { - // Only happens for this length ! - if ( gr != 0x0008 || ( elem != 0x0070 && elem != 0x0080 ) ) - { - foundLength = 10; - entry->SetReadLength(10); /// \todo a bug is to be fixed !? - } - } - - //////// Fix for some brain-dead 'Leonardo' Siemens images. - // Occurence of such images is quite low (unless one leaves close to a - // 'Leonardo' source. Hence, one might consider commenting out the - // following fix on efficiency reasons. - else if ( gr == 0x0009 && ( elem == 0x1113 || elem == 0x1114 ) ) - { - foundLength = 4; - entry->SetReadLength(4); /// \todo a bug is to be fixed !? - } - - else if ( entry->GetVR() == "SQ" ) - { - foundLength = 0; // ReadLength is unchanged - } - - //////// We encountered a 'delimiter' element i.e. a tag of the form - // "fffe|xxxx" which is just a marker. Delimiters length should not be - // taken into account. - else if( gr == 0xfffe ) - { - // According to the norm, fffe|0000 shouldn't exist. BUT the Philips - // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to - // causes extra troubles... - if( entry->GetElement() != 0x0000 ) - { - foundLength = 0; - } - } - - entry->SetLength(foundLength); -} - -/** - * \brief Apply some heuristics to predict whether the considered - * element value contains/represents an integer or not. - * @param entry The element value on which to apply the predicate. - * @return The result of the heuristical predicate. - */ -bool Document::IsDocEntryAnInteger(DocEntry *entry) -{ - uint16_t elem = entry->GetElement(); - uint16_t group = entry->GetGroup(); - const std::string &vr = entry->GetVR(); - uint32_t length = entry->GetLength(); - - // When we have some semantics on the element we just read, and if we - // a priori know we are dealing with an integer, then we shall be - // able to swap it's element value properly. - if ( elem == 0 ) // This is the group length of the group - { - if ( length == 4 ) - { - return true; - } - else - { - // Allthough this should never happen, still some images have a - // corrupted group length [e.g. have a glance at offset x(8336) of - // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm]. - // Since for dicom compliant and well behaved headers, the present - // test is useless (and might even look a bit paranoid), when we - // encounter such an ill-formed image, we simply display a warning - // message and proceed on parsing (while crossing fingers). - long filePosition = Fp->tellg(); - gdcmVerboseMacro( "Erroneous Group Length element length on : (" - << std::hex << group << " , " << elem - << ") -before- position x(" << filePosition << ")" - << "lgt : " << length ); - } - } - - if ( vr == "UL" || vr == "US" || vr == "SL" || vr == "SS" ) - { - return true; - } - return false; -} - -/** - * \brief Find the Length till the next sequence delimiter - * \warning NOT end user intended method ! - * @return - */ - -uint32_t Document::FindDocEntryLengthOBOrOW() - throw( FormatUnexpected ) -{ - // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data. - long positionOnEntry = Fp->tellg(); - bool foundSequenceDelimiter = false; - uint32_t totalLength = 0; - - while ( !foundSequenceDelimiter ) - { - uint16_t group; - uint16_t elem; - try - { - group = ReadInt16(); - elem = ReadInt16(); - } - catch ( FormatError ) - { - throw FormatError("Unexpected end of file encountered during ", - "Document::FindDocEntryLengthOBOrOW()"); - } - - // We have to decount the group and element we just read - totalLength += 4; - - if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) ) - { - long filePosition = Fp->tellg(); - gdcmVerboseMacro( "Neither an Item tag nor a Sequence delimiter tag on :" - << std::hex << group << " , " << elem - << ") -before- position x(" << filePosition << ")" ); - - Fp->seekg(positionOnEntry, std::ios::beg); - throw FormatUnexpected( "Neither an Item tag nor a Sequence delimiter tag."); - } - - if ( elem == 0xe0dd ) - { - foundSequenceDelimiter = true; - } - - uint32_t itemLength = ReadInt32(); - // We add 4 bytes since we just read the ItemLength with ReadInt32 - totalLength += itemLength + 4; - SkipBytes(itemLength); - - if ( foundSequenceDelimiter ) - { - break; - } - } - Fp->seekg( positionOnEntry, std::ios::beg); - return totalLength; -} - -/** - * \brief Reads a supposed to be 16 Bits integer - * (swaps it depending on processor endianity) - * @return read value - */ -uint16_t Document::ReadInt16() - throw( FormatError ) -{ - uint16_t g; - Fp->read ((char*)&g, (size_t)2); - if ( Fp->fail() ) + if ( foundLength == 0xffffffff) { - throw FormatError( "Document::ReadInt16()", " file error." ); + foundLength = 0; } - if( Fp->eof() ) + + uint16_t gr = entry->GetGroup(); + uint16_t elem = entry->GetElement(); + + if ( foundLength % 2) { - throw FormatError( "Document::ReadInt16()", "EOF." ); + gdcmWarningMacro( "Warning : Tag with uneven length " << foundLength + << " in x(" << std::hex << gr << "," << elem <<")"); } - g = SwapShort(g); - return g; -} - -/** - * \brief Reads a supposed to be 32 Bits integer - * (swaps it depending on processor endianity) - * @return read value - */ -uint32_t Document::ReadInt32() - throw( FormatError ) -{ - uint32_t g; - Fp->read ((char*)&g, (size_t)4); - if ( Fp->fail() ) + + //////// Fix for some naughty General Electric images. + // Allthough not recent many such GE corrupted images are still present + // on Creatis hard disks. Hence this fix shall remain when such images + // are no longer in use (we are talking a few years, here)... + // Note: XMedCon probably uses such a trick since it is able to read + // those pesky GE images ... + if ( foundLength == 13) { - throw FormatError( "Document::ReadInt32()", " file error." ); + // Only happens for this length ! + if ( gr != 0x0008 || ( elem != 0x0070 && elem != 0x0080 ) ) + { + foundLength = 10; + entry->SetReadLength(10); // a bug is to be fixed !? + } } - if( Fp->eof() ) + + //////// Fix for some brain-dead 'Leonardo' Siemens images. + // Occurence of such images is quite low (unless one leaves close to a + // 'Leonardo' source. Hence, one might consider commenting out the + // following fix on efficiency reasons. + else if ( gr == 0x0009 && ( elem == 0x1113 || elem == 0x1114 ) ) { - throw FormatError( "Document::ReadInt32()", "EOF." ); - } - g = SwapLong(g); - return g; + foundLength = 4; + entry->SetReadLength(4); // a bug is to be fixed !? + } + + else if ( entry->GetVR() == "SQ" ) + { + foundLength = 0; // ReadLength is unchanged + } + + //////// We encountered a 'delimiter' element i.e. a tag of the form + // "fffe|xxxx" which is just a marker. Delimiters length should not be + // taken into account. + else if ( gr == 0xfffe ) + { + // According to the norm, fffe|0000 shouldn't exist. BUT the Philips + // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to + // causes extra troubles... + if ( entry->GetElement() != 0x0000 ) + { + foundLength = 0; + } + } + entry->SetLength(foundLength); } /** - * \brief skips bytes inside the source file - * \warning NOT end user intended method ! - * @return + * \brief Apply some heuristics to predict whether the considered + * element value contains/represents an integer or not. + * @param entry The element value on which to apply the predicate. + * @return The result of the heuristical predicate. */ -void Document::SkipBytes(uint32_t nBytes) +bool Document::IsDocEntryAnInteger(DocEntry *entry) { - //FIXME don't dump the returned value - Fp->seekg((long)nBytes, std::ios::cur); -} + uint16_t elem = entry->GetElement(); + uint16_t group = entry->GetGroup(); + const std::string &vr = entry->GetVR(); + uint32_t length = entry->GetLength(); -/** - * \brief Loads all the needed Dictionaries - * \warning NOT end user intended method ! - */ -void Document::Initialise() -{ - RefPubDict = Global::GetDicts()->GetDefaultPubDict(); - RefShaDict = NULL; - RLEInfo = new RLEFramesInfo; - JPEGInfo = new JPEGFragmentsInfo; - Filetype = Unknown; + // When we have some semantics on the element we just read, and if we + // a priori know we are dealing with an integer, then we shall be + // able to swap it's element value properly. + if ( elem == 0 ) // This is the group length of the group + { + if ( length == 4 ) + { + return true; + } + else + { + // Allthough this should never happen, still some images have a + // corrupted group length [e.g. have a glance at offset x(8336) of + // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm]. + // Since for dicom compliant and well behaved headers, the present + // test is useless (and might even look a bit paranoid), when we + // encounter such an ill-formed image, we simply display a warning + // message and proceed on parsing (while crossing fingers). + long filePosition = Fp->tellg(); + gdcmWarningMacro( "Erroneous Group Length element length on : (" + << std::hex << group << " , " << elem + << ") -before- position x(" << filePosition << ")" + << "lgt : " << length ); + } + } + + if ( vr == "UL" || vr == "US" || vr == "SL" || vr == "SS" ) + { + return true; + } + return false; } /** @@ -2096,14 +1994,7 @@ void Document::Initialise() * true when we hope ours assuptions are OK */ bool Document::CheckSwap() -{ - // The only guaranted way of finding the swap code is to find a - // group tag since we know it's length has to be of four bytes i.e. - // 0x00000004. Finding the swap code in then straigthforward. Trouble - // occurs when we can't find such group... - - uint32_t x = 4; // x : for ntohs - bool net2host; // true when HostByteOrder is the same as NetworkByteOrder +{ uint32_t s32; uint16_t s16; @@ -2111,23 +2002,16 @@ bool Document::CheckSwap() // First, compare HostByteOrder and NetworkByteOrder in order to // determine if we shall need to swap bytes (i.e. the Endian type). - if ( x == ntohs(x) ) - { - net2host = true; - } - else - { - net2host = false; - } + bool net2host = Util::IsCurrentProcessorBigEndian(); // The easiest case is the one of a 'true' DICOM header, we just have // to look for the string "DICM" inside the file preamble. Fp->read(deb, 256); char *entCur = deb + 128; - if( memcmp(entCur, "DICM", (size_t)4) == 0 ) + if ( memcmp(entCur, "DICM", (size_t)4) == 0 ) { - gdcmVerboseMacro( "Looks like DICOM Version3 (preamble + DCM)" ); + gdcmWarningMacro( "Looks like DICOM Version3 (preamble + DCM)" ); // Group 0002 should always be VR, and the first element 0000 // Let's be carefull (so many wrong headers ...) @@ -2149,51 +2033,88 @@ bool Document::CheckSwap() // even if elem 0002,0010 (Transfer Syntax) tells us the file is // *Implicit* VR (see former 'gdcmData/icone.dcm') - if( memcmp(entCur, "UL", (size_t)2) == 0 || - memcmp(entCur, "OB", (size_t)2) == 0 || - memcmp(entCur, "UI", (size_t)2) == 0 || - memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later - // when Write DCM *adds* + if ( memcmp(entCur, "UL", (size_t)2) == 0 || + memcmp(entCur, "OB", (size_t)2) == 0 || + memcmp(entCur, "UI", (size_t)2) == 0 || + memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later + // when Write DCM *adds* // FIXME // Use Document::dicom_vr to test all the possibilities // instead of just checking for UL, OB and UI !? group 0000 { Filetype = ExplicitVR; - gdcmVerboseMacro( "Group 0002 : Explicit Value Representation"); + gdcmWarningMacro( "Group 0002 : Explicit Value Representation"); } else { Filetype = ImplicitVR; - gdcmVerboseMacro( "Group 0002 :Not an explicit Value Representation;" + gdcmWarningMacro( "Group 0002 :Not an explicit Value Representation;" << "Looks like a bugged Header!"); } if ( net2host ) { SwapCode = 4321; - gdcmVerboseMacro( "HostByteOrder != NetworkByteOrder"); + gdcmWarningMacro( "HostByteOrder != NetworkByteOrder"); } else { SwapCode = 1234; - gdcmVerboseMacro( "HostByteOrder = NetworkByteOrder"); + gdcmWarningMacro( "HostByteOrder = NetworkByteOrder"); } // Position the file position indicator at first tag // (i.e. after the file preamble and the "DICM" string). - Fp->seekg(0, std::ios::beg); + + Fp->seekg(0, std::ios::beg); // FIXME : Is it usefull? + Fp->seekg ( 132L, std::ios::beg); return true; - } // End of DicomV3 + } // ------------------------------- End of DicomV3 ---------------- // Alas, this is not a DicomV3 file and whatever happens there is no file // preamble. We can reset the file position indicator to where the data // is (i.e. the beginning of the file). - gdcmVerboseMacro( "Not a DICOM Version3 file"); + + gdcmWarningMacro( "Not a Kosher DICOM Version3 file (no preamble)"); + Fp->seekg(0, std::ios::beg); + // Let's check 'No Preamble Dicom File' : + // Should start with group 0x0002 + // and be Explicit Value Representation + + s16 = *((uint16_t *)(deb)); + SwapCode = 0; + switch ( s16 ) + { + case 0x0002 : + SwapCode = 1234; + entCur = deb + 4; + break; + case 0x0200 : + SwapCode = 4321; + entCur = deb + 6; + } + + if ( SwapCode != 0 ) + { + if ( memcmp(entCur, "UL", (size_t)2) == 0 || + memcmp(entCur, "OB", (size_t)2) == 0 || + memcmp(entCur, "UI", (size_t)2) == 0 || + memcmp(entCur, "SH", (size_t)2) == 0 || + memcmp(entCur, "AE", (size_t)2) == 0 || + memcmp(entCur, "OB", (size_t)2) == 0 ) + { + Filetype = ExplicitVR; + gdcmWarningMacro( "Group 0002 : Explicit Value Representation"); + return true; + } + } +// ------------------------------- End of 'No Preamble' DicomV3 ------------- + // Our next best chance would be to be considering a 'clean' ACR/NEMA file. - // By clean we mean that the length of the first tag is written down. + // By clean we mean that the length of the first group is written down. // If this is the case and since the length of the first group HAS to be // four (bytes), then determining the proper swap code is straightforward. @@ -2202,7 +2123,6 @@ bool Document::CheckSwap() // representation of a 32 bits integer. Hence the following dirty // trick : s32 = *((uint32_t *)(entCur)); - switch( s32 ) { case 0x00040000 : @@ -2264,21 +2184,20 @@ bool Document::CheckSwap() Filetype = ACR; return true; default : - gdcmVerboseMacro( "ACR/NEMA unfound swap info (Really hopeless !)"); + gdcmWarningMacro("ACR/NEMA unfound swap info (Really hopeless !)"); Filetype = Unknown; return false; } } } - - /** * \brief Change the Byte Swap code. */ void Document::SwitchByteSwapCode() { - gdcmVerboseMacro( "Switching Byte Swap code from "<< SwapCode); + gdcmWarningMacro( "Switching Byte Swap code from "<< SwapCode + << " at :" <tellg() ); if ( SwapCode == 1234 ) { SwapCode = 4321; @@ -2293,148 +2212,34 @@ void Document::SwitchByteSwapCode() } else if ( SwapCode == 2143 ) { - SwapCode = 3412; - } -} - -/** - * \brief during parsing, Header Elements too long are not loaded in memory - * @param newSize - */ -void Document::SetMaxSizeLoadEntry(long newSize) -{ - if ( newSize < 0 ) - { - return; - } - if ((uint32_t)newSize >= (uint32_t)0xffffffff ) - { - MaxSizeLoadEntry = 0xffffffff; - return; - } - MaxSizeLoadEntry = newSize; -} - - -/** - * \brief Header Elements too long will not be printed - * \todo See comments of \ref Document::MAX_SIZE_PRINT_ELEMENT_VALUE - * @param newSize - */ -void Document::SetMaxSizePrintEntry(long newSize) -{ - //DOH !! This is exactly SetMaxSizeLoadEntry FIXME FIXME - if ( newSize < 0 ) - { - return; - } - if ((uint32_t)newSize >= (uint32_t)0xffffffff ) - { - MaxSizePrintEntry = 0xffffffff; - return; - } - MaxSizePrintEntry = newSize; -} - - - -/** - * \brief Handle broken private tag from Philips NTSCAN - * where the endianess is being switch to BigEndian for no - * apparent reason - * @return no return - */ -void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem) -{ - // Endian reversion. Some files contain groups of tags with reversed endianess. - static int reversedEndian = 0; - // try to fix endian switching in the middle of headers - if ((group == 0xfeff) && (elem == 0x00e0)) - { - // start endian swap mark for group found - reversedEndian++; - SwitchByteSwapCode(); - // fix the tag - group = 0xfffe; - elem = 0xe000; - } - else if (group == 0xfffe && elem == 0xe00d && reversedEndian) - { - // end of reversed endian group - reversedEndian--; - SwitchByteSwapCode(); - } -} - -/** - * \brief Accesses the info from 0002,0010 : Transfer Syntax and TS - * else 1. - * @return The full Transfer Syntax Name (as opposed to Transfer Syntax UID) - */ -std::string Document::GetTransferSyntaxName() -{ - // use the TS (TS : Transfer Syntax) - std::string transferSyntax = GetEntry(0x0002,0x0010); - - if ( (transferSyntax.find(GDCM_NOTLOADED) < transferSyntax.length()) ) - { - gdcmErrorMacro( "Transfer Syntax not loaded. " << std::endl - << "Better you increase MAX_SIZE_LOAD_ELEMENT_VALUE" ); - return "Uncompressed ACR-NEMA"; - } - if ( transferSyntax == GDCM_UNFOUND ) - { - gdcmVerboseMacro( "Unfound Transfer Syntax (0002,0010)"); - return "Uncompressed ACR-NEMA"; - } - - // we do it only when we need it - const TSKey &tsName = Global::GetTS()->GetValue( transferSyntax ); - - // Global::GetTS() is a global static you shall never try to delete it! - return tsName; -} - -/** - * \brief Group 0002 is always coded Little Endian - * whatever Transfer Syntax is - * @return no return - */ -void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) -{ - // Endian reversion. Some files contain groups of tags with reversed endianess. - if ( !Group0002Parsed && group != 0x0002) - { - Group0002Parsed = true; - // we just came out of group 0002 - // if Transfer syntax is Big Endian we have to change CheckSwap - - std::string ts = GetTransferSyntax(); - if ( !Global::GetTS()->IsTransferSyntax(ts) ) - { - gdcmVerboseMacro("True DICOM File, with NO Tansfer Syntax: " << ts ); - return; - } - - // FIXME Strangely, this works with - //'Implicit VR Transfer Syntax (GE Private) - if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ExplicitVRBigEndian ) - { - gdcmVerboseMacro("Transfer Syntax Name = [" - << GetTransferSyntaxName() << "]" ); - SwitchByteSwapCode(); - group = SwapShort(group); - elem = SwapShort(elem); - } + SwapCode = 3412; + } +} + +/** + * \brief during parsing, Header Elements too long are not loaded in memory + * @param newSize new size + */ +void Document::SetMaxSizeLoadEntry(long newSize) +{ + if ( newSize < 0 ) + { + return; + } + if ((uint32_t)newSize >= (uint32_t)0xffffffff ) + { + MaxSizeLoadEntry = 0xffffffff; + return; } + MaxSizeLoadEntry = newSize; } /** - * \brief Read the next tag but WITHOUT loading it's value + * \brief Read the next tag WITHOUT loading it's value * (read the 'Group Number', the 'Element Number', * gets the Dict Entry * gets the VR, gets the length, gets the offset value) - * @return On succes the newly created DocEntry, NULL on failure. + * @return On succes : the newly created DocEntry, NULL on failure. */ DocEntry *Document::ReadNextDocEntry() { @@ -2446,51 +2251,66 @@ DocEntry *Document::ReadNextDocEntry() group = ReadInt16(); elem = ReadInt16(); } - catch ( FormatError e ) + catch ( FormatError ) { // We reached the EOF (or an error occured) therefore // header parsing has to be considered as finished. - //std::cout << e; return 0; } // Sometimes file contains groups of tags with reversed endianess. HandleBrokenEndian(group, elem); -// In 'true DICOM' files Group 0002 is always little endian + // In 'true DICOM' files Group 0002 is always little endian if ( HasDCMPreamble ) HandleOutOfGroup0002(group, elem); std::string vr = FindDocEntryVR(); std::string realVR = vr; - if( vr == GDCM_UNKNOWN) + if ( vr == GDCM_UNKNOWN ) { - DictEntry *dictEntry = GetDictEntry(group,elem); - if( dictEntry ) - realVR = dictEntry->GetVR(); + if ( elem == 0x0000 ) // Group Length + { + realVR = "UL"; // must be UL + } + else if (group%2 == 1 && (elem >= 0x0010 && elem <=0x00ff )) + { + // DICOM PS 3-5 7.8.1 a) states that those + // (gggg-0010->00FF where gggg is odd) attributes have to be LO + realVR = "LO"; + } + else + { + DictEntry *dictEntry = GetDictEntry(group,elem); + if ( dictEntry ) + { + realVR = dictEntry->GetVR(); + } + } } DocEntry *newEntry; - if( Global::GetVR()->IsVROfSequence(realVR) ) + if ( Global::GetVR()->IsVROfSequence(realVR) ) newEntry = NewSeqEntry(group, elem); - else if( Global::GetVR()->IsVROfStringRepresentable(realVR) ) - newEntry = NewValEntry(group, elem,vr); + else if ( Global::GetVR()->IsVROfStringRepresentable(realVR) ) + newEntry = NewValEntry(group, elem, realVR); else - newEntry = NewBinEntry(group, elem,vr); + newEntry = NewBinEntry(group, elem, realVR); - if( vr == GDCM_UNKNOWN ) + if ( vr == GDCM_UNKNOWN ) { - if( Filetype == ExplicitVR ) + if ( Filetype == ExplicitVR ) { // We thought this was explicit VR, but we end up with an // implicit VR tag. Let's backtrack. if ( newEntry->GetGroup() != 0xfffe ) { std::string msg; - msg = Util::Format("Falsely explicit vr file (%04x,%04x)\n", - newEntry->GetGroup(), newEntry->GetElement()); - gdcmVerboseMacro( msg.c_str() ); + int offset = Fp->tellg(); + msg = Util::Format("Entry (%04x,%04x) at 0x(%x) should be Explicit VR\n", + newEntry->GetGroup(), newEntry->GetElement(), offset ); + gdcmWarningMacro( msg.c_str() ); } } newEntry->SetImplicitVR(); @@ -2500,497 +2320,89 @@ DocEntry *Document::ReadNextDocEntry() { FindDocEntryLength(newEntry); } - catch ( FormatError e ) + catch ( FormatError ) { // Call it quits - //std::cout << e; delete newEntry; return 0; } newEntry->SetOffset(Fp->tellg()); - + return newEntry; } - -/** - * \brief Generate a free TagKey i.e. a TagKey that is not present - * in the TagHt dictionary. - * @param group The generated tag must belong to this group. - * @return The element of tag with given group which is fee. - */ -uint32_t Document::GenerateFreeTagKeyInGroup(uint16_t group) -{ - for (uint32_t elem = 0; elem < UINT32_MAX; elem++) - { - TagKey key = DictEntry::TranslateToKey(group, elem); - if (TagHT.count(key) == 0) - { - return elem; - } - } - return UINT32_MAX; -} - -/** - * \brief Assuming the internal file pointer \ref Document::Fp - * is placed at the beginning of a tag check whether this - * tag is (TestGroup, TestElement). - * \warning On success the internal file pointer \ref Document::Fp - * is modified to point after the tag. - * On failure (i.e. when the tag wasn't the expected tag - * (TestGroup, TestElement) the internal file pointer - * \ref Document::Fp is restored to it's original position. - * @param testGroup The expected group of the tag. - * @param testElement The expected Element of the tag. - * @return True on success, false otherwise. - */ -bool Document::ReadTag(uint16_t testGroup, uint16_t testElement) -{ - long positionOnEntry = Fp->tellg(); - long currentPosition = Fp->tellg(); // On debugging purposes - - //// Read the Item Tag group and element, and make - // sure they are what we expected: - uint16_t itemTagGroup; - uint16_t itemTagElement; - try - { - itemTagGroup = ReadInt16(); - itemTagElement = ReadInt16(); - } - catch ( FormatError e ) - { - //std::cerr << e << std::endl; - return false; - } - if ( itemTagGroup != testGroup || itemTagElement != testElement ) - { - gdcmVerboseMacro( "Wrong Item Tag found:" - << " We should have found tag (" - << std::hex << testGroup << "," << testElement << ")" << std::endl - << " but instead we encountered tag (" - << std::hex << itemTagGroup << "," << itemTagElement << ")" - << " at address: " << " 0x(" << (unsigned int)currentPosition << ")" - ) ; - Fp->seekg(positionOnEntry, std::ios::beg); - - return false; - } - return true; -} - -/** - * \brief Assuming the internal file pointer \ref Document::Fp - * is placed at the beginning of a tag (TestGroup, TestElement), - * read the length associated to the Tag. - * \warning On success the internal file pointer \ref Document::Fp - * is modified to point after the tag and it's length. - * On failure (i.e. when the tag wasn't the expected tag - * (TestGroup, TestElement) the internal file pointer - * \ref Document::Fp is restored to it's original position. - * @param testGroup The expected group of the tag. - * @param testElement The expected Element of the tag. - * @return On success returns the length associated to the tag. On failure - * returns 0. - */ -uint32_t Document::ReadTagLength(uint16_t testGroup, uint16_t testElement) -{ - long positionOnEntry = Fp->tellg(); - (void)positionOnEntry; - - if ( !ReadTag(testGroup, testElement) ) - { - return 0; - } - - //// Then read the associated Item Length - long currentPosition = Fp->tellg(); - uint32_t itemLength = ReadInt32(); - { - gdcmVerboseMacro( "Basic Item Length is: " - << itemLength << std::endl - << " at address: " << std::hex << (unsigned int)currentPosition); - } - return itemLength; -} - -/** - * \brief When parsing the Pixel Data of an encapsulated file, read - * the basic offset table (when present, and BTW dump it). - */ -void Document::ReadAndSkipEncapsulatedBasicOffsetTable() -{ - //// Read the Basic Offset Table Item Tag length... - uint32_t itemLength = ReadTagLength(0xfffe, 0xe000); - - // When present, read the basic offset table itself. - // Notes: - since the presence of this basic offset table is optional - // we can't rely on it for the implementation, and we will simply - // trash it's content (when present). - // - still, when present, we could add some further checks on the - // lengths, but we won't bother with such fuses for the time being. - if ( itemLength != 0 ) - { - char *basicOffsetTableItemValue = new char[itemLength + 1]; - Fp->read(basicOffsetTableItemValue, itemLength); - -#ifdef GDCM_DEBUG - for (unsigned int i=0; i < itemLength; i += 4 ) - { - uint32_t individualLength = str2num( &basicOffsetTableItemValue[i], - uint32_t); - gdcmVerboseMacro( "Read one length: " << - std::hex << individualLength ); - } -#endif //GDCM_DEBUG - - delete[] basicOffsetTableItemValue; - } -} - /** - * \brief Parse pixel data from disk of [multi-]fragment RLE encoding. - * Compute the RLE extra information and store it in \ref RLEInfo - * for later pixel retrieval usage. + * \brief Handle broken private tag from Philips NTSCAN + * where the endianess is being switched to BigEndian + * for no apparent reason + * @return no return */ -void Document::ComputeRLEInfo() +void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem) { - std::string ts = GetTransferSyntax(); - if ( !Global::GetTS()->IsRLELossless(ts) ) + // Endian reversion. Some files contain groups of tags with reversed endianess. + static int reversedEndian = 0; + // try to fix endian switching in the middle of headers + if ((group == 0xfeff) && (elem == 0x00e0)) { - return; - } - - // Encoded pixel data: for the time being we are only concerned with - // Jpeg or RLE Pixel data encodings. - // As stated in PS 3.5-2003, section 8.2 p44: - // "If sent in Encapsulated Format (i.e. other than the Native Format) the - // value representation OB is used". - // Hence we expect an OB value representation. Concerning OB VR, - // the section PS 3.5-2003, section A.4.c p 58-59, states: - // "For the Value Representations OB and OW, the encoding shall meet the - // following specifications depending on the Data element tag:" - // [...snip...] - // - the first item in the sequence of items before the encoded pixel - // data stream shall be basic offset table item. The basic offset table - // item value, however, is not required to be present" - - ReadAndSkipEncapsulatedBasicOffsetTable(); - - // Encapsulated RLE Compressed Images (see PS 3.5-2003, Annex G) - // Loop on the individual frame[s] and store the information - // on the RLE fragments in a RLEFramesInfo. - // Note: - when only a single frame is present, this is a - // classical image. - // - when more than one frame are present, then we are in - // the case of a multi-frame image. - long frameLength; - while ( (frameLength = ReadTagLength(0xfffe, 0xe000)) ) - { - // Parse the RLE Header and store the corresponding RLE Segment - // Offset Table information on fragments of this current Frame. - // Note that the fragment pixels themselves are not loaded - // (but just skipped). - long frameOffset = Fp->tellg(); - - uint32_t nbRleSegments = ReadInt32(); - if ( nbRleSegments > 16 ) - { - // There should be at most 15 segments (refer to RLEFrame class) - gdcmVerboseMacro( "Too many segments."); - } - - uint32_t rleSegmentOffsetTable[16]; - for( int k = 1; k <= 15; k++ ) - { - rleSegmentOffsetTable[k] = ReadInt32(); - } - - // Deduce from both the RLE Header and the frameLength the - // fragment length, and again store this info in a - // RLEFramesInfo. - long rleSegmentLength[15]; - // skipping (not reading) RLE Segments - if ( nbRleSegments > 1) - { - for(unsigned int k = 1; k <= nbRleSegments-1; k++) - { - rleSegmentLength[k] = rleSegmentOffsetTable[k+1] - - rleSegmentOffsetTable[k]; - SkipBytes(rleSegmentLength[k]); - } - } - - rleSegmentLength[nbRleSegments] = frameLength - - rleSegmentOffsetTable[nbRleSegments]; - SkipBytes(rleSegmentLength[nbRleSegments]); - - // Store the collected info - RLEFrame *newFrameInfo = new RLEFrame; - newFrameInfo->NumberFragments = nbRleSegments; - for( unsigned int uk = 1; uk <= nbRleSegments; uk++ ) - { - newFrameInfo->Offset[uk] = frameOffset + rleSegmentOffsetTable[uk]; - newFrameInfo->Length[uk] = rleSegmentLength[uk]; - } - RLEInfo->Frames.push_back( newFrameInfo ); - } - - // Make sure that at the end of the item we encounter a 'Sequence - // Delimiter Item': - if ( !ReadTag(0xfffe, 0xe0dd) ) + // start endian swap mark for group found + reversedEndian++; + SwitchByteSwapCode(); + // fix the tag + group = 0xfffe; + elem = 0xe000; + } + else if (group == 0xfffe && elem == 0xe00d && reversedEndian) { - gdcmVerboseMacro( "No sequence delimiter item at end of RLE item sequence"); + // end of reversed endian group + reversedEndian--; + SwitchByteSwapCode(); } } /** - * \brief Parse pixel data from disk of [multi-]fragment Jpeg encoding. - * Compute the jpeg extra information (fragment[s] offset[s] and - * length) and store it[them] in \ref JPEGInfo for later pixel - * retrieval usage. + * \brief Group 0002 is always coded Little Endian + * whatever Transfer Syntax is + * @return no return */ -void Document::ComputeJPEGFragmentInfo() +void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) { - // If you need to, look for comments of ComputeRLEInfo(). - std::string ts = GetTransferSyntax(); - if ( ! Global::GetTS()->IsJPEG(ts) ) - { - return; - } - - ReadAndSkipEncapsulatedBasicOffsetTable(); - - // Loop on the fragments[s] and store the parsed information in a - // JPEGInfo. - long fragmentLength; - while ( (fragmentLength = ReadTagLength(0xfffe, 0xe000)) ) - { - long fragmentOffset = Fp->tellg(); - - // Store the collected info - JPEGFragment *newFragment = new JPEGFragment; - newFragment->Offset = fragmentOffset; - newFragment->Length = fragmentLength; - JPEGInfo->Fragments.push_back( newFragment ); - - SkipBytes( fragmentLength ); - } - - // Make sure that at the end of the item we encounter a 'Sequence - // Delimiter Item': - if ( !ReadTag(0xfffe, 0xe0dd) ) - { - gdcmVerboseMacro( "No sequence delimiter item at end of JPEG item sequence"); - } -} - -/** - * \brief Walk recursively the given \ref DocEntrySet, and feed - * the given hash table (\ref TagDocEntryHT) with all the - * \ref DocEntry (Dicom entries) encountered. - * This method does the job for \ref BuildFlatHashTable. - * @param builtHT Where to collect all the \ref DocEntry encountered - * when recursively walking the given set. - * @param set The structure to be traversed (recursively). - */ -void Document::BuildFlatHashTableRecurse( TagDocEntryHT &builtHT, - DocEntrySet *set ) -{ - if (ElementSet *elementSet = dynamic_cast< ElementSet* > ( set ) ) + // Endian reversion. Some files contain groups of tags with reversed endianess. + if ( !Group0002Parsed && group != 0x0002) { - TagDocEntryHT const ¤tHT = elementSet->GetTagHT(); - for( TagDocEntryHT::const_iterator i = currentHT.begin(); - i != currentHT.end(); - ++i) - { - DocEntry *entry = i->second; - if ( SeqEntry *seqEntry = dynamic_cast(entry) ) - { - const ListSQItem& items = seqEntry->GetSQItems(); - for( ListSQItem::const_iterator item = items.begin(); - item != items.end(); - ++item) - { - BuildFlatHashTableRecurse( builtHT, *item ); - } - continue; - } - builtHT[entry->GetKey()] = entry; - } - return; - } + Group0002Parsed = true; + // we just came out of group 0002 + // if Transfer syntax is Big Endian we have to change CheckSwap - if (SQItem *SQItemSet = dynamic_cast< SQItem* > ( set ) ) - { - const ListDocEntry& currentList = SQItemSet->GetDocEntries(); - for (ListDocEntry::const_iterator i = currentList.begin(); - i != currentList.end(); - ++i) + std::string ts = GetTransferSyntax(); + if ( !Global::GetTS()->IsTransferSyntax(ts) ) { - DocEntry *entry = *i; - if ( SeqEntry *seqEntry = dynamic_cast(entry) ) - { - const ListSQItem& items = seqEntry->GetSQItems(); - for( ListSQItem::const_iterator item = items.begin(); - item != items.end(); - ++item) - { - BuildFlatHashTableRecurse( builtHT, *item ); - } - continue; - } - builtHT[entry->GetKey()] = entry; + gdcmWarningMacro("True DICOM File, with NO Tansfer Syntax: " << ts ); + return; } - } -} - -/** - * \brief Build a \ref TagDocEntryHT (i.e. a std::map<>) from the current - * Document. - * - * The structure used by a Document (through \ref ElementSet), - * in order to hold the parsed entries of a Dicom header, is a recursive - * one. This is due to the fact that the sequences (when present) - * can be nested. Additionaly, the sequence items (represented in - * gdcm as \ref SQItem) add an extra complexity to the data - * structure. Hence, a gdcm user whishing to visit all the entries of - * a Dicom header will need to dig in the gdcm internals (which - * implies exposing all the internal data structures to the API). - * In order to avoid this burden to the user, \ref BuildFlatHashTable - * recursively builds a temporary hash table, which holds all the - * Dicom entries in a flat structure (a \ref TagDocEntryHT i.e. a - * std::map<>). - * \warning Of course there is NO integrity constrain between the - * returned \ref TagDocEntryHT and the \ref ElementSet used - * to build it. Hence if the underlying \ref ElementSet is - * altered, then it is the caller responsability to invoke - * \ref BuildFlatHashTable again... - * @return The flat std::map<> we juste build. - */ -TagDocEntryHT *Document::BuildFlatHashTable() -{ - TagDocEntryHT *FlatHT = new TagDocEntryHT; - BuildFlatHashTableRecurse( *FlatHT, this ); - return FlatHT; -} + // Group 0002 is always 'Explicit ...' enven when Transfer Syntax says 'Implicit ..." - - -/** - * \brief Compares two documents, according to \ref DicomDir rules - * \warning Does NOT work with ACR-NEMA files - * \todo Find a trick to solve the pb (use RET fields ?) - * @param document - * @return true if 'smaller' - */ -bool Document::operator<(Document &document) -{ - // Patient Name - std::string s1 = GetEntry(0x0010,0x0010); - std::string s2 = document.GetEntry(0x0010,0x0010); - if(s1 < s2) - { - return true; - } - else if( s1 > s2 ) - { - return false; - } - else - { - // Patient ID - s1 = GetEntry(0x0010,0x0020); - s2 = document.GetEntry(0x0010,0x0020); - if ( s1 < s2 ) - { - return true; - } - else if ( s1 > s2 ) - { - return false; - } - else - { - // Study Instance UID - s1 = GetEntry(0x0020,0x000d); - s2 = document.GetEntry(0x0020,0x000d); - if ( s1 < s2 ) - { - return true; - } - else if( s1 > s2 ) - { - return false; - } - else + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ImplicitVRLittleEndian ) { - // Serie Instance UID - s1 = GetEntry(0x0020,0x000e); - s2 = document.GetEntry(0x0020,0x000e); - if ( s1 < s2 ) - { - return true; - } - else if( s1 > s2 ) - { - return false; - } + Filetype = ImplicitVR; } - } - } - return false; -} - - -/** - * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader - * @param filetype Type of the File to be written - */ -int Document::ComputeGroup0002Length( FileType filetype ) -{ - uint16_t gr, el; - std::string vr; - - int groupLength = 0; - bool found0002 = false; - - // for each zero-level Tag in the DCM Header - DocEntry *entry; - - Initialize(); - entry = GetNextEntry(); - while(entry) - { - gr = entry->GetGroup(); - - if (gr == 0x0002) + + // FIXME Strangely, this works with + //'Implicit VR Transfer Syntax (GE Private) + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ExplicitVRBigEndian ) { - found0002 = true; - - el = entry->GetElement(); - vr = entry->GetVR(); - - if (filetype == ExplicitVR) - { - if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") ) - { - groupLength += 4; // explicit VR AND OB, OW, SQ : 4 more bytes - } - } - groupLength += 2 + 2 + 4 + entry->GetLength(); + gdcmWarningMacro("Transfer Syntax Name = [" + << GetTransferSyntaxName() << "]" ); + SwitchByteSwapCode(); + group = SwapShort(group); + elem = SwapShort(elem); } - else if (found0002 ) - break; - - entry = GetNextEntry(); } - return groupLength; } -} // end namespace gdcm +//----------------------------------------------------------------------------- +// Print //----------------------------------------------------------------------------- +} // end namespace gdcm