X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=src%2FgdcmDocument.cxx;h=557eb4e6b6a8ee1d20e998d485cc7fc0857c3823;hb=173aecc6a03f633d2a5a3f2dbb73a464c01335c0;hp=9976959e156995dc10708b731ed4b5cd920a26ac;hpb=44a9814938a6824d3a66b932429084517da6dadf;p=gdcm.git diff --git a/src/gdcmDocument.cxx b/src/gdcmDocument.cxx index 9976959e..557eb4e6 100644 --- a/src/gdcmDocument.cxx +++ b/src/gdcmDocument.cxx @@ -3,12 +3,12 @@ Program: gdcm Module: $RCSfile: gdcmDocument.cxx,v $ Language: C++ - Date: $Date: 2004/09/13 12:10:53 $ - Version: $Revision: 1.76 $ + Date: $Date: 2005/08/30 15:34:35 $ + Version: $Revision: 1.274 $ Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de l'Image). All rights reserved. See Doc/License.txt or - http://www.creatis.insa-lyon.fr/Public/Gdcm/License.htm for details. + http://www.creatis.insa-lyon.fr/Public/Gdcm/License.html for details. This software is distributed WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR @@ -20,219 +20,319 @@ #include "gdcmValEntry.h" #include "gdcmBinEntry.h" #include "gdcmSeqEntry.h" - #include "gdcmGlobal.h" #include "gdcmUtil.h" #include "gdcmDebug.h" +#include "gdcmTS.h" +#include "gdcmDictSet.h" +#include "gdcmDocEntrySet.h" +#include "gdcmSQItem.h" #include +#include +#include +#include // for isdigit +#include // for atoi -// For nthos: -#ifdef _MSC_VER - #include -#else - #include -#endif - -# include - -// Implicit VR Little Endian -#define UI1_2_840_10008_1_2 "1.2.840.10008.1.2" -// Explicit VR Little Endian -#define UI1_2_840_10008_1_2_1 "1.2.840.10008.1.2.1" -// Deflated Explicit VR Little Endian -#define UI1_2_840_10008_1_2_1_99 "1.2.840.10008.1.2.1.99" -// Explicit VR Big Endian -#define UI1_2_840_10008_1_2_2 "1.2.840.10008.1.2.2" -// JPEG Baseline (Process 1) -#define UI1_2_840_10008_1_2_4_50 "1.2.840.10008.1.2.4.50" -// JPEG Extended (Process 2 & 4) -#define UI1_2_840_10008_1_2_4_51 "1.2.840.10008.1.2.4.51" -// JPEG Extended (Process 3 & 5) -#define UI1_2_840_10008_1_2_4_52 "1.2.840.10008.1.2.4.52" -// JPEG Spectral Selection, Non-Hierarchical (Process 6 & 8) -#define UI1_2_840_10008_1_2_4_53 "1.2.840.10008.1.2.4.53" -// JPEG Full Progression, Non-Hierarchical (Process 10 & 12) -#define UI1_2_840_10008_1_2_4_55 "1.2.840.10008.1.2.4.55" -// JPEG Lossless, Non-Hierarchical (Process 14) -#define UI1_2_840_10008_1_2_4_57 "1.2.840.10008.1.2.4.57" -// JPEG Lossless, Hierarchical, First-Order Prediction (Process 14, -// [Selection Value 1]) -#define UI1_2_840_10008_1_2_4_70 "1.2.840.10008.1.2.4.70" -// JPEG 2000 Lossless -#define UI1_2_840_10008_1_2_4_90 "1.2.840.10008.1.2.4.90" -// JPEG 2000 -#define UI1_2_840_10008_1_2_4_91 "1.2.840.10008.1.2.4.91" -// RLE Lossless -#define UI1_2_840_10008_1_2_5 "1.2.840.10008.1.2.5" -// UI1_1_2_840_10008_1_2_5 -#define str2num(str, typeNum) *((typeNum *)(str)) - +namespace gdcm +{ //----------------------------------------------------------------------------- -// Refer to gdcmDocument::CheckSwap() -const unsigned int gdcmDocument::HEADER_LENGTH_TO_READ = 256; -// Refer to gdcmDocument::SetMaxSizeLoadEntry() -const unsigned int gdcmDocument::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096 -const unsigned int gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff; +// Refer to Document::SetMaxSizeLoadEntry() +const unsigned int Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096 //----------------------------------------------------------------------------- // 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. */ -gdcmDocument::gdcmDocument( std::string const & filename ) - : gdcmElementSet(-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; } - dbg.Verbose(0, "gdcmDocument::gdcmDocument: starting parsing of file: ", - Filename.c_str()); - rewind(Fp); - - fseek(Fp,0L,SEEK_END); - long lgt = ftell(Fp); - - rewind(Fp); - CheckSwap(); - long beg = ftell(Fp); - lgt -= beg; - - SQDepthLevel = 0; - - (void)ParseDES( this, beg, lgt, false); // le Load sera fait a la volee + Group0002Parsed = false; + + gdcmWarningMacro( "Starting parsing of file: " << Filename.c_str()); + + 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 - rewind(Fp); + // 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); // Load 'non string' values - std::string PhotometricInterpretation = GetEntryByNumber(0x0028,0x0004); - if( PhotometricInterpretation == "PALETTE COLOR " ) + std::string PhotometricInterpretation = GetEntryValue(0x0028,0x0004); + if ( PhotometricInterpretation == "PALETTE COLOR " ) { - LoadEntryVoidArea(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 gdcmBinEntry. - /// In order to fix things "Quick and Dirty" the dictionnary was - /// altered on PURPOUS but now contains a WRONG value. + /// the ones of a BinEntry. + /// 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: - /// - multivaluated US, and hence loaded as gdcmValEntry, but afterwards - /// also used as gdcmBinEntry, which requires the proper conversion, - /// - OW, and hence loaded as gdcmBinEntry, but afterwards also used - /// as gdcmValEntry, which requires the proper conversion. - LoadEntryVoidArea(0x0028,0x1201); // R LUT - LoadEntryVoidArea(0x0028,0x1202); // G LUT - LoadEntryVoidArea(0x0028,0x1203); // B LUT + /// 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 + /// as ValEntry, which requires the proper conversion. + LoadEntryBinArea(0x0028,0x1201); // R LUT + LoadEntryBinArea(0x0028,0x1202); // G LUT + LoadEntryBinArea(0x0028,0x1203); // B LUT // Segmented Red Palette Color LUT Data - LoadEntryVoidArea(0x0028,0x1221); + LoadEntryBinArea(0x0028,0x1221); // Segmented Green Palette Color LUT Data - LoadEntryVoidArea(0x0028,0x1222); + LoadEntryBinArea(0x0028,0x1222); // Segmented Blue Palette Color LUT Data - LoadEntryVoidArea(0x0028,0x1223); - } + LoadEntryBinArea(0x0028,0x1223); + } + //FIXME later : how to use it? - LoadEntryVoidArea(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) + { + d = GetDocEntry( (*it).Group, (*it).Elem); + + if ( d == NULL) + continue; + + if ( dynamic_cast(d) ) + { + LoadDocEntry(d, true); + continue; + } + + if ( dynamic_cast(d) ) + { + LoadEntryBinArea((*it).Group, (*it).Elem); + continue; + } + + if ( dynamic_cast(d) ) + { + gdcmWarningMacro( "You cannot 'ForceLoad' a SeqEntry "); + 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 = GetEntryByNumber(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 = gdcmACR_LIBIDO; - std::string rows = GetEntryByNumber(0x0028, 0x0010); - std::string columns = GetEntryByNumber(0x0028, 0x0011); - SetEntryByNumber(columns, 0x0028, 0x0010); - SetEntryByNumber(rows , 0x0028, 0x0011); + Filetype = ACR_LIBIDO; + 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 ------------------ - - PrintLevel = 1; // 'Medium' print level by default + // --- End of ACR-LibIDO kludge --- + return true; } -/** - * \brief This default constructor doesn't parse the file. You should - * then invoke \ref gdcmDocument::SetFileName and then the parsing. - */ -gdcmDocument::gdcmDocument() - :gdcmElementSet(-1) -{ - SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); - Initialise(); - PrintLevel = 1; // 'Medium' print level by default -} /** - * \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. */ -gdcmDocument::~gdcmDocument () -{ - RefPubDict = NULL; - RefShaDict = NULL; - - // Recursive clean up of sequences - for (TagDocEntryHT::const_iterator it = TagHT.begin(); - it != TagHT.end(); ++it ) - { - delete it->second; - } - TagHT.clear(); -} - -//----------------------------------------------------------------------------- -// Print - -/** - * \brief Prints The Dict Entries of THE public Dicom Dictionary - * @return - */ -void gdcmDocument::PrintPubDict(std::ostream & os) -{ - RefPubDict->Print(os); -} - -/** - * \brief Prints The Dict Entries of THE shadow Dicom Dictionary - * @return - */ -void gdcmDocument::PrintShaDict(std::ostream & os) -{ - 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 */ -gdcmDict *gdcmDocument::GetPubDict() +Dict *Document::GetPubDict() { return RefPubDict; } @@ -240,16 +340,16 @@ gdcmDict *gdcmDocument::GetPubDict() /** * \brief Get the shadow dictionary used */ -gdcmDict *gdcmDocument::GetShaDict() +Dict *Document::GetShaDict() { return RefShaDict; } /** * \brief Set the shadow dictionary used - * \param dict dictionary to use in shadow + * @param dict dictionary to use in shadow */ -bool gdcmDocument::SetShaDict(gdcmDict *dict) +bool Document::SetShaDict(Dict *dict) { RefShaDict = dict; return !RefShaDict; @@ -257,264 +357,267 @@ bool gdcmDocument::SetShaDict(gdcmDict *dict) /** * \brief Set the shadow dictionary used - * \param dictName name of the dictionary to use in shadow + * @param dictName name of the dictionary to use in shadow */ -bool gdcmDocument::SetShaDict(DictKey const & dictName) +bool Document::SetShaDict(DictKey const &dictName) { - RefShaDict = gdcmGlobal::GetDicts()->GetDict(dictName); + RefShaDict = Global::GetDicts()->GetDict(dictName); return !RefShaDict; } /** - * \brief This predicate, based on hopefully reasonable heuristics, - * decides whether or not the current gdcmDocument was properly parsed - * and contains the mandatory information for being considered as - * a well formed and usable Dicom/Acr File. - * @return true when gdcmDocument 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 gdcmDocument::IsReadable() +bool Document::IsReadable() { - if( Filetype == gdcmUnknown) + if ( Filetype == Unknown ) { - dbg.Verbose(0, "gdcmDocument::IsReadable: wrong filetype"); + gdcmWarningMacro( "Wrong filetype"); return false; } - if( TagHT.empty() ) - { - dbg.Verbose(0, "gdcmDocument::IsReadable: 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 Internal function that checks whether the Transfer Syntax given - * as argument is the one present in the current document. - * @param syntaxToCheck The transfert syntax we need to check against. - * @return True when SyntaxToCheck corresponds to the Transfer Syntax of - * the current document. False either when the document contains - * no Transfer Syntax, or when the Tranfer Syntaxes doesn't match. + * \brief Accessor to the Transfer Syntax (when present) of the + * current document (it internally handles reading the + * value from disk when only parsing occured). + * @return The encountered Transfer Syntax of the current document. */ -bool gdcmDocument::IsGivenTransferSyntax(std::string const & syntaxToCheck) +std::string Document::GetTransferSyntax() { - gdcmDocEntry *entry = GetDocEntryByNumber(0x0002, 0x0010); + DocEntry *entry = GetDocEntry(0x0002, 0x0010); if ( !entry ) { - return false; + return GDCM_UNKNOWN; } // The entry might be present but not loaded (parsing and loading // happen at different stages): try loading and proceed with check... LoadDocEntrySafe(entry); - if (gdcmValEntry* valEntry = dynamic_cast< gdcmValEntry* >(entry) ) + if (ValEntry *valEntry = dynamic_cast< ValEntry* >(entry) ) { std::string transfer = valEntry->GetValue(); // The actual transfer (as read from disk) might be padded. We // first need to remove the potential padding. We can make the // weak assumption that padding was not executed with digits... - if ( transfer.length() == 0 ) { // for brain damaged headers - return false; - } - while ( ! isdigit(transfer[transfer.length()-1]) ) + if ( transfer.length() == 0 ) { - transfer.erase(transfer.length()-1, 1); + // for brain damaged headers + return GDCM_UNKNOWN; } - if ( transfer == syntaxToCheck ) + while ( !isdigit((unsigned char)transfer[transfer.length()-1]) ) { - return true; + transfer.erase(transfer.length()-1, 1); } + return transfer; } - return false; -} - -/** - * \brief Determines if the Transfer Syntax of the present document - * corresponds to a Implicit Value Representation of - * Little Endian. - * \sa \ref gdcmDocument::IsGivenTransferSyntax. - * @return True when ImplicitVRLittleEndian found. False in all other cases. - */ -bool gdcmDocument::IsImplicitVRLittleEndianTransferSyntax() -{ - return IsGivenTransferSyntax(UI1_2_840_10008_1_2); -} - -/** - * \brief Determines if the Transfer Syntax was already encountered - * and if it corresponds to a ExplicitVRLittleEndian one. - * @return True when ExplicitVRLittleEndian found. False in all other cases. - */ -bool gdcmDocument::IsExplicitVRLittleEndianTransferSyntax() -{ - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1); -} - -/** - * \brief Determines if the Transfer Syntax was already encountered - * and if it corresponds to a DeflatedExplicitVRLittleEndian one. - * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases. - */ -bool gdcmDocument::IsDeflatedExplicitVRLittleEndianTransferSyntax() -{ - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1_99); -} - -/** - * \brief Determines if the Transfer Syntax was already encountered - * and if it corresponds to a Explicit VR Big Endian one. - * @return True when big endian found. False in all other cases. - */ -bool gdcmDocument::IsExplicitVRBigEndianTransferSyntax() -{ - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_2); + return GDCM_UNKNOWN; } /** - * \brief Determines if the Transfer Syntax was already encountered - * and if it corresponds to a JPEGBaseLineProcess1 one. - * @return True when JPEGBaseLineProcess1found. False in all other cases. - */ -bool gdcmDocument::IsJPEGBaseLineProcess1TransferSyntax() -{ - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_50); -} - -/** - * \brief Determines if the Transfer Syntax was already encountered - * and if it corresponds to a JPEGExtendedProcess2-4 one. - * @return True when JPEGExtendedProcess2-4 found. False in all other cases. - */ -bool gdcmDocument::IsJPEGExtendedProcess2_4TransferSyntax() -{ - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_51); -} - -/** - * \brief Determines if the Transfer Syntax was already encountered - * and if it corresponds to a JPEGExtendeProcess3-5 one. - * @return True when JPEGExtendedProcess3-5 found. False in all other cases. + * \brief Accesses the info from 0002,0010 : Transfer Syntax and TS + * @return The full Transfer Syntax Name (as opposed to Transfer Syntax UID) */ -bool gdcmDocument::IsJPEGExtendedProcess3_5TransferSyntax() +std::string Document::GetTransferSyntaxName() { - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_52); -} + // use the TS (TS : Transfer Syntax) + std::string transferSyntax = GetEntryValue(0x0002,0x0010); -/** - * \brief Determines if the Transfer Syntax was already encountered - * and if it corresponds to a JPEGSpectralSelectionProcess6-8 one. - * @return True when JPEGSpectralSelectionProcess6-8 found. False in all - * other cases. - */ -bool gdcmDocument::IsJPEGSpectralSelectionProcess6_8TransferSyntax() -{ - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_53); -} + 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"; + } -/** - * \brief Determines if the Transfer Syntax was already encountered - * and if it corresponds to a RLE Lossless one. - * @return True when RLE Lossless found. False in all - * other cases. - */ -bool gdcmDocument::IsRLELossLessTransferSyntax() -{ - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_5); -} + // we do it only when we need it + const TSKey &tsName = Global::GetTS()->GetValue( transferSyntax ); -/** - * \brief Determines if Transfer Syntax was already encountered - * and if it corresponds to a JPEG Lossless one. - * @return True when RLE Lossless found. False in all - * other cases. - */ - -bool gdcmDocument::IsJPEGLossless() -{ - return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_55) - || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_57) - || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_70) ); -} - -/** - * \brief Determines if the Transfer Syntax was already encountered - * and if it corresponds to a JPEG2000 one - * @return True when JPEG2000 (Lossly or LossLess) found. False in all - * other cases. - */ -bool gdcmDocument::IsJPEG2000() -{ - return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_90) - || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_91) ); + // Global::GetTS() is a global static you shall never try to delete it! + return tsName; } - +// +// --------------- Swap Code ------------------ /** - * \brief Predicate for dicom version 3 file. - * @return True when the file is a dicom version 3. + * \brief Swaps the bytes so they agree with the processor order + * @return The properly swaped 16 bits integer. */ -bool gdcmDocument::IsDicomV3() +uint16_t Document::SwapShort(uint16_t a) { - // Checking if Transfert 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 GetDocEntryByNumber(0x0002, 0x0010) != NULL; + if ( SwapCode == 4321 || SwapCode == 2143 ) + { + //a = ((( a << 8 ) & 0xff00 ) | (( a >> 8 ) & 0x00ff ) ); + // Save CPU time + a = ( a << 8 ) | ( a >> 8 ); + } + return a; } /** - * \brief returns the File Type - * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown) - * @return the FileType code + * \brief Swaps back the bytes of 4-byte long integer accordingly to + * processor order. + * @return The properly swaped 32 bits integer. */ -FileType gdcmDocument::GetFileType() +uint32_t Document::SwapLong(uint32_t a) { - return Filetype; -} + 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 gdcmDocument::Filename and + * \brief Tries to open the file \ref Document::Filename and * checks the preamble when existing. * @return The FILE pointer on success. */ -FILE* gdcmDocument::OpenFile() +std::ifstream *Document::OpenFile() { - Fp = fopen(Filename.c_str(),"rb"); + HasDCMPreamble = false; + if (Filename.length() == 0) + { + return 0; + } + + if ( Fp ) + { + gdcmWarningMacro( "File already open: " << Filename.c_str()); + CloseFile(); + } - if(!Fp) + Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary); + if ( ! *Fp ) { - dbg.Verbose( 0, - "gdcmDocument::OpenFile 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; - fread(&zero, (size_t)2, (size_t)1, Fp); - - //ACR -- or DICOM with no Preamble -- - if( zero == 0x0008 || zero == 0x0800 || zero == 0x0002 || zero == 0x0200 ) + uint16_t zero = 0; + Fp->read((char*)&zero, (size_t)2); + if ( Fp->eof() ) { + CloseFile(); + return 0; + } + + //-- 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 starting at the begining of the file:(%04x)\n", zero); + gdcmWarningMacro( msg.c_str() ); return Fp; } - //DICOM - fseek(Fp, 126L, SEEK_CUR); - char dicm[4]; - fread(dicm, (size_t)4, (size_t)1, Fp); - if( memcmp(dicm, "DICM", 4) == 0 ) + //-- DICOM -- + Fp->seekg(126L, std::ios::cur); + char dicm[4] = {' ',' ',' ',' '}; + Fp->read(dicm, (size_t)4); + if ( Fp->eof() ) + { + CloseFile(); + return 0; + } + if ( memcmp(dicm, "DICM", 4) == 0 ) { + HasDCMPreamble = true; return Fp; } - - fclose(Fp); - dbg.Verbose( 0, - "gdcmDocument::OpenFile not DICOM/ACR (missing preamble)", - Filename.c_str()); - + + // -- Neither ACR/No Preamble Dicom nor DICOMV3 file + CloseFile(); + gdcmWarningMacro( "Neither ACR/No Preamble Dicom nor DICOMV3 file: " + << Filename.c_str()); return 0; } @@ -522,957 +625,628 @@ FILE* gdcmDocument::OpenFile() * \brief closes the file * @return TRUE if the close was successfull */ -bool gdcmDocument::CloseFile() +bool Document::CloseFile() { - int closed = fclose(Fp); - Fp = 0; - - return closed; + if ( Fp ) + { + Fp->close(); + delete Fp; + Fp = 0; + } + return true; } /** * \brief Writes in a file all the Header Entries (Dicom Elements) - * @param fp file pointer on an already open file + * @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. + * @return Always true. */ -void gdcmDocument::Write(FILE* fp,FileType filetype) +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 - - /// WARNING : Si on veut ecrire du DICOM V3 a partir d'un DcmHeader ACR-NEMA - /// no way (check : FileType est un champ de gdcmDocument ...) - /// a moins de se livrer a un tres complique ajout des champs manquants. - /// faire un CheckAndCorrectHeader (?) - - if (filetype == gdcmImplicitVR) + // \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 + + if ( filetype == ImplicitVR || filetype == ExplicitVR ) { - std::string implicitVRTransfertSyntax = UI1_2_840_10008_1_2; - ReplaceOrCreateByNumber(implicitVRTransfertSyntax,0x0002, 0x0010); - - /// \todo Refer to standards on page 21, chapter 6.2 - /// "Value representation": values with a VR of UI shall be - /// padded with a single trailing null - /// in the following case we have to padd manually with a 0 - - SetEntryLengthByNumber(18, 0x0002, 0x0010); - } - - if (filetype == gdcmExplicitVR) - { - std::string explicitVRTransfertSyntax = UI1_2_840_10008_1_2_1; - ReplaceOrCreateByNumber(explicitVRTransfertSyntax,0x0002, 0x0010); - - /// \todo Refer to standards on page 21, chapter 6.2 - /// "Value representation": values with a VR of UI shall be - /// padded with a single trailing null - /// Dans le cas suivant on doit pader manuellement avec un 0 - - SetEntryLengthByNumber(20, 0x0002, 0x0010); + // writing Dicom File Preamble + char filePreamble[128]; + memset(filePreamble, 0, 128); + fp->write(filePreamble, 128); + 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); - */ - - gdcmElementSet::Write(fp, filetype); // This one is recursive + /* + * \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 Header 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 - * \return pointer to the modified/created Header Entry (NULL when creation - * failed). - */ - -gdcmValEntry * gdcmDocument::ReplaceOrCreateByNumber( - std::string const & value, - uint16_t group, - uint16_t elem ) -{ - gdcmValEntry* valEntry = 0; - gdcmDocEntry* currentEntry = GetDocEntryByNumber( group, elem); - - if (!currentEntry) - { - // The entry wasn't present and we simply create the required ValEntry: - currentEntry = NewDocEntryByNumber(group, elem); - if (!currentEntry) - { - dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: call to" - " NewDocEntryByNumber failed."); - return NULL; - } - valEntry = new gdcmValEntry(currentEntry); - if ( !AddEntry(valEntry)) - { - dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: AddEntry" - " failed allthough this is a creation."); - } - } - else - { - valEntry = dynamic_cast< gdcmValEntry* >(currentEntry); - if ( !valEntry ) // Euuuuh? It wasn't a ValEntry - // then we change it to a ValEntry ? - // Shouldn't it be considered as an error ? - { - // We need to promote the gdcmDocEntry to a gdcmValEntry: - valEntry = new gdcmValEntry(currentEntry); - if (!RemoveEntry(currentEntry)) - { - dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: removal" - " of previous DocEntry failed."); - return NULL; - } - if ( !AddEntry(valEntry)) - { - dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: adding" - " promoted ValEntry failed."); - return NULL; - } - } - } - - SetEntryByNumber(value, group, elem); - - return valEntry; -} - -/* - * \brief Modifies the value of a given Header Entry (Dicom Element) - * when it exists. Create it with the given value when unexistant. - * @param voidArea (binary) value to be set - * @param Group Group number of the Entry - * @param Elem Element number of the Entry - * \return pointer to the modified/created Header Entry (NULL when creation - * failed). + * \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 */ -gdcmBinEntry * gdcmDocument::ReplaceOrCreateByNumber( - void *voidArea, - int lgth, - uint16_t group, - uint16_t elem) +void Document::LoadEntryBinArea(uint16_t group, uint16_t elem) { - gdcmBinEntry* b = 0; - gdcmDocEntry* a = GetDocEntryByNumber( group, elem); - if (!a) - { - a = NewBinEntryByNumber(group, elem); - if (!a) - { - return 0; - } - - b = new gdcmBinEntry(a); - AddEntry(b); - b->SetVoidArea(voidArea); - } - - SetEntryByNumber(voidArea, lgth, group, elem); - //b->SetVoidArea(voidArea); //what if b == 0 !! - - return b; -} + // Search the corresponding DocEntry + DocEntry *docElement = GetDocEntry(group, elem); + if ( !docElement ) + return; + BinEntry *binElement = dynamic_cast(docElement); + if ( !binElement ) + return; -/* - * \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). - */ -gdcmSeqEntry * gdcmDocument::ReplaceOrCreateByNumber( - uint16_t group, - uint16_t elem) -{ - gdcmSeqEntry* b = 0; - gdcmDocEntry* a = GetDocEntryByNumber( group, elem); - if (!a) - { - a = NewSeqEntryByNumber(group, elem); - if (!a) - { - return 0; - } + LoadEntryBinArea(binElement); +} - b = new gdcmSeqEntry(a, 1); // FIXME : 1 (Depth) - AddEntry(b); - } - return b; -} - /** - * \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 + * \brief Loads (from disk) the element content + * when a string is not suitable + * @param elem Entry whose binArea is going to be loaded */ -bool gdcmDocument::ReplaceIfExistByNumber(std::string const & value, - uint16_t group, uint16_t elem ) +void Document::LoadEntryBinArea(BinEntry *elem) { - SetEntryByNumber(value, group, elem); - - return true; -} + if (elem->GetBinArea() ) + return; -//----------------------------------------------------------------------------- -// Protected + bool openFile = !Fp; + if ( openFile ) + OpenFile(); -/** - * \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 gdcmDocument::CheckIfEntryExistByNumber(uint16_t group, uint16_t element ) -{ - const std::string &key = gdcmDictEntry::TranslateToKey(group, element ); - return TagHT.count(key); -} + size_t o =(size_t)elem->GetOffset(); + Fp->seekg(o, std::ios::beg); -/** - * \brief Searches within Header Entries (Dicom Elements) parsed with - * the public and private dictionaries - * for the element value of a given tag. - * \warning Don't use any longer : use GetPubEntryByName - * @param tagName name of the searched element. - * @return Corresponding element value when it exists, - * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. - */ -std::string gdcmDocument::GetEntryByName(TagName const & tagName) -{ - gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName); - if( !dictEntry ) + size_t l = elem->GetLength(); + uint8_t *a = new uint8_t[l]; + if ( !a ) { - return GDCM_UNFOUND; + gdcmWarningMacro( "Cannot allocate BinEntry content for : " + << std::hex << elem->GetGroup() + << "|" << elem->GetElement() ); + return; } - return GetEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement()); -} - -/** - * \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 tagName name of the searched element. - * @return Corresponding element value representation when it exists, - * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. - */ -std::string gdcmDocument::GetEntryVRByName(TagName const & tagName) -{ - gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName); - if( dictEntry == NULL) + Fp->read((char*)a, l); + if ( Fp->fail() || Fp->eof() ) { - return GDCM_UNFOUND; + delete[] a; + return; } - gdcmDocEntry* elem = GetDocEntryByNumber(dictEntry->GetGroup(), - dictEntry->GetElement()); - return elem->GetVR(); -} - + elem->SetBinArea(a); -/** - * \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 gdcmDocument::GetEntryByNumber(uint16_t group, uint16_t element) -{ - gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element); - /// \todo use map methods, instead of multimap JPR - if ( !TagHT.count(key)) - { - return GDCM_UNFOUND; - } - - return ((gdcmValEntry *)TagHT.find(key)->second)->GetValue(); + if ( openFile ) + CloseFile(); } /** - * \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. + * \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. */ -std::string gdcmDocument::GetEntryVRByNumber(uint16_t group, uint16_t element) +void Document::LoadDocEntrySafe(DocEntry *entry) { - gdcmDocEntry* elem = GetDocEntryByNumber(group, element); - if ( !elem ) + if ( Fp ) { - return GDCM_UNFOUND; + long PositionOnEntry = Fp->tellg(); + LoadDocEntry(entry); + Fp->seekg(PositionOnEntry, std::ios::beg); } - 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 gdcmDocument::GetEntryLengthByNumber(uint16_t group, uint16_t element) -{ - gdcmDocEntry* elem = GetDocEntryByNumber(group, element); - if ( !elem ) - { - return -2; //magic number - } - return elem->GetLength(); -} -/** - * \brief Sets the value (string) of the Header Entry (Dicom Element) - * @param content string value of the Dicom Element - * @param tagName name of the searched Dicom Element. - * @return true when found + * \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' */ -bool gdcmDocument::SetEntryByName(std::string const & content,std::string const & tagName) +bool Document::operator<(Document &document) { - gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName); - if( !dictEntry ) + // Patient Name + std::string s1 = GetEntryValue(0x0010,0x0010); + std::string s2 = document.GetEntryValue(0x0010,0x0010); + if (s1 < s2) { - return false; + return true; } - - return SetEntryByNumber(content,dictEntry->GetGroup(), - dictEntry->GetElement()); -} - -/** - * \brief Accesses an existing gdcmDocEntry (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 gdcmDocument::SetEntryByNumber(std::string const & content, - uint16_t group, - uint16_t element) -{ - int c; - int l; - - gdcmValEntry* valEntry = GetValEntryByNumber(group, element); - if (!valEntry ) + else if ( s1 > s2 ) { - dbg.Verbose(0, "gdcmDocument::SetEntryByNumber: no corresponding", - " ValEntry (try promotion first)."); return false; } - // Non even content must be padded with a space (020H)... - std::string finalContent = content; - if( finalContent.length() % 2 ) + else { - finalContent += '\0'; // ... therefore we padd with (000H) .!?! - } - valEntry->SetValue(finalContent); - - // Integers have a special treatement for their length: - - l = finalContent.length(); - if ( l != 0) // To avoid to be cheated by 'zero length' integers - { - gdcmVRKey vr = valEntry->GetVR(); - if( vr == "US" || vr == "SS" ) + // Patient ID + s1 = GetEntryValue(0x0010,0x0020); + s2 = document.GetEntryValue(0x0010,0x0020); + if ( s1 < s2 ) { - c = CountSubstring(content, "\\") + 1; // for multivaluated items - l = c*2; + return true; } - else if( vr == "UL" || vr == "SL" ) + else if ( s1 > s2 ) { - c = CountSubstring(content, "\\") + 1; // for multivaluated items - l = c*4;; + 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; + } + } } } - valEntry->SetLength(l); - return true; -} - -/** - * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element) - * through it's (group, element) and modifies it's content with - * the given value. - * @param content new value (void *) 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 gdcmDocument::SetEntryByNumber(void *content, - int lgth, - uint16_t group, - uint16_t element) -{ - (void)lgth; //not used - gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element); - if ( !TagHT.count(key)) - { - return false; - } - -/* 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? - } -*/ - gdcmBinEntry* a = (gdcmBinEntry *)TagHT[key]; - a->SetVoidArea(content); - a->SetLength(lgth); - - return true; -} - -/** - * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element) - * in the PubDocEntrySet of this instance - * through it's (group, element) and modifies it's length with - * the given value. - * \warning Use with extreme caution. - * @param l new length to substitute with - * @param group group number of the Entry to modify - * @param element element number of the Entry to modify - * @return true on success, false otherwise. - */ -bool gdcmDocument::SetEntryLengthByNumber(uint32_t l, - uint16_t group, - uint16_t element) -{ - /// \todo use map methods, instead of multimap JPR - gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element); - if ( !TagHT.count(key) ) - { - return false; - } - if ( l % 2 ) - { - l++; // length must be even - } - ( ((TagHT.equal_range(key)).first)->second )->SetLength(l); - - return true ; + return false; } +//----------------------------------------------------------------------------- +// Protected /** - * \brief Gets (from Header) the offset of 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 File Offset of the Element Value + * \brief Reads a supposed to be 16 Bits integer + * (swaps it depending on processor endianness) + * @return read value */ -size_t gdcmDocument::GetEntryOffsetByNumber(uint16_t group, uint16_t elem) +uint16_t Document::ReadInt16() + throw( FormatError ) { - gdcmDocEntry* entry = GetDocEntryByNumber(group, elem); - if (!entry) + uint16_t g; + Fp->read ((char*)&g, (size_t)2); + if ( Fp->fail() ) { - dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry present."); - return 0; + throw FormatError( "Document::ReadInt16()", " file error." ); } - return entry->GetOffset(); -} - -/** - * \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 * gdcmDocument::GetEntryVoidAreaByNumber(uint16_t group, uint16_t elem) -{ - gdcmDocEntry* entry = GetDocEntryByNumber(group, elem); - if (!entry) + if ( Fp->eof() ) { - dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry"); - return 0; + throw FormatError( "Document::ReadInt16()", "EOF." ); } - return ((gdcmBinEntry *)entry)->GetVoidArea(); + g = SwapShort(g); + return g; } /** - * \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* gdcmDocument::LoadEntryVoidArea(uint16_t group, uint16_t elem) -{ - gdcmDocEntry *docElement = GetDocEntryByNumber(group, elem); - if ( !docElement ) - { - return NULL; - } - size_t o =(size_t)docElement->GetOffset(); - fseek(Fp, o, SEEK_SET); - size_t l = docElement->GetLength(); - char* a = new char[l]; - if(!a) - { - dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a"); - return NULL; - } - size_t l2 = fread(a, 1, l , Fp); - if( l != l2 ) - { - delete[] a; - return NULL; - } - /// \TODO Drop any already existing void area! JPR - if( !SetEntryVoidAreaByNumber( a, group, elem ) ); - { - dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea setting failed."); - } - - return a; -} -/** - * \brief Loads (from disk) the element content - * when a string is not suitable - * @param element Entry whose voidArea is going to be loaded + * \brief Reads a supposed to be 32 Bits integer + * (swaps it depending on processor endianness) + * @return read value */ -void *gdcmDocument::LoadEntryVoidArea(gdcmBinEntry *element) +uint32_t Document::ReadInt32() + throw( FormatError ) { - size_t o =(size_t)element->GetOffset(); - fseek(Fp, o, SEEK_SET); - size_t l = element->GetLength(); - char* a = new char[l]; - if( !a ) + uint32_t g; + Fp->read ((char*)&g, (size_t)4); + if ( Fp->fail() ) { - dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a"); - return NULL; + throw FormatError( "Document::ReadInt32()", " file error." ); } - element->SetVoidArea((void *)a); - /// \todo check the result - size_t l2 = fread(a, 1, l , Fp); - if( l != l2 ) + if ( Fp->eof() ) { - delete[] a; - return NULL; + throw FormatError( "Document::ReadInt32()", "EOF." ); } - - return a; + g = SwapLong(g); + return g; } /** - * \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 skips bytes inside the source file + * \warning NOT end user intended method ! + * @return */ -bool gdcmDocument::SetEntryVoidAreaByNumber(void * area, - uint16_t group, - uint16_t element) +void Document::SkipBytes(uint32_t nBytes) { - gdcmDocEntry* currentEntry = GetDocEntryByNumber(group, element); - if ( !currentEntry ) - { - return false; - } - if ( gdcmBinEntry* binEntry = dynamic_cast(currentEntry) ) - { - binEntry->SetVoidArea( area ); - return true; - } - return true; + //FIXME don't dump the returned value + Fp->seekg((long)nBytes, std::ios::cur); } /** - * \brief Update the entries with the shadow dictionary. - * Only non even entries are analyzed + * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader + * @param filetype Type of the File to be written */ -void gdcmDocument::UpdateShaEntries() +int Document::ComputeGroup0002Length( FileType filetype ) { - //gdcmDictEntry *entry; + uint16_t gr; std::string vr; - /// \todo TODO : still any use to explore recursively the whole structure? -/* - for(ListTag::iterator it=listEntries.begin(); - it!=listEntries.end(); - ++it) + int groupLength = 0; + bool found0002 = false; + + // for each zero-level Tag in the DCM Header + DocEntry *entry = GetFirstEntry(); + while( entry ) { - // Odd group => from public dictionary - if((*it)->GetGroup()%2==0) - continue; + gr = entry->GetGroup(); - // Peer group => search the corresponding dict entry - if(RefShaDict) - entry=RefShaDict->GetDictEntryByNumber((*it)->GetGroup(),(*it)->GetElement()); - else - entry=NULL; - - if((*it)->IsImplicitVR()) - vr="Implicit"; - else - vr=(*it)->GetVR(); - - (*it)->SetValue(GetDocEntryUnvalue(*it)); // to go on compiling - if(entry){ - // Set the new entry and the new value - (*it)->SetDictEntry(entry); - CheckDocEntryVR(*it,vr); + if ( gr == 0x0002 ) + { + found0002 = true; - (*it)->SetValue(GetDocEntryValue(*it)); // to go on compiling + if ( entry->GetElement() != 0x0000 ) + { + vr = entry->GetVR(); + if ( filetype == ExplicitVR ) + { + if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") || (vr == "UT") ) + { + // explicit VR AND OB, OW, SQ, UT : 4 more bytes + groupLength += 4; + } + } + groupLength += 2 + 2 + 4 + entry->GetLength(); + } } - else - { - // Remove precedent value transformation - (*it)->SetDictEntry(NewVirtualDictEntry((*it)->GetGroup(),(*it)->GetElement(),vr)); - } - } -*/ -} - -/** - * \brief Searches within the Header Entries for a Dicom Element of - * a given tag. - * @param tagName name of the searched Dicom Element. - * @return Corresponding Dicom Element when it exists, and NULL - * otherwise. - */ -gdcmDocEntry* gdcmDocument::GetDocEntryByName(std::string const & tagName) -{ - gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName); - if( !dictEntry ) - { - return NULL; - } - - return GetDocEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement()); -} - -/** - * \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 - */ -gdcmDocEntry* gdcmDocument::GetDocEntryByNumber(uint16_t group, - uint16_t element) -{ - gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element); - if ( !TagHT.count(key)) - { - return NULL; - } - return TagHT.find(key)->second; -} - -/** - * \brief Same as \ref gdcmDocument::GetDocEntryByNumber except it only - * returns a result when the corresponding entry is of type - * ValEntry. - * @return When present, the corresponding ValEntry. - */ -gdcmValEntry* gdcmDocument::GetValEntryByNumber(uint16_t group, - uint16_t element) -{ - gdcmDocEntry* currentEntry = GetDocEntryByNumber(group, element); - if ( !currentEntry ) - { - return 0; - } - if ( gdcmValEntry* valEntry = dynamic_cast(currentEntry) ) - { - return valEntry; - } - dbg.Verbose(0, "gdcmDocument::GetValEntryByNumber: unfound ValEntry."); - - return 0; -} - -/** - * \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 shall be loaded. - * @return - */ -void gdcmDocument::LoadDocEntrySafe(gdcmDocEntry * entry) -{ - long PositionOnEntry = ftell(Fp); - LoadDocEntry(entry); - fseek(Fp, PositionOnEntry, SEEK_SET); -} - -/** - * \brief Swaps back the bytes of 4-byte long integer accordingly to - * processor order. - * @return The properly swaped 32 bits integer. - */ -uint32_t gdcmDocument::SwapLong(uint32_t a) -{ - switch (SwapCode) - { - case 0 : + else if (found0002 ) 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 : - //std::cout << "swapCode= " << SwapCode << std::endl; - dbg.Error(" gdcmDocument::SwapLong : unset swap code"); - a = 0; - } - return a; -} -/** - * \brief Unswaps back the bytes of 4-byte long integer accordingly to - * processor order. - * @return The properly unswaped 32 bits integer. - */ -uint32_t gdcmDocument::UnswapLong(uint32_t a) -{ - return SwapLong(a); -} - -/** - * \brief Swaps the bytes so they agree with the processor order - * @return The properly swaped 16 bits integer. - */ -uint16_t gdcmDocument::SwapShort(uint16_t a) -{ - if ( SwapCode == 4321 || SwapCode == 2143 ) - { - a = ((( a << 8 ) & 0x0ff00 ) | (( a >> 8 ) & 0x00ff ) ); + 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 gdcmDocument::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 */ - -long gdcmDocument::ParseDES(gdcmDocEntrySet *set, - long offset, - long l_max, - bool delim_mode) -{ - gdcmDocEntry *newDocEntry = 0; - unsigned long l = 0; - - int depth = set->GetDepthLevel(); +void Document::ParseDES(DocEntrySet *set, long offset, + long l_max, bool delim_mode) +{ + DocEntry *newDocEntry; + ValEntry *newValEntry; + BinEntry *newBinEntry; + SeqEntry *newSeqEntry; + VRKey vr; + bool used; + bool delim_mode_intern = delim_mode; + bool first = true; while (true) - { - if ( !delim_mode && (ftell(Fp)-offset) >= l_max) + { + if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) { break; } + newDocEntry = ReadNextDocEntry( ); + + // 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; } - gdcmVRKey vr = newDocEntry->GetVR(); - if ( vr != "SQ" ) + // an Item Starter found elsewhere but the first postition + // 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); + + if ( newValEntry || newBinEntry ) { - - if ( gdcmGlobal::GetVR()->IsVROfGdcmStringRepresentable(vr) ) + //////////////////////////// ContentEntry + if ( newBinEntry ) { - /////// ValEntry - gdcmValEntry* newValEntry = - new gdcmValEntry( newDocEntry->GetDictEntry() ); - newValEntry->Copy( newDocEntry ); - newValEntry->SetKey( set->GetBaseTagKey() + newValEntry->GetKey() ); - set->AddEntry( newValEntry ); - LoadDocEntry( newValEntry ); - if (newValEntry->IsItemDelimitor()) + vr = newDocEntry->GetVR(); + if ( Filetype == ExplicitVR && + !Global::GetVR()->IsVROfBinaryRepresentable(vr) ) + { + ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR + 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 ( 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 key has the + // generalized form (refer to \ref BaseTagKey): + + // time waste hunting + //if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + //{ + // newBinEntry->SetKey( parentSQItem->GetBaseTagKey() + // + newBinEntry->GetKey() ); + //} + + if ( !set->AddEntry( newBinEntry ) ) { - break; + gdcmWarningMacro( "in ParseDES : cannot add a BinEntry " + << newBinEntry->GetKey() + << " (at offset : " + << newBinEntry->GetOffset() << " )" ); + used=false; } - if ( !delim_mode && (ftell(Fp)-offset) >= l_max) + else { - break; + // Load only if we can add (not a duplicate key) + LoadDocEntry( newBinEntry ); } - } + } // end BinEntry else { - if ( ! gdcmGlobal::GetVR()->IsVROfGdcmBinaryRepresentable(vr) ) - { - ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR - dbg.Verbose(0, "gdcmDocument::ParseDES: neither Valentry, " - "nor BinEntry. Probably unknown VR."); + /////////////////////// 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 ( 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): + + // time waste hunting + //if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + //{ + // newValEntry->SetKey( parentSQItem->GetBaseTagKey() + // + newValEntry->GetKey() ); + //} + + if ( !set->AddEntry( newValEntry ) ) + { + gdcmWarningMacro( "in ParseDES : cannot add a ValEntry " + << newValEntry->GetKey() + << " (at offset : " + << newValEntry->GetOffset() << " )" ); + used=false; + } + else + { + // Load only if we can add (not a duplicate key) + LoadDocEntry( newValEntry ); } - ////// BinEntry or UNKOWN VR: - gdcmBinEntry* newBinEntry = - new gdcmBinEntry( newDocEntry->GetDictEntry() ); - newBinEntry->Copy( newDocEntry ); - newBinEntry->SetKey( set->GetBaseTagKey() + newBinEntry->GetKey() ); - set->AddEntry( newBinEntry ); - LoadDocEntry( newBinEntry ); - } + if ( newValEntry->GetElement() == 0x0000 ) // if on group length + { + 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 ); + newDocEntry = 0; + continue; + } + } + } + } - if (newDocEntry->GetGroup() == 0x7fe0 && - newDocEntry->GetElement() == 0x0010 ) - { - if (newDocEntry->GetReadLength()==0xffffffff) - { - // Broken US.3405.1.dcm - Parse7FE0(); // to skip the pixels - // (multipart JPEG/RLE are trouble makers) - } - else - { - SkipToNextDocEntry(newDocEntry); - l = newDocEntry->GetFullLength(); - } - } - else - { - // to be sure we are at the beginning - SkipToNextDocEntry(newDocEntry); - l = newDocEntry->GetFullLength(); + bool delimitor=newValEntry->IsItemDelimitor(); + + if ( (delimitor) || + (!delim_mode && ((long)(Fp->tellg())-offset) >= l_max) ) + { + if ( !used ) + delete newDocEntry; + break; + } } + + // Just to make sure we are at the beginning of next entry. + SkipToNextDocEntry(newDocEntry); } else { - // VR = "SQ" - 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 ... - gdcmSeqEntry* newSeqEntry = - new gdcmSeqEntry( newDocEntry->GetDictEntry(), - set->GetDepthLevel() ); - newSeqEntry->Copy( newDocEntry ); - newSeqEntry->SetDelimitorMode( delim_mode ); - newSeqEntry->SetDepthLevel( depth ); - newSeqEntry->SetKey( set->GetBaseTagKey() + newSeqEntry->GetKey() ); + + 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: +// SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ); + if ( set == this ) // ( dynamic_cast< Document* > ( set ) ) + { + newSeqEntry->SetDepthLevel( 1 ); + // 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: + + // time waste hunting + else if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + { + newSeqEntry->SetDepthLevel( parentSQItem->GetDepthLevel() + 1 ); + + // newSeqEntry->SetKey( parentSQItem->GetBaseTagKey() + // + newSeqEntry->GetKey() ); + } if ( l != 0 ) { // Don't try to parse zero-length sequences - (void)ParseSQ( newSeqEntry, - newDocEntry->GetOffset(), - l, delim_mode); + ParseSQ( newSeqEntry, + newDocEntry->GetOffset(), + l, delim_mode_intern); } - set->AddEntry( newSeqEntry ); - if ( !delim_mode && (ftell(Fp)-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 ) + { + delete newDocEntry; } - delete newDocEntry; - } - return l; // Probably useless + first = false; + } // end While } /** * \brief Parses a Sequence ( SeqEntry after SeqEntry) * @return parsed length for this level */ -long gdcmDocument::ParseSQ( gdcmSeqEntry* seqEntry, - long offset, long l_max, bool delim_mode) +void Document::ParseSQ( SeqEntry *seqEntry, + long offset, long l_max, bool delim_mode) { int SQItemNumber = 0; bool dlm_mod; + long offsetStartCurrentSQItem = offset; while (true) { - gdcmDocEntry* 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; } } - if ( !delim_mode && (ftell(Fp)-offset) >= l_max) + if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) { - break; + delete newDocEntry; + break; } - - gdcmSQItem *itemSQ = new gdcmSQItem( seqEntry->GetDepthLevel() ); + // 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 ) @@ -1483,40 +1257,82 @@ long gdcmDocument::ParseSQ( gdcmSeqEntry* seqEntry, { dlm_mod = false; } - - (void)ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod); - - seqEntry->AddEntry( itemSQ, SQItemNumber ); + + // Let's try :------------ + // remove fff0,e000, created out of the SQItem + delete newDocEntry; + 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 && ( ftell(Fp) - offset ) >= l_max ) + if ( !delim_mode && ((long)(Fp->tellg())-offset ) >= l_max ) { break; } } +} + +/** + * \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 ); - int lgth = ftell(Fp) - offset; - return lgth; + // 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 the element content if its length doesn't exceed - * the value specified with gdcmDocument::SetMaxSizeLoadEntry() - * @param entry Header Entry (Dicom Element) to be dealt with + * \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 + * @param forceLoad wheter we want to load its content even if its length + * exceeds the value specified with Document::SetMaxSizeLoadEntry() */ -void gdcmDocument::LoadDocEntry(gdcmDocEntry* entry) +void Document::LoadDocEntry(DocEntry *entry, bool forceLoad) { - size_t item_read; uint16_t group = entry->GetGroup(); std::string vr = entry->GetVR(); uint32_t length = entry->GetLength(); - fseek(Fp, (long)entry->GetOffset(), SEEK_SET); + Fp->seekg((long)entry->GetOffset(), std::ios::beg); // A SeQuence "contains" a set of Elements. // (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; @@ -1525,64 +1341,67 @@ void gdcmDocument::LoadDocEntry(gdcmDocEntry* entry) // When the length is zero things are easy: if ( length == 0 ) { - ((gdcmValEntry *)entry)->SetValue(""); + ((ValEntry *)entry)->SetValue(""); return; } // 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 (gdcmBinEntry* binEntryPtr = dynamic_cast< gdcmBinEntry* >(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 (gdcmValEntry* valEntryPtr = dynamic_cast< gdcmValEntry* >(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 (!forceLoad) + { + if (length > MaxSizeLoadEntry) { - // fusible - std::cout<< "MaxSizeLoadEntry exceeded, neither a BinEntry " - << "nor a ValEntry ?! Should never print that !" << std::endl; - } + 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 ... - fseek(Fp,(long)entry->GetOffset()+(long)entry->GetLength(),SEEK_SET); - 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 : - if (gdcmBinEntry* binEntryPtr = dynamic_cast< gdcmBinEntry* >(entry) ) + if (BinEntry *binEntryPtr = dynamic_cast< BinEntry* >(entry) ) { - s << "gdcm::Loaded (BinEntry)"; + s << GDCM_BINLOADED; binEntryPtr->SetValue(s.str()); - LoadEntryVoidArea(binEntryPtr); // last one, not to erase length ! + LoadEntryBinArea(binEntryPtr); // last one, not to erase length ! 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) @@ -1621,30 +1440,45 @@ void gdcmDocument::LoadDocEntry(gdcmDocEntry* entry) s << std::ends; // to avoid oddities on Solaris #endif //GDCM_NO_ANSI_STRING_STREAM - ((gdcmValEntry *)entry)->SetValue(s.str()); + ((ValEntry *)entry)->SetValue(s.str()); return; } - // We need an additional byte for storing \0 that is not on disk - //std::string newValue(length,0); - //item_read = fread(&(newValue[0]), (size_t)length, (size_t)1, Fp); - //rah !! I can't believe it could work, normally this is a const char* !!! + // FIXME: We need an additional byte for storing \0 that is not on disk char *str = new char[length+1]; - item_read = fread(str, (size_t)length, (size_t)1, Fp); - str[length] = '\0'; - std::string newValue = str; + Fp->read(str, (size_t)length); + 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 ) + { + newValue = Util::DicomString(str, length+1); + gdcmWarningMacro("Warning: bad length: " << length << + " For string :" << newValue.c_str()); + // Since we change the length of string update it length + //entry->SetReadLength(length+1); + } + else + { + newValue = Util::DicomString(str, length); + } delete[] str; - if ( gdcmValEntry* valEntry = dynamic_cast(entry) ) - { - if ( item_read != 1 ) + + if ( ValEntry *valEntry = dynamic_cast(entry) ) + { + if ( Fp->fail() || Fp->eof()) { - dbg.Verbose(1, "gdcmDocument::LoadDocEntry", - "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); @@ -1656,54 +1490,57 @@ void gdcmDocument::LoadDocEntry(gdcmDocEntry* entry) } else { - dbg.Error(true, "gdcmDocument::LoadDocEntry" - "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 * @param entry Header Entry whose length of the value shall be loaded. */ -void gdcmDocument::FindDocEntryLength( gdcmDocEntry *entry ) - throw ( gdcmFormatError ) +void Document::FindDocEntryLength( DocEntry *entry ) + throw ( FormatError ) { - uint16_t element = entry->GetElement(); std::string vr = entry->GetVR(); uint16_t length16; - if ( Filetype == gdcmExplicitVR && !entry->IsImplicitVR() ) + 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 3.5-2001, section - // 7.1.2 Data element structure with explicit vr p27) must be + // The following reserved two bytes (see PS 3.5-2003, section + // "7.1.2 Data element structure with explicit vr", p 27) must be // skipped before proceeding on reading the length on 4 bytes. - fseek(Fp, 2L, SEEK_CUR); + Fp->seekg( 2L, std::ios::cur); uint32_t length32 = ReadInt32(); - if ( vr == "OB" && length32 == 0xffffffff ) + if ( (vr == "OB" || vr == "OW") && length32 == 0xffffffff ) { uint32_t lengthOB; try { - lengthOB = FindDocEntryLengthOB(); + lengthOB = FindDocEntryLengthOBOrOW(); } - catch ( gdcmFormatUnexpected ) + catch ( FormatUnexpected ) { // Computing the length failed (this happens with broken // files like gdcm-JPEG-LossLess3a.dcm). We still have a // 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. - long currentPosition = ftell(Fp); - fseek(Fp,0L,SEEK_END); - long lengthUntilEOF = ftell(Fp) - currentPosition; - fseek(Fp, currentPosition, SEEK_SET); + long currentPosition = Fp->tellg(); + Fp->seekg(0L,std::ios::end); + + long lengthUntilEOF = (long)(Fp->tellg())-currentPosition; + Fp->seekg(currentPosition, std::ios::beg); + + entry->SetReadLength(lengthUntilEOF); entry->SetLength(lengthUntilEOF); return; } + entry->SetReadLength(lengthOB); entry->SetLength(lengthOB); return; } @@ -1713,64 +1550,11 @@ void gdcmDocument::FindDocEntryLength( gdcmDocEntry *entry ) // Length is encoded on 2 bytes. length16 = ReadInt16(); - - // 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. - // 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 ) - { - if ( !IsExplicitVRBigEndianTransferSyntax() ) - { - throw gdcmFormatError( "gdcmDocument::FindDocEntryLength()", - " not explicit VR." ); - return; - } - length16 = 4; - SwitchSwapToBigEndian(); - // 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() ); - gdcmDictEntry* newTag = GetDictEntryByNumber( 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 ); - } - - // Heuristic: well, some files are really ill-formed. + + // 0xffff means that we deal with 'No Length' Sequence + // or 'No Length' SQItem if ( length16 == 0xffff) - { - // 0xffff means that we deal with 'Unknown Length' Sequence + { length16 = 0; } FixDocEntryFoundLength( entry, (uint32_t)length16 ); @@ -1784,6 +1568,9 @@ void gdcmDocument::FindDocEntryLength( gdcmDocEntry *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; @@ -1791,19 +1578,74 @@ void gdcmDocument::FindDocEntryLength( gdcmDocEntry *entry ) } /** - * \brief Find the Value Representation of the current Dicom Element. - * @param entry + * \brief Find the Length till the next sequence delimiter + * \warning NOT end user intended method ! + * @return */ -void gdcmDocument::FindDocEntryVR( gdcmDocEntry *entry ) +uint32_t Document::FindDocEntryLengthOBOrOW() + throw( FormatUnexpected ) { - if ( Filetype != gdcmExplicitVR ) + // 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 ) { - return; + 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; +} - char vr[3]; +/** + * \brief Find the Value Representation of the current Dicom Element. + * @return Value Representation of the current Entry + */ +std::string Document::FindDocEntryVR() +{ + if ( Filetype != ExplicitVR ) + return GDCM_UNKNOWN; - long positionOnEntry = ftell(Fp); + long positionOnEntry = Fp->tellg(); // Warning: we believe this is explicit VR (Value Representation) because // we used a heuristic that found "UL" in the first tag. Alas this // doesn't guarantee that all the tags will be in explicit VR. In some @@ -1811,105 +1653,31 @@ void gdcmDocument::FindDocEntryVR( gdcmDocEntry *entry ) // within an explicit VR file. Hence we make sure the present tag // is in explicit VR and try to fix things if it happens not to be // the case. - - fread (vr, (size_t)2,(size_t)1, Fp); + + char vr[3]; + Fp->read (vr, (size_t)2); vr[2] = 0; - if( !CheckDocEntryVR(entry, vr) ) + if ( !CheckDocEntryVR(vr) ) { - fseek(Fp, positionOnEntry, SEEK_SET); - // When this element is known in the dictionary we shall use, e.g. for - // the semantics (see the usage of IsAnInteger), the VR proposed by the - // dictionary entry. Still we have to flag the element as implicit since - // we know now our assumption on expliciteness is not furfilled. - // avoid . - if ( entry->IsVRUnknown() ) - { - entry->SetVR("Implicit"); - } - entry->SetImplicitVR(); + Fp->seekg(positionOnEntry, std::ios::beg); + return GDCM_UNKNOWN; } + return vr; } /** * \brief Check the correspondance between the VR of the header entry * and the taken VR. If they are different, the header entry is * updated with the new VR. - * @param entry Header Entry to check * @param vr Dicom Value Representation * @return false if the VR is incorrect of if the VR isn't referenced * otherwise, it returns true */ -bool gdcmDocument::CheckDocEntryVR(gdcmDocEntry *entry, gdcmVRKey vr) +bool Document::CheckDocEntryVR(VRKey vr) { - char msg[100]; // for sprintf - bool realExplicit = true; - - // Assume we are reading a falsely explicit VR file i.e. we reached - // a tag where we expect reading a VR but are in fact we read the - // first to bytes of the length. Then we will interogate (through find) - // the dicom_vr dictionary with oddities like "\004\0" which crashes - // both GCC and VC++ implementations of the STL map. Hence when the - // expected VR read happens to be non-ascii characters we consider - // we hit falsely explicit VR tag. - - if ( !isalpha(vr[0]) && !isalpha(vr[1]) ) - { - realExplicit = false; - } - - // 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 ( realExplicit && !gdcmGlobal::GetVR()->Count(vr) ) - { - realExplicit = false; - } - - if ( !realExplicit ) - { - // We thought this was explicit VR, but we end up with an - // implicit VR tag. Let's backtrack. - sprintf(msg,"Falsely explicit vr file (%04x,%04x)\n", - entry->GetGroup(), entry->GetElement()); - dbg.Verbose(1, "gdcmDocument::FindVR: ",msg); - - if( entry->GetGroup() % 2 && entry->GetElement() == 0x0000) - { - // Group length is UL ! - gdcmDictEntry* newEntry = NewVirtualDictEntry( - entry->GetGroup(), entry->GetElement(), - "UL", "FIXME", "Group Length"); - entry->SetDictEntry( newEntry ); - } + if ( !Global::GetVR()->IsValidVR(vr) ) return false; - } - - if ( entry->IsVRUnknown() ) - { - // When not a dictionary entry, we can safely overwrite the VR. - if( entry->GetElement() == 0x0000 ) - { - // Group length is UL ! - entry->SetVR("UL"); - } - else - { - entry->SetVR(vr); - } - } - else if ( entry->GetVR() != vr ) - { - // The VR present in the file and the dictionary disagree. We assume - // the file writer knew best and use the VR of the file. Since it would - // be unwise to overwrite the VR of a dictionary (since it would - // compromise it's next user), we need to clone the actual DictEntry - // and change the VR for the read one. - gdcmDictEntry* newEntry = NewVirtualDictEntry( - entry->GetGroup(), entry->GetElement(), - vr, "FIXME", entry->GetName()); - entry->SetDictEntry(newEntry); - } return true; } @@ -1921,11 +1689,11 @@ bool gdcmDocument::CheckDocEntryVR(gdcmDocEntry *entry, gdcmVRKey vr) * @param entry entry to tranform * @return Transformed entry value */ -std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *entry) +std::string Document::GetDocEntryValue(DocEntry *entry) { if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() ) { - std::string val = ((gdcmValEntry *)entry)->GetValue(); + std::string val = ((ValEntry *)entry)->GetValue(); std::string vr = entry->GetVR(); uint32_t length = entry->GetLength(); std::ostringstream s; @@ -1938,14 +1706,14 @@ std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *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 << '\\'; } @@ -1961,14 +1729,14 @@ std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *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 << '\\'; } @@ -1985,8 +1753,7 @@ std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *entry) #endif //GDCM_NO_ANSI_STRING_STREAM return s.str(); } - - return ((gdcmValEntry *)entry)->GetValue(); + return ((ValEntry *)entry)->GetValue(); } /** @@ -1997,7 +1764,7 @@ std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *entry) * @param entry Entry to reverse transform * @return Reverse transformed entry value */ -std::string gdcmDocument::GetDocEntryUnvalue(gdcmDocEntry* entry) +std::string Document::GetDocEntryUnvalue(DocEntry *entry) { if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() ) { @@ -2010,7 +1777,7 @@ std::string gdcmDocument::GetDocEntryUnvalue(gdcmDocEntry* entry) uint16_t newInt16; tokens.erase( tokens.begin(), tokens.end()); // clean any previous value - Tokenize (((gdcmValEntry *)entry)->GetValue(), tokens, "\\"); + Util::Tokenize (((ValEntry *)entry)->GetValue(), tokens, "\\"); for (unsigned int i=0; iGetValue(), tokens, "\\"); + Util::Tokenize (((ValEntry *)entry)->GetValue(), tokens, "\\"); for (unsigned int i=0; iGetValue(); + return ((ValEntry *)entry)->GetValue(); } /** @@ -2050,20 +1817,27 @@ std::string gdcmDocument::GetDocEntryUnvalue(gdcmDocEntry* entry) * \warning NOT end user intended method ! * @param entry entry to skip */ -void gdcmDocument::SkipDocEntry(gdcmDocEntry *entry) +void Document::SkipDocEntry(DocEntry *entry) { SkipBytes(entry->GetLength()); } /** - * \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 entry entry to skip + * @param currentDocEntry entry to skip */ -void gdcmDocument::SkipToNextDocEntry(gdcmDocEntry *entry) +void Document::SkipToNextDocEntry(DocEntry *currentDocEntry) { - fseek(Fp, (long)(entry->GetOffset()), SEEK_SET); - fseek(Fp, (long)(entry->GetReadLength()), SEEK_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); + } } /** @@ -2071,10 +1845,10 @@ void gdcmDocument::SkipToNextDocEntry(gdcmDocEntry *entry) * the parser went Jabberwocky) one can hope improving things by * applying some heuristics. * @param entry entry to check - * @param foundLength fist assumption about length + * @param foundLength first assumption about length */ -void gdcmDocument::FixDocEntryFoundLength(gdcmDocEntry *entry, - uint32_t foundLength) +void Document::FixDocEntryFoundLength(DocEntry *entry, + uint32_t foundLength) { entry->SetReadLength( foundLength ); // will be updated only if a bug is found if ( foundLength == 0xffffffff) @@ -2082,243 +1856,106 @@ void gdcmDocument::FixDocEntryFoundLength(gdcmDocEntry *entry, foundLength = 0; } - uint16_t gr = entry->GetGroup(); - uint16_t el = entry->GetElement(); + uint16_t gr = entry->GetGroup(); + uint16_t elem = entry->GetElement(); if ( foundLength % 2) { - std::ostringstream s; - s << "Warning : Tag with uneven length " - << foundLength - << " in x(" << std::hex << gr << "," << el <<")" << std::dec; - dbg.Verbose(0, s.str().c_str()); + gdcmWarningMacro( "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 user (we are talking a few years, here)... - // Note: XMedCom probably uses such a trick since it is able to read + // 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) { - // Only happens for this length ! - if ( entry->GetGroup() != 0x0008 - || ( entry->GetElement() != 0x0070 - && entry->GetElement() != 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 ( entry->GetGroup() == 0x0009 - && ( entry->GetElement() == 0x1113 - || entry->GetElement() == 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( entry->GetGroup() == 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->SetUsableLength(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 gdcmDocument::IsDocEntryAnInteger(gdcmDocEntry *entry) -{ - uint16_t element = entry->GetElement(); - uint16_t group = entry->GetGroup(); - 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 ( element == 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). - std::ostringstream s; - int filePosition = ftell(Fp); - s << "Erroneous Group Length element length on : (" \ - << std::hex << group << " , " << element - << ") -before- position x(" << filePosition << ")" - << "lgt : " << length; - dbg.Verbose(0, "gdcmDocument::IsDocEntryAnInteger", s.str().c_str() ); - } - } - - 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 gdcmDocument::FindDocEntryLengthOB() - throw( gdcmFormatUnexpected ) -{ - // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data. - long positionOnEntry = ftell(Fp); - bool foundSequenceDelimiter = false; - uint32_t totalLength = 0; - - while ( !foundSequenceDelimiter ) - { - uint16_t group; - uint16_t elem; - try - { - group = ReadInt16(); - elem = ReadInt16(); - } - catch ( gdcmFormatError ) - { - throw gdcmFormatError("gdcmDocument::FindDocEntryLengthOB()", - " group or element not present."); - } - - // We have to decount the group and element we just read - totalLength += 4; - - if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) ) - { - dbg.Verbose(1, "gdcmDocument::FindDocEntryLengthOB: neither an Item " - "tag nor a Sequence delimiter tag."); - fseek(Fp, positionOnEntry, SEEK_SET); - throw gdcmFormatUnexpected("gdcmDocument::FindDocEntryLengthOB()", - "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 ) + // Only happens for this length ! + if ( gr != 0x0008 || ( elem != 0x0070 && elem != 0x0080 ) ) { - break; + foundLength = 10; + entry->SetReadLength(10); // a bug is to be fixed !? } } - fseek(Fp, positionOnEntry, SEEK_SET); - return totalLength; -} -/** - * \brief Reads a supposed to be 16 Bits integer - * (swaps it depending on processor endianity) - * @return read value - */ -uint16_t gdcmDocument::ReadInt16() - throw( gdcmFormatError ) -{ - uint16_t g; - size_t item_read = fread (&g, (size_t)2,(size_t)1, Fp); - if ( item_read != 1 ) + //////// 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 ) ) { - if( ferror(Fp) ) - { - throw gdcmFormatError( "gdcmDocument::ReadInt16()", " file error." ); - } - throw gdcmFormatError( "gdcmDocument::ReadInt16()", "EOF." ); - } - g = SwapShort(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 Reads a supposed to be 32 Bits integer - * (swaps it depending on processor endianity) - * @return read value + * \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. */ -uint32_t gdcmDocument::ReadInt32() - throw( gdcmFormatError ) +bool Document::IsDocEntryAnInteger(DocEntry *entry) { - uint32_t g; - size_t item_read = fread (&g, (size_t)4,(size_t)1, Fp); - if ( item_read != 1 ) - { - if( ferror(Fp) ) + 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 { - throw gdcmFormatError( "gdcmDocument::ReadInt16()", " file error." ); + // 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 ); } - throw gdcmFormatError( "gdcmDocument::ReadInt32()", "EOF." ); } - g = SwapLong(g); - return g; -} - -/** - * \brief skips bytes inside the source file - * \warning NOT end user intended method ! - * @return - */ -void gdcmDocument::SkipBytes(uint32_t nBytes) -{ - //FIXME don't dump the returned value - (void)fseek(Fp, (long)nBytes, SEEK_CUR); -} -/** - * \brief Loads all the needed Dictionaries - * \warning NOT end user intended method ! - */ -void gdcmDocument::Initialise() -{ - RefPubDict = gdcmGlobal::GetDicts()->GetDefaultPubDict(); - RefShaDict = NULL; + if ( vr == "UL" || vr == "US" || vr == "SL" || vr == "SS" ) + { + return true; + } + return false; } /** @@ -2329,51 +1966,35 @@ void gdcmDocument::Initialise() * it's neither ACR-NEMA nor DICOM * true when we hope ours assuptions are OK */ -bool gdcmDocument::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 +bool Document::CheckSwap() +{ uint32_t s32; uint16_t s16; - char deb[HEADER_LENGTH_TO_READ]; + char deb[256]; // 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 DICOM header, since it possesses a - // file preamble where it suffice to look for the string "DICM". - int lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, Fp); - (void)lgrLue; //FIXME not used + // 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 ) { - dbg.Verbose(1, "gdcmDocument::CheckSwap:", "looks like DICOM Version3"); + gdcmWarningMacro( "Looks like DICOM Version3 (preamble + DCM)" ); - // Next, determine the value representation (VR). Let's skip to the - // first element (0002, 0000) and check there if we find "UL" - // - or "OB" if the 1st one is (0002,0001) -, + // Group 0002 should always be VR, and the first element 0000 + // Let's be carefull (so many wrong headers ...) + // and determine the value representation (VR) : + // Let's skip to the first element (0002,0000) and check there if we find + // "UL" - or "OB" if the 1st one is (0002,0001) -, // in which case we (almost) know it is explicit VR. // WARNING: if it happens to be implicit VR then what we will read // is the length of the group. If this ascii representation of this // length happens to be "UL" then we shall believe it is explicit VR. - // FIXME: in order to fix the above warning, we could read the next - // element value (or a couple of elements values) in order to make - // sure we are not commiting a big mistake. // We need to skip : // * the 128 bytes of File Preamble (often padded with zeroes), // * the 4 bytes of "DICM" string, @@ -2381,59 +2002,92 @@ bool gdcmDocument::CheckSwap() // i.e. a total of 136 bytes. entCur = deb + 136; - // FIXME : FIXME: - // Sometimes (see : gdcmData/icone.dcm) group 0x0002 *is* Explicit VR, - // but elem 0002,0010 (Transfert Syntax) tells us the file is - // *Implicit* VR. -and it is !- + // group 0x0002 *is always* Explicit VR Sometimes , + // 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 gdcmDocument::dicom_vr to test all the possibilities + // Use Document::dicom_vr to test all the possibilities // instead of just checking for UL, OB and UI !? group 0000 { - Filetype = gdcmExplicitVR; - dbg.Verbose(1, "gdcmDocument::CheckSwap:", - "explicit Value Representation"); + Filetype = ExplicitVR; + gdcmWarningMacro( "Group 0002 : Explicit Value Representation"); } else { - Filetype = gdcmImplicitVR; - dbg.Verbose(1, "gdcmDocument::CheckSwap:", - "not an explicit Value Representation"); + Filetype = ImplicitVR; + gdcmWarningMacro( "Group 0002 :Not an explicit Value Representation;" + << "Looks like a bugged Header!"); } if ( net2host ) { SwapCode = 4321; - dbg.Verbose(1, "gdcmDocument::CheckSwap:", - "HostByteOrder != NetworkByteOrder"); + gdcmWarningMacro( "HostByteOrder != NetworkByteOrder"); } else { - SwapCode = 0; - dbg.Verbose(1, "gdcmDocument::CheckSwap:", - "HostByteOrder = NetworkByteOrder"); + SwapCode = 1234; + gdcmWarningMacro( "HostByteOrder = NetworkByteOrder"); } - // Position the file position indicator at first tag (i.e. - // after the file preamble and the "DICM" string). - rewind(Fp); - fseek (Fp, 132L, SEEK_SET); + // Position the file position indicator at first tag + // (i.e. after the file preamble and the "DICM" string). + + 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). - dbg.Verbose(1, "gdcmDocument::CheckSwap:", "not a DICOM Version3 file"); - rewind(Fp); + + 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. @@ -2442,24 +2096,23 @@ bool gdcmDocument::CheckSwap() // representation of a 32 bits integer. Hence the following dirty // trick : s32 = *((uint32_t *)(entCur)); - switch( s32 ) { case 0x00040000 : SwapCode = 3412; - Filetype = gdcmACR; + Filetype = ACR; return true; case 0x04000000 : SwapCode = 4321; - Filetype = gdcmACR; + Filetype = ACR; return true; case 0x00000400 : SwapCode = 2143; - Filetype = gdcmACR; + Filetype = ACR; return true; case 0x00000004 : - SwapCode = 0; - Filetype = gdcmACR; + SwapCode = 1234; + Filetype = ACR; return true; default : // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file. @@ -2467,13 +2120,13 @@ bool gdcmDocument::CheckSwap() // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA, // i.e. the 'group length' element is not present : - // check the supposed to be 'group number' - // 0x0002 or 0x0004 or 0x0008 + // check the supposed-to-be 'group number' + // in ( 0x0001 .. 0x0008 ) // to determine ' SwapCode' value . // Only 0 or 4321 will be possible // (no oportunity to check for the formerly well known // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian' - // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -4, 8-) + // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -3, 4, ..., 8-) // the file IS NOT ACR-NEMA nor DICOM V3 // Find a trick to tell it the caller... @@ -2481,48 +2134,50 @@ bool gdcmDocument::CheckSwap() switch ( s16 ) { + case 0x0001 : case 0x0002 : + case 0x0003 : case 0x0004 : - case 0x0008 : - SwapCode = 0; - Filetype = gdcmACR; + case 0x0005 : + case 0x0006 : + case 0x0007 : + case 0x0008 : + SwapCode = 1234; + Filetype = ACR; return true; + case 0x0100 : case 0x0200 : + case 0x0300 : case 0x0400 : - case 0x0800 : + case 0x0500 : + case 0x0600 : + case 0x0700 : + case 0x0800 : SwapCode = 4321; - Filetype = gdcmACR; + Filetype = ACR; return true; default : - dbg.Verbose(0, "gdcmDocument::CheckSwap:", - "ACR/NEMA unfound swap info (Really hopeless !)"); - Filetype = gdcmUnknown; + gdcmWarningMacro( "ACR/NEMA unfound swap info (Really hopeless !)"); + Filetype = Unknown; return false; } - // Then the only info we have is the net2host one. - //if (! net2host ) - // SwapCode = 0; - //else - // SwapCode = 4321; - //return; } } /** - * \brief Restore the unproperly loaded values i.e. the group, the element - * and the dictionary entry depending on them. + * \brief Change the Byte Swap code. */ -void gdcmDocument::SwitchSwapToBigEndian() +void Document::SwitchByteSwapCode() { - dbg.Verbose(1, "gdcmDocument::SwitchSwapToBigEndian", - "Switching to BigEndian mode."); - if ( SwapCode == 0 ) + gdcmWarningMacro( "Switching Byte Swap code from "<< SwapCode + << " at :" <tellg() ); + if ( SwapCode == 1234 ) { SwapCode = 4321; } else if ( SwapCode == 4321 ) { - SwapCode = 0; + SwapCode = 1234; } else if ( SwapCode == 3412 ) { @@ -2536,9 +2191,9 @@ void gdcmDocument::SwitchSwapToBigEndian() /** * \brief during parsing, Header Elements too long are not loaded in memory - * @param newSize + * @param newSize new size */ -void gdcmDocument::SetMaxSizeLoadEntry(long newSize) +void Document::SetMaxSizeLoadEntry(long newSize) { if ( newSize < 0 ) { @@ -2552,37 +2207,14 @@ void gdcmDocument::SetMaxSizeLoadEntry(long newSize) MaxSizeLoadEntry = newSize; } - -/** - * \brief Header Elements too long will not be printed - * \todo See comments of \ref gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE - * @param newSize - */ -void gdcmDocument::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 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 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. */ -gdcmDocEntry* gdcmDocument::ReadNextDocEntry() +DocEntry *Document::ReadNextDocEntry() { uint16_t group; uint16_t elem; @@ -2592,302 +2224,150 @@ gdcmDocEntry* gdcmDocument::ReadNextDocEntry() group = ReadInt16(); elem = ReadInt16(); } - catch ( gdcmFormatError e ) + catch ( FormatError e ) { // We reached the EOF (or an error occured) therefore // header parsing has to be considered as finished. - //std::cout << e; return 0; } - gdcmDocEntry *newEntry = NewDocEntryByNumber(group, elem); - FindDocEntryVR(newEntry); + // Sometimes file contains groups of tags with reversed endianess. + HandleBrokenEndian(group, elem); - try - { - FindDocEntryLength(newEntry); - } - catch ( gdcmFormatError e ) + // 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 ) { - // Call it quits - //std::cout << e; - delete newEntry; - return 0; + if ( elem == 0x0000 ) // Group Length + realVR = "UL"; // must be UL + else + { + DictEntry *dictEntry = GetDictEntry(group,elem); + if ( dictEntry ) + { + realVR = dictEntry->GetVR(); + } + } } - newEntry->SetOffset(ftell(Fp)); - - return newEntry; -} - + DocEntry *newEntry; + if ( Global::GetVR()->IsVROfSequence(realVR) ) + newEntry = NewSeqEntry(group, elem); + else if ( Global::GetVR()->IsVROfStringRepresentable(realVR) ) + newEntry = NewValEntry(group, elem,vr); + else + newEntry = NewBinEntry(group, elem,vr); -/** - * \brief Generate a free gdcmTagKey i.e. a gdcmTagKey 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 gdcmDocument::GenerateFreeTagKeyInGroup(uint16_t group) -{ - for (uint32_t elem = 0; elem < UINT32_MAX; elem++) + if ( vr == GDCM_UNKNOWN ) { - gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, elem); - if (TagHT.count(key) == 0) + if ( Filetype == ExplicitVR ) { - return elem; + // 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; + 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(); } - return UINT32_MAX; -} - -/** - * \brief Assuming the internal file pointer \ref gdcmDocument::Fp - * is placed at the beginning of a tag check whether this - * tag is (TestGroup, TestElement). - * \warning On success the internal file pointer \ref gdcmDocument::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 gdcmDocument::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 gdcmDocument::ReadTag(uint16_t testGroup, uint16_t testElement) -{ - long positionOnEntry = ftell(Fp); - long currentPosition = ftell(Fp); // On debugging purposes - //// Read the Item Tag group and element, and make - // sure they are what we expected: - uint16_t itemTagGroup = ReadInt16(); - uint16_t itemTagElement = ReadInt16(); - if ( itemTagGroup != testGroup || itemTagElement != testElement ) + try { - std::ostringstream s; - s << " We should have found tag ("; - s << std::hex << testGroup << "," << testElement << ")" << std::endl; - s << " but instead we encountered tag ("; - s << std::hex << itemTagGroup << "," << itemTagElement << ")" - << std::endl; - s << " at address: " << (unsigned)currentPosition << std::endl; - dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: wrong Item Tag found:"); - dbg.Verbose(0, s.str().c_str()); - fseek(Fp, positionOnEntry, SEEK_SET); - - return false; + FindDocEntryLength(newEntry); } - return true; + catch ( FormatError e ) + { + // Call it quits + delete newEntry; + return 0; + } + + newEntry->SetOffset(Fp->tellg()); + + return newEntry; } /** - * \brief Assuming the internal file pointer \ref gdcmDocument::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 gdcmDocument::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 gdcmDocument::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. + * \brief Handle broken private tag from Philips NTSCAN + * where the endianess is being switched to BigEndian + * for no apparent reason + * @return no return */ -uint32_t gdcmDocument::ReadTagLength(uint16_t testGroup, uint16_t testElement) +void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem) { - long positionOnEntry = ftell(Fp); - (void)positionOnEntry; - - if ( !ReadTag(testGroup, testElement) ) + // 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 0; - } - - //// Then read the associated Item Length - long currentPosition = ftell(Fp); - uint32_t itemLength = ReadInt32(); + // start endian swap mark for group found + reversedEndian++; + SwitchByteSwapCode(); + // fix the tag + group = 0xfffe; + elem = 0xe000; + } + else if (group == 0xfffe && elem == 0xe00d && reversedEndian) { - std::ostringstream s; - s << "Basic Item Length is: " - << itemLength << std::endl; - s << " at address: " << (unsigned)currentPosition << std::endl; - dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: ", s.str().c_str()); + // end of reversed endian group + reversedEndian--; + SwitchByteSwapCode(); } - return itemLength; } /** - * \brief Parse pixel data from disk for multi-fragment Jpeg/Rle files - * No other way so 'skip' the Data - * + * \brief Group 0002 is always coded Little Endian + * whatever Transfer Syntax is + * @return no return */ -void gdcmDocument::Parse7FE0 () +void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) { - gdcmDocEntry* element = GetDocEntryByNumber(0x0002, 0x0010); - if ( !element ) - { - // Should warn user FIXME - return; - } - - if ( IsImplicitVRLittleEndianTransferSyntax() - || IsExplicitVRLittleEndianTransferSyntax() - || IsExplicitVRBigEndianTransferSyntax() /// \todo 1.2.2 ??? A verifier ! - || IsDeflatedExplicitVRLittleEndianTransferSyntax() ) - { - return; - } - - // ---------------- for Parsing : Position on begining of Jpeg/RLE Pixels - - //// Read the Basic Offset Table Item Tag length... - uint32_t itemLength = ReadTagLength(0xfffe, 0xe000); - - //// ... and then read length[s] itself[themselves]. We don't use - // the values read (BTW what is the purpous of those lengths ?) - if ( itemLength != 0 ) + // Endian reversion. Some files contain groups of tags with reversed endianess. + if ( !Group0002Parsed && group != 0x0002) { - // BTW, what is the purpous of those length anyhow !? - char* basicOffsetTableItemValue = new char[itemLength + 1]; - fread(basicOffsetTableItemValue, itemLength, 1, Fp); + Group0002Parsed = true; + // we just came out of group 0002 + // if Transfer syntax is Big Endian we have to change CheckSwap - for (unsigned int i=0; i < itemLength; i += 4 ) + std::string ts = GetTransferSyntax(); + if ( !Global::GetTS()->IsTransferSyntax(ts) ) { - uint32_t individualLength = str2num(&basicOffsetTableItemValue[i],uint32_t); - std::ostringstream s; - s << " Read one length: "; - s << std::hex << individualLength << std::endl; - dbg.Verbose(0, "gdcmDocument::Parse7FE0: ", s.str().c_str()); + gdcmWarningMacro("True DICOM File, with NO Tansfer Syntax: " << ts ); + return; } - delete[] basicOffsetTableItemValue; - } - if ( ! IsRLELossLessTransferSyntax() ) - { - // JPEG Image - - //// We then skip (not reading them) all the fragments of images: - while ( (itemLength = ReadTagLength(0xfffe, 0xe000)) ) - { - SkipBytes(itemLength); - } - } - else - { - // RLE Image - long ftellRes; - long rleSegmentLength[15], fragmentLength; + // Group 0002 is always 'Explicit ...' enven when Transfer Syntax says 'Implicit ..." - // While we find some items: - while ( (fragmentLength = ReadTagLength(0xfffe, 0xe000)) ) - { - // Parse fragments of the current Fragment (Frame) - //------------------ scanning (not reading) fragment pixels - uint32_t nbRleSegments = ReadInt32(); - - //// Reading RLE Segments Offset Table - uint32_t rleSegmentOffsetTable[15]; - for(int k=1; k<=15; k++) + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ImplicitVRLittleEndian ) { - ftellRes = ftell(Fp); - rleSegmentOffsetTable[k] = ReadInt32(); + Filetype = ImplicitVR; } - - // skipping (not reading) RLE Segments - if ( nbRleSegments > 1) - { - for(unsigned int k = 1; k <= nbRleSegments-1; k++) - { - rleSegmentLength[k] = rleSegmentOffsetTable[k+1] - - rleSegmentOffsetTable[k]; - ftellRes = ftell(Fp); - SkipBytes(rleSegmentLength[k]); - } - } - - rleSegmentLength[nbRleSegments] = fragmentLength - - rleSegmentOffsetTable[nbRleSegments]; - ftellRes = ftell(Fp); - SkipBytes(rleSegmentLength[nbRleSegments]); - } - - // Make sure that at the end of the item we encounter a 'Sequence - // Delimiter Item': - if ( !ReadTag(0xfffe, 0xe0dd) ) - { - dbg.Verbose(0, "gdcmDocument::Parse7FE0: no sequence delimiter item"); - dbg.Verbose(0, " at end of RLE item sequence"); - } - } -} - - - -/** - * \brief Compares two documents, according to \ref gdcmDicomDir 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 gdcmDocument::operator<(gdcmDocument &document) -{ - // Patient Name - std::string s1 = GetEntryByNumber(0x0010,0x0010); - std::string s2 = document.GetEntryByNumber(0x0010,0x0010); - if(s1 < s2) - { - return true; - } - else if( s1 > s2 ) - { - return false; - } - else - { - // Patient ID - s1 = GetEntryByNumber(0x0010,0x0020); - s2 = document.GetEntryByNumber(0x0010,0x0020); - if ( s1 < s2 ) - { - return true; - } - else if ( s1 > s2 ) - { - return false; - } - else + + // FIXME Strangely, this works with + //'Implicit VR Transfer Syntax (GE Private) + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ExplicitVRBigEndian ) { - // Study Instance UID - s1 = GetEntryByNumber(0x0020,0x000d); - s2 = document.GetEntryByNumber(0x0020,0x000d); - if ( s1 < s2 ) - { - return true; - } - else if( s1 > s2 ) - { - return false; - } - else - { - // Serie Instance UID - s1 = GetEntryByNumber(0x0020,0x000e); - s2 = document.GetEntryByNumber(0x0020,0x000e); - if ( s1 < s2 ) - { - return true; - } - else if( s1 > s2 ) - { - return false; - } - } + gdcmWarningMacro("Transfer Syntax Name = [" + << GetTransferSyntaxName() << "]" ); + SwitchByteSwapCode(); + group = SwapShort(group); + elem = SwapShort(elem); } } - return false; } +//----------------------------------------------------------------------------- +// Print //----------------------------------------------------------------------------- +} // end namespace gdcm