X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=src%2FgdcmDocument.cxx;h=f87337747ee26e66f5ff2c46f3f66208e3463044;hb=985954a7af00b467b183bb3f5db9dc11f89fcaf5;hp=66abee0e052cfcdbb41ab2ea11c53301bc54d119;hpb=e40fc77cef3155aab87305ce2f8f14d1acbf158f;p=gdcm.git diff --git a/src/gdcmDocument.cxx b/src/gdcmDocument.cxx index 66abee0e..f8733774 100644 --- a/src/gdcmDocument.cxx +++ b/src/gdcmDocument.cxx @@ -3,12 +3,12 @@ Program: gdcm Module: $RCSfile: gdcmDocument.cxx,v $ Language: C++ - Date: $Date: 2004/07/02 13:55:27 $ - Version: $Revision: 1.46 $ + Date: $Date: 2005/04/05 10:20:53 $ + Version: $Revision: 1.234 $ 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 @@ -20,215 +20,179 @@ #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 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 Document::CheckSwap() +//const unsigned int Document::HEADER_LENGTH_TO_READ = 256; -// Refer to gdcmDocument::SetMaxSizeLoadEntry() -const unsigned int gdcmDocument::MAX_SIZE_LOAD_ELEMENT_VALUE = 0x7fffffff;// 4096;// -const unsigned int gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff;//64; +// 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 constructor - * @param inFilename file to be opened for parsing - * @param exception_on_error whether we throw an exception or not - * @param enable_sequences = true to allow the header - * to be parsed *inside* the SeQuences, - * when they have an actual length - * \warning enable_sequences *has to be* true for reading PAPYRUS 3.0 files - * @param ignore_shadow to allow skipping the shadow elements, - * to save memory space. - * \warning The TRUE value for this param has to be used - * with a FALSE value for the 'enable_sequence' param. - * ('public elements' may be embedded in 'shadow Sequences') + * \brief This default constructor neither loads nor parses the file. + * You should then invoke \ref Document::Load. + * */ -gdcmDocument::gdcmDocument(std::string const & inFilename, - bool exception_on_error, - bool enable_sequences, - bool ignore_shadow) - : gdcmElementSet(-1) { - IgnoreShadow =ignore_shadow; - //enableSequences=enable_sequences; - (void)enable_sequences; - enableSequences=true; // JPR // TODO : remove params out of the constructor - SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); - Filename = inFilename; - Initialise(); - - if ( !OpenFile(exception_on_error)) - return; - - 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; - - long l=ParseDES( this, beg, lgt, false); // le Load sera fait a la volee - (void)l; //is l used anywhere ? - - rewind(fp); - - // Load 'non string' values - - std::string PhotometricInterpretation = GetEntryByNumber(0x0028,0x0004); - if( PhotometricInterpretation == "PALETTE COLOR " ) { - LoadEntryVoidArea(0x0028,0x1200); // gray LUT - LoadEntryVoidArea(0x0028,0x1201); // R LUT - LoadEntryVoidArea(0x0028,0x1202); // G LUT - LoadEntryVoidArea(0x0028,0x1203); // B LUT - - LoadEntryVoidArea(0x0028,0x1221); // Segmented Red Palette Color LUT Data - LoadEntryVoidArea(0x0028,0x1222); // Segmented Green Palette Color LUT Data - LoadEntryVoidArea(0x0028,0x1223); // Segmented Blue Palette Color LUT Data - } - //FIXME later : how to use it? - LoadEntryVoidArea(0x0028,0x3006); //LUT Data (CTX dependent) - - CloseFile(); - - // -------------------------------------------------------------- - // Special Patch to allow gdcm to read ACR-LibIDO formated 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 - 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); - } - // ----------------- End of Special Patch ---------------- +Document::Document() + :ElementSet(-1) +{ + Fp = 0; - printLevel = 1; // 'Medium' print level by default + SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); + Initialize(); + SwapCode = 1234; + Filetype = ExplicitVR; + Group0002Parsed = false; + LoadMode = 0x00000000; // default : load everything, later } /** - * \brief constructor - * @param exception_on_error + * \brief Constructor (not to break the API) + * @param filename 'Document' (File or DicomDir) to be opened for parsing */ -gdcmDocument::gdcmDocument(bool exception_on_error) - :gdcmElementSet(-1) +Document::Document( std::string const &filename ) + :ElementSet(-1) { - (void)exception_on_error; - //enableSequences=0; // ?!? JPR + Fp = 0; SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); - Initialise(); + Initialize(); + SwapCode = 1234; + Filetype = ExplicitVR; + Group0002Parsed = false; + LoadMode = 0x00000000; // Load everything - printLevel = 1; // 'Medium' print level by default + Load(filename); } - /** * \brief Canonical destructor. */ -gdcmDocument::~gdcmDocument () +Document::~Document () { RefPubDict = NULL; RefShaDict = NULL; - - // Recursive clean up of sequences - for (TagDocEntryHT::iterator it = tagHT.begin(); it != tagHT.end(); ++it ) - { - delete it->second; - } - tagHT.clear(); } //----------------------------------------------------------------------------- -// Print +// Public /** - * \brief Prints The Dict Entries of THE public Dicom Dictionary - * @return - */ -void gdcmDocument::PrintPubDict(std::ostream & os) + * \brief Loader + * @param filename 'Document' (File or DicomDir) to be opened for parsing + */ +void Document::Load( std::string const &filename ) { - RefPubDict->Print(os); -} + Filename = filename; -/** - * \brief Prints The Dict Entries of THE shadow Dicom Dictionary - * @return - */ -void gdcmDocument::PrintShaDict(std::ostream & os) -{ - RefShaDict->Print(os); + // We should clean out anything that already exists. + + Fp = 0; + if ( !OpenFile() ) + { + return; + } + + 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(); + long beg = Fp->tellg(); // just after DICOM preamble (if any) + + lgt -= beg; // remaining length to parse + + ParseDES( this, beg, lgt, false); // Loading is done during parsing + + Fp->seekg( 0, std::ios::beg); + + // Load 'non string' values + + std::string PhotometricInterpretation = GetEntryValue(0x0028,0x0004); + if( PhotometricInterpretation == "PALETTE COLOR " ) + { + LoadEntryBinArea(0x0028,0x1200); // gray LUT + /// FIXME FIXME FIXME + /// The tags refered by the three following lines used to be CORRECTLY + /// defined as having an US Value Representation in the public + /// dictionary. BUT the semantics implied by the three following + /// lines state that the corresponding tag contents are in fact + /// the ones of a BinEntry. + /// In order to fix things "Quick and Dirty" the 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 + /// 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 + LoadEntryBinArea(0x0028,0x1221); + // Segmented Green Palette Color LUT Data + LoadEntryBinArea(0x0028,0x1222); + // Segmented Blue Palette Color LUT Data + LoadEntryBinArea(0x0028,0x1223); + } + //FIXME later : how to use it? + LoadEntryBinArea(0x0028,0x3006); //LUT Data (CTX dependent) + + 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 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 = GetEntryValue(0x0008, 0x0010); // recognition code (RET) + if (RecCode == "ACRNEMA_LIBIDO_1.1" || + RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares + // with "little-endian strings" + { + Filetype = ACR_LIBIDO; + std::string rows = GetEntryValue(0x0028, 0x0010); + std::string columns = GetEntryValue(0x0028, 0x0011); + SetValEntry(columns, 0x0028, 0x0010); + SetValEntry(rows , 0x0028, 0x0011); + } + // --- End of ACR-LibIDO kludge --- } -//----------------------------------------------------------------------------- -// Public /** * \brief Get the public dictionary used */ -gdcmDict *gdcmDocument::GetPubDict() +Dict *Document::GetPubDict() { return RefPubDict; } @@ -236,1164 +200,885 @@ 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; + RefShaDict = dict; return !RefShaDict; } /** * \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 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 + * decides whether or not the current Document was properly parsed * and contains the mandatory information for being considered as * a well formed and usable Dicom/Acr File. - * @return true when gdcmDocument is the one of a reasonable Dicom/Acr file, + * @return true when Document is the one of a reasonable Dicom/Acr file, * false otherwise. */ -bool gdcmDocument::IsReadable() { - - if(Filetype==gdcmUnknown) { - std::cout << " gdcmDocument::IsReadable: Filetype " << Filetype - << " " << "gdcmUnknown " << gdcmUnknown << std::endl; //JPR - dbg.Verbose(0, "gdcmDocument::IsReadable: wrong filetype"); +bool Document::IsReadable() +{ + if( Filetype == Unknown) + { + 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); + return true; } - /** - * \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 don't match. + * \brief Predicate for dicom version 3 file. + * @return True when the file is a dicom version 3. */ -bool gdcmDocument::IsGivenTransferSyntax(const std::string & SyntaxToCheck) +bool Document::IsDicomV3() { - gdcmDocEntry *Entry = GetDocEntryByNumber(0x0002, 0x0010); - if ( !Entry ) - return false; - - // The entry might be present but not loaded (parsing and loading - // happen at differente stages): try loading and proceed with check... - LoadDocEntrySafe(Entry); - if (gdcmValEntry* ValEntry = dynamic_cast< gdcmValEntry* >(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... - while ( ! isdigit(Transfer[Transfer.length()-1]) ) - { - Transfer.erase(Transfer.length()-1, 1); - } - if ( Transfer == SyntaxToCheck ) - return true; - } - return false; + // Checking if Transfer Syntax exists is enough + // Anyway, it's to late check if the 'Preamble' was found ... + // And ... would it be a rich idea to check ? + // (some 'no Preamble' DICOM images exist !) + return GetDocEntry(0x0002, 0x0010) != NULL; } /** - * \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. + * \brief Predicate for Papyrus file + * Dedicated to whomsoever it may concern + * @return True when the file is a Papyrus file. */ -bool gdcmDocument::IsImplicitVRLittleEndianTransferSyntax() +bool Document::IsPapyrus() { - return IsGivenTransferSyntax(UI1_2_840_10008_1_2); + // 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 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. + * \brief returns the File Type + * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown) + * @return the FileType code */ -bool gdcmDocument::IsExplicitVRLittleEndianTransferSyntax() +FileType Document::GetFileType() { - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1); + return Filetype; } /** - * \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. + * \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::IsDeflatedExplicitVRLittleEndianTransferSyntax() +std::string Document::GetTransferSyntax() { - return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1_99); -} + DocEntry *entry = GetDocEntry(0x0002, 0x0010); + if ( !entry ) + { + return GDCM_UNKNOWN; + } -/** - * \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); + // The entry might be present but not loaded (parsing and loading + // happen at different stages): try loading and proceed with check... + LoadDocEntrySafe(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 GDCM_UNKNOWN; + } + while ( !isdigit((unsigned char)transfer[transfer.length()-1]) ) + { + transfer.erase(transfer.length()-1, 1); + } + return transfer; + } + 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) ); + // Global::GetTS() is a global static you shall never try to delete it! + return tsName; } - +// +// --------------- Swap Code ------------------ /** - * \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. + * \brief Swaps the bytes so they agree with the processor order + * @return The properly swaped 16 bits integer. */ -bool gdcmDocument::IsJPEG2000() +uint16_t Document::SwapShort(uint16_t a) { - return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_90) - || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_91) ); + if ( SwapCode == 4321 || SwapCode == 2143 ) + { + //a = ((( a << 8 ) & 0xff00 ) | (( a >> 8 ) & 0x00ff ) ); + // Save CPU time + a = ( a << 8 ) | ( a >> 8 ); + } + return a; } /** - * \brief Predicate for dicom version 3 file. - * @return True when the file is a dicom version 3. + * \brief Swaps back the bytes of 4-byte long integer accordingly to + * processor order. + * @return The properly swaped 32 bits integer. */ -bool gdcmDocument::IsDicomV3() +uint32_t Document::SwapLong(uint32_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); -} + 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 returns the File Type - * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown) - * @return the FileType code + * \brief Tries to open the file \ref Document::Filename and + * checks the preamble when existing. + * @return The FILE pointer on success. */ -FileType gdcmDocument::GetFileType() +std::ifstream *Document::OpenFile() { - return Filetype; -} + HasDCMPreamble = false; + if (Filename.length() == 0) + { + return 0; + } -/** - * \brief opens the file - * @param exception_on_error - * @return - */ -FILE *gdcmDocument::OpenFile(bool exception_on_error) - throw(gdcmFileError) -{ - fp=fopen(Filename.c_str(),"rb"); + if(Fp) + { + gdcmWarningMacro( "File already open: " << Filename.c_str()); + CloseFile(); + } - if(!fp) - { - if(exception_on_error) - throw gdcmFileError("gdcmDocument::gdcmDocument(const char *, bool)"); - else - { - dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file: ", - Filename.c_str()); - return NULL; - } - } - - if ( fp ) - { - 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) - 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 ) - return fp; - - fclose(fp); - dbg.Verbose(0, "gdcmDocument::OpenFile not DICOM/ACR", Filename.c_str()); - } - else { - dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file", Filename.c_str()); - } - return NULL; + Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary); + if( ! *Fp ) + { + gdcmDebugMacro( "Cannot open file: " << Filename.c_str()); + delete Fp; + Fp = 0; + return 0; + } + + 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 with no preamble: (%04x)\n", zero); + gdcmWarningMacro( msg.c_str() ); + return Fp; + } + + //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; + } + + CloseFile(); + gdcmWarningMacro( "Not DICOM/ACR (missing preamble)" << Filename.c_str()); + + return 0; } /** * \brief closes the file * @return TRUE if the close was successfull */ -bool gdcmDocument::CloseFile() +bool Document::CloseFile() { - int closed = fclose(fp); - fp = (FILE *)0; - if (! closed) - return false; - return true; + 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 - /// \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) - { - 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); - } + // (necessary if user wants to write a DICOM V3 file + // starting from an ACR-NEMA (V2) Header - if (filetype == gdcmExplicitVR) + if ( filetype == ImplicitVR || filetype == ExplicitVR ) { - 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); + */ -/** - * \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 Value, - uint16_t Group, - uint16_t Elem ) -{ - gdcmDocEntry* CurrentEntry; - gdcmValEntry* ValEntry; - - 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 ) - { - // 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). - */ -gdcmBinEntry * gdcmDocument::ReplaceOrCreateByNumber( - void *voidArea, - int lgth, - uint16_t Group, - uint16_t Elem) -{ - gdcmDocEntry* a; - gdcmBinEntry* b = 0; - a = GetDocEntryByNumber( Group, Elem); - if (a == NULL) { - a =NewBinEntryByNumber(Group, Elem); - if (a == NULL) - return NULL; - - b = new gdcmBinEntry(a); - AddEntry(b); - } - SetEntryByNumber(voidArea, lgth, Group, Elem); - b->SetVoidArea(voidArea); - - return b; -} + ElementSet::WriteContent(fp, filetype); // This one is recursive +} +// ----------------------------------------- +// Content entries /** - * \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 group group number of the Entry + * @param elem element number of the Entry */ -bool gdcmDocument::ReplaceIfExistByNumber(char* Value, uint16_t Group, - uint16_t Elem ) +void Document::LoadEntryBinArea(uint16_t group, uint16_t elem) { - std::string v = Value; - SetEntryByNumber(v, Group, Elem); - return true; -} + // Search the corresponding DocEntry + DocEntry *docElement = GetDocEntry(group, elem); + if ( !docElement ) + return; -//----------------------------------------------------------------------------- -// Protected + BinEntry *binElement = dynamic_cast(docElement); + if( !binElement ) + return; -/** - * \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 number of occurences - */ -int gdcmDocument::CheckIfEntryExistByNumber(uint16_t group, uint16_t element ) -{ - std::string key = gdcmDictEntry::TranslateToKey(group, element ); - return tagHT.count(key); + LoadEntryBinArea(binElement); } /** - * \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. + * \brief Loads (from disk) the element content + * when a string is not suitable + * @param elem Entry whose binArea is going to be loaded */ -std::string gdcmDocument::GetEntryByName(TagName tagName) +void Document::LoadEntryBinArea(BinEntry *elem) { - gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName); - if( dictEntry == NULL) - return GDCM_UNFOUND; + if(elem->GetBinArea()) + return; - return GetEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement()); -} + bool openFile = !Fp; + if(openFile) + OpenFile(); -/** - * \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 tagName) -{ - gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName); - if( dictEntry == NULL) - return GDCM_UNFOUND; + size_t o =(size_t)elem->GetOffset(); + Fp->seekg(o, std::ios::beg); - gdcmDocEntry* elem = GetDocEntryByNumber(dictEntry->GetGroup(), - dictEntry->GetElement()); - return elem->GetVR(); -} + size_t l = elem->GetLength(); + uint8_t *a = new uint8_t[l]; + if( !a ) + { + gdcmWarningMacro( "Cannot allocate BinEntry content"); + return; + } + Fp->read((char*)a, l); + if( Fp->fail() || Fp->eof()) + { + delete[] a; + return; + } -/** - * \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(); -} + 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.. - * - * Obtaining the VR (Value Representation) might be needed by caller - * to convert the string typed content to caller's native type - * (think of C++ vs Python). The VR is actually of a higher level - * of semantics than just the native C++ type. - * @param group Group number of the searched tag. - * @param element Element number of the searched tag. - * @return Corresponding element value representation when it exists, - * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. - */ -std::string gdcmDocument::GetEntryVRByNumber(uint16_t group, uint16_t element) -{ - gdcmDocEntry* elem = GetDocEntryByNumber(group, element); - if ( !elem ) - return GDCM_UNFOUND; - return elem->GetVR(); + if(openFile) + CloseFile(); } /** - * \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; - 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 Loads the element while preserving the current + * underlying file position indicator as opposed to + * LoadDocEntry that modifies it. + * @param entry DocEntry whose value will be loaded. */ -bool gdcmDocument::SetEntryByName(std::string content,std::string tagName) +void Document::LoadDocEntrySafe(DocEntry *entry) { - gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName); - if( dictEntry == NULL) - return false; - - return SetEntryByNumber(content,dictEntry->GetGroup(), - dictEntry->GetElement()); + if(Fp) + { + long PositionOnEntry = Fp->tellg(); + LoadDocEntry(entry); + Fp->seekg(PositionOnEntry, std::ios::beg); + } } /** - * \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 + * \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::SetEntryByNumber(std::string content, - uint16_t group, - uint16_t element) +bool Document::operator<(Document &document) { - gdcmValEntry* ValEntry = GetValEntryByNumber(group, element); - if (!ValEntry) + // Patient Name + std::string s1 = GetEntryValue(0x0010,0x0010); + std::string s2 = document.GetEntryValue(0x0010,0x0010); + if(s1 < s2) + { + return true; + } + 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). - if((content.length())%2) { - content = content + '\0'; - } - ValEntry->SetValue(content); - - // Integers have a special treatement for their length: - gdcmVRKey vr = ValEntry->GetVR(); - if( (vr == "US") || (vr == "SS") ) - ValEntry->SetLength(2); - else if( (vr == "UL") || (vr == "SL") ) - ValEntry->SetLength(4); else - ValEntry->SetLength(content.length()); - - 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 Binaray 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; - 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 ; -} - -/** - * \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 - */ -size_t gdcmDocument::GetEntryOffsetByNumber(uint16_t Group, uint16_t Elem) -{ - gdcmDocEntry* Entry = GetDocEntryByNumber(Group, Elem); - if (!Entry) { - dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry present."); - return (size_t)0; + // Patient ID + s1 = GetEntryValue(0x0010,0x0020); + s2 = document.GetEntryValue(0x0010,0x0020); + if ( s1 < s2 ) + { + return true; + } + else if ( s1 > s2 ) + { + return false; + } + else + { + // Study Instance UID + s1 = GetEntryValue(0x0020,0x000d); + s2 = document.GetEntryValue(0x0020,0x000d); + if ( s1 < s2 ) + { + return true; + } + else if( s1 > s2 ) + { + return false; + } + else + { + // Serie Instance UID + s1 = GetEntryValue(0x0020,0x000e); + s2 = document.GetEntryValue(0x0020,0x000e); + if ( s1 < s2 ) + { + return true; + } + else if( s1 > s2 ) + { + return false; + } + } + } } - return Entry->GetOffset(); + return false; } +//----------------------------------------------------------------------------- +// Protected /** - * \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 + * \brief Reads a supposed to be 16 Bits integer + * (swaps it depending on processor endianness) + * @return read value */ -void * gdcmDocument::GetEntryVoidAreaByNumber(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() ) + { + throw FormatError( "Document::ReadInt16()", " file error." ); + } + if( Fp->eof() ) { - dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry"); - return (NULL); + 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 + * \brief Reads a supposed to be 32 Bits integer + * (swaps it depending on processor endianness) + * @return read value */ -void *gdcmDocument::LoadEntryVoidArea(uint16_t Group, uint16_t Elem) +uint32_t Document::ReadInt32() + throw( FormatError ) { - gdcmDocEntry * Element= GetDocEntryByNumber(Group, Elem); - if ( !Element ) - return NULL; - size_t o =(size_t)Element->GetOffset(); - fseek(fp, o, SEEK_SET); - size_t l = Element->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) + uint32_t g; + Fp->read ((char*)&g, (size_t)4); + if ( Fp->fail() ) { - delete[] a; - return NULL; + throw FormatError( "Document::ReadInt32()", " file error." ); } - /// \todo Drop any allready existing void area! JPR - SetEntryVoidAreaByNumber(a, Group, Elem); - 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 - */ -void *gdcmDocument::LoadEntryVoidArea(gdcmBinEntry *Element) -{ - size_t o =(size_t)Element->GetOffset(); - fseek(fp, o, SEEK_SET); - size_t l = Element->GetLength(); - char* a = new char[l]; - if(!a) { - dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a"); - return NULL; - } - 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) { - gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element); - if ( ! tagHT.count(key)) - return false; - // This was for multimap ? - (( gdcmBinEntry *)( ((tagHT.equal_range(key)).first)->second ))->SetVoidArea(area); - - 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() { - //gdcmDictEntry *entry; +int Document::ComputeGroup0002Length( FileType filetype ) +{ + 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) - { - // Odd group => from public dictionary - if((*it)->GetGroup()%2==0) - continue; - - // 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(); + int groupLength = 0; + bool found0002 = false; + + // for each zero-level Tag in the DCM Header + DocEntry *entry = GetFirstEntry(); + while( entry ) + { + gr = entry->GetGroup(); - (*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") ) + { + // explicit VR AND OB, OW, SQ : 4 more bytes + groupLength += 4; + } + } + groupLength += 2 + 2 + 4 + entry->GetLength(); + } } - else - { - // Remove precedent value transformation - (*it)->SetDictEntry(NewVirtualDictEntry((*it)->GetGroup(),(*it)->GetElement(),vr)); - } + else if (found0002 ) + break; + + entry = GetNextEntry(); } -*/ + return groupLength; } +//----------------------------------------------------------------------------- +// Private /** - * \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. + * \brief Loads all the needed Dictionaries + * \warning NOT end user intended method ! */ -gdcmDocEntry* gdcmDocument::GetDocEntryByName(std::string tagName) +void Document::Initialize() { - gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName); - if( dictEntry == NULL) - return NULL; - - return(GetDocEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement())); + RefPubDict = Global::GetDicts()->GetDefaultPubDict(); + RefShaDict = NULL; + Filetype = Unknown; } /** - * \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) + * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries) + * @param set DocEntrySet we are going to parse ('zero level' or a SQItem) + * @param offset start of parsing + * @param l_max length to parse + * @param delim_mode : whether we are in 'delimitor mode' (l=0xffffff) or not + */ +void Document::ParseDES(DocEntrySet *set, long offset, + long l_max, bool delim_mode) { - gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element); - if ( ! tagHT.count(key)) - return NULL; - return tagHT.find(key)->second; -} + DocEntry *newDocEntry; + ValEntry *newValEntry; + BinEntry *newBinEntry; + SeqEntry *newSeqEntry; + VRKey vr; + bool used; -/** - * \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 (gdcmValEntry*)0; - if ( gdcmValEntry* ValEntry = dynamic_cast(CurrentEntry) ) - { - return ValEntry; - } - dbg.Verbose(0, "gdcmDocument::GetValEntryByNumber: unfound ValEntry."); - return (gdcmValEntry*)0; -} + while (true) + { + if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) + { + break; + } -/** - * \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); -} + used = true; + newDocEntry = ReadNextDocEntry( ); -/** - * \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 (sw) { - case 0 : - 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) ); + if ( !newDocEntry ) + { break; - default : - std::cout << "swapCode= " << sw << 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); -} + vr = newDocEntry->GetVR(); + newValEntry = dynamic_cast(newDocEntry); + newBinEntry = dynamic_cast(newDocEntry); + newSeqEntry = dynamic_cast(newDocEntry); -/** - * \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 ( (sw==4321) || (sw==2143) ) - a =(((a<<8) & 0x0ff00) | ((a>>8)&0x00ff)); - return a; -} + if ( newValEntry || newBinEntry ) + { + if ( newBinEntry ) + { + 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."); + } -/** - * \brief Unswaps the bytes so they agree with the processor order - * @return The properly unswaped 16 bits integer. - */ -uint16_t gdcmDocument::UnswapShort(uint16_t a) -{ - return SwapShort(a); -} + //////////////////// BinEntry or UNKOWN VR: + // When "this" is a Document the Key is simply of the + // form ( group, elem )... + if ( 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): + if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + { + newBinEntry->SetKey( parentSQItem->GetBaseTagKey() + + newBinEntry->GetKey() ); + } -//----------------------------------------------------------------------------- -// Private + LoadDocEntry( newBinEntry ); + if( !set->AddEntry( newBinEntry ) ) + { + //Expect big troubles if here + //delete newBinEntry; + used=false; + } + } + else + { + /////////////////////// ValEntry -/** - * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries) - * @return length of the parsed set. - */ + // 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 ( dynamic_cast< Document* > ( set ) ) + { + newValEntry->SetKey( newValEntry->GetKey() ); + } + // ...but when "set" is a SQItem, we are inserting this new + // valEntry in a sequence item. Hence the key has the + // generalized form (refer to \ref BaseTagKey): + if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + { + newValEntry->SetKey( parentSQItem->GetBaseTagKey() + + newValEntry->GetKey() ); + } + + LoadDocEntry( newValEntry ); + bool delimitor=newValEntry->IsItemDelimitor(); -long gdcmDocument::ParseDES(gdcmDocEntrySet *set, - long offset, - long l_max, - bool delim_mode) -{ - gdcmDocEntry *NewDocEntry = (gdcmDocEntry *)0; - gdcmValEntry *NewValEntry = (gdcmValEntry *)0; - gdcmBinEntry *bn; - gdcmSeqEntry *sq; - gdcmVRKey vr; - unsigned long l = 0; - int depth; - - depth = set->GetDepthLevel(); - while (true) { - - if ( !delim_mode && ftell(fp)-offset >= l_max) { - break; - } - NewDocEntry = ReadNextDocEntry( ); - if (!NewDocEntry) - break; + if ( LoadMode & NO_SHADOW ) // User asked to skip, if possible, + // shadow groups ( if possible : + // whether element 0x0000 exits) + { + if ( newValEntry->GetGroup()%2 != 0 ) + { + if ( newValEntry->GetElement() == 0x0000 ) + { + std::string strLgrGroup = newValEntry->GetValue(); + int lgrGroup; + if ( strLgrGroup != GDCM_UNFOUND) + { + lgrGroup = atoi(strLgrGroup.c_str()); + Fp->seekg(lgrGroup , std::ios::cur); + used = false; + continue; + } + } + } + } - vr = NewDocEntry->GetVR(); - if (vr!="SQ") - { - - if ( gdcmGlobal::GetVR()->IsVROfGdcmStringRepresentable(vr) ) - { - /////// ValEntry - NewValEntry = new gdcmValEntry(NewDocEntry->GetDictEntry()); - NewValEntry->Copy(NewDocEntry); - NewValEntry->SetDepthLevel(depth); - set->AddEntry(NewValEntry); - LoadDocEntry(NewValEntry); - if (NewValEntry->isItemDelimitor()) + if( !set->AddEntry( newValEntry ) ) + { + // If here expect big troubles + // delete newValEntry; //otherwise mem leak + used=false; + } + + if (delimitor) + { + if(!used) + delete newDocEntry; break; - if ( !delim_mode && ftell(fp)-offset >= l_max) + } + if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) { + if(!used) + delete newDocEntry; break; } } - else + + // Just to make sure we are at the beginning of next entry. + SkipToNextDocEntry(newDocEntry); + } + else + { + // VR = "SQ" + unsigned long l = newDocEntry->GetReadLength(); + if ( l != 0 ) // don't mess the delim_mode for zero-length sequence { - 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."); + if ( l == 0xffffffff ) + { + delim_mode = true; + } + else + { + delim_mode = false; } - - ////// BinEntry or UNKOWN VR: - bn = new gdcmBinEntry(NewDocEntry->GetDictEntry()); - bn->Copy(NewDocEntry); - set->AddEntry(bn); - LoadDocEntry(bn); } + + if ( (LoadMode & NO_SEQ) && ! delim_mode ) // User asked to skip SQ + { + Fp->seekg( l, std::ios::cur); + used = false; + continue; + } + + // no other way to create it ... + newSeqEntry->SetDelimitorMode( delim_mode ); - if (NewDocEntry->GetGroup() == 0x7fe0 && - NewDocEntry->GetElement() == 0x0010 ) + // At the top of the hierarchy, stands a Document. When "set" + // is a Document, then we are building the first depth level. + // Hence the SeqEntry we are building simply has a depth + // level of one: + if ( dynamic_cast< Document* > ( set ) ) { - 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(); - } + newSeqEntry->SetDepthLevel( 1 ); + newSeqEntry->SetKey( newSeqEntry->GetKey() ); } - else + // But when "set" is already a SQItem, we are building a nested + // sequence, and hence the depth level of the new SeqEntry + // we are building, is one level deeper: + if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) { - // to be sure we are at the beginning - SkipToNextDocEntry(NewDocEntry); - l = NewDocEntry->GetFullLength(); + newSeqEntry->SetDepthLevel( parentSQItem->GetDepthLevel() + 1 ); + newSeqEntry->SetKey( parentSQItem->GetBaseTagKey() + + newSeqEntry->GetKey() ); } - } - else - { // VR = "SQ" - - l=NewDocEntry->GetReadLength(); - if (l != 0) // don't mess the delim_mode for zero-length sequence - if (l == 0xffffffff) - delim_mode = true; - else - delim_mode = false; - // no other way to create it ... - sq = new gdcmSeqEntry(NewDocEntry->GetDictEntry(), - set->GetDepthLevel()); - sq->Copy(NewDocEntry); - sq->SetDelimitorMode(delim_mode); - sq->SetDepthLevel(depth); - if (l != 0) + if ( l != 0 ) { // Don't try to parse zero-length sequences - long lgt = ParseSQ( sq, - NewDocEntry->GetOffset(), - l, delim_mode); - (void)lgt; //not used... + ParseSQ( newSeqEntry, + newDocEntry->GetOffset(), + l, delim_mode); + } + if( !set->AddEntry( newSeqEntry ) ) + { + used = false; } - set->AddEntry(sq); - if ( !delim_mode && ftell(fp)-offset >= l_max) + + if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) { + if( !used ) + delete newDocEntry; break; } } - delete NewDocEntry; + + if( !used ) + delete newDocEntry; } - return l; // Probably useless } /** * \brief Parses a Sequence ( SeqEntry after SeqEntry) * @return parsed length for this level */ -long gdcmDocument::ParseSQ(gdcmSeqEntry *set, - long offset, long l_max, bool delim_mode) +void Document::ParseSQ( SeqEntry *seqEntry, + long offset, long l_max, bool delim_mode) { int SQItemNumber = 0; - - gdcmDocEntry *NewDocEntry = (gdcmDocEntry *)0; - gdcmSQItem *itemSQ; bool dlm_mod; - int lgr, lgth; - unsigned int l; - int depth = set->GetDepthLevel(); - (void)depth; //not used + long offsetStartCurrentSQItem = offset; - while (true) { + while (true) + { + // the first time, we read the fff0,e000 of the first SQItem + DocEntry *newDocEntry = ReadNextDocEntry(); - NewDocEntry = ReadNextDocEntry(); - if (!NewDocEntry) + if ( !newDocEntry ) + { + // FIXME Should warn user break; - if(delim_mode) { - if (NewDocEntry->isSequenceDelimitor()) { - set->SetSequenceDelimitationItem(NewDocEntry); + } + if( delim_mode ) + { + if ( newDocEntry->IsSequenceDelimitor() ) + { + seqEntry->SetDelimitationItem( newDocEntry ); break; - } + } } - if (!delim_mode && (ftell(fp)-offset) >= l_max) { - break; + if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) + { + delete newDocEntry; + break; } - - itemSQ = new gdcmSQItem(set->GetDepthLevel()); - itemSQ->AddEntry(NewDocEntry); - l= NewDocEntry->GetReadLength(); + // 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) + if ( l == 0xffffffff ) + { dlm_mod = true; + } else - dlm_mod=false; - - lgr=ParseDES(itemSQ, NewDocEntry->GetOffset(), l, dlm_mod); - - set->AddEntry(itemSQ,SQItemNumber); - SQItemNumber ++; - if (!delim_mode && (ftell(fp)-offset) >= l_max) { + { + dlm_mod = false; + } + + // 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 && ((long)(Fp->tellg())-offset ) >= l_max ) + { break; } } - lgth = ftell(fp) - offset; - return lgth; } /** - * \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 the element content if its length doesn't exceed + * the value specified with Document::SetMaxSizeLoadEntry() + * @param entry Header Entry (Dicom Element) to be dealt with */ -void gdcmDocument::LoadDocEntry(gdcmDocEntry* Entry) +void Document::LoadDocEntry(DocEntry *entry) { - size_t item_read; - uint16_t group = Entry->GetGroup(); - std::string vr = Entry->GetVR(); - uint32_t length = Entry->GetLength(); + 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 @@ -1408,7 +1093,7 @@ void gdcmDocument::LoadDocEntry(gdcmDocEntry* Entry) // When the length is zero things are easy: if ( length == 0 ) { - ((gdcmValEntry *)Entry)->SetValue(""); + ((ValEntry *)entry)->SetValue(""); return; } @@ -1419,46 +1104,50 @@ void gdcmDocument::LoadDocEntry(gdcmDocEntry* Entry) std::ostringstream s; if (length > MaxSizeLoadEntry) { - if (gdcmBinEntry* BinEntryPtr = dynamic_cast< gdcmBinEntry* >(Entry) ) - { - s << "gdcm::NotLoaded (BinEntry)"; - s << " Address:" << (long)Entry->GetOffset(); - s << " Length:" << Entry->GetLength(); - s << " x(" << std::hex << Entry->GetLength() << ")"; - BinEntryPtr->SetValue(s.str()); + if (BinEntry *binEntryPtr = dynamic_cast< BinEntry* >(entry) ) + { + //s << "gdcm::NotLoaded (BinEntry)"; + s << GDCM_NOTLOADED; + s << " Address:" << (long)entry->GetOffset(); + s << " Length:" << entry->GetLength(); + s << " x(" << std::hex << entry->GetLength() << ")"; + binEntryPtr->SetValue(s.str()); } - // to be sure we are at the end of the value ... - fseek(fp, (long)Entry->GetOffset()+(long)Entry->GetLength(), SEEK_SET); - return; - // Be carefull : a BinEntry IS_A ValEntry ... - if (gdcmValEntry* ValEntryPtr = dynamic_cast< gdcmValEntry* >(Entry) ) + // Be carefull : a BinEntry IS_A ValEntry ... + else if (ValEntry *valEntryPtr = dynamic_cast< ValEntry* >(entry) ) { - s << "gdcm::NotLoaded. (ValEntry)"; - s << " Address:" << (long)Entry->GetOffset(); - s << " Length:" << Entry->GetLength(); - s << " x(" << std::hex << Entry->GetLength() << ")"; - ValEntryPtr->SetValue(s.str()); + // 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 + { + // 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); + 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) ) { - - LoadEntryVoidArea(BinEntryPtr); - s << "gdcm::Loaded (BinEntry)"; - BinEntryPtr->SetValue(s.str()); + s << GDCM_BINLOADED; + binEntryPtr->SetValue(s.str()); + LoadEntryBinArea(binEntryPtr); // last one, not to erase length ! return; } - - /// \todo Any compacter code suggested (?) - if ( IsDocEntryAnInteger(Entry) ) + + if ( IsDocEntryAnInteger(entry) ) { uint32_t NewInt; - std::ostringstream s; 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 @@ -1500,174 +1189,205 @@ 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); - if (gdcmValEntry* ValEntry = dynamic_cast< gdcmValEntry* >(Entry) ) - { - if ( item_read != 1 ) + // FIXME: We need an additional byte for storing \0 that is not on disk + char *str = new char[length+1]; + 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 ( ValEntry *valEntry = dynamic_cast(entry) ) + { + if ( Fp->fail() || Fp->eof()) { - dbg.Verbose(1, "gdcmDocument::LoadDocEntry", - "unread element value"); - ValEntry->SetValue("gdcm::UnRead"); + gdcmWarningMacro("Unread element value"); + valEntry->SetValue(GDCM_UNREAD); return; } - if( (vr == "UI") ) // Because of correspondance with the VR dic - ValEntry->SetValue(NewValue.c_str()); + if( vr == "UI" ) + { + // Because of correspondance with the VR dic + valEntry->SetValue(newValue); + } else - ValEntry->SetValue(NewValue); + { + valEntry->SetValue(newValue); + } } else { - dbg.Error(true, "gdcmDocument::LoadDocEntry" - "Should have a ValEntry, here !"); + gdcmErrorMacro( "Should have a ValEntry, here !"); } } - /** * \brief Find the value Length of the passed Header Entry - * @param Entry Header Entry whose length of the value shall be loaded. + * @param entry Header Entry whose length of the value shall be loaded. */ -void gdcmDocument::FindDocEntryLength (gdcmDocEntry *Entry) +void Document::FindDocEntryLength( DocEntry *entry ) + throw ( FormatError ) { - uint16_t element = Entry->GetElement(); - std::string vr = Entry->GetVR(); - uint16_t length16; - + 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 == "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 = FindDocEntryLengthOB(); - if ( errno == 1 ) + uint32_t lengthOB; + try + { + lengthOB = FindDocEntryLengthOBOrOW(); + } + 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); - Entry->SetLength(LengthUntilEOF); - errno = 0; + 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->SetLength(LengthOB); + entry->SetReadLength(lengthOB); + entry->SetLength(lengthOB); return; } - FixDocEntryFoundLength(Entry, length32); + FixDocEntryFoundLength(entry, length32); return; } // Length is encoded on 2 bytes. length16 = ReadInt16(); + + // 0xffff means that we deal with 'No Length' Sequence + // or 'No Length' SQItem + if ( length16 == 0xffff) + { + length16 = 0; + } + FixDocEntryFoundLength( entry, (uint32_t)length16 ); + return; + } + else + { + // Either implicit VR or a non DICOM conformal (see note below) explicit + // VR that ommited the VR of (at least) this element. Farts happen. + // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25 + // 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 ...' - // 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) ) + FixDocEntryFoundLength( entry, ReadInt32() ); + return; + } +} + +/** + * \brief Find the Length till the next sequence delimiter + * \warning NOT end user intended method ! + * @return + */ +uint32_t Document::FindDocEntryLengthOBOrOW() + throw( FormatUnexpected ) +{ + // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data. + long positionOnEntry = Fp->tellg(); + bool foundSequenceDelimiter = false; + uint32_t totalLength = 0; + + while ( !foundSequenceDelimiter ) + { + uint16_t group; + uint16_t elem; + try { - if ( ! IsExplicitVRBigEndianTransferSyntax() ) - { - dbg.Verbose(0, "gdcmDocument::FindLength", "not explicit VR"); - errno = 1; - 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); + 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."); } - - // Heuristic: well, some files are really ill-formed. - if ( length16 == 0xffff) + if ( elem == 0xe0dd ) { - length16 = 0; - // Length16= 0xffff means that we deal with - // 'Unknown Length' Sequence + foundSequenceDelimiter = true; } - FixDocEntryFoundLength(Entry, (uint32_t)length16); - return; - } - else - { - // Either implicit VR or a non DICOM conformal (see note below) explicit - // VR that ommited the VR of (at least) this element. Farts happen. - // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25 - // 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. + uint32_t itemLength = ReadInt32(); + // We add 4 bytes since we just read the ItemLength with ReadInt32 + totalLength += itemLength + 4; + SkipBytes(itemLength); - FixDocEntryFoundLength(Entry, ReadInt32()); - return; + if ( foundSequenceDelimiter ) + { + break; + } } + Fp->seekg( positionOnEntry, std::ios::beg); + return totalLength; } /** * \brief Find the Value Representation of the current Dicom Element. - * @param Entry + * @return Value Representation of the current Entry */ -void gdcmDocument::FindDocEntryVR( gdcmDocEntry *Entry) +std::string Document::FindDocEntryVR() { - if (Filetype != gdcmExplicitVR) - return; - - char VR[3]; + 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 @@ -1675,91 +1395,32 @@ 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. - - (void)fread (&VR, (size_t)2,(size_t)1, fp); - VR[2]=0; - if(!CheckDocEntryVR(Entry,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(); + + char vr[3]; + Fp->read (vr, (size_t)2); + vr[2] = 0; + + if( !CheckDocEntryVR(vr) ) + { + 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; } @@ -1767,60 +1428,66 @@ bool gdcmDocument::CheckDocEntryVR(gdcmDocEntry *Entry, gdcmVRKey vr) * \brief Get the transformed value of the header entry. The VR value * is used to define the transformation to operate on the value * \warning NOT end user intended method ! - * @param Entry + * @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()) ) + if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() ) { - std::string val=((gdcmValEntry *)Entry)->GetValue(); - std::string vr=Entry->GetVR(); - uint32_t length = Entry->GetLength(); + std::string val = ((ValEntry *)entry)->GetValue(); + std::string vr = entry->GetVR(); + uint32_t length = entry->GetLength(); std::ostringstream s; int nbInt; - // When short integer(s) are expected, read and convert the following - // n * 2 bytes properly i.e. as a multivaluated strings - // (each single value is separated fromthe next one by '\' - // as usual for standard multivaluated filels - // Elements with Value Multiplicity > 1 - // contain a set of short integers (not a single one) + // When short integer(s) are expected, read and convert the following + // n * 2 bytes properly i.e. as a multivaluated strings + // (each single value is separated fromthe next one by '\' + // as usual for standard multivaluated filels + // 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; + uint16_t newInt16; nbInt = length / 2; for (int i=0; i < nbInt; i++) { - if(i!=0) + if( i != 0 ) + { s << '\\'; - NewInt16 = (val[2*i+0]&0xFF)+((val[2*i+1]&0xFF)<<8); - NewInt16 = SwapShort(NewInt16); - s << NewInt16; + } + newInt16 = ( val[2*i+0] & 0xFF ) + ( ( val[2*i+1] & 0xFF ) << 8); + newInt16 = SwapShort( newInt16 ); + s << newInt16; } } - // When integer(s) are expected, read and convert the following - // n * 4 bytes properly i.e. as a multivaluated strings - // (each single value is separated fromthe next one by '\' - // 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") + // When integer(s) are expected, read and convert the following + // n * 4 bytes properly i.e. as a multivaluated strings + // (each single value is separated fromthe next one by '\' + // 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" ) { - uint32_t NewInt32; + uint32_t newInt32; nbInt = length / 4; for (int i=0; i < nbInt; i++) { - if(i!=0) + if( i != 0) + { s << '\\'; - NewInt32= (val[4*i+0]&0xFF)+((val[4*i+1]&0xFF)<<8)+ - ((val[4*i+2]&0xFF)<<16)+((val[4*i+3]&0xFF)<<24); - NewInt32=SwapLong(NewInt32); - s << NewInt32; + } + newInt32 = ( val[4*i+0] & 0xFF ) + + (( val[4*i+1] & 0xFF ) << 8 ) + + (( val[4*i+2] & 0xFF ) << 16 ) + + (( val[4*i+3] & 0xFF ) << 24 ); + newInt32 = SwapLong( newInt32 ); + s << newInt32; } } #ifdef GDCM_NO_ANSI_STRING_STREAM @@ -1828,8 +1495,7 @@ std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *Entry) #endif //GDCM_NO_ANSI_STRING_STREAM return s.str(); } - - return ((gdcmValEntry *)Entry)->GetValue(); + return ((ValEntry *)entry)->GetValue(); } /** @@ -1837,41 +1503,44 @@ std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *Entry) * value is used to define the reverse transformation to operate on * the value * \warning NOT end user intended method ! - * @param 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()) ) + if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() ) { - std::string vr=Entry->GetVR(); - std::ostringstream s; + std::string vr = entry->GetVR(); std::vector tokens; + std::ostringstream s; - if (vr == "US" || vr == "SS") + if ( vr == "US" || vr == "SS" ) { - uint16_t NewInt16; + uint16_t newInt16; - tokens.erase(tokens.begin(),tokens.end()); // clean any previous value - Tokenize (((gdcmValEntry *)Entry)->GetValue(), tokens, "\\"); - for (unsigned int i=0; iGetValue(), tokens, "\\"); + for (unsigned int i=0; i>8)&0xFF); + newInt16 = atoi(tokens[i].c_str()); + s << ( newInt16 & 0xFF ) + << (( newInt16 >> 8 ) & 0xFF ); } tokens.clear(); } - if (vr == "UL" || vr == "SL") + if ( vr == "UL" || vr == "SL") { - uint32_t NewInt32; + uint32_t newInt32; 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; i>8)&0xFF) - <<(char)((NewInt32>>16)&0xFF)<<(char)((NewInt32>>24)&0xFF); + newInt32 = atoi(tokens[i].c_str()); + s << (char)( newInt32 & 0xFF ) + << (char)(( newInt32 >> 8 ) & 0xFF ) + << (char)(( newInt32 >> 16 ) & 0xFF ) + << (char)(( newInt32 >> 24 ) & 0xFF ); } tokens.clear(); } @@ -1882,67 +1551,69 @@ std::string gdcmDocument::GetDocEntryUnvalue(gdcmDocEntry *Entry) return s.str(); } - return ((gdcmValEntry *)Entry)->GetValue(); + return ((ValEntry *)entry)->GetValue(); } /** * \brief Skip a given Header Entry * \warning NOT end user intended method ! - * @param entry + * @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 + * @param currentDocEntry entry to skip */ -void gdcmDocument::SkipToNextDocEntry(gdcmDocEntry *entry) +void Document::SkipToNextDocEntry(DocEntry *currentDocEntry) { - (void)fseek(fp, (long)(entry->GetOffset()), SEEK_SET); - (void)fseek(fp, (long)(entry->GetReadLength()), SEEK_CUR); + Fp->seekg((long)(currentDocEntry->GetOffset()), std::ios::beg); + if (currentDocEntry->GetGroup() != 0xfffe) // for fffe pb + Fp->seekg( (long)(currentDocEntry->GetReadLength()),std::ios::cur); } /** * \brief When the length of an element value is obviously wrong (because * the parser went Jabberwocky) one can hope improving things by * applying some heuristics. + * @param entry entry to check + * @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) { - FoundLength = 0; + entry->SetReadLength( foundLength ); // will be updated only if a bug is found + if ( foundLength == 0xffffffff) + { + 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()); + if ( foundLength % 2) + { + 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)... + // are no longer in use (we are talking a few years, here)... // Note: XMedCom probably uses such a trick since it is able to read // those pesky GE images ... - if (FoundLength == 13) { // Only happens for this length ! - if ( (Entry->GetGroup() != 0x0008) - || ( (Entry->GetElement() != 0x0070) - && (Entry->GetElement() != 0x0080) ) ) + if ( foundLength == 13) + { + // Only happens for this length ! + if ( gr != 0x0008 || ( elem != 0x0070 && elem != 0x0080 ) ) { - FoundLength = 10; - Entry->SetReadLength(10); /// \todo a bug is to be fixed !? + foundLength = 10; + entry->SetReadLength(10); // a bug is to be fixed !? } } @@ -1950,59 +1621,55 @@ void gdcmDocument::FixDocEntryFoundLength(gdcmDocEntry *Entry, // 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) ) ) + else if ( gr == 0x0009 && ( elem == 0x1113 || elem == 0x1114 ) ) { - FoundLength = 4; - Entry->SetReadLength(4); /// \todo a bug is to be fixed !? + foundLength = 4; + entry->SetReadLength(4); // a bug is to be fixed !? } - //////// Deal with sequences, but only on users request: - else - if ( ( Entry->GetVR() == "SQ") && enableSequences) + else if ( entry->GetVR() == "SQ" ) { - FoundLength = 0; // ReadLength is unchanged + 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) + 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 ) + if( entry->GetElement() != 0x0000 ) { - FoundLength = 0; + foundLength = 0; } - } - - Entry->SetUsableLength(FoundLength); + } + entry->SetLength(foundLength); } /** * \brief Apply some heuristics to predict whether the considered * element value contains/represents an integer or not. - * @param Entry The element value on which to apply the predicate. + * @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(); +bool Document::IsDocEntryAnInteger(DocEntry *entry) +{ + uint16_t elem = entry->GetElement(); + uint16_t group = entry->GetGroup(); + const std::string &vr = entry->GetVR(); + uint32_t length = entry->GetLength(); // When we have some semantics on the element we just read, and if we // a priori know we are dealing with an integer, then we shall be // able to swap it's element value properly. - if ( element == 0 ) // This is the group length of the group + if ( elem == 0 ) // This is the group length of the group { - if (length == 4) + if ( length == 4 ) + { return true; + } else { // Allthough this should never happen, still some images have a @@ -2012,132 +1679,21 @@ bool gdcmDocument::IsDocEntryAnInteger(gdcmDocEntry *Entry) { // 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 + long filePosition = Fp->tellg(); + gdcmWarningMacro( "Erroneous Group Length element length on : (" + << std::hex << group << " , " << elem << ") -before- position x(" << filePosition << ")" - << "lgt : " << length; - dbg.Verbose(0, "gdcmDocument::IsDocEntryAnInteger", s.str().c_str() ); + << "lgt : " << length ); } } - if ( (vr == "UL") || (vr == "US") || (vr == "SL") || (vr == "SS") ) + 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() { - // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data. - uint16_t g; - uint16_t n; - long PositionOnEntry = ftell(fp); - bool FoundSequenceDelimiter = false; - uint32_t TotalLength = 0; - uint32_t ItemLength; - - while ( ! FoundSequenceDelimiter) - { - g = ReadInt16(); - n = ReadInt16(); - if (errno == 1) - return 0; - - // We have to decount the group and element we just read - TotalLength += 4; - - if ( ( g != 0xfffe ) - || ( ( n != 0xe0dd ) && ( n != 0xe000 ) ) ) - { - dbg.Verbose(1, "gdcmDocument::FindLengthOB: neither an Item tag " - "nor a Sequence delimiter tag."); - fseek(fp, PositionOnEntry, SEEK_SET); - errno = 1; - return 0; - } - - if ( n == 0xe0dd ) - FoundSequenceDelimiter = true; - - ItemLength = ReadInt32(); - // We add 4 bytes since we just read the ItemLength with ReadInt32 - TotalLength += ItemLength + 4; - SkipBytes(ItemLength); - - if ( FoundSequenceDelimiter ) - break; - } - 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() { - uint16_t g; - size_t item_read; - item_read = fread (&g, (size_t)2,(size_t)1, fp); - if ( item_read != 1 ) { - if(ferror(fp)) - dbg.Verbose(0, "gdcmDocument::ReadInt16", " File Error"); - errno = 1; - return 0; - } - errno = 0; - g = SwapShort(g); - return g; -} - -/** - * \brief Reads a supposed to be 32 Bits integer - * (swaps it depending on processor endianity) - * @return read value - */ -uint32_t gdcmDocument::ReadInt32() { - uint32_t g; - size_t item_read; - item_read = fread (&g, (size_t)4,(size_t)1, fp); - if ( item_read != 1 ) { - if(ferror(fp)) - dbg.Verbose(0, "gdcmDocument::ReadInt32", " File Error"); - errno = 1; - return 0; - } - errno = 0; - 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; -} - /** * \brief Discover what the swap code is (among little endian, big endian, * bad little endian, bad big endian). @@ -2146,47 +1702,40 @@ void gdcmDocument::Initialise() * it's neither ACR-NEMA nor DICOM * true when we hope ours assuptions are OK */ -bool gdcmDocument::CheckSwap() { - +bool Document::CheckSwap() +{ // The only guaranted way of finding the swap code is to find a // group tag since we know it's length has to be of four bytes i.e. // 0x00000004. Finding the swap code in then straigthforward. Trouble // occurs when we can't find such group... - uint32_t x=4; // x : for ntohs - bool net2host; // true when HostByteOrder is the same as NetworkByteOrder uint32_t s32; uint16_t s16; - int lgrLue; - char *entCur; - 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". - lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, fp); + // 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); - entCur = deb + 128; - if(memcmp(entCur, "DICM", (size_t)4) == 0) { - dbg.Verbose(1, "gdcmDocument::CheckSwap:", "looks like DICOM Version3"); + char *entCur = deb + 128; + if( memcmp(entCur, "DICM", (size_t)4) == 0 ) + { + 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, @@ -2194,56 +1743,52 @@ 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) + if ( net2host ) { - sw = 4321; - dbg.Verbose(1, "gdcmDocument::CheckSwap:", - "HostByteOrder != NetworkByteOrder"); - } + SwapCode = 4321; + gdcmWarningMacro( "HostByteOrder != NetworkByteOrder"); + } else { - sw = 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); + Fp->seekg ( 132L, std::ios::beg); return true; } // 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 DICOM Version3 file"); + Fp->seekg(0, std::ios::beg); // 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. @@ -2255,408 +1800,275 @@ bool gdcmDocument::CheckSwap() { // representation of a 32 bits integer. Hence the following dirty // trick : s32 = *((uint32_t *)(entCur)); - - switch (s32) { + + switch( s32 ) + { case 0x00040000 : - sw = 3412; - Filetype = gdcmACR; + SwapCode = 3412; + Filetype = ACR; return true; case 0x04000000 : - sw = 4321; - Filetype = gdcmACR; + SwapCode = 4321; + Filetype = ACR; return true; case 0x00000400 : - sw = 2143; - Filetype = gdcmACR; + SwapCode = 2143; + Filetype = ACR; return true; case 0x00000004 : - sw = 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. - // It is time for despaired wild guesses. - // 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 - // to determine ' sw' 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-) - // the file IS NOT ACR-NEMA nor DICOM V3 - // Find a trick to tell it the caller... - - s16 = *((uint16_t *)(deb)); + // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file. + // It is time for despaired wild guesses. + // 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' + // 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 -3, 4, ..., 8-) + // the file IS NOT ACR-NEMA nor DICOM V3 + // Find a trick to tell it the caller... - switch (s16) { - case 0x0002 : - case 0x0004 : - case 0x0008 : - sw = 0; - Filetype = gdcmACR; - return true; - case 0x0200 : - case 0x0400 : - case 0x0800 : - sw = 4321; - Filetype = gdcmACR; - return true; - default : - dbg.Verbose(0, "gdcmDocument::CheckSwap:", - "ACR/NEMA unfound swap info (Really hopeless !)"); - Filetype = gdcmUnknown; - return false; - } + s16 = *((uint16_t *)(deb)); - // Then the only info we have is the net2host one. - //if (! net2host ) - // sw = 0; - //else - // sw = 4321; - //return; + switch ( s16 ) + { + case 0x0001 : + case 0x0002 : + case 0x0003 : + case 0x0004 : + case 0x0005 : + case 0x0006 : + case 0x0007 : + case 0x0008 : + SwapCode = 1234; + Filetype = ACR; + return true; + case 0x0100 : + case 0x0200 : + case 0x0300 : + case 0x0400 : + case 0x0500 : + case 0x0600 : + case 0x0700 : + case 0x0800 : + SwapCode = 4321; + Filetype = ACR; + return true; + default : + gdcmWarningMacro( "ACR/NEMA unfound swap info (Really hopeless !)"); + Filetype = Unknown; + return false; + } } } /** - * \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 ( sw == 0 ) + gdcmWarningMacro( "Switching Byte Swap code from "<< SwapCode); + if ( SwapCode == 1234 ) { - sw = 4321; - return; + SwapCode = 4321; } - if ( sw == 4321 ) + else if ( SwapCode == 4321 ) { - sw = 0; - return; + SwapCode = 1234; } - if ( sw == 3412 ) + else if ( SwapCode == 3412 ) { - sw = 2143; - return; + SwapCode = 2143; + } + else if ( SwapCode == 2143 ) + { + SwapCode = 3412; } - if ( sw == 2143 ) - sw = 3412; } /** * \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) - return; - if ((uint32_t)NewSize >= (uint32_t)0xffffffff) + if ( newSize < 0 ) { - MaxSizeLoadEntry = 0xffffffff; return; } - 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) -{ - if (NewSize < 0) - return; - if ((uint32_t)NewSize >= (uint32_t)0xffffffff) + if ((uint32_t)newSize >= (uint32_t)0xffffffff ) { - MaxSizePrintEntry = 0xffffffff; + MaxSizeLoadEntry = 0xffffffff; return; } - MaxSizePrintEntry = NewSize; + MaxSizeLoadEntry = newSize; } - - /** * \brief Read the next tag but 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. */ -gdcmDocEntry *gdcmDocument::ReadNextDocEntry() { - uint16_t g = ReadInt16(); - uint16_t n = ReadInt16(); - gdcmDocEntry *NewEntry; - - if (errno == 1) +DocEntry *Document::ReadNextDocEntry() +{ + uint16_t group; + uint16_t elem; + + try + { + group = ReadInt16(); + elem = ReadInt16(); + } + catch ( FormatError e ) + { // We reached the EOF (or an error occured) therefore // header parsing has to be considered as finished. - return (gdcmDocEntry *)0; + return 0; + } - NewEntry = NewDocEntryByNumber(g, n); - FindDocEntryVR(NewEntry); - FindDocEntryLength(NewEntry); + // Sometimes file contains groups of tags with reversed endianess. + HandleBrokenEndian(group, elem); + + // In 'true DICOM' files Group 0002 is always little endian + if ( HasDCMPreamble ) + HandleOutOfGroup0002(group, elem); + + std::string vr = FindDocEntryVR(); + std::string realVR = vr; - if (errno == 1) + if( vr == GDCM_UNKNOWN) { - // Call it quits - delete NewEntry; - return NULL; + 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) - return elem; + if( Filetype == ExplicitVR ) + { + // We thought this was explicit VR, but we end up with an + // implicit VR tag. Let's backtrack. + if ( newEntry->GetGroup() != 0xfffe ) + { + std::string msg; + msg = Util::Format("Entry (%04x,%04x) should be Explicit VR\n", + newEntry->GetGroup(), newEntry->GetElement()); + 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) -{ - uint16_t ItemTagGroup; - uint16_t ItemTagElement; - 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: - ItemTagGroup = ReadInt16(); - 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 switch 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); - - 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; - 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 ) - 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) { - // BTW, what is the purpous of those length anyhow !? - char * BasicOffsetTableItemValue = new char[ItemLength + 1]; - fread(BasicOffsetTableItemValue, ItemLength, 1, fp); - for (unsigned int i=0; i < ItemLength; i += 4){ - uint32_t IndividualLength; - 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()); - } - } - - if ( ! IsRLELossLessTransferSyntax() ) + // Endian reversion. Some files contain groups of tags with reversed endianess. + if ( !Group0002Parsed && group != 0x0002) { - // 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; - - // 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++) { - ftellRes=ftell(fp); - RleSegmentOffsetTable[k] = ReadInt32(); - } - - // 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]); - } - } + Group0002Parsed = true; + // we just came out of group 0002 + // if Transfer syntax is Big Endian we have to change CheckSwap - 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) ) + std::string ts = GetTransferSyntax(); + if ( !Global::GetTS()->IsTransferSyntax(ts) ) { - dbg.Verbose(0, "gdcmDocument::Parse7FE0: no sequence delimiter item"); - dbg.Verbose(0, " at end of RLE item sequence"); + gdcmWarningMacro("True DICOM File, with NO Tansfer Syntax: " << ts ); + return; } - } -} - + // Group 0002 is always 'Explicit ...' enven when Transfer Syntax says 'Implicit ..." -/** - * \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) -{ - std::string s1,s2; - - // Patient Name - s1=this->GetEntryByNumber(0x0010,0x0010); - s2=document.GetEntryByNumber(0x0010,0x0010); - if(s1 < s2) - return true; - else if(s1 > s2) - return false; - else - { - // Patient ID - s1=this->GetEntryByNumber(0x0010,0x0020); - s2=document.GetEntryByNumber(0x0010,0x0020); - if (s1 < s2) - return true; - else if (s1 > s2) - return true; - else - { - // Study Instance UID - s1=this->GetEntryByNumber(0x0020,0x000d); - s2=document.GetEntryByNumber(0x0020,0x000d); - if (s1 < s2) - return true; - else if(s1 > s2) - return false; - else + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ImplicitVRLittleEndian ) { - // Serie Instance UID - s1=this->GetEntryByNumber(0x0020,0x000e); - s2=document.GetEntryByNumber(0x0020,0x000e); - if (s1 < s2) - return true; - else if(s1 > s2) - return false; + Filetype = ImplicitVR; } + + // FIXME Strangely, this works with + //'Implicit VR Transfer Syntax (GE Private) + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ExplicitVRBigEndian ) + { + gdcmWarningMacro("Transfer Syntax Name = [" + << GetTransferSyntaxName() << "]" ); + SwitchByteSwapCode(); + group = SwapShort(group); + elem = SwapShort(elem); } } - return false; } +//----------------------------------------------------------------------------- +// Print //----------------------------------------------------------------------------- +} // end namespace gdcm