X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=src%2FgdcmDocument.cxx;h=1a3f177c3525696ae6a6126d146b0625c3f84943;hb=e46fd685cb5a48ed7604c523cba93359003f5984;hp=0147abbe90c9d4b0434b6cfa21f4f9350043bbba;hpb=7ad80376db5efb1b6f98e910a82886bf3cc68a26;p=gdcm.git diff --git a/src/gdcmDocument.cxx b/src/gdcmDocument.cxx index 0147abbe..1a3f177c 100644 --- a/src/gdcmDocument.cxx +++ b/src/gdcmDocument.cxx @@ -3,8 +3,8 @@ Program: gdcm Module: $RCSfile: gdcmDocument.cxx,v $ Language: C++ - Date: $Date: 2005/03/22 11:29:37 $ - Version: $Revision: 1.233 $ + Date: $Date: 2005/11/21 16:28:06 $ + Version: $Revision: 1.331 $ Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de l'Image). All rights reserved. See Doc/License.txt or @@ -17,8 +17,6 @@ =========================================================================*/ #include "gdcmDocument.h" -#include "gdcmValEntry.h" -#include "gdcmBinEntry.h" #include "gdcmSeqEntry.h" #include "gdcmGlobal.h" #include "gdcmUtil.h" @@ -27,16 +25,17 @@ #include "gdcmDictSet.h" #include "gdcmDocEntrySet.h" #include "gdcmSQItem.h" +#include "gdcmDataEntry.h" #include #include #include +#include // for isdigit +#include // for atoi namespace gdcm { //----------------------------------------------------------------------------- -// Refer to Document::CheckSwap() -//const unsigned int Document::HEADER_LENGTH_TO_READ = 256; // Refer to Document::SetMaxSizeLoadEntry() const unsigned int Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096 @@ -47,12 +46,11 @@ const unsigned int Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096 /** * \brief This default constructor neither loads nor parses the file. - * You should then invoke \ref Document::SetFileName - * and \ref Document::Load. + * You should then invoke \ref Document::Load. * */ Document::Document() - :ElementSet(-1) + :ElementSet() { Fp = 0; @@ -60,81 +58,125 @@ Document::Document() Initialize(); SwapCode = 1234; Filetype = ExplicitVR; + // Load will set it to true if sucessfull Group0002Parsed = false; - LoadMode = 0x00000000; // Load everything + IsDocumentAlreadyLoaded = false; + IsDocumentModified = true; + LoadMode = LD_ALL; // default : load everything, later + SetFileName(""); } -/** - * \brief Constructor (not to break the API) - * @param filename 'Document' (File or DicomDir) to be opened for parsing - */ -Document::Document( std::string const &filename ) - :ElementSet(-1) -{ - Fp = 0; - - SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); - Initialize(); - SwapCode = 1234; - Filetype = ExplicitVR; - Group0002Parsed = false; - LoadMode = 0x00000000; // Load everything - - Load(filename); -} /** * \brief Canonical destructor. */ Document::~Document () { - RefPubDict = NULL; - RefShaDict = NULL; + CloseFile(); } //----------------------------------------------------------------------------- // Public +/** + * \brief Loader. use SetLoadMode(), SetFileName() before ! + * @return false if file cannot be open or no swap info was found, + * or no tag was found. + */ +bool Document::Load( ) +{ + if ( GetFileName() == "" ) + { + gdcmWarningMacro( "Use SetFileName, before !" ); + return false; + } + return DoTheLoadingDocumentJob( ); +} + +/* +//#ifndef GDCM_LEGACY_REMOVE +bool Document::Load( std::string const &fileName ) +{ + Filename = fileName; + return DoTheLoadingDocumentJob( ); +} +//#endif +*/ /** - * \brief Loader - * @param filename 'Document' (File or DicomDir) to be opened for parsing + * \brief Performs the Loading Job (internal use only) + * @return false if file cannot be open or no swap info was found, + * or no tag was found. */ -void Document::Load( std::string const &filename ) +bool Document::DoTheLoadingDocumentJob( ) { - Filename = filename; + if ( ! IsDocumentModified ) // Nothing to do ! + return true; - // We should clean out anything that already exists. + ClearEntry(); Fp = 0; if ( !OpenFile() ) { - return; + // warning already performed in OpenFile() + Filetype = Unknown; + return false; } Group0002Parsed = false; - gdcmWarningMacro( "Starting parsing of file: " << Filename.c_str()); + gdcmDebugMacro( "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(); + // CheckSwap returns a boolean + // (false if no swap info of any kind was found) + if (! CheckSwap() ) + { + gdcmWarningMacro( "Neither a DICOM V3 nor an ACR-NEMA file: " + << Filename.c_str()); + CloseFile(); + return false; + } + long beg = Fp->tellg(); // just after DICOM preamble (if any) lgt -= beg; // remaining length to parse - ParseDES( this, beg, lgt, false); // Loading is done during parsing + // Recursive call. + // Loading is done during parsing + ParseDES( this, beg, lgt, false); // delim_mode is first defaulted to false + + if ( IsEmpty() ) + { + gdcmErrorMacro( "No tag in internal hash table for: " + << Filename.c_str()); + CloseFile(); + return false; + } + IsDocumentAlreadyLoaded = true; Fp->seekg( 0, std::ios::beg); // Load 'non string' values - std::string PhotometricInterpretation = GetEntryValue(0x0028,0x0004); - if( PhotometricInterpretation == "PALETTE COLOR " ) + std::string PhotometricInterpretation = GetEntryString(0x0028,0x0004); + if ( PhotometricInterpretation == "PALETTE COLOR " ) { - LoadEntryBinArea(0x0028,0x1200); // gray LUT - /// FIXME FIXME FIXME + // FIXME + // Probabely this line should be outside the 'if' + // Try to find an image sample holding a 'gray LUT' + LoadEntryBinArea(0x0028,0x1200); // gray LUT + + /// FIXME + /// --> FIXME : The difference between BinEntry and DataEntry + /// --> no longer exists, but the alteration of Dicom Dictionary remains. + /// --> Old comment restored on purpose. + /// --> New one (replacing both BinEntry and ValEntry by DataEntry) + /// --> had absolutely no meaning. + /// --> The whole comment will be removed when the stuff is cleaned ! + /// --> /// 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 @@ -143,12 +185,18 @@ void Document::Load( std::string const &filename ) /// 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: + /// correct value, one needs to decide 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. + + // --> OB (byte aray) or OW (short int aray) + // The actual VR has to be deduced from other entries. + // Our way of loading them may fail in some cases : + // We must or not SwapByte depending on other field values. + LoadEntryBinArea(0x0028,0x1201); // R LUT LoadEntryBinArea(0x0028,0x1202); // G LUT LoadEntryBinArea(0x0028,0x1203); // B LUT @@ -159,16 +207,57 @@ void Document::Load( std::string const &filename ) LoadEntryBinArea(0x0028,0x1222); // Segmented Blue Palette Color LUT Data LoadEntryBinArea(0x0028,0x1223); - } + } + //FIXME later : how to use it? - LoadEntryBinArea(0x0028,0x3006); //LUT Data (CTX dependent) + SeqEntry *modLutSeq = GetSeqEntry(0x0028,0x3000); + if ( modLutSeq !=0 ) + { + SQItem *sqi= modLutSeq->GetFirstSQItem(); + if ( sqi != 0 ) + { + DataEntry *dataEntry = sqi->GetDataEntry(0x0028,0x3006); + if ( dataEntry != 0 ) + { + if ( dataEntry->GetLength() != 0 ) + { + // FIXME : CTX dependent means : contexted dependant. + // see upper comment. + LoadEntryBinArea(dataEntry); //LUT Data (CTX dependent) + } + } + } + } + + // Force Loading some more elements if user asked to. + + gdcm::DocEntry *d; + for (ListElements::iterator it = UserForceLoadList.begin(); + it != UserForceLoadList.end(); + ++it) + { + gdcmDebugMacro( "Force Load " << std::hex + << (*it).Group << "|" <<(*it).Elem ); + + d = GetDocEntry( (*it).Group, (*it).Elem); + + if ( d == NULL) + { + gdcmWarningMacro( "You asked toForce Load " << std::hex + << (*it).Group <<"|"<< (*it).Elem + << " that doesn't exist" ); + continue; + } + + LoadDocEntry(d, true); + } 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) + // used at CREATIS. For the time being (say a couple of years) // we keep this kludge to allow CREATIS users // reading their old images. // @@ -176,20 +265,34 @@ void Document::Load( std::string const &filename ) // we switch lineNumber and columnNumber // std::string RecCode; - RecCode = GetEntryValue(0x0008, 0x0010); // recognition code (RET) + RecCode = GetEntryString(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); + std::string rows = GetEntryString(0x0028, 0x0010); + std::string columns = GetEntryString(0x0028, 0x0011); + SetEntryString(columns, 0x0028, 0x0010); + SetEntryString(rows , 0x0028, 0x0011); } // --- End of ACR-LibIDO kludge --- + return true; } + +/** + * \brief Adds a new element we want to load anyway + * @param group Group number of the target tag. + * @param elem Element number of the target tag. + */ +void Document::AddForceLoadElement (uint16_t group, uint16_t elem) +{ + DicomElement el; + el.Group = group; + el.Elem = elem; + UserForceLoadList.push_back(el); +} /** * \brief Get the public dictionary used */ @@ -227,18 +330,17 @@ bool Document::SetShaDict(DictKey const &dictName) } /** - * \brief This predicate, based on hopefully reasonable heuristics, - * decides whether or not the current Document was properly parsed - * and contains the mandatory information for being considered as - * a well formed and usable Dicom/Acr File. - * @return true when Document is the one of a reasonable Dicom/Acr file, - * false otherwise. + * \brief This predicate tells us whether or not the current Document + * was properly parsed and contains at least *one* Dicom Element + * (and nothing more, sorry). + * @return false when we're 150 % sure it's NOT a Dicom/Acr file, + * true otherwise. */ -bool Document::IsReadable() +bool Document::IsParsable() { - if( Filetype == Unknown) + if ( Filetype == Unknown ) { - gdcmWarningMacro( "Wrong filetype"); + gdcmWarningMacro( "Wrong filetype for " << GetFileName()); return false; } @@ -250,6 +352,17 @@ bool Document::IsReadable() return true; } +/** + * \brief This predicate tells us whether or not the current Document + * was properly parsed and contains at least *one* Dicom Element + * (and nothing more, sorry). + * @return false when we're 150 % sure it's NOT a Dicom/Acr file, + * true otherwise. + */ +bool Document::IsReadable() +{ + return IsParsable(); +} /** * \brief Predicate for dicom version 3 file. @@ -258,7 +371,7 @@ bool Document::IsReadable() bool Document::IsDicomV3() { // Checking if Transfer Syntax exists is enough - // Anyway, it's to late check if the 'Preamble' was found ... + // Anyway, it's too late check if the 'Preamble' was found ... // And ... would it be a rich idea to check ? // (some 'no Preamble' DICOM images exist !) return GetDocEntry(0x0002, 0x0010) != NULL; @@ -295,7 +408,8 @@ FileType Document::GetFileType() * \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. + * @return The encountered Transfer Syntax of the current document, if DICOM. + * GDCM_UNKNOWN for ACR-NEMA files (or broken headers ...) */ std::string Document::GetTransferSyntax() { @@ -308,20 +422,27 @@ std::string Document::GetTransferSyntax() // 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) ) + if (DataEntry *dataEntry = dynamic_cast(entry) ) { - std::string transfer = valEntry->GetValue(); + std::string transfer = dataEntry->GetString(); // 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 + gdcmWarningMacro( "Transfer Syntax has length = 0."); return GDCM_UNKNOWN; } while ( !isdigit((unsigned char)transfer[transfer.length()-1]) ) { transfer.erase(transfer.length()-1, 1); + if ( transfer.length() == 0 ) + { + // for brain damaged headers + gdcmWarningMacro( "Transfer Syntax contains no valid character."); + return GDCM_UNKNOWN; + } } return transfer; } @@ -335,7 +456,7 @@ std::string Document::GetTransferSyntax() std::string Document::GetTransferSyntaxName() { // use the TS (TS : Transfer Syntax) - std::string transferSyntax = GetEntryValue(0x0002,0x0010); + std::string transferSyntax = GetEntryString(0x0002,0x0010); if ( (transferSyntax.find(GDCM_NOTLOADED) < transferSyntax.length()) ) { @@ -345,7 +466,7 @@ std::string Document::GetTransferSyntaxName() } if ( transferSyntax == GDCM_UNFOUND ) { - gdcmWarningMacro( "Unfound Transfer Syntax (0002,0010)"); + gdcmDebugMacro( "Unfound Transfer Syntax (0002,0010)"); return "Uncompressed ACR-NEMA"; } @@ -398,12 +519,48 @@ uint32_t Document::SwapLong(uint32_t a) a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) ); break; default : - gdcmErrorMacro( "Unset swap code:" << SwapCode ); + gdcmErrorMacro( "Unexpected swap code:" << SwapCode ); a = 0; } return a; } +/** + * \brief Swaps back the bytes of 8-byte long 'double' accordingly to + * processor order. + * @return The properly swaped 64 bits double. + */ +double Document::SwapDouble(double a) +{ + switch (SwapCode) + { + // There were no 'double' at ACR-NEMA time. + // We just have to deal with 'straight Little Endian' and + // 'straight Big Endian' + case 1234 : + break; + case 4321 : + { + char *beg = (char *)&a; + char *end = beg + 7; + char t; + for (unsigned int i = 0; i<7; i++) + { + t = *beg; + *beg = *end; + *end = t; + beg++, + end--; + } + } + break; + default : + gdcmErrorMacro( "Unexpected swap code:" << SwapCode ); + a = 0.; + } + return a; +} + // // -----------------File I/O --------------- /** @@ -419,60 +576,77 @@ std::ifstream *Document::OpenFile() return 0; } - if(Fp) + if ( Fp ) { - gdcmWarningMacro( "File already open: " << Filename.c_str()); + gdcmDebugMacro( "File already open: " << Filename.c_str()); CloseFile(); } Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary); - if( ! *Fp ) + if ( ! *Fp ) { - gdcmDebugMacro( "Cannot open file: " << Filename.c_str()); + // Don't user gdcmErrorMacro : + // a spurious message will appear when you use, for instance + // gdcm::FileHelper *fh = new gdcm::FileHelper( outputFileName ); + // to create outputFileName. + + // FIXME : if the upper comment is still usefull + // --> the constructor is not so good ... + + gdcmWarningMacro( "Cannot open file: " << Filename.c_str()); delete Fp; Fp = 0; return 0; + //exit(1); // No function is allowed to leave the application instead + // of warning the caller } uint16_t zero = 0; Fp->read((char*)&zero, (size_t)2); - if( Fp->eof() ) + if ( Fp->eof() ) { CloseFile(); return 0; } - //ACR -- or DICOM with no Preamble; may start with a Shadow Group -- - if( + //-- Broken ACR or DICOM with no Preamble; may start with a Shadow Group -- + + // FIXME : We cannot be sure the preable is only zeroes.. + // (see ACUSON-24-YBR_FULL-RLE.dcm ) + 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); + std::string msg = Util::Format( + "ACR/DICOM starting by 0x(%04x) at the beginning of the file\n", zero); + // FIXME : is it a Warning message, or a Debug message? gdcmWarningMacro( msg.c_str() ); return Fp; } - //DICOM + //-- DICOM -- Fp->seekg(126L, std::ios::cur); - char dicm[4] = {' ',' ',' ',' '}; + char dicm[4]; // = {' ',' ',' ',' '}; Fp->read(dicm, (size_t)4); - if( Fp->eof() ) + if ( Fp->eof() ) { CloseFile(); return 0; } - if( memcmp(dicm, "DICM", 4) == 0 ) + if ( memcmp(dicm, "DICM", 4) == 0 ) { HasDCMPreamble = true; return Fp; } - + + // -- Neither ACR/No Preamble Dicom nor DICOMV3 file CloseFile(); - gdcmWarningMacro( "Not DICOM/ACR (missing preamble)" << Filename.c_str()); - + // Don't user Warning nor Error, not to polute the output + // while directory recursive parsing ... + gdcmDebugMacro( "Neither ACR/No Preamble Dicom nor DICOMV3 file: " + << Filename.c_str()); return 0; } @@ -482,7 +656,7 @@ std::ifstream *Document::OpenFile() */ bool Document::CloseFile() { - if( Fp ) + if ( Fp ) { Fp->close(); delete Fp; @@ -492,21 +666,17 @@ bool Document::CloseFile() } /** - * \brief Writes in a file all the Header Entries (Dicom Elements) + * \brief Writes in a file all the Entries (Dicom Elements) * @param fp file pointer on an already open file (actually: Output File Stream) * @param filetype Type of the File to be written * (ACR-NEMA, ExplicitVR, ImplicitVR) - * @return Always true. */ void Document::WriteContent(std::ofstream *fp, FileType filetype) { - // \TODO move the following lines (and a lot of others, to be written) - // to a future function CheckAndCorrectHeader - - // (necessary if user wants to write a DICOM V3 file - // starting from an ACR-NEMA (V2) Header + // Skip if user wants to write an ACR-NEMA file - if ( filetype == ImplicitVR || filetype == ExplicitVR ) + if ( filetype == ImplicitVR || filetype == ExplicitVR || + filetype == JPEG ) { // writing Dicom File Preamble char filePreamble[128]; @@ -525,6 +695,10 @@ void Document::WriteContent(std::ofstream *fp, FileType filetype) * UpdateGroupLength(false,filetype); * if ( filetype == ACR) * UpdateGroupLength(true,ACR); + * + * --> Computing group length for groups with embeded Sequences + * --> was too much tricky / we were [in a hurry / too lazy] + * --> We don't write the element 0x0000 (group length) */ ElementSet::WriteContent(fp, filetype); // This one is recursive @@ -541,52 +715,100 @@ void Document::WriteContent(std::ofstream *fp, FileType filetype) void Document::LoadEntryBinArea(uint16_t group, uint16_t elem) { // Search the corresponding DocEntry - DocEntry *docElement = GetDocEntry(group, elem); - if ( !docElement ) + DocEntry *docEntry = GetDocEntry(group, elem); + if ( !docEntry ) + { + gdcmDebugMacro(std::hex << group << "|" << elem + << " doesn't exist" ); return; - - BinEntry *binElement = dynamic_cast(docElement); - if( !binElement ) + } + DataEntry *dataEntry = dynamic_cast(docEntry); + if ( !dataEntry ) + { + gdcmWarningMacro(std::hex << group << "|" << elem + << " is NOT a DataEntry"); return; - - LoadEntryBinArea(binElement); + } + LoadEntryBinArea(dataEntry); } /** * \brief Loads (from disk) the element content * when a string is not suitable - * @param elem Entry whose binArea is going to be loaded + * @param entry Entry whose binArea is going to be loaded */ -void Document::LoadEntryBinArea(BinEntry *elem) +void Document::LoadEntryBinArea(DataEntry *entry) { - if(elem->GetBinArea()) + if( entry->GetBinArea() ) return; bool openFile = !Fp; - if(openFile) + if ( openFile ) OpenFile(); - size_t o =(size_t)elem->GetOffset(); + size_t o =(size_t)entry->GetOffset(); Fp->seekg(o, std::ios::beg); - size_t l = elem->GetLength(); - uint8_t *a = new uint8_t[l]; - if( !a ) + size_t l = entry->GetLength(); + uint8_t *data = new uint8_t[l]; + if ( !data ) { - gdcmWarningMacro( "Cannot allocate BinEntry content"); + gdcmWarningMacro( "Cannot allocate DataEntry content for : " + << std::hex << entry->GetGroup() + << "|" << entry->GetElement() ); return; } - Fp->read((char*)a, l); - if( Fp->fail() || Fp->eof()) + // Read the data + Fp->read((char*)data, l); + if ( Fp->fail() || Fp->eof() ) { - delete[] a; + delete[] data; + entry->SetState(DataEntry::STATE_UNREAD); return; } - elem->SetBinArea(a); + // Swap the data content if necessary + uint32_t i; + unsigned short vrLgth = + Global::GetVR()->GetAtomicElementLength(entry->GetVR()); + +// FIXME : trouble expected if we read an ... OW Entry (LUT, etc ..) +// if( entry->GetVR() == "OW" ) +// vrLgth = 1; + + switch(vrLgth) + { + case 1: + { + break; + } + case 2: + { + uint16_t *data16 = (uint16_t *)data; + for(i=0;iSetBinArea(data); - if(openFile) + if ( openFile ) CloseFile(); } @@ -598,7 +820,7 @@ void Document::LoadEntryBinArea(BinEntry *elem) */ void Document::LoadDocEntrySafe(DocEntry *entry) { - if(Fp) + if ( Fp ) { long PositionOnEntry = Fp->tellg(); LoadDocEntry(entry); @@ -616,21 +838,21 @@ void Document::LoadDocEntrySafe(DocEntry *entry) bool Document::operator<(Document &document) { // Patient Name - std::string s1 = GetEntryValue(0x0010,0x0010); - std::string s2 = document.GetEntryValue(0x0010,0x0010); - if(s1 < s2) + std::string s1 = GetEntryString(0x0010,0x0010); + std::string s2 = document.GetEntryString(0x0010,0x0010); + if (s1 < s2) { return true; } - else if( s1 > s2 ) + else if ( s1 > s2 ) { return false; } else { // Patient ID - s1 = GetEntryValue(0x0010,0x0020); - s2 = document.GetEntryValue(0x0010,0x0020); + s1 = GetEntryString(0x0010,0x0020); + s2 = document.GetEntryString(0x0010,0x0020); if ( s1 < s2 ) { return true; @@ -642,26 +864,26 @@ bool Document::operator<(Document &document) else { // Study Instance UID - s1 = GetEntryValue(0x0020,0x000d); - s2 = document.GetEntryValue(0x0020,0x000d); + s1 = GetEntryString(0x0020,0x000d); + s2 = document.GetEntryString(0x0020,0x000d); if ( s1 < s2 ) { return true; } - else if( s1 > s2 ) + else if ( s1 > s2 ) { return false; } else { // Serie Instance UID - s1 = GetEntryValue(0x0020,0x000e); - s2 = document.GetEntryValue(0x0020,0x000e); + s1 = GetEntryString(0x0020,0x000e); + s2 = document.GetEntryString(0x0020,0x000e); if ( s1 < s2 ) { return true; } - else if( s1 > s2 ) + else if ( s1 > s2 ) { return false; } @@ -687,7 +909,7 @@ uint16_t Document::ReadInt16() { throw FormatError( "Document::ReadInt16()", " file error." ); } - if( Fp->eof() ) + if ( Fp->eof() ) { throw FormatError( "Document::ReadInt16()", "EOF." ); } @@ -709,7 +931,7 @@ uint32_t Document::ReadInt32() { throw FormatError( "Document::ReadInt32()", " file error." ); } - if( Fp->eof() ) + if ( Fp->eof() ) { throw FormatError( "Document::ReadInt32()", "EOF." ); } @@ -719,7 +941,6 @@ uint32_t Document::ReadInt32() /** * \brief skips bytes inside the source file - * \warning NOT end user intended method ! * @return */ void Document::SkipBytes(uint32_t nBytes) @@ -729,13 +950,12 @@ void Document::SkipBytes(uint32_t nBytes) } /** - * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader - * @param filetype Type of the File to be written + * \brief Re-computes the length of the Dicom group 0002. */ -int Document::ComputeGroup0002Length( FileType filetype ) +int Document::ComputeGroup0002Length( ) { uint16_t gr; - std::string vr; + VRKey vr; int groupLength = 0; bool found0002 = false; @@ -746,22 +966,22 @@ int Document::ComputeGroup0002Length( FileType filetype ) { gr = entry->GetGroup(); - if( gr == 0x0002 ) + if ( gr == 0x0002 ) { found0002 = true; - if( entry->GetElement() != 0x0000 ) + if ( entry->GetElement() != 0x0000 ) { vr = entry->GetVR(); - - if( filetype == ExplicitVR ) - { - if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") ) + + //if ( (vr == "OB")||(vr == "OW")||(vr == "UT")||(vr == "SQ")) + // (no SQ, OW, UT in group 0x0002;) + if ( vr == "OB" ) { - // explicit VR AND OB, OW, SQ : 4 more bytes + // explicit VR AND (OB, OW, SQ, UT) : 4 more bytes groupLength += 4; } - } + groupLength += 2 + 2 + 4 + entry->GetLength(); } } @@ -777,7 +997,6 @@ int Document::ComputeGroup0002Length( FileType filetype ) // Private /** * \brief Loads all the needed Dictionaries - * \warning NOT end user intended method ! */ void Document::Initialize() { @@ -788,21 +1007,24 @@ void Document::Initialize() /** * \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 set DocEntrySet we are going to parse ('zero level' or a SQItem) * @param offset start of parsing - * @param l_max length to parse + * @param l_max length to parse (meaningless when we are in 'delimitor mode') * @param delim_mode : whether we are in 'delimitor mode' (l=0xffffff) or not */ void Document::ParseDES(DocEntrySet *set, long offset, long l_max, bool delim_mode) { DocEntry *newDocEntry; - ValEntry *newValEntry; - BinEntry *newBinEntry; + DataEntry *newDataEntry; SeqEntry *newSeqEntry; VRKey vr; - bool used; - + bool used; // will be set to false when something wrong happens to an Entry. + // (Entry will then be deleted) + bool delim_mode_intern = delim_mode; + bool first = true; + gdcmDebugMacro( "Enter in ParseDES, delim-mode " << delim_mode + << " at offset " << std::hex << offset ); while (true) { if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) @@ -810,119 +1032,86 @@ void Document::ParseDES(DocEntrySet *set, long offset, break; } - used = true; newDocEntry = ReadNextDocEntry( ); + // Uncoment this cerr line to be able to 'follow' the DocEntries + // when something *very* strange happens + if( Debug::GetDebugFlag() ) + std::cerr<GetKey()<<" "<GetVR()<GetVR(); - newValEntry = dynamic_cast(newDocEntry); - newBinEntry = dynamic_cast(newDocEntry); - newSeqEntry = dynamic_cast(newDocEntry); + // an Item Starter found elsewhere but the first position + // of a SeqEntry means previous entry was a Sequence + // but we didn't get it (private Sequence + Implicit VR) + // we have to backtrack. + if ( !first && newDocEntry->IsItemStarter() ) + { + // Debug message within the method ! + newDocEntry = Backtrack(newDocEntry); + } + else + { + PreviousDocEntry = newDocEntry; + } + + used = true; + newDataEntry = dynamic_cast(newDocEntry); - if ( newValEntry || newBinEntry ) + if ( newDataEntry ) { - 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."); - } - - //////////////////// 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() ); - } + //////////////////////////// DataEntry + + vr = newDocEntry->GetVR(); - LoadDocEntry( newBinEntry ); - if( !set->AddEntry( newBinEntry ) ) - { - //Expect big troubles if here - //delete newBinEntry; - used=false; - } + if ( !set->AddEntry( newDataEntry ) ) + { + gdcmDebugMacro( "in ParseDES : cannot add a DataEntry " + << newDataEntry->GetKey() + << " (at offset : " + << newDataEntry->GetOffset() << " )" ); + used=false; } else { - /////////////////////// ValEntry - - // When "set" is a Document, then we are at the top of the - // hierarchy and the Key is simply of the form ( group, elem )... - if ( 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(); + newDataEntry->Delete(); + // Load only if we can add (not a duplicate key) + LoadDocEntry( newDataEntry ); + } - if ( LoadMode & NO_SHADOW ) // User asked to skip, if possible, - // shadow groups ( if possible : - // whether element 0x0000 exits) + if ( newDataEntry->GetElement() == 0x0000 ) // if on group length + { + if ( newDataEntry->GetGroup()%2 != 0 ) // if Shadow Group { - if ( newValEntry->GetGroup()%2 != 0 ) + if ( LoadMode & LD_NOSHADOW ) // if user asked to skip shad.gr { - if ( newValEntry->GetElement() == 0x0000 ) + std::string strLgrGroup = newDataEntry->GetString(); + int lgrGroup; + if ( newDataEntry->IsUnfound() ) { - 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; - } + lgrGroup = atoi(strLgrGroup.c_str()); + Fp->seekg(lgrGroup, std::ios::cur); + //used = false; // never used + RemoveEntry( newDocEntry ); // Remove and delete + // bcc 5.5 is right "assigned a value that's never used" + // newDocEntry = 0; + continue; } } - } - - 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 && ((long)(Fp->tellg())-offset) >= l_max) - { - if(!used) - delete newDocEntry; - break; - } + bool delimitor = newDataEntry->IsItemDelimitor(); + + if ( (delimitor) || + (!delim_mode && ((long)(Fp->tellg())-offset) >= l_max) ) + { + if ( !used ) + newDocEntry->Delete(); + break; } // Just to make sure we are at the beginning of next entry. @@ -930,71 +1119,104 @@ void Document::ParseDES(DocEntrySet *set, long offset, } else { - // VR = "SQ" + /////////////////////// SeqEntry : VR = "SQ" + unsigned long l = newDocEntry->GetReadLength(); - if ( l != 0 ) // don't mess the delim_mode for zero-length sequence + if ( l != 0 ) // don't mess the delim_mode for 'zero-length sequence' { if ( l == 0xffffffff ) { - delim_mode = true; + delim_mode_intern = true; } else { - delim_mode = false; + delim_mode_intern = false; } } - if ( (LoadMode & NO_SEQ) && ! delim_mode ) // User asked to skip SQ + if ( (LoadMode & LD_NOSHADOWSEQ) && ! delim_mode_intern ) + { + // User asked to skip SeQuences *only* if they belong to Shadow Group + if ( newDocEntry->GetGroup()%2 != 0 ) + { + Fp->seekg( l, std::ios::cur); + newDocEntry->Delete(); // Delete, not in the set + continue; + } + } + if ( (LoadMode & LD_NOSEQ) && ! delim_mode_intern ) { + // User asked to skip *any* SeQuence Fp->seekg( l, std::ios::cur); - used = false; + newDocEntry->Delete(); // Delete, not in the set continue; - } + } + // delay the dynamic cast as late as possible + newSeqEntry = dynamic_cast(newDocEntry); - // no other way to create it ... - newSeqEntry->SetDelimitorMode( delim_mode ); + // no other way to create the Delimitor ... + newSeqEntry->SetDelimitorMode( delim_mode_intern ); // At the top of the hierarchy, stands a Document. When "set" // is a Document, then we are building the first depth level. // Hence the SeqEntry we are building simply has a depth // level of one: - if ( dynamic_cast< Document* > ( set ) ) + if ( set == this ) // ( dynamic_cast< Document* > ( set ) ) { newSeqEntry->SetDepthLevel( 1 ); - newSeqEntry->SetKey( newSeqEntry->GetKey() ); } // But when "set" is already a SQItem, we are building a nested // sequence, and hence the depth level of the new SeqEntry // we are building, is one level deeper: - if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) + + // time waste hunting + else if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) { newSeqEntry->SetDepthLevel( parentSQItem->GetDepthLevel() + 1 ); - newSeqEntry->SetKey( parentSQItem->GetBaseTagKey() - + newSeqEntry->GetKey() ); } if ( l != 0 ) { // Don't try to parse zero-length sequences + + gdcmDebugMacro( "Entry in ParseSQ, delim " << delim_mode_intern + << " at offset " << std::hex + << newDocEntry->GetOffset() ); + ParseSQ( newSeqEntry, newDocEntry->GetOffset(), - l, delim_mode); + l, delim_mode_intern); + + gdcmDebugMacro( "Exit from ParseSQ, delim " << delim_mode_intern); + } - if( !set->AddEntry( newSeqEntry ) ) + if ( !set->AddEntry( newSeqEntry ) ) { + gdcmWarningMacro( "in ParseDES : cannot add a SeqEntry " + << newSeqEntry->GetKey() + << " (at offset : " + << newSeqEntry->GetOffset() << " )" ); used = false; } - + else + { + newDocEntry->Delete(); + } + if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) { - if( !used ) - delete newDocEntry; + if ( !used ) + newDocEntry->Delete(); break; } - } + } // end SeqEntry : VR = "SQ" - if( !used ) - delete newDocEntry; - } + if ( !used ) + { + newDocEntry->Delete(); + } + first = false; + } // end While + gdcmDebugMacro( "Exit from ParseDES, delim-mode " << delim_mode ); } /** @@ -1015,30 +1237,25 @@ void Document::ParseSQ( SeqEntry *seqEntry, if ( !newDocEntry ) { - // FIXME Should warn user + gdcmWarningMacro("in ParseSQ : should never get here!"); break; } - if( delim_mode ) + if ( delim_mode ) { if ( newDocEntry->IsSequenceDelimitor() ) { seqEntry->SetDelimitationItem( newDocEntry ); + newDocEntry->Delete(); break; } } if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) { - delete newDocEntry; + newDocEntry->Delete(); break; } // create the current SQItem - SQItem *itemSQ = new SQItem( seqEntry->GetDepthLevel() ); - std::ostringstream newBase; - newBase << seqEntry->GetKey() - << "/" - << SQItemNumber - << "#"; - itemSQ->SetBaseTagKey( newBase.str() ); + SQItem *itemSQ = SQItem::New( seqEntry->GetDepthLevel() ); unsigned int l = newDocEntry->GetReadLength(); if ( l == 0xffffffff ) @@ -1050,16 +1267,17 @@ void Document::ParseSQ( SeqEntry *seqEntry, 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 ); + itemSQ->Delete(); + newDocEntry->Delete(); SQItemNumber++; if ( !delim_mode && ((long)(Fp->tellg())-offset ) >= l_max ) { @@ -1069,14 +1287,51 @@ void Document::ParseSQ( SeqEntry *seqEntry, } /** - * \brief Loads the element content if its length doesn't exceed - * the value specified with Document::SetMaxSizeLoadEntry() + * \brief When a private Sequence + Implicit VR is encountered + * we cannot guess it's a Sequence till we find the first + * Item Starter. We then backtrack to do the job. + * @param docEntry Item Starter that warned us + */ +DocEntry *Document::Backtrack(DocEntry *docEntry) +{ + // delete the Item Starter, built erroneously out of any Sequence + // it's not yet in the HTable/chained list + docEntry->Delete(); + + // Get all info we can from PreviousDocEntry + uint16_t group = PreviousDocEntry->GetGroup(); + uint16_t elem = PreviousDocEntry->GetElement(); + uint32_t lgt = PreviousDocEntry->GetLength(); + long offset = PreviousDocEntry->GetOffset(); + + gdcmDebugMacro( "Backtrack :" << std::hex << group + << "|" << elem + << " at offset " << offset ); + RemoveEntry( PreviousDocEntry ); + + // forge the Seq Entry + DocEntry *newEntry = NewSeqEntry(group, elem); + newEntry->SetLength(lgt); + newEntry->SetOffset(offset); + + // Move back to the beginning of the Sequence + Fp->seekg( 0, std::ios::beg); + Fp->seekg(offset, std::ios::cur); + + return newEntry; +} + +/** + * \brief Loads (or not) the element content depending if its length exceeds + * or not the value specified with Document::SetMaxSizeLoadEntry() * @param entry Header Entry (Dicom Element) to be dealt with + * @param forceLoad whether you want to force loading of 'long' elements */ -void Document::LoadDocEntry(DocEntry *entry) +void Document::LoadDocEntry(DocEntry *entry, bool forceLoad) { uint16_t group = entry->GetGroup(); - std::string vr = entry->GetVR(); + uint16_t elem = entry->GetElement(); + const VRKey &vr = entry->GetVR(); uint32_t length = entry->GetLength(); Fp->seekg((long)entry->GetOffset(), std::ios::beg); @@ -1085,173 +1340,65 @@ void Document::LoadDocEntry(DocEntry *entry) // (fffe e000) tells us an Element is beginning // (fffe e00d) tells us an Element just ended // (fffe e0dd) tells us the current SeQuence just ended - if( group == 0xfffe ) + // + // (fffe 0000) is an 'impossible' tag value, + // found in MR-PHILIPS-16-Multi-Seq.dcm + + if ( (group == 0xfffe && elem != 0x0000 ) || vr == "SQ" ) { // NO more value field for SQ ! return; } + DataEntry *dataEntryPtr = dynamic_cast< DataEntry* >(entry); + if( !dataEntryPtr ) + { + return; + } + // When the length is zero things are easy: if ( length == 0 ) { - ((ValEntry *)entry)->SetValue(""); + dataEntryPtr->SetBinArea(NULL,true); return; } // The elements whose length is bigger than the specified upper bound - // are not loaded. Instead we leave a short notice of the offset of - // the element content and it's length. + // are not loaded. std::ostringstream s; - if (length > MaxSizeLoadEntry) - { - if (BinEntry *binEntryPtr = dynamic_cast< BinEntry* >(entry) ) - { - //s << "gdcm::NotLoaded (BinEntry)"; - s << GDCM_NOTLOADED; - s << " Address:" << (long)entry->GetOffset(); - s << " Length:" << entry->GetLength(); - s << " x(" << std::hex << entry->GetLength() << ")"; - binEntryPtr->SetValue(s.str()); - } - // Be carefull : a BinEntry IS_A ValEntry ... - else if (ValEntry *valEntryPtr = dynamic_cast< ValEntry* >(entry) ) - { - // s << "gdcm::NotLoaded. (ValEntry)"; - s << GDCM_NOTLOADED; - s << " Address:" << (long)entry->GetOffset(); - s << " Length:" << entry->GetLength(); - s << " x(" << std::hex << entry->GetLength() << ")"; - valEntryPtr->SetValue(s.str()); - } - else - { - // fusible - gdcmErrorMacro( "MaxSizeLoadEntry exceeded, neither a BinEntry " - << "nor a ValEntry ?! Should never print that !" ); - } - - // to be sure we are at the end of the value ... - Fp->seekg((long)entry->GetOffset()+(long)entry->GetLength(), - std::ios::beg); - return; - } - // When we find a BinEntry not very much can be done : - if (BinEntry *binEntryPtr = dynamic_cast< BinEntry* >(entry) ) + if (!forceLoad) { - s << GDCM_BINLOADED; - binEntryPtr->SetValue(s.str()); - LoadEntryBinArea(binEntryPtr); // last one, not to erase length ! - return; - } - - if ( IsDocEntryAnInteger(entry) ) - { - uint32_t NewInt; - int nbInt; - // When short integer(s) are expected, read and convert the following - // n *two characters properly i.e. consider them as short integers as - // opposed to strings. - // Elements with Value Multiplicity > 1 - // contain a set of integers (not a single one) - if (vr == "US" || vr == "SS") + if (length > MaxSizeLoadEntry) { - nbInt = length / 2; - NewInt = ReadInt16(); - s << NewInt; - if (nbInt > 1) - { - for (int i=1; i < nbInt; i++) - { - s << '\\'; - NewInt = ReadInt16(); - s << NewInt; - } - } - } - // See above comment on multiple integers (mutatis mutandis). - else if (vr == "UL" || vr == "SL") - { - nbInt = length / 4; - NewInt = ReadInt32(); - s << NewInt; - if (nbInt > 1) - { - for (int i=1; i < nbInt; i++) - { - s << '\\'; - NewInt = ReadInt32(); - s << NewInt; - } - } - } -#ifdef GDCM_NO_ANSI_STRING_STREAM - s << std::ends; // to avoid oddities on Solaris -#endif //GDCM_NO_ANSI_STRING_STREAM + dataEntryPtr->SetBinArea(NULL,true); + dataEntryPtr->SetState(DataEntry::STATE_NOTLOADED); - ((ValEntry *)entry)->SetValue(s.str()); - return; - } - - // 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()) - { - gdcmWarningMacro("Unread element value"); - valEntry->SetValue(GDCM_UNREAD); + // to be sure we are at the end of the value ... + Fp->seekg((long)entry->GetOffset()+(long)entry->GetLength(), + std::ios::beg); return; } - - if( vr == "UI" ) - { - // Because of correspondance with the VR dic - valEntry->SetValue(newValue); - } - else - { - valEntry->SetValue(newValue); - } - } - else - { - gdcmErrorMacro( "Should have a ValEntry, here !"); } + + LoadEntryBinArea(dataEntryPtr); // last one, not to erase length ! } /** - * \brief Find the value Length of the passed Header Entry + * \brief Find the value Length of the passed Doc Entry * @param entry Header Entry whose length of the value shall be loaded. */ void Document::FindDocEntryLength( DocEntry *entry ) throw ( FormatError ) { - std::string vr = entry->GetVR(); + const VRKey &vr = entry->GetVR(); uint16_t length16; if ( Filetype == ExplicitVR && !entry->IsImplicitVR() ) { - if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UN" ) + if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UT" + || vr == "UN" ) { // The following reserved two bytes (see PS 3.5-2003, section // "7.1.2 Data element structure with explicit vr", p 27) must be @@ -1273,6 +1420,9 @@ void Document::FindDocEntryLength( DocEntry *entry ) // chance to get the pixels by deciding the element goes // until the end of the file. Hence we artificially fix the // the length and proceed. + gdcmWarningMacro( " Computing the length failed for " << + entry->GetKey() <<" in " <tellg(); Fp->seekg(0L,std::ios::end); @@ -1313,7 +1463,8 @@ void Document::FindDocEntryLength( DocEntry *entry ) // Length is on 4 bytes. // Well ... group 0002 is always coded in 'Explicit VR Litle Endian' - // even if Transfer Syntax is 'Implicit VR ...' + // even if Transfer Syntax is 'Implicit VR ...' + // --> Except for 'Implicit VR Big Endian Transfer Syntax GE Private' FixDocEntryFoundLength( entry, ReadInt32() ); return; @@ -1322,7 +1473,6 @@ void Document::FindDocEntryLength( DocEntry *entry ) /** * \brief Find the Length till the next sequence delimiter - * \warning NOT end user intended method ! * @return */ uint32_t Document::FindDocEntryLengthOBOrOW() @@ -1383,28 +1533,33 @@ uint32_t Document::FindDocEntryLengthOBOrOW() * \brief Find the Value Representation of the current Dicom Element. * @return Value Representation of the current Entry */ -std::string Document::FindDocEntryVR() +VRKey Document::FindDocEntryVR() { if ( Filetype != ExplicitVR ) - return GDCM_UNKNOWN; + return GDCM_VRUNKNOWN; 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 - // cases (see e-film filtered files) one finds implicit VR tags mixed - // 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. + // we used a heuristic that found "UL" in the first tag and/or + // 'Transfer Syntax' told us it is. + // Alas this doesn't guarantee that all the tags will be in explicit VR. + // In some cases one finds implicit VR tags mixed 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. - char vr[3]; - Fp->read (vr, (size_t)2); - vr[2] = 0; + VRKey vr; + Fp->read(&(vr[0]),(size_t)2); - if( !CheckDocEntryVR(vr) ) + if ( !CheckDocEntryVR(vr) ) { + // Don't warn user with useless messages + // Often, delimiters (0xfffe), are not explicit VR ... + if ( CurrentGroup != 0xfffe ) + gdcmWarningMacro( "Unknown VR " << std::hex << "0x(" + << (unsigned int)vr[0] << "|" << (unsigned int)vr[1] + << ") at offset :" << positionOnEntry ); Fp->seekg(positionOnEntry, std::ios::beg); - return GDCM_UNKNOWN; + return GDCM_VRUNKNOWN; } return vr; } @@ -1414,150 +1569,16 @@ std::string Document::FindDocEntryVR() * and the taken VR. If they are different, the header entry is * updated with the new VR. * @param vr Dicom Value Representation - * @return false if the VR is incorrect of if the VR isn't referenced + * @return false if the VR is incorrect or if the VR isn't referenced * otherwise, it returns true */ -bool Document::CheckDocEntryVR(VRKey vr) +bool Document::CheckDocEntryVR(const VRKey &vr) { - if ( !Global::GetVR()->IsValidVR(vr) ) - return false; - - return true; -} - -/** - * \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 entry to tranform - * @return Transformed entry value - */ -std::string Document::GetDocEntryValue(DocEntry *entry) -{ - if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() ) - { - 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) - - if( vr == "US" || vr == "SS" ) - { - uint16_t newInt16; - - nbInt = length / 2; - for (int i=0; i < nbInt; i++) - { - if( i != 0 ) - { - s << '\\'; - } - 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" ) - { - uint32_t newInt32; - - nbInt = length / 4; - for (int i=0; i < nbInt; i++) - { - 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; - } - } -#ifdef GDCM_NO_ANSI_STRING_STREAM - s << std::ends; // to avoid oddities on Solaris -#endif //GDCM_NO_ANSI_STRING_STREAM - return s.str(); - } - return ((ValEntry *)entry)->GetValue(); -} - -/** - * \brief Get the reverse transformed value of the header entry. The VR - * value is used to define the reverse transformation to operate on - * the value - * \warning NOT end user intended method ! - * @param entry Entry to reverse transform - * @return Reverse transformed entry value - */ -std::string Document::GetDocEntryUnvalue(DocEntry *entry) -{ - if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() ) - { - std::string vr = entry->GetVR(); - std::vector tokens; - std::ostringstream s; - - if ( vr == "US" || vr == "SS" ) - { - uint16_t newInt16; - - tokens.erase( tokens.begin(), tokens.end()); // clean any previous value - Util::Tokenize (((ValEntry *)entry)->GetValue(), tokens, "\\"); - for (unsigned int i=0; i> 8 ) & 0xFF ); - } - tokens.clear(); - } - if ( vr == "UL" || vr == "SL") - { - uint32_t newInt32; - - tokens.erase(tokens.begin(),tokens.end()); // clean any previous value - Util::Tokenize (((ValEntry *)entry)->GetValue(), tokens, "\\"); - for (unsigned int i=0; i> 8 ) & 0xFF ) - << (char)(( newInt32 >> 16 ) & 0xFF ) - << (char)(( newInt32 >> 24 ) & 0xFF ); - } - tokens.clear(); - } - -#ifdef GDCM_NO_ANSI_STRING_STREAM - s << std::ends; // to avoid oddities on Solaris -#endif //GDCM_NO_ANSI_STRING_STREAM - return s.str(); - } - - return ((ValEntry *)entry)->GetValue(); + return Global::GetVR()->IsValidVR(vr); } /** * \brief Skip a given Header Entry - * \warning NOT end user intended method ! * @param entry entry to skip */ void Document::SkipDocEntry(DocEntry *entry) @@ -1567,14 +1588,19 @@ void Document::SkipDocEntry(DocEntry *entry) /** * \brief Skips to the beginning of the next Header Entry - * \warning NOT end user intended method ! * @param currentDocEntry entry to skip */ void Document::SkipToNextDocEntry(DocEntry *currentDocEntry) { - Fp->seekg((long)(currentDocEntry->GetOffset()), std::ios::beg); + int l = currentDocEntry->GetReadLength(); + if ( l == -1 ) // length = 0xffff shouldn't appear here ... + // ... but PMS imagers happen ! + return; + Fp->seekg((long)(currentDocEntry->GetOffset()), std::ios::beg); if (currentDocEntry->GetGroup() != 0xfffe) // for fffe pb + { Fp->seekg( (long)(currentDocEntry->GetReadLength()),std::ios::cur); + } } /** @@ -1587,7 +1613,7 @@ void Document::SkipToNextDocEntry(DocEntry *currentDocEntry) void Document::FixDocEntryFoundLength(DocEntry *entry, uint32_t foundLength) { - entry->SetReadLength( foundLength ); // will be updated only if a bug is found + entry->SetReadLength( foundLength );// will be updated only if a bug is found if ( foundLength == 0xffffffff) { foundLength = 0; @@ -1598,7 +1624,7 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, if ( foundLength % 2) { - gdcmWarningMacro( "Warning : Tag with uneven length " << foundLength + gdcmWarningMacro( "Warning : Tag with uneven length " << foundLength << " in x(" << std::hex << gr << "," << elem <<")"); } @@ -1606,7 +1632,7 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, // Allthough not recent many such GE corrupted images are still present // on Creatis hard disks. Hence this fix shall remain when such images // are no longer in use (we are talking a few years, here)... - // Note: XMedCom probably uses such a trick since it is able to read + // Note: XMedCon probably uses such a trick since it is able to read // those pesky GE images ... if ( foundLength == 13) { @@ -1625,7 +1651,7 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, else if ( gr == 0x0009 && ( elem == 0x1113 || elem == 0x1114 ) ) { foundLength = 4; - entry->SetReadLength(4); // a bug is to be fixed !? + entry->SetReadLength(4); // a bug is to be fixed ! } else if ( entry->GetVR() == "SQ" ) @@ -1636,16 +1662,20 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, //////// We encountered a 'delimiter' element i.e. a tag of the form // "fffe|xxxx" which is just a marker. Delimiters length should not be // taken into account. - else if( gr == 0xfffe ) + 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; } - } + else + { + foundLength=12; // to skip the mess that follows this bugged Tag ! + } + } entry->SetLength(foundLength); } @@ -1659,7 +1689,7 @@ bool Document::IsDocEntryAnInteger(DocEntry *entry) { uint16_t elem = entry->GetElement(); uint16_t group = entry->GetGroup(); - const std::string &vr = entry->GetVR(); + const VRKey &vr = entry->GetVR(); uint32_t length = entry->GetLength(); // When we have some semantics on the element we just read, and if we @@ -1673,9 +1703,9 @@ bool Document::IsDocEntryAnInteger(DocEntry *entry) } else { - // Allthough this should never happen, still some images have a + // Although this should never happen, still some images have a // corrupted group length [e.g. have a glance at offset x(8336) of - // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm]. + // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm. // Since for dicom compliant and well behaved headers, the present // test is useless (and might even look a bit paranoid), when we // encounter such an ill-formed image, we simply display a warning @@ -1704,12 +1734,7 @@ bool Document::IsDocEntryAnInteger(DocEntry *entry) * true when we hope ours assuptions are OK */ bool Document::CheckSwap() -{ - // The only guaranted way of finding the swap code is to find a - // group tag since we know it's length has to be of four bytes i.e. - // 0x00000004. Finding the swap code in then straigthforward. Trouble - // occurs when we can't find such group... - +{ uint32_t s32; uint16_t s16; @@ -1724,9 +1749,9 @@ bool Document::CheckSwap() Fp->read(deb, 256); char *entCur = deb + 128; - if( memcmp(entCur, "DICM", (size_t)4) == 0 ) + if ( memcmp(entCur, "DICM", (size_t)4) == 0 ) { - gdcmWarningMacro( "Looks like DICOM Version3 (preamble + DCM)" ); + gdcmDebugMacro( "Looks like DICOM Version3 (preamble + DCM)" ); // Group 0002 should always be VR, and the first element 0000 // Let's be carefull (so many wrong headers ...) @@ -1748,17 +1773,17 @@ bool Document::CheckSwap() // even if elem 0002,0010 (Transfer Syntax) tells us the file is // *Implicit* VR (see former 'gdcmData/icone.dcm') - if( memcmp(entCur, "UL", (size_t)2) == 0 || - memcmp(entCur, "OB", (size_t)2) == 0 || - memcmp(entCur, "UI", (size_t)2) == 0 || - memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later - // when Write DCM *adds* + if ( memcmp(entCur, "UL", (size_t)2) == 0 || + memcmp(entCur, "OB", (size_t)2) == 0 || + memcmp(entCur, "UI", (size_t)2) == 0 || + memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later + // when Write DCM *adds* // FIXME // Use Document::dicom_vr to test all the possibilities // instead of just checking for UL, OB and UI !? group 0000 { Filetype = ExplicitVR; - gdcmWarningMacro( "Group 0002 : Explicit Value Representation"); + gdcmDebugMacro( "Group 0002 : Explicit Value Representation"); } else { @@ -1770,29 +1795,66 @@ bool Document::CheckSwap() if ( net2host ) { SwapCode = 4321; - gdcmWarningMacro( "HostByteOrder != NetworkByteOrder"); + gdcmDebugMacro( "HostByteOrder != NetworkByteOrder, SwapCode = 4321"); } else { SwapCode = 1234; - gdcmWarningMacro( "HostByteOrder = NetworkByteOrder"); + gdcmDebugMacro( "HostByteOrder = NetworkByteOrder, SwapCode = 1234"); } // Position the file position indicator at first tag // (i.e. after the file preamble and the "DICM" string). - Fp->seekg(0, std::ios::beg); + + Fp->seekg(0, std::ios::beg); // FIXME : Is it usefull? + Fp->seekg ( 132L, std::ios::beg); return true; - } // End of DicomV3 + } // ------------------------------- End of DicomV3 ---------------- // Alas, this is not a DicomV3 file and whatever happens there is no file // preamble. We can reset the file position indicator to where the data // is (i.e. the beginning of the file). - gdcmWarningMacro( "Not a DICOM Version3 file"); + + gdcmWarningMacro( "Not a Kosher DICOM Version3 file (no preamble)"); + Fp->seekg(0, std::ios::beg); + // Let's check 'No Preamble Dicom File' : + // Should start with group 0x0002 + // and be Explicit Value Representation + + s16 = *((uint16_t *)(deb)); + SwapCode = 0; + switch ( s16 ) + { + case 0x0002 : + SwapCode = 1234; + entCur = deb + 4; + break; + case 0x0200 : + SwapCode = 4321; + entCur = deb + 6; + } + + if ( SwapCode != 0 ) + { + if ( memcmp(entCur, "UL", (size_t)2) == 0 || + memcmp(entCur, "OB", (size_t)2) == 0 || + memcmp(entCur, "UI", (size_t)2) == 0 || + memcmp(entCur, "SH", (size_t)2) == 0 || + memcmp(entCur, "AE", (size_t)2) == 0 || + memcmp(entCur, "OB", (size_t)2) == 0 ) + { + Filetype = ExplicitVR; + gdcmDebugMacro( "Group 0002 : Explicit Value Representation"); + return true; + } + } +// ------------------------------- End of 'No Preamble' DicomV3 ------------- + // Our next best chance would be to be considering a 'clean' ACR/NEMA file. - // By clean we mean that the length of the first tag is written down. + // By clean we mean that the length of the first group is written down. // If this is the case and since the length of the first group HAS to be // four (bytes), then determining the proper swap code is straightforward. @@ -1801,7 +1863,6 @@ bool Document::CheckSwap() // representation of a 32 bits integer. Hence the following dirty // trick : s32 = *((uint32_t *)(entCur)); - switch( s32 ) { case 0x00040000 : @@ -1832,7 +1893,7 @@ bool Document::CheckSwap() // 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-) + // 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... @@ -1863,7 +1924,7 @@ bool Document::CheckSwap() Filetype = ACR; return true; default : - gdcmWarningMacro( "ACR/NEMA unfound swap info (Really hopeless !)"); + gdcmWarningMacro("ACR/NEMA unfound swap info (Hopeless !)"); Filetype = Unknown; return false; } @@ -1875,7 +1936,8 @@ bool Document::CheckSwap() */ void Document::SwitchByteSwapCode() { - gdcmWarningMacro( "Switching Byte Swap code from "<< SwapCode); + gdcmDebugMacro( "Switching Byte Swap code from "<< SwapCode + << " at: 0x" << std::hex << Fp->tellg() ); if ( SwapCode == 1234 ) { SwapCode = 4321; @@ -1892,10 +1954,11 @@ void Document::SwitchByteSwapCode() { SwapCode = 3412; } + gdcmDebugMacro( " Into: "<< SwapCode ); } /** - * \brief during parsing, Header Elements too long are not loaded in memory + * \brief during parsing, Header Elements too long are not loaded in memory * @param newSize new size */ void Document::SetMaxSizeLoadEntry(long newSize) @@ -1913,23 +1976,20 @@ void Document::SetMaxSizeLoadEntry(long newSize) } /** - * \brief Read the next tag but WITHOUT loading it's value + * \brief Read the next tag WITHOUT loading it's value * (read the 'Group Number', the 'Element Number', * gets the Dict Entry * gets the VR, gets the length, gets the offset value) - * @return On succes the newly created DocEntry, NULL on failure. + * @return On succes : the newly created DocEntry, NULL on failure. */ DocEntry *Document::ReadNextDocEntry() { - uint16_t group; - uint16_t elem; - try { - group = ReadInt16(); - elem = ReadInt16(); + CurrentGroup = ReadInt16(); + CurrentElem = ReadInt16(); } - catch ( FormatError e ) + catch ( FormatError ) { // We reached the EOF (or an error occured) therefore // header parsing has to be considered as finished. @@ -1937,46 +1997,63 @@ DocEntry *Document::ReadNextDocEntry() } // Sometimes file contains groups of tags with reversed endianess. - HandleBrokenEndian(group, elem); + HandleBrokenEndian(CurrentGroup, CurrentElem); // In 'true DICOM' files Group 0002 is always little endian if ( HasDCMPreamble ) - HandleOutOfGroup0002(group, elem); + HandleOutOfGroup0002(CurrentGroup, CurrentElem); - std::string vr = FindDocEntryVR(); - std::string realVR = vr; + VRKey vr = FindDocEntryVR(); + + VRKey realVR = vr; - if( vr == GDCM_UNKNOWN) + if ( vr == GDCM_VRUNKNOWN ) { - if ( elem == 0x0000 ) // Group Length + if ( CurrentElem == 0x0000 ) // Group Length + { realVR = "UL"; // must be UL + } + else if (CurrentGroup%2 == 1 && + (CurrentElem >= 0x0010 && CurrentElem <=0x00ff )) + { + // DICOM PS 3-5 7.8.1 a) states that those + // (gggg-0010->00FF where gggg is odd) attributes have to be LO + realVR = "LO"; + } else { - DictEntry *dictEntry = GetDictEntry(group,elem); - if( dictEntry ) + DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem); + if ( dictEntry ) + { realVR = dictEntry->GetVR(); + dictEntry->Unregister(); + } } } + // gdcmDebugMacro( "Found VR: " << vr << " / Real VR: " << realVR ); 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); + if ( Global::GetVR()->IsVROfSequence(realVR) ) + newEntry = NewSeqEntry(CurrentGroup, CurrentElem); + else + { + newEntry = NewDataEntry(CurrentGroup, CurrentElem, realVR); + static_cast(newEntry)->SetState(DataEntry::STATE_NOTLOADED); + } - if( vr == GDCM_UNKNOWN ) + if ( vr == GDCM_VRUNKNOWN ) { - if( Filetype == ExplicitVR ) + if ( Filetype == ExplicitVR ) { // We thought this was explicit VR, but we end up with an // implicit VR tag. Let's backtrack. if ( newEntry->GetGroup() != 0xfffe ) { std::string msg; - msg = Util::Format("Entry (%04x,%04x) should be Explicit VR\n", - newEntry->GetGroup(), newEntry->GetElement()); + int offset = Fp->tellg(); + msg = Util::Format( + "Entry (%04x,%04x) at x(%x) should be Explicit VR\n", + newEntry->GetGroup(), newEntry->GetElement(), offset ); gdcmWarningMacro( msg.c_str() ); } } @@ -1987,32 +2064,34 @@ DocEntry *Document::ReadNextDocEntry() { FindDocEntryLength(newEntry); } - catch ( FormatError e ) + catch ( FormatError ) { // Call it quits - delete newEntry; + newEntry->Delete(); return 0; } newEntry->SetOffset(Fp->tellg()); - + return newEntry; } /** * \brief Handle broken private tag from Philips NTSCAN - * where the endianess is being switch to BigEndian for no - * apparent reason + * where the endianess is being switched to BigEndian + * for no apparent reason * @return no return */ void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem) { - // Endian reversion. Some files contain groups of tags with reversed endianess. + // Endian reversion. + // Some files contain groups of tags with reversed endianess. static int reversedEndian = 0; // try to fix endian switching in the middle of headers if ((group == 0xfeff) && (elem == 0x00e0)) { // start endian swap mark for group found + gdcmDebugMacro( "Start endian swap mark found." ); reversedEndian++; SwitchByteSwapCode(); // fix the tag @@ -2022,9 +2101,26 @@ void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem) else if (group == 0xfffe && elem == 0xe00d && reversedEndian) { // end of reversed endian group + gdcmDebugMacro( "End of reversed endian." ); reversedEndian--; SwitchByteSwapCode(); } + else if (group == 0xfeff && elem == 0xdde0) + { + // reversed Sequence Terminator found + // probabely a bug in the header ! + // Do what you want, it breaks ! + //reversedEndian--; + //SwitchByteSwapCode(); + gdcmWarningMacro( "Should never get here! reversed Sequence Terminator!" ); + // fix the tag + group = 0xfffe; + elem = 0xe0dd; + } + else if (group == 0xfffe && elem == 0xe0dd) + { + gdcmDebugMacro( "Straight Sequence Terminator." ); + } } /** @@ -2034,32 +2130,47 @@ void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem) */ void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) { - // Endian reversion. Some files contain groups of tags with reversed endianess. + // Endian reversion. + // Some files contain groups of tags with reversed endianess. if ( !Group0002Parsed && group != 0x0002) { Group0002Parsed = true; // we just came out of group 0002 - // if Transfer syntax is Big Endian we have to change CheckSwap + // if Transfer Syntax is Big Endian we have to change CheckSwap std::string ts = GetTransferSyntax(); + if ( ts == GDCM_UNKNOWN ) + { + gdcmDebugMacro("True DICOM File, with NO Transfer Syntax (?!) " ); + return; + } if ( !Global::GetTS()->IsTransferSyntax(ts) ) { - gdcmWarningMacro("True DICOM File, with NO Tansfer Syntax: " << ts ); + gdcmWarningMacro("True DICOM File, with illegal Transfer Syntax: [" + << ts << "]"); return; } - // Group 0002 is always 'Explicit ...' enven when Transfer Syntax says 'Implicit ..." + // Group 0002 is always 'Explicit ...' + // even when Transfer Syntax says 'Implicit ..." - if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ImplicitVRLittleEndian ) - { - Filetype = ImplicitVR; - } + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == + TS::ImplicitVRLittleEndian ) + { + Filetype = ImplicitVR; + } // FIXME Strangely, this works with - //'Implicit VR Transfer Syntax (GE Private) - if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ExplicitVRBigEndian ) + //'Implicit VR BigEndian Transfer Syntax (GE Private) + // + // --> Probabely normal, since we considered we never have + // to trust manufacturers. + // (we find very often 'Implicit VR' tag, + // even when Transfer Syntax tells us it's Explicit ... + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == + TS::ExplicitVRBigEndian ) { - gdcmWarningMacro("Transfer Syntax Name = [" + gdcmDebugMacro("Transfer Syntax Name = [" << GetTransferSyntaxName() << "]" ); SwitchByteSwapCode(); group = SwapShort(group);