X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=src%2FgdcmDocument.cxx;h=8ac4b2296ae0d43209ef487b7bfed0319bf1e90c;hb=880f6dfe2e061712fafbf75ab9547aadf170fc40;hp=dbeb89c48896adaa8015f07f721f06e0fc55983b;hpb=0493aec929fbf327a5ead989191b312edab84b9b;p=gdcm.git diff --git a/src/gdcmDocument.cxx b/src/gdcmDocument.cxx index dbeb89c4..8ac4b229 100644 --- a/src/gdcmDocument.cxx +++ b/src/gdcmDocument.cxx @@ -3,8 +3,8 @@ Program: gdcm Module: $RCSfile: gdcmDocument.cxx,v $ Language: C++ - Date: $Date: 2005/11/04 15:33:35 $ - Version: $Revision: 1.324 $ + Date: $Date: 2006/05/05 22:13:55 $ + Version: $Revision: 1.347 $ Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de l'Image). All rights reserved. See Doc/License.txt or @@ -33,6 +33,10 @@ #include // for isdigit #include // for atoi +#if defined(__BORLANDC__) + #include // for memset +#endif + namespace gdcm { //----------------------------------------------------------------------------- @@ -89,7 +93,9 @@ bool Document::Load( ) return false; } return DoTheLoadingDocumentJob( ); -} +} + +#ifndef GDCM_LEGACY_REMOVE /** * \brief Loader. (DEPRECATED : not to break the API) * @param fileName 'Document' (File or DicomDir) to be open for parsing @@ -101,6 +107,7 @@ bool Document::Load( std::string const &fileName ) Filename = fileName; return DoTheLoadingDocumentJob( ); } +#endif /** * \brief Performs the Loading Job (internal use only) @@ -126,10 +133,10 @@ bool Document::DoTheLoadingDocumentJob( ) 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); + // Computes the total length of the file + Fp->seekg(0, std::ios::end); // Once per Document ! + long lgt = Fp->tellg(); // Once per Document ! + Fp->seekg(0, std::ios::beg); // Once per Document ! // CheckSwap returns a boolean // (false if no swap info of any kind was found) @@ -158,7 +165,7 @@ bool Document::DoTheLoadingDocumentJob( ) } IsDocumentAlreadyLoaded = true; - Fp->seekg( 0, std::ios::beg); + //Fp->seekg(0, std::ios::beg); // Once per Document! // Load 'non string' values @@ -211,13 +218,13 @@ bool Document::DoTheLoadingDocumentJob( ) } //FIXME later : how to use it? - SeqEntry *modLutSeq = GetSeqEntry(0x0028,0x3000); + SeqEntry *modLutSeq = GetSeqEntry(0x0028,0x3000); // Modality LUT Sequence if ( modLutSeq !=0 ) { SQItem *sqi= modLutSeq->GetFirstSQItem(); if ( sqi != 0 ) { - DataEntry *dataEntry = sqi->GetDataEntry(0x0028,0x3006); + DataEntry *dataEntry = sqi->GetDataEntry(0x0028,0x3006); // LUT Data if ( dataEntry != 0 ) { if ( dataEntry->GetLength() != 0 ) @@ -244,7 +251,7 @@ bool Document::DoTheLoadingDocumentJob( ) if ( d == NULL) { - gdcmWarningMacro( "You asked toForce Load " << std::hex + gdcmWarningMacro( "You asked to ForceLoad " << std::hex << (*it).Group <<"|"<< (*it).Elem << " that doesn't exist" ); continue; @@ -337,22 +344,33 @@ bool Document::SetShaDict(DictKey const &dictName) * @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 ) { - gdcmErrorMacro( "Wrong filetype for " << GetFileName()); + gdcmWarningMacro( "Wrong filetype for " << GetFileName()); return false; } if ( IsEmpty() ) { - gdcmErrorMacro( "No tag in internal hash table."); + gdcmWarningMacro( "No tag in internal hash table."); return false; } 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. @@ -411,7 +429,10 @@ 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); + + // Well ... + // (parsing and loading happen at the very same stage!) + //LoadDocEntrySafe(entry); //JPRx if (DataEntry *dataEntry = dynamic_cast(entry) ) { std::string transfer = dataEntry->GetString(); @@ -599,7 +620,10 @@ std::ifstream *Document::OpenFile() return 0; } - //-- ACR or DICOM with no Preamble; may start with a Shadow Group -- + //-- 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 || @@ -614,7 +638,7 @@ std::ifstream *Document::OpenFile() } //-- DICOM -- - Fp->seekg(126L, std::ios::cur); + Fp->seekg(126L, std::ios::cur); // Once per Document char dicm[4]; // = {' ',' ',' ',' '}; Fp->read(dicm, (size_t)4); if ( Fp->eof() ) @@ -630,7 +654,7 @@ std::ifstream *Document::OpenFile() // -- Neither ACR/No Preamble Dicom nor DICOMV3 file CloseFile(); - // Don't user Warning nor Error, not to polute the output + // Don't user Warning nor Error, not to pollute the output // while directory recursive parsing ... gdcmDebugMacro( "Neither ACR/No Preamble Dicom nor DICOMV3 file: " << Filename.c_str()); @@ -671,7 +695,6 @@ void Document::WriteContent(std::ofstream *fp, FileType filetype) fp->write(filePreamble, 128); fp->write("DICM", 4); } - /* * \todo rewrite later, if really usefull * - 'Group Length' element is optional in DICOM @@ -705,15 +728,15 @@ void Document::LoadEntryBinArea(uint16_t group, uint16_t elem) DocEntry *docEntry = GetDocEntry(group, elem); if ( !docEntry ) { - gdcmWarningMacro(std::hex << group << "|" << elem - << "doesn't exist" ); + gdcmDebugMacro(std::hex << group << "|" << elem + << " doesn't exist" ); return; } DataEntry *dataEntry = dynamic_cast(docEntry); if ( !dataEntry ) { gdcmWarningMacro(std::hex << group << "|" << elem - << "is NOT a DataEntry"); + << " is NOT a DataEntry"); return; } LoadEntryBinArea(dataEntry); @@ -725,7 +748,7 @@ void Document::LoadEntryBinArea(uint16_t group, uint16_t elem) * @param entry Entry whose binArea is going to be loaded */ void Document::LoadEntryBinArea(DataEntry *entry) -{ +{ if( entry->GetBinArea() ) return; @@ -733,8 +756,8 @@ void Document::LoadEntryBinArea(DataEntry *entry) if ( openFile ) OpenFile(); - size_t o =(size_t)entry->GetOffset(); - Fp->seekg(o, std::ios::beg); + //size_t o =(size_t)entry->GetOffset(); + Fp->seekg((size_t)entry->GetOffset(), std::ios::beg); // FIXME : for each DataEntry ! size_t l = entry->GetLength(); uint8_t *data = new uint8_t[l]; @@ -795,25 +818,26 @@ void Document::LoadEntryBinArea(DataEntry *entry) entry->SetBinArea(data); - if ( openFile ) + if ( openFile ) // The file is left in the state (open/close) it was at entrance CloseFile(); } /** * \brief Loads the element while preserving the current * underlying file position indicator as opposed to - * LoadDocEntry that modifies it. + * LoadDocEntry that modifies it + * \note seems to be unused!. * @param entry DocEntry whose value will be loaded. */ -void Document::LoadDocEntrySafe(DocEntry *entry) -{ - if ( Fp ) - { - long PositionOnEntry = Fp->tellg(); - LoadDocEntry(entry); - Fp->seekg(PositionOnEntry, std::ios::beg); - } -} +//void Document::LoadDocEntrySafe(DocEntry *entry) +//{ +// if ( Fp ) +// { +// long PositionOnEntry = Fp->tellg(); // LoadDocEntrySafe is not used +// LoadDocEntry(entry); +// Fp->seekg(PositionOnEntry, std::ios::beg); // LoadDocEntrySafe is not used +// } +//} /** * \brief Compares two documents, according to \ref DicomDir rules @@ -926,17 +950,6 @@ uint32_t Document::ReadInt32() return g; } -/** - * \brief skips bytes inside the source file - * \warning NOT end user intended method ! - * @return - */ -void Document::SkipBytes(uint32_t nBytes) -{ - //FIXME don't dump the returned value - Fp->seekg((long)nBytes, std::ios::cur); -} - /** * \brief Re-computes the length of the Dicom group 0002. */ @@ -969,7 +982,6 @@ int Document::ComputeGroup0002Length( ) // explicit VR AND (OB, OW, SQ, UT) : 4 more bytes groupLength += 4; } - groupLength += 2 + 2 + 4 + entry->GetLength(); } } @@ -981,11 +993,38 @@ int Document::ComputeGroup0002Length( ) return groupLength; } +/** + * \brief CallStartMethod + */ +void Document::CallStartMethod() +{ + Progress = 0.0f; + Abort = false; + CommandManager::ExecuteCommand(this,CMD_STARTPROGRESS); +} + +/** + * \brief CallProgressMethod + */ +void Document::CallProgressMethod() +{ + CommandManager::ExecuteCommand(this,CMD_PROGRESS); +} + +/** + * \brief CallEndMethod + */ +void Document::CallEndMethod() +{ + Progress = 1.0f; + CommandManager::ExecuteCommand(this,CMD_ENDPROGRESS); +} + //----------------------------------------------------------------------------- // Private /** * \brief Loads all the needed Dictionaries - * \warning NOT end user intended method ! + * \warning NOT end user intended method ! */ void Document::Initialize() { @@ -1013,14 +1052,26 @@ void Document::ParseDES(DocEntrySet *set, long offset, bool delim_mode_intern = delim_mode; bool first = true; gdcmDebugMacro( "Enter in ParseDES, delim-mode " << delim_mode - << " at offset " << std::hex << offset ); + << " at offset " << std::hex << "0x(" << offset << ")" ); while (true) { - if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) - { - break; - } - + + ///\todo FIXME : On 64 bits processors, tellg gives unexpected results after a while ? + /// Probabely a bug in gdcm code somwhere (some memory erased ?) + +// Uncomment to track the bug +/* + if( Debug::GetDebugFlag() ) + std::cout << std::dec <<"(long)(Fp->tellg()) " << (long)(Fp->tellg()) // in Debug mode + << std::hex << " 0x(" <<(long)(Fp->tellg()) << ")" << std::endl; +*/ + + // if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) // Once per DocEntry + if ( !delim_mode ) // 'and then' doesn't exist in C++ :-( + if ( ((long)(Fp->tellg())-offset) >= l_max) // Once per DocEntry, when no delim mode + { + break; + } newDocEntry = ReadNextDocEntry( ); // Uncoment this cerr line to be able to 'follow' the DocEntries @@ -1034,13 +1085,13 @@ void Document::ParseDES(DocEntrySet *set, long offset, } // an Item Starter found elsewhere but the first position - // of a SeqEntry means previous entry was a Sequence + // 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); + { + // Debug message within the method ! + newDocEntry = Backtrack(newDocEntry); } else { @@ -1059,9 +1110,9 @@ void Document::ParseDES(DocEntrySet *set, long offset, if ( !set->AddEntry( newDataEntry ) ) { gdcmDebugMacro( "in ParseDES : cannot add a DataEntry " - << newDataEntry->GetKey() - << " (at offset : " - << newDataEntry->GetOffset() << " )" ); + << newDataEntry->GetKey() + << " (at offset : 0x(" + << newDataEntry->GetOffset() << ") )" ); used=false; } else @@ -1070,7 +1121,6 @@ void Document::ParseDES(DocEntrySet *set, long offset, // Load only if we can add (not a duplicate key) LoadDocEntry( newDataEntry ); } - if ( newDataEntry->GetElement() == 0x0000 ) // if on group length { if ( newDataEntry->GetGroup()%2 != 0 ) // if Shadow Group @@ -1078,25 +1128,30 @@ void Document::ParseDES(DocEntrySet *set, long offset, if ( LoadMode & LD_NOSHADOW ) // if user asked to skip shad.gr { std::string strLgrGroup = newDataEntry->GetString(); + int lgrGroup; - if ( newDataEntry->IsUnfound() ) + //if ( newDataEntry->IsUnfound() ) /?!? JPR { lgrGroup = atoi(strLgrGroup.c_str()); - Fp->seekg(lgrGroup, std::ios::cur); - //used = false; // never used + Fp->seekg(lgrGroup, std::ios::cur); // Once per Shadow group, when NOSHADOW RemoveEntry( newDocEntry ); // Remove and delete - // bcc 5.5 is right "assigned a value that's never used" - // newDocEntry = 0; continue; } } } } - bool delimitor = newDataEntry->IsItemDelimitor(); + bool delimitor = newDataEntry->IsItemDelimitor(); + bool outOfBounds = false; + if (!delim_mode ) + if ( ((long)(Fp->tellg())-offset) >= l_max ) //Once per DataEntry when no delim mode + outOfBounds = true; - if ( (delimitor) || - (!delim_mode && ((long)(Fp->tellg())-offset) >= l_max) ) + // 'and then', 'or else' don't exist in C++ :-( + // if ( (delimitor) || + // (!delim_mode && ((long)(Fp->tellg())-offset) >= l_max) ) // Once per DataEntry + + if ( delimitor || outOfBounds ) { if ( !used ) newDocEntry->Delete(); @@ -1104,7 +1159,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, } // Just to make sure we are at the beginning of next entry. - SkipToNextDocEntry(newDocEntry); + SkipToNextDocEntry(newDocEntry); // FIXME : once per DocEntry, segfault if commented out } else { @@ -1128,7 +1183,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, // User asked to skip SeQuences *only* if they belong to Shadow Group if ( newDocEntry->GetGroup()%2 != 0 ) { - Fp->seekg( l, std::ios::cur); + Fp->seekg( l, std::ios::cur); // once per SQITEM, when NOSHADOWSEQ newDocEntry->Delete(); // Delete, not in the set continue; } @@ -1136,7 +1191,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, if ( (LoadMode & LD_NOSEQ) && ! delim_mode_intern ) { // User asked to skip *any* SeQuence - Fp->seekg( l, std::ios::cur); + Fp->seekg( l, std::ios::cur); // Once per SQ, when NOSEQ newDocEntry->Delete(); // Delete, not in the set continue; } @@ -1168,30 +1223,33 @@ void Document::ParseDES(DocEntrySet *set, long offset, { // Don't try to parse zero-length sequences gdcmDebugMacro( "Entry in ParseSQ, delim " << delim_mode_intern - << " at offset " << std::hex - << newDocEntry->GetOffset() ); + << " at offset 0x(" << std::hex + << newDocEntry->GetOffset() << ")"); ParseSQ( newSeqEntry, newDocEntry->GetOffset(), l, delim_mode_intern); gdcmDebugMacro( "Exit from ParseSQ, delim " << delim_mode_intern); - } if ( !set->AddEntry( newSeqEntry ) ) { gdcmWarningMacro( "in ParseDES : cannot add a SeqEntry " << newSeqEntry->GetKey() - << " (at offset : " - << newSeqEntry->GetOffset() << " )" ); + << " (at offset : 0x(" + << newSeqEntry->GetOffset() << ") )" ); used = false; } else { newDocEntry->Delete(); } + + // if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) // Once per SeqEntry - if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) + if ( !delim_mode ) // 'and then' doesn't exist in C++ :-( + if ( ((long)(Fp->tellg())-offset) >= l_max) // Once per SeqEntry when no delim mode + { if ( !used ) newDocEntry->Delete(); @@ -1238,10 +1296,13 @@ void Document::ParseSQ( SeqEntry *seqEntry, break; } } - if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) + else // ! delim_mode { - newDocEntry->Delete(); - break; + if ( ((long)(Fp->tellg())-offset) >= l_max) // Once per SQItem when no delim mode + { + newDocEntry->Delete(); + break; + } } // create the current SQItem SQItem *itemSQ = SQItem::New( seqEntry->GetDepthLevel() ); @@ -1255,20 +1316,19 @@ void Document::ParseSQ( SeqEntry *seqEntry, { dlm_mod = false; } - - // remove fff0,e000, created out of the SQItem - Fp->seekg(offsetStartCurrentSQItem, std::ios::beg); + // fill up the current SQItem, starting at the beginning of fff0,e000 - + + Fp->seekg(offsetStartCurrentSQItem, std::ios::beg); // Once per SQItem ParseDES(itemSQ, offsetStartCurrentSQItem, l+8, dlm_mod); - - offsetStartCurrentSQItem = Fp->tellg(); + offsetStartCurrentSQItem = Fp->tellg(); // Once per SQItem seqEntry->AddSQItem( itemSQ, SQItemNumber ); itemSQ->Delete(); newDocEntry->Delete(); SQItemNumber++; - if ( !delim_mode && ((long)(Fp->tellg())-offset ) >= l_max ) + //if ( !delim_mode && ((long)(Fp->tellg())-offset ) >= l_max ) //JPRx + if ( !delim_mode && (offsetStartCurrentSQItem-offset ) >= l_max ) { break; } @@ -1294,8 +1354,8 @@ DocEntry *Document::Backtrack(DocEntry *docEntry) long offset = PreviousDocEntry->GetOffset(); gdcmDebugMacro( "Backtrack :" << std::hex << group - << "|" << elem - << " at offset " << offset ); + << "|" << elem + << " at offset 0x(" <SetOffset(offset); // Move back to the beginning of the Sequence - Fp->seekg( 0, std::ios::beg); - Fp->seekg(offset, std::ios::cur); + Fp->seekg(offset, std::ios::beg); // Only for Shadow Implicit VR SQ return newEntry; } @@ -1323,13 +1382,12 @@ void Document::LoadDocEntry(DocEntry *entry, bool forceLoad) const VRKey &vr = entry->GetVR(); uint32_t length = entry->GetLength(); - Fp->seekg((long)entry->GetOffset(), std::ios::beg); + // Fp->seekg((long)entry->GetOffset(), std::ios::beg); // JPRx // A SeQuence "contains" a set of Elements. // (fffe e000) tells us an Element is beginning // (fffe e00d) tells us an Element just ended // (fffe e0dd) tells us the current SeQuence just ended - // // (fffe 0000) is an 'impossible' tag value, // found in MR-PHILIPS-16-Multi-Seq.dcm @@ -1355,8 +1413,6 @@ void Document::LoadDocEntry(DocEntry *entry, bool forceLoad) // The elements whose length is bigger than the specified upper bound // are not loaded. - std::ostringstream s; - if (!forceLoad) { if (length > MaxSizeLoadEntry) @@ -1364,13 +1420,20 @@ void Document::LoadDocEntry(DocEntry *entry, bool forceLoad) dataEntryPtr->SetBinArea(NULL,true); dataEntryPtr->SetState(DataEntry::STATE_NOTLOADED); - // to be sure we are at the end of the value ... - Fp->seekg((long)entry->GetOffset()+(long)entry->GetLength(), - std::ios::beg); + // to be sure we are at the end of the value ... + // Fp->seekg((long)entry->GetOffset()+(long)entry->GetLength(), + // std::ios::beg); //JPRx return; } } - + + /// \todo: a method that *doesn't* load anything (maybe with MaxSizeLoadEntry=0 ?) + /// + a ForceLoad call on the +/- 20 'usefull' fields + /// Allow user to tell the fields he wants to ForceLoad + /// during initial stage. + /// Later, a GetString or GetBinArea will load the value from disk, if not loaded + /// + a method that load *everything* that's not yet loaded + LoadEntryBinArea(dataEntryPtr); // last one, not to erase length ! } @@ -1392,7 +1455,8 @@ void Document::FindDocEntryLength( DocEntry *entry ) // 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. - Fp->seekg( 2L, std::ios::cur); + + Fp->seekg( 2L, std::ios::cur); // Once per OW,OB,SQ DocEntry uint32_t length32 = ReadInt32(); if ( (vr == "OB" || vr == "OW") && length32 == 0xffffffff ) @@ -1400,7 +1464,7 @@ void Document::FindDocEntryLength( DocEntry *entry ) uint32_t lengthOB; try { - lengthOB = FindDocEntryLengthOBOrOW(); + lengthOB = FindDocEntryLengthOBOrOW();// for encapsulation of encoded pixel } catch ( FormatUnexpected ) { @@ -1412,11 +1476,11 @@ void Document::FindDocEntryLength( DocEntry *entry ) gdcmWarningMacro( " Computing the length failed for " << entry->GetKey() <<" in " <tellg(); - Fp->seekg(0L,std::ios::end); + long currentPosition = Fp->tellg(); // Only for gdcm-JPEG-LossLess3a.dcm-like + Fp->seekg(0L,std::ios::end); // Only for gdcm-JPEG-LossLess3a.dcm-like - long lengthUntilEOF = (long)(Fp->tellg())-currentPosition; - Fp->seekg(currentPosition, std::ios::beg); + long lengthUntilEOF = (long)(Fp->tellg())-currentPosition; // Only for gdcm-JPEG-LossLess3a.dcm-like + Fp->seekg(currentPosition, std::ios::beg); // Only for gdcm-JPEG-LossLess3a.dcm-like entry->SetReadLength(lengthUntilEOF); entry->SetLength(lengthUntilEOF); @@ -1454,6 +1518,7 @@ void Document::FindDocEntryLength( DocEntry *entry ) // Well ... group 0002 is always coded in 'Explicit VR Litle Endian' // even if Transfer Syntax is 'Implicit VR ...' // --> Except for 'Implicit VR Big Endian Transfer Syntax GE Private' + // where Group 0x0002 is *also* encoded in Implicit VR ! FixDocEntryFoundLength( entry, ReadInt32() ); return; @@ -1469,7 +1534,8 @@ 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(); + long positionOnEntry = Fp->tellg(); // Only for OB,OW DataElements + bool foundSequenceDelimiter = false; uint32_t totalLength = 0; @@ -1491,13 +1557,12 @@ uint32_t Document::FindDocEntryLengthOBOrOW() 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); + Fp->seekg(positionOnEntry, std::ios::beg); // Once per fragment (if any) of OB,OW DataElements throw FormatUnexpected( "Neither an Item tag nor a Sequence delimiter tag."); } @@ -1515,7 +1580,7 @@ uint32_t Document::FindDocEntryLengthOBOrOW() break; } } - Fp->seekg( positionOnEntry, std::ios::beg); + Fp->seekg( positionOnEntry, std::ios::beg); // Only for OB,OW DataElements return totalLength; } @@ -1526,14 +1591,26 @@ uint32_t Document::FindDocEntryLengthOBOrOW() VRKey Document::FindDocEntryVR() { if ( Filetype != ExplicitVR ) + { return GDCM_VRUNKNOWN; + } - long positionOnEntry = Fp->tellg(); + // Delimiters (0xfffe), are not explicit VR ... + if ( CurrentGroup == 0xfffe ) + return GDCM_VRUNKNOWN; + + long positionOnEntry; + if( Debug::GetWarningFlag() ) + positionOnEntry = Fp->tellg(); // Only in Warning Mode + // Warning: we believe this is explicit VR (Value Representation) because // 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. + // In some cases one finds implicit VR tags mixed within an explicit VR file + // Well... + // 'Normaly' the only case is : group 0002 Explicit, and other groups Implicit + // // Hence we make sure the present tag is in explicit VR and try to fix things // if it happens not to be the case. @@ -1542,12 +1619,23 @@ VRKey Document::FindDocEntryVR() if ( !CheckDocEntryVR(vr) ) { - // Don't warn user with useless messages - if ( (unsigned char)vr[0] != 0xff || (unsigned char)vr[1] != 0xff ) - gdcmWarningMacro( "Unknown VR " << std::hex << "0x(" +/* +// std::cout << "================================================================Unknown VR" + << std::hex << "0x(" << (unsigned int)vr[0] << "|" << (unsigned int)vr[1] - << ") at offset :" << positionOnEntry ); - Fp->seekg(positionOnEntry, std::ios::beg); + << ")" << "for : " << CurrentGroup + << " at offset : 0x(" << positionOnEntry << ")" + << std::endl; +*/ + gdcmWarningMacro( "Unknown VR " << std::hex << "0x(" + << (unsigned int)vr[0] << "|" << (unsigned int)vr[1] + << ")" + << " at offset : 0x(" << positionOnEntry<< ") for group " << CurrentGroup + ); + + //Fp->seekg(positionOnEntry, std::ios::beg); //JPRx + Fp->seekg((long)-2, std::ios::cur);// only for unrecognized VR (?!?) + //see :MR_Philips_Intera_PrivateSequenceExplicitVR.dcm return GDCM_VRUNKNOWN; } return vr; @@ -1568,7 +1656,6 @@ bool Document::CheckDocEntryVR(const VRKey &vr) /** * \brief Skip a given Header Entry - * \warning NOT end user intended method ! * @param entry entry to skip */ void Document::SkipDocEntry(DocEntry *entry) @@ -1578,19 +1665,18 @@ 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) { - int l = currentDocEntry->GetReadLength(); + long 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); + Fp->seekg((size_t)(currentDocEntry->GetOffset()), std::ios::beg); //FIXME :each DocEntry if (currentDocEntry->GetGroup() != 0xfffe) // for fffe pb { - Fp->seekg( (long)(currentDocEntry->GetReadLength()),std::ios::cur); + Fp->seekg( l,std::ios::cur); //FIXME :each DocEntry } } @@ -1599,17 +1685,21 @@ void Document::SkipToNextDocEntry(DocEntry *currentDocEntry) * the parser went Jabberwocky) one can hope improving things by * applying some heuristics. * @param entry entry to check - * @param foundLength first assumption about length + * @param foundLength first assumption about length (before bug fix, or set to zero if =0xffffffff) */ 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; + //foundLength = 0; + //entry->SetLength(foundLength); + entry->SetLength(0); + return; // return ASAP; don't waist time on useless tests } - + uint16_t gr = entry->GetGroup(); uint16_t elem = entry->GetElement(); @@ -1639,26 +1729,37 @@ void Document::FixDocEntryFoundLength(DocEntry *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 ( gr == 0x0009 && ( elem == 0x1113 || elem == 0x1114 ) ) - { - foundLength = 4; - entry->SetReadLength(4); // a bug is to be fixed ! - } - + else if ( gr == 0x0009 && ( elem == 0x1113 || elem == 0x1114 ) ) + { + // Ideally we should check we are in Explicit and double check + // that VR=UL... this is done properly in gdcm2 + if( foundLength == 6 ) + { + gdcmWarningMacro( "Replacing Length from 6 into 4" ); + foundLength = 4; + entry->SetReadLength(4); // a bug is to be fixed ! + } + else if ( foundLength%4 ) + { + gdcmErrorMacro( "This looks like to a buggy Siemens DICOM file." + "The length of this tag seems to be wrong" ); + } + } + else if ( entry->GetVR() == "SQ" ) { - foundLength = 0; // ReadLength is unchanged - } - - //////// We encountered a 'delimiter' element i.e. a tag of the form + foundLength = 0; // ReadLength is unchanged + } + + //////// We encountered a 'delimiter' element i.e. a tag of the form // "fffe|xxxx" which is just a marker. Delimiters length should not be // taken into account. else if ( gr == 0xfffe ) - { + { // According to the norm, fffe|0000 shouldn't exist. BUT the Philips // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to // causes extra troubles... - if ( entry->GetElement() != 0x0000 ) + if ( elem != 0x0000 ) { foundLength = 0; } @@ -1666,7 +1767,7 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, { foundLength=12; // to skip the mess that follows this bugged Tag ! } - } + } entry->SetLength(foundLength); } @@ -1701,7 +1802,7 @@ bool Document::IsDocEntryAnInteger(DocEntry *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). - long filePosition = Fp->tellg(); + long filePosition = Fp->tellg(); // Only when elem 0x0000 length is not 4 (?!?) gdcmWarningMacro( "Erroneous Group Length element length on : (" << std::hex << group << " , " << elem << ") -before- position x(" << filePosition << ")" @@ -1797,9 +1898,7 @@ bool Document::CheckSwap() // Position the file position indicator at first tag // (i.e. after the file preamble and the "DICM" string). - Fp->seekg(0, std::ios::beg); // FIXME : Is it usefull? - - Fp->seekg ( 132L, std::ios::beg); + Fp->seekg ( 132L, std::ios::beg); // Once per Document return true; } // ------------------------------- End of DicomV3 ---------------- @@ -1809,7 +1908,7 @@ bool Document::CheckSwap() gdcmWarningMacro( "Not a Kosher DICOM Version3 file (no preamble)"); - Fp->seekg(0, std::ios::beg); + Fp->seekg(0, std::ios::beg); // Once per ACR-NEMA Document // Let's check 'No Preamble Dicom File' : // Should start with group 0x0002 @@ -1837,7 +1936,8 @@ bool Document::CheckSwap() memcmp(entCur, "AE", (size_t)2) == 0 || memcmp(entCur, "OB", (size_t)2) == 0 ) { - Filetype = ExplicitVR; + Filetype = ExplicitVR; // FIXME : not enough to say it's Explicit + // Wait untill reading Transfer Syntax gdcmDebugMacro( "Group 0002 : Explicit Value Representation"); return true; } @@ -1928,7 +2028,7 @@ bool Document::CheckSwap() void Document::SwitchByteSwapCode() { gdcmDebugMacro( "Switching Byte Swap code from "<< SwapCode - << " at: 0x" << std::hex << Fp->tellg() ); + << " at: 0x" << std::hex << Fp->tellg() ); // Only when DEBUG if ( SwapCode == 1234 ) { SwapCode = 4321; @@ -1952,7 +2052,7 @@ void Document::SwitchByteSwapCode() * \brief during parsing, Header Elements too long are not loaded in memory * @param newSize new size */ -void Document::SetMaxSizeLoadEntry(long newSize) +void Document::SetMaxSizeLoadEntry(long newSize) { if ( newSize < 0 ) { @@ -1971,66 +2071,78 @@ void Document::SetMaxSizeLoadEntry(long newSize) * (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 ) { - // We reached the EOF (or an error occured) therefore + // We reached the EOF (or an error occured) therefore // header parsing has to be considered as finished. return 0; } - // 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); - + { + if ( !Group0002Parsed && CurrentGroup != 0x0002) // avoid calling a function when useless + HandleOutOfGroup0002(CurrentGroup, CurrentElem); + else + // Sometimes file contains groups of tags with reversed endianess. + HandleBrokenEndian(CurrentGroup, CurrentElem); + } + VRKey vr = FindDocEntryVR(); - VRKey realVR = vr; if ( vr == GDCM_VRUNKNOWN ) { - if ( elem == 0x0000 ) // Group Length + if ( CurrentElem == 0x0000 ) // Group Length { realVR = "UL"; // must be UL } - else if (group%2 == 1 && (elem >= 0x0010 && elem <=0x00ff )) + else if (CurrentGroup == 0xfffe) // Don't get DictEntry for Delimitors + { + realVR = "UL"; + } + + // Commented out in order not to generate 'Shadow Groups' where some + // Data Elements are Explicit VR and some other ones Implicit VR + // (Stupid MatLab DICOM Reader couldn't read gdcm-written images) + /* + 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); + DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem);//only when ImplicitVR if ( dictEntry ) { realVR = dictEntry->GetVR(); - dictEntry->Unregister(); + dictEntry->Unregister(); // GetDictEntry registered it } } } - // gdcmDebugMacro( "Found VR: " << vr << " / Real VR: " << realVR ); DocEntry *newEntry; - if ( Global::GetVR()->IsVROfSequence(realVR) ) - newEntry = NewSeqEntry(group, elem); - else + //if ( Global::GetVR()->IsVROfSequence(realVR) ) + if (realVR == "SQ") { - newEntry = NewDataEntry(group, elem, realVR); + newEntry = NewSeqEntry(CurrentGroup, CurrentElem); + } + else + { + newEntry = NewDataEntry(CurrentGroup, CurrentElem, realVR); static_cast(newEntry)->SetState(DataEntry::STATE_NOTLOADED); } @@ -2040,14 +2152,14 @@ DocEntry *Document::ReadNextDocEntry() { // We thought this was explicit VR, but we end up with an // implicit VR tag. Let's backtrack. - if ( newEntry->GetGroup() != 0xfffe ) + + //if ( newEntry->GetGroup() != 0xfffe ) + if (CurrentGroup != 0xfffe ) { - std::string msg; - 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() ); + int offset = Fp->tellg();//Only when heuristic for Explicit/Implicit was wrong + + gdcmWarningMacro("Entry (" << newEntry->GetKey() << ") at x(" + << offset << ") should be Explicit VR"); } } newEntry->SetImplicitVR(); @@ -2064,7 +2176,7 @@ DocEntry *Document::ReadNextDocEntry() return 0; } - newEntry->SetOffset(Fp->tellg()); + newEntry->SetOffset(Fp->tellg()); // for each DocEntry return newEntry; } @@ -2084,6 +2196,7 @@ void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem) if ((group == 0xfeff) && (elem == 0x00e0)) { // start endian swap mark for group found + gdcmDebugMacro( "Start endian swap mark found." ); reversedEndian++; SwitchByteSwapCode(); // fix the tag @@ -2093,6 +2206,7 @@ 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(); } @@ -2123,43 +2237,34 @@ void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) { // Endian reversion. // Some files contain groups of tags with reversed endianess. - if ( !Group0002Parsed && group != 0x0002) - { + Group0002Parsed = true; // we just came out of group 0002 // if Transfer Syntax is Big Endian we have to change CheckSwap std::string ts = GetTransferSyntax(); - if ( ts == GDCM_UNKNOWN ) - { - gdcmDebugMacro("True DICOM File, with NO Transfer Syntax (?!) " ); - return; - } - if ( !Global::GetTS()->IsTransferSyntax(ts) ) - { - gdcmWarningMacro("True DICOM File, with illegal Transfer Syntax: [" - << ts << "]"); - return; - } + TS::SpecialType s = Global::GetTS()->GetSpecialTransferSyntax(ts); // Group 0002 is always 'Explicit ...' // even when Transfer Syntax says 'Implicit ..." - if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == - TS::ImplicitVRLittleEndian ) + if ( s == TS::ImplicitVRLittleEndian + || + s == TS::ImplicitVRBigEndianPrivateGE + ) { Filetype = ImplicitVR; } // FIXME Strangely, this works with - //'Implicit VR BigEndian Transfer Syntax (GE Private) + //'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 ) + + if ( s == TS::ExplicitVRBigEndian ) { gdcmDebugMacro("Transfer Syntax Name = [" << GetTransferSyntaxName() << "]" ); @@ -2167,7 +2272,32 @@ void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) group = SwapShort(group); elem = SwapShort(elem); } - } + + /// \todo find a trick to warn user and stop processing + + if ( s == TS::DeflatedExplicitVRLittleEndian) + { + gdcmWarningMacro("Transfer Syntax [" + << GetTransferSyntaxName() << "] :" + << " not yet dealt with "); + return; + } + + // The following shouldn't occur very often + // Let's check at the very end. + + if ( ts == GDCM_UNKNOWN ) + { + gdcmDebugMacro("True DICOM File, with NO Transfer Syntax (?!) " ); + return; + } + + if ( !Global::GetTS()->IsTransferSyntax(ts) ) + { + gdcmWarningMacro("True DICOM File, with illegal Transfer Syntax: [" + << ts << "]"); + return; + } } //-----------------------------------------------------------------------------