X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=src%2FgdcmDocument.cxx;h=284fe5e6fee561ab73a129a052ac542318acbb84;hb=fd5c14391627c6e42283ced90561271af2a161a2;hp=d2991e701c874f6862f9ab85fb4bbd6e267ddc96;hpb=17ebda2271dd65565f512e835546b22c22ebbbb0;p=gdcm.git diff --git a/src/gdcmDocument.cxx b/src/gdcmDocument.cxx index d2991e70..284fe5e6 100644 --- a/src/gdcmDocument.cxx +++ b/src/gdcmDocument.cxx @@ -3,8 +3,8 @@ Program: gdcm Module: $RCSfile: gdcmDocument.cxx,v $ Language: C++ - Date: $Date: 2006/04/11 16:03:26 $ - Version: $Revision: 1.346 $ + Date: $Date: 2011/03/29 07:36:00 $ + Version: $Revision: 1.386 $ Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de l'Image). All rights reserved. See Doc/License.txt or @@ -37,7 +37,7 @@ #include // for memset #endif -namespace gdcm +namespace GDCM_NAME_SPACE { //----------------------------------------------------------------------------- @@ -50,7 +50,7 @@ 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::Load. + * You should then invoke Document::Load. * */ Document::Document() @@ -62,12 +62,17 @@ Document::Document() Initialize(); SwapCode = 1234; Filetype = ExplicitVR; + CurrentOffsetPosition = 0; + OffsetOfPreviousParseDES =0; // Load will set it to true if sucessfull Group0002Parsed = false; IsDocumentAlreadyLoaded = false; IsDocumentModified = true; LoadMode = LD_ALL; // default : load everything, later + SetFileName(""); + changeFromUN=false; + UnexpectedEOF=false; } /** @@ -95,19 +100,22 @@ bool Document::Load( ) return DoTheLoadingDocumentJob( ); } -#ifndef GDCM_LEGACY_REMOVE + +//#ifndef GDCM_LEGACY_REMOVE /** * \brief Loader. (DEPRECATED : not to break the API) * @param fileName 'Document' (File or DicomDir) to be open for parsing * @return false if file cannot be open or no swap info was found, * or no tag was found. */ + /* bool Document::Load( std::string const &fileName ) { Filename = fileName; return DoTheLoadingDocumentJob( ); } -#endif +*/ +//#endif /** * \brief Performs the Loading Job (internal use only) @@ -154,6 +162,7 @@ bool Document::DoTheLoadingDocumentJob( ) // Recursive call. // Loading is done during parsing + OffsetOfPreviousParseDES = beg; ParseDES( this, beg, lgt, false); // delim_mode is first defaulted to false if ( IsEmpty() ) @@ -239,7 +248,7 @@ bool Document::DoTheLoadingDocumentJob( ) // Force Loading some more elements if user asked to. - gdcm::DocEntry *d; + GDCM_NAME_SPACE::DocEntry *d; for (ListElements::iterator it = UserForceLoadList.begin(); it != UserForceLoadList.end(); ++it) @@ -260,7 +269,7 @@ bool Document::DoTheLoadingDocumentJob( ) LoadDocEntry(d, true); } - CloseFile(); + CloseFile(); // ---------------------------- // Specific code to allow gdcm to read ACR-LibIDO formated images @@ -274,10 +283,13 @@ bool Document::DoTheLoadingDocumentJob( ) // std::string RecCode; 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" + + + if(RecCode.find("ACRNEMA_LIBIDO") == 0 || // any version + RecCode.find("CANRME_AILIBOD") == 0) // for brain-damaged softwares + // with "little-endian strings" { + Filetype = ACR_LIBIDO; std::string rows = GetEntryString(0x0028, 0x0010); std::string columns = GetEntryString(0x0028, 0x0011); @@ -431,7 +443,7 @@ std::string Document::GetTransferSyntax() // happen at different stages): try loading and proceed with check... // Well ... - // (parsing and loading happen at the very same stage!) + // (parsing and loading happen at the very same stage!) //LoadDocEntrySafe(entry); //JPRx if (DataEntry *dataEntry = dynamic_cast(entry) ) { @@ -521,11 +533,11 @@ uint32_t Document::SwapLong(uint32_t a) // save CPU time a=( ( a<<24) | ((a<<8) & 0x00ff0000) | ((a>>8) & 0x0000ff00) | (a>>24) ); - break; + break; case 3412 : // a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) ); a=( (a<<16) | (a>>16) ); - break; + break; case 2143 : a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) ); break; @@ -534,7 +546,7 @@ uint32_t Document::SwapLong(uint32_t a) a = 0; } return a; -} +} /** * \brief Swaps back the bytes of 8-byte long 'double' accordingly to @@ -575,14 +587,15 @@ double Document::SwapDouble(double a) // // -----------------File I/O --------------- /** - * \brief Tries to open the file \ref Document::Filename and - * checks the preamble when existing. - * @return The FILE pointer on success. + * \brief Tries to open the file Document::Filename and + * checks the preamble when existing, + * or if the file starts with an ACR-NEMA look-like element. + * @return The FILE pointer on success, 0 on failure. */ std::ifstream *Document::OpenFile() { HasDCMPreamble = false; - if (Filename.length() == 0) + if (Filename.length() == 0) { return 0; } @@ -597,13 +610,13 @@ std::ifstream *Document::OpenFile() if ( ! *Fp ) { // Don't user gdcmErrorMacro : - // a spurious message will appear when you use, for instance + // 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 + + // 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; @@ -611,7 +624,7 @@ std::ifstream *Document::OpenFile() //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() ) @@ -619,24 +632,7 @@ std::ifstream *Document::OpenFile() CloseFile(); return 0; } - - //-- 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 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 -- Fp->seekg(126L, std::ios::cur); // Once per Document char dicm[4]; // = {' ',' ',' ',' '}; @@ -646,24 +642,44 @@ std::ifstream *Document::OpenFile() CloseFile(); return 0; } + if ( memcmp(dicm, "DICM", 4) == 0 ) { HasDCMPreamble = true; return Fp; } + //-- 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 || + zero == 0x0028 || 0x2800 // worse : some ACR-NEMA like files + // start 00028 group ?!? + ) + { + 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; + } + // -- Neither ACR/No Preamble Dicom nor DICOMV3 file CloseFile(); // 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()); + << Filename.c_str()); return 0; } /** - * \brief closes the file - * @return TRUE if the close was successfull + * \brief closes the file + * @return TRUE if the close was successfull */ bool Document::CloseFile() { @@ -677,17 +693,17 @@ bool Document::CloseFile() } /** - * \brief Writes in a file all the 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 + * @param filetype Type of the File to be written * (ACR-NEMA, ExplicitVR, ImplicitVR) */ -void Document::WriteContent(std::ofstream *fp, FileType filetype) +void Document::WriteContent(std::ofstream *fp, FileType filetype, bool, bool) { // Skip if user wants to write an ACR-NEMA file if ( filetype == ImplicitVR || filetype == ExplicitVR || - filetype == JPEG ) + filetype == JPEG || filetype == JPEG2000 ) { // writing Dicom File Preamble char filePreamble[128]; @@ -710,8 +726,10 @@ void Document::WriteContent(std::ofstream *fp, FileType filetype) * --> 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 + // This one is recursive + // false : outside MetaElements + // false : outside Sequence + ElementSet::WriteContent(fp, filetype, false, false); } // ----------------------------------------- @@ -840,7 +858,7 @@ void Document::LoadEntryBinArea(DataEntry *entry) //} /** - * \brief Compares two documents, according to \ref DicomDir rules + * \brief Compares two documents, according to DicomDir rules * \warning Does NOT work with ACR-NEMA files * \todo Find a trick to solve the pb (use RET fields ?) * @param document to compare with current one @@ -906,6 +924,27 @@ bool Document::operator<(Document &document) //----------------------------------------------------------------------------- // Protected + +/** + * \brief Reads a given length of bytes + * (in order to avoid to many CPU time-consuming fread-s) + * @param l length to read + */ +void Document::ReadBegBuffer(size_t l) + throw( FormatError ) +{ + Fp->read (BegBuffer, (size_t)l); + if ( Fp->fail() ) + { + throw FormatError( "Document::ReadBegBuffer()", " file error." ); + } + if ( Fp->eof() ) + { + throw FormatError( "Document::ReadBegBuffer()", "EOF." ); + } + PtrBegBuffer = BegBuffer; + CurrentOffsetPosition+=l; +} /** * \brief Reads a supposed to be 16 Bits integer * (swaps it depending on processor endianness) @@ -928,6 +967,18 @@ uint16_t Document::ReadInt16() return g; } +/** + * \brief Gets from BegBuffer a supposed to be 16 Bits integer + * (swaps it depending on processor endianness) + * @return read value + */ +uint16_t Document::GetInt16() +{ + uint16_t g = *((uint16_t*)PtrBegBuffer); + g = SwapShort(g); + PtrBegBuffer+=2; + return g; +} /** * \brief Reads a supposed to be 32 Bits integer * (swaps it depending on processor endianness) @@ -950,10 +1001,23 @@ uint32_t Document::ReadInt32() return g; } +/** + * \brief Gets from BegBuffer a supposed to be 32 Bits integer + * (swaps it depending on processor endianness) + * @return read value + */ +uint32_t Document::GetInt32() +{ + uint32_t g = *((uint32_t*)PtrBegBuffer); + g = SwapLong(g); + PtrBegBuffer+=4; + return g; +} + /** * \brief Re-computes the length of the Dicom group 0002. */ -int Document::ComputeGroup0002Length( ) +int Document::ComputeGroup0002Length( ) { uint16_t gr; VRKey vr; @@ -976,10 +1040,10 @@ int Document::ComputeGroup0002Length( ) vr = entry->GetVR(); //if ( (vr == "OB")||(vr == "OW")||(vr == "UT")||(vr == "SQ")) - // (no SQ, OW, UT in group 0x0002;) + // (no SQ, OW, OL, UT in group 0x0002;) if ( vr == "OB" ) { - // explicit VR AND (OB, OW, SQ, UT) : 4 more bytes + // explicit VR AND (OB, OW, OL, SQ, UT, UN) : 4 more bytes groupLength += 4; } groupLength += 2 + 2 + 4 + entry->GetLength(); @@ -1046,66 +1110,66 @@ void Document::ParseDES(DocEntrySet *set, long offset, DocEntry *newDocEntry; DataEntry *newDataEntry; SeqEntry *newSeqEntry; - VRKey vr; + //VRKey vr; 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 << "0x(" << offset << ")" ); + << " at offset " << std::hex << "0x(" << offset << ")" ); while (true) { ///\todo FIXME : On 64 bits processors, tellg gives unexpected results after a while ? - /// Probabely a bug in gdcm code somwhere (some memory erased ?) + /// Probabely a bug in gdcm code somewhere (some memory erased ?) // Uncomment to track the bug -/* - if( Debug::GetDebugFlag() ) + + 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 - // when something *very* strange happens - if( Debug::GetDebugFlag() ) - std::cerr<GetKey()<<" "<GetVR()<GetKey()<<" "<GetVR()<IsItemStarter() ) { // Debug message within the method ! - newDocEntry = Backtrack(newDocEntry); + newDocEntry = Backtrack(newDocEntry, set); } else - { - PreviousDocEntry = newDocEntry; + { + PreviousDocEntry = newDocEntry; } - + used = true; newDataEntry = dynamic_cast(newDocEntry); - if ( newDataEntry ) + if ( newDataEntry ) { //////////////////////////// DataEntry - - vr = newDocEntry->GetVR(); + + //vr = newDocEntry->GetVR(); // useless ? if ( !set->AddEntry( newDataEntry ) ) { @@ -1141,7 +1205,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, } } - 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 @@ -1165,7 +1229,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, { /////////////////////// SeqEntry : VR = "SQ" - unsigned long l = newDocEntry->GetReadLength(); + unsigned long l = newDocEntry->GetReadLength(); if ( l != 0 ) // don't mess the delim_mode for 'zero-length sequence' { if ( l == 0xffffffff ) @@ -1226,11 +1290,11 @@ void Document::ParseDES(DocEntrySet *set, long offset, << " at offset 0x(" << std::hex << newDocEntry->GetOffset() << ")"); - ParseSQ( newSeqEntry, - newDocEntry->GetOffset(), - l, delim_mode_intern); + bool res = ParseSQ( newSeqEntry, + newDocEntry->GetOffset(), + l, delim_mode_intern); - gdcmDebugMacro( "Exit from ParseSQ, delim " << delim_mode_intern); + gdcmDebugMacro( "Exit from ParseSQ, delim " << delim_mode_intern << " -->return : " << res); } if ( !set->AddEntry( newSeqEntry ) ) { @@ -1249,7 +1313,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, 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(); @@ -1262,15 +1326,20 @@ void Document::ParseDES(DocEntrySet *set, long offset, newDocEntry->Delete(); } first = false; + + if (UnexpectedEOF) // some terminator was missing + { + break; + } } // end While gdcmDebugMacro( "Exit from ParseDES, delim-mode " << delim_mode ); } /** * \brief Parses a Sequence ( SeqEntry after SeqEntry) - * @return parsed length for this level + * @return false if expected fff0,e000 not found */ -void Document::ParseSQ( SeqEntry *seqEntry, +bool Document::ParseSQ( SeqEntry *seqEntry, long offset, long l_max, bool delim_mode) { int SQItemNumber = 0; @@ -1284,8 +1353,10 @@ void Document::ParseSQ( SeqEntry *seqEntry, if ( !newDocEntry ) { + // The most frequent is when a SQ terminator is missing (?!?) gdcmWarningMacro("in ParseSQ : should never get here!"); - break; + UnexpectedEOF = true; + return false; } if ( delim_mode ) { @@ -1316,9 +1387,23 @@ void Document::ParseSQ( SeqEntry *seqEntry, { dlm_mod = false; } - + + // avoid infinite loop when Bad assumption was made on illegal 'unknown length' UN //JPRx + + if (offsetStartCurrentSQItem <= OffsetOfPreviousParseDES) + { + gdcmWarningMacro("Bad assumption was made on illegal 'unknown length' UN!" << std::endl << + "OffsetOfPreviousParseDES " << std::hex << OffsetOfPreviousParseDES + << " offsetStartCurrentSQItem " << offsetStartCurrentSQItem); + /// \todo when "Bad assumption (SQ) on illegal 'unknown length' UN", Backtrack again + try OB + return false; + } + else + { + OffsetOfPreviousParseDES = offsetStartCurrentSQItem; + } + // 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(); // Once per SQItem @@ -1333,15 +1418,17 @@ void Document::ParseSQ( SeqEntry *seqEntry, break; } } + return true; } /** * \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 + * @param docEntry Item Starter that warned us + * @param set DocEntrySet (ElementSet/SQItem) the DocEntry will belong */ -DocEntry *Document::Backtrack(DocEntry *docEntry) +DocEntry *Document::Backtrack(DocEntry *docEntry, DocEntrySet *set) { // delete the Item Starter, built erroneously out of any Sequence // it's not yet in the HTable/chained list @@ -1356,7 +1443,8 @@ DocEntry *Document::Backtrack(DocEntry *docEntry) gdcmDebugMacro( "Backtrack :" << std::hex << group << "|" << elem << " at offset 0x(" <RemoveEntry( PreviousDocEntry ); // forge the Seq Entry DocEntry *newEntry = NewSeqEntry(group, elem); @@ -1366,7 +1454,7 @@ DocEntry *Document::Backtrack(DocEntry *docEntry) // Move back to the beginning of the Sequence Fp->seekg(offset, std::ios::beg); // Only for Shadow Implicit VR SQ - return newEntry; + return newEntry; // It will added where it has to be! } /** @@ -1377,10 +1465,10 @@ DocEntry *Document::Backtrack(DocEntry *docEntry) */ void Document::LoadDocEntry(DocEntry *entry, bool forceLoad) { - uint16_t group = entry->GetGroup(); - uint16_t elem = entry->GetElement(); + uint16_t group = entry->GetGroup(); + uint16_t elem = entry->GetElement(); const VRKey &vr = entry->GetVR(); - uint32_t length = entry->GetLength(); + uint32_t length = entry->GetLength(); // Fp->seekg((long)entry->GetOffset(), std::ios::beg); // JPRx @@ -1426,7 +1514,7 @@ void Document::LoadDocEntry(DocEntry *entry, bool forceLoad) 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 @@ -1446,20 +1534,30 @@ void Document::FindDocEntryLength( DocEntry *entry ) { const VRKey &vr = entry->GetVR(); uint16_t length16; - if ( Filetype == ExplicitVR && !entry->IsImplicitVR() ) { - if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UT" - || vr == "UN" ) + + // WARNING : + // + // For some images, length of UN elements is coded on 2 bytes (instead of 4) + // There are *not* readable ! + // You can make a quick and dirty patch, commenting out + //| vr == "UN" + // in the following line. + // (the 'straight' images will no longer be readable ...) + + if ( vr == "OB" || vr == "OW" || vr == "OL" || vr == "SQ" || vr == "UT" + || vr == "UN" || changeFromUN == true) { + changeFromUN = false; // 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); // Once per OW,OB,SQ DocEntry - uint32_t length32 = ReadInt32(); - - if ( (vr == "OB" || vr == "OW") && length32 == 0xffffffff ) + //Fp->seekg( 2L, std::ios::cur); // Once per OB,OW,OL,UT,UN,SQ DocEntry + uint32_t length32 = ReadInt32(); // Once per OB,OW,OL,UT,UN,SQ DocEntry + CurrentOffsetPosition+=4; + if ( (vr == "OB" || vr == "OW" || vr == "OL") && length32 == 0xffffffff ) { uint32_t lengthOB; try @@ -1493,10 +1591,9 @@ void Document::FindDocEntryLength( DocEntry *entry ) FixDocEntryFoundLength(entry, length32); return; } - // Length is encoded on 2 bytes. - length16 = ReadInt16(); - + //length16 = ReadInt16(); + length16 = GetInt16(); // 0xffff means that we deal with 'No Length' Sequence // or 'No Length' SQItem if ( length16 == 0xffff) @@ -1519,8 +1616,8 @@ void Document::FindDocEntryLength( DocEntry *entry ) // 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() ); + + FixDocEntryFoundLength( entry, GetInt32() /*ReadInt32()*/ ); return; } } @@ -1534,6 +1631,7 @@ 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(); // Only for OB,OW DataElements bool foundSequenceDelimiter = false; @@ -1543,24 +1641,30 @@ uint32_t Document::FindDocEntryLengthOBOrOW() { uint16_t group; uint16_t elem; + try - { - group = ReadInt16(); - elem = ReadInt16(); + { ///\todo make sure there is never OL encoded pixel data! + + //group = ReadInt16(); // Once per fragment (if any) of OB,OW DataElements + //elem = ReadInt16(); // Once per fragment (if any) of OB,OW DataElements + ReadBegBuffer(4); // Once per fragment (if any) of OB,OW DataElements } catch ( FormatError ) { throw FormatError("Unexpected end of file encountered during ", "Document::FindDocEntryLengthOBOrOW()"); } + group = GetInt16(); + elem = GetInt16(); + // We have to decount the group and element we just read totalLength += 4; if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) ) { gdcmWarningMacro( "Neither an Item tag nor a Sequence delimiter tag on :" - << std::hex << group << " , " << elem - << ")" ); + << std::hex << group << "|" << elem << ") Pos. on entry was 0x(" <seekg(positionOnEntry, std::ios::beg); // Once per fragment (if any) of OB,OW DataElements throw FormatUnexpected( @@ -1570,7 +1674,7 @@ uint32_t Document::FindDocEntryLengthOBOrOW() { foundSequenceDelimiter = true; } - uint32_t itemLength = ReadInt32(); + uint32_t itemLength = ReadInt32(); // Once per fragment (if any) of OB,OW DataElements // We add 4 bytes since we just read the ItemLength with ReadInt32 totalLength += itemLength + 4; SkipBytes(itemLength); @@ -1580,7 +1684,7 @@ uint32_t Document::FindDocEntryLengthOBOrOW() break; } } - Fp->seekg( positionOnEntry, std::ios::beg); // Only for OB,OW DataElements + Fp->seekg( positionOnEntry, std::ios::beg); // Only once for OB,OW DataElements return totalLength; } @@ -1598,11 +1702,11 @@ VRKey Document::FindDocEntryVR() // Delimiters (0xfffe), are not explicit VR ... if ( CurrentGroup == 0xfffe ) return GDCM_VRUNKNOWN; - - long positionOnEntry; - if( Debug::GetWarningFlag() ) - positionOnEntry = Fp->tellg(); // Only in Warning Mode - + + //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. @@ -1615,27 +1719,22 @@ VRKey Document::FindDocEntryVR() // if it happens not to be the case. VRKey vr; - Fp->read(&(vr[0]),(size_t)2); + //Fp->read(&(vr[0]),(size_t)2); + vr[0] = *PtrBegBuffer++; + vr[1] = *PtrBegBuffer++; - if ( !CheckDocEntryVR(vr) ) + //if ( !CheckDocEntryVR(vr) ) // avoid useless function call + if ( !Global::GetVR()->IsValidVR(vr) ) { -/* -// std::cout << "================================================================Unknown VR" - << std::hex << "0x(" - << (unsigned int)vr[0] << "|" << (unsigned int)vr[1] - << ")" << "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 - ); + + gdcmWarningMacro( "Unknown VR " << vr.GetHexaRepresentation() << std::hex + << " at offset : 0x(" << CurrentOffsetPosition-4 + << ") for group " << std::hex << CurrentGroup ); //Fp->seekg(positionOnEntry, std::ios::beg); //JPRx - Fp->seekg((long)-2, std::ios::cur);// only for unrecognized VR (?!?) + //Fp->seekg((long)-2, std::ios::cur);// only for unrecognized VR (?!?) //see :MR_Philips_Intera_PrivateSequenceExplicitVR.dcm + PtrBegBuffer-=2; return GDCM_VRUNKNOWN; } return vr; @@ -1670,7 +1769,7 @@ void Document::SkipDocEntry(DocEntry *entry) void Document::SkipToNextDocEntry(DocEntry *currentDocEntry) { long l = currentDocEntry->GetReadLength(); - if ( l == -1 ) // length = 0xffff shouldn't appear here ... + if ( (uint32_t) l == (uint32_t)-1 ) // length = 0xffff shouldn't appear here ... // ... but PMS imagers happen ! return; Fp->seekg((size_t)(currentDocEntry->GetOffset()), std::ios::beg); //FIXME :each DocEntry @@ -1699,16 +1798,18 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, entry->SetLength(0); return; // return ASAP; don't waist time on useless tests } - + uint16_t gr = entry->GetGroup(); uint16_t elem = entry->GetElement(); if ( foundLength % 2) { - gdcmWarningMacro( "Warning : Tag with uneven length " << foundLength - << " in x(" << std::hex << gr << "," << elem <<")"); + gdcmWarningMacro( "Warning : Tag (" << std::hex << gr << "|" << elem << ") with uneven length " + << std::dec << foundLength << " 0x(" << std::hex << foundLength << ") " + //<< " at offset x(" << offset << ")" + ); } - + //////// Fix for some naughty General Electric images. // Allthough not recent many such GE corrupted images are still present // on Creatis hard disks. Hence this fix shall remain when such images @@ -1730,7 +1831,7 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, // 'Leonardo' source. Hence, one might consider commenting out the // following fix on efficiency reasons. 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 ) @@ -1738,24 +1839,24 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, 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... @@ -1767,7 +1868,7 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, { foundLength=12; // to skip the mess that follows this bugged Tag ! } - } + } entry->SetLength(foundLength); } @@ -1786,7 +1887,7 @@ bool Document::IsDocEntryAnInteger(DocEntry *entry) // When we have some semantics on the element we just read, and if we // a priori know we are dealing with an integer, then we shall be - // able to swap it's element value properly. + // able to swap its element value properly. if ( elem == 0 ) // This is the group length of the group { if ( length == 4 ) @@ -1803,7 +1904,8 @@ bool Document::IsDocEntryAnInteger(DocEntry *entry) // encounter such an ill-formed image, we simply display a warning // message and proceed on parsing (while crossing fingers). long filePosition = Fp->tellg(); // Only when elem 0x0000 length is not 4 (?!?) - gdcmWarningMacro( "Erroneous Group Length element length on : (" + (void)filePosition; + gdcmWarningMacro( "Erroneous Group Length element length on : (" << std::hex << group << " , " << elem << ") -before- position x(" << filePosition << ")" << "lgt : " << length ); @@ -1861,7 +1963,7 @@ bool Document::CheckSwap() // i.e. a total of 136 bytes. entCur = deb + 136; - // group 0x0002 *is always* Explicit VR Sometimes , + // group 0x0002 *is always* Explicit VR Sometimes, // even if elem 0002,0010 (Transfer Syntax) tells us the file is // *Implicit* VR (see former 'gdcmData/icone.dcm') @@ -1884,6 +1986,8 @@ bool Document::CheckSwap() << "Looks like a bugged Header!"); } + // Here, we assume that the file IS kosher Dicom ! + // (The meta elements - group 0x0002 - ARE little endian !) if ( net2host ) { SwapCode = 4321; @@ -1899,6 +2003,7 @@ bool Document::CheckSwap() // (i.e. after the file preamble and the "DICM" string). Fp->seekg ( 132L, std::ios::beg); // Once per Document + CurrentOffsetPosition = 132; return true; } // ------------------------------- End of DicomV3 ---------------- @@ -1909,7 +2014,7 @@ bool Document::CheckSwap() gdcmWarningMacro( "Not a Kosher DICOM Version3 file (no preamble)"); Fp->seekg(0, std::ios::beg); // Once per ACR-NEMA Document - + CurrentOffsetPosition = 0; // Let's check 'No Preamble Dicom File' : // Should start with group 0x0002 // and be Explicit Value Representation @@ -1989,7 +2094,9 @@ bool Document::CheckSwap() // Find a trick to tell it the caller... s16 = *((uint16_t *)(deb)); - + + gdcmDebugMacro("not a DicomV3 nor a 'clean' ACR/NEMA;" + << " (->despaired wild guesses !)"); switch ( s16 ) { case 0x0001 : @@ -2000,8 +2107,11 @@ bool Document::CheckSwap() case 0x0006 : case 0x0007 : case 0x0008 : + case 0x0028 : SwapCode = 1234; - Filetype = ACR; + // Brute hack to allow reading DICOM RT files + //Filetype = ACR; // DICOM RT are *not* ACR-Nema files! + Filetype = ExplicitVR; return true; case 0x0100 : case 0x0200 : @@ -2011,13 +2121,53 @@ bool Document::CheckSwap() case 0x0600 : case 0x0700 : case 0x0800 : + case 0x2800 : SwapCode = 4321; Filetype = ACR; return true; default : - gdcmWarningMacro("ACR/NEMA unfound swap info (Hopeless !)"); - Filetype = Unknown; - return false; + + s16 = *((uint16_t *)(deb)); + if (s16 != 0x0000) + return false; + s16 = *((uint16_t *)(deb+2)); + + Fp->seekg ( 0L, std::ios::beg); // Once per Document + CurrentOffsetPosition = 0; + switch(s16) // try an other trick! + // -> to be able to decode 0029|1010 DataElement + // -> and be not less cleaver than dcmdump ;-) + { + case 0x0004 : + SwapCode = 1234; + break; + case 0x0400 : + SwapCode = 3412; + break; + default: + gdcmWarningMacro("ACR/NEMA unfound swap info (Hopeless !)"); + Filetype = Unknown; + return false; + } + // Check if next 2 bytes are a VR + // Probabely something more time-consuming exists with std::string + const char VRvalues[] = "AEASATCSDADTFLFDISLOLTPNSHSLSSSTTMUIULUSUTOBOWOLOFATUNSQRT"; + int nbVal = 29; + const char *pt = VRvalues; + for (int i=0;i= 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"; + // -> Better we fix the problem at Write time + + else if (CurrentGroup%2 == 1 ) + { + if (CurrentElem >= 0x0010 && CurrentElem <=0x00ff ) + { + // DICOM PS 3-5 7.8.1 a) states that : + // Private Creator Data Elements numbered (gggg,0010-00FF) (gggg is odd) + // attributes have to be LO (Long String) and the VM shall be equal to 1 + realVR = "LO"; + + // Seems not to be true + // Still in gdcmtk, David Clunnie disagrees, Marco Eichelberg says it's OK ... + // We let it for a while? + //(We should check length==4, for more security, but we don't have it yet !) + } + else if ( CurrentElem == 0x0001) + { + realVR = "UL"; // Private Group Length To End + } + else // check the private dictionary for shadow elements when Implicit VR! + { + DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem); + if ( dictEntry ) + { + realVR = dictEntry->GetVR(); + dictEntry->Unregister(); // GetDictEntry registered it + } + } } - */ + else { DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem);//only when ImplicitVR @@ -2134,6 +2310,28 @@ DocEntry *Document::ReadNextDocEntry() } } + // if UN found, let's check the dictionary, and trust it! + // (maybe a private dictionary exists?) + else if (vr == "UN") + { + DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem); + if ( dictEntry ) + { + realVR = dictEntry->GetVR(); + dictEntry->Unregister(); // GetDictEntry registered it + + // for VR = "UN", length is always stored on 4 bytes. + // remember this info, in order not to crash later + changeFromUN=true; + /// \todo : fixme If inside a vr = "UN" DataElement (but SQ according to a private dictionnary) + /// there is some more vr = "UN" DataElements, it will probabely fail. + /// --> find a -non time consuming- trick to store changeFromUN info at DataElement level, + /// not at the Document level. + /// --> ?!? JPR + + } + } + DocEntry *newEntry; //if ( Global::GetVR()->IsVROfSequence(realVR) ) if (realVR == "SQ") @@ -2152,14 +2350,14 @@ 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 (CurrentGroup != 0xfffe ) - { + if (CurrentGroup != 0xfffe) + { int offset = Fp->tellg();//Only when heuristic for Explicit/Implicit was wrong gdcmWarningMacro("Entry (" << newEntry->GetKey() << ") at x(" - << offset << ") should be Explicit VR"); + << std::hex << offset << ") should be Explicit VR"); } } newEntry->SetImplicitVR(); @@ -2177,7 +2375,6 @@ DocEntry *Document::ReadNextDocEntry() } newEntry->SetOffset(Fp->tellg()); // for each DocEntry - return newEntry; } @@ -2189,6 +2386,23 @@ DocEntry *Document::ReadNextDocEntry() */ void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem) { + // for strange PMS Gyroscan Intera images + // Item 'starter' has a tag : 0x3f3f,0x3f00, for no apparent reason + + // --- Feel free to remove this test *on your own coy of gdcm* + // if you are sure you'll never face this problem. + + if ((group == 0x3f3f) && (elem == 0x3f00)) + { + // start endian swap mark for group found + gdcmDebugMacro( " delimiter 0x3f3f found." ); + // fix the tag + group = 0xfffe; + elem = 0xe000; + return; + } + // --- End of removable code + // Endian reversion. // Some files contain groups of tags with reversed endianess. static int reversedEndian = 0; @@ -2250,7 +2464,7 @@ void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) if ( s == TS::ImplicitVRLittleEndian || - s == TS::ImplicitVRBigEndianPrivateGE + s == TS::ImplicitVRBigEndianPrivateGE ) { Filetype = ImplicitVR; @@ -2261,9 +2475,13 @@ void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) // // --> Probabely normal, since we considered we never have // to trust manufacturers. - // (we find very often 'Implicit VR' tag, + // (we often find 'Implicit VR' tag, // even when Transfer Syntax tells us it's Explicit ... - + + // NEVER trust the meta elements! + // (see what ezDICOM does ...) + + /* if ( s == TS::ExplicitVRBigEndian ) { gdcmDebugMacro("Transfer Syntax Name = [" @@ -2272,32 +2490,45 @@ void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) group = SwapShort(group); elem = SwapShort(elem); } + */ + //-- Broken ACR may start with a Shadow Group -- + // worse : some ACR-NEMA like files start 00028 group ?!? + if ( !( (group >= 0x0001 && group <= 0x0008) || group == 0x0028 ) ) + { + // We trust what we see. + SwitchByteSwapCode(); + group = SwapShort(group); + elem = SwapShort(elem); + // not what we where told (by meta elements) ! + gdcmDebugMacro("Transfer Syntax Name = [" + << GetTransferSyntaxName() << "]" ); + } /// \todo find a trick to warn user and stop processing - + if ( s == TS::DeflatedExplicitVRLittleEndian) { gdcmWarningMacro("Transfer Syntax [" << GetTransferSyntaxName() << "] :" << " not yet dealt with "); - return; + 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; + return; } - + if ( !Global::GetTS()->IsTransferSyntax(ts) ) { gdcmWarningMacro("True DICOM File, with illegal Transfer Syntax: [" << ts << "]"); return; - } + } } //-----------------------------------------------------------------------------