X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=src%2FgdcmDocument.cxx;h=284ea27e28f463d729f456e532f8abfd7a98c562;hb=3f95c0f72b077bed808f8d70d3edff265d7403b5;hp=01300cb2562c512f389ccfdabb1d57d79aa53f3a;hpb=9ad2753ce67285c38abd453362eb59b1f38142d7;p=gdcm.git diff --git a/src/gdcmDocument.cxx b/src/gdcmDocument.cxx index 01300cb2..284ea27e 100644 --- a/src/gdcmDocument.cxx +++ b/src/gdcmDocument.cxx @@ -3,8 +3,8 @@ Program: gdcm Module: $RCSfile: gdcmDocument.cxx,v $ Language: C++ - Date: $Date: 2005/02/25 15:49:55 $ - Version: $Revision: 1.231 $ + Date: $Date: 2005/04/26 16:18:23 $ + Version: $Revision: 1.237 $ Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de l'Image). All rights reserved. See Doc/License.txt or @@ -40,25 +40,91 @@ namespace gdcm // Refer to Document::SetMaxSizeLoadEntry() const unsigned int Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096 -const unsigned int Document::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff; //----------------------------------------------------------------------------- // Constructor / Destructor // Constructors and destructors are protected to avoid user to invoke directly + /** - * \brief constructor - * @param filename 'Document' (File or DicomDir) to be opened for parsing + * \brief This default constructor neither loads nor parses the file. + * You should then invoke \ref Document::Load. + * */ -Document::Document( std::string const &filename ) +Document::Document() :ElementSet(-1) { - SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); - Filename = filename; + Fp = 0; + + SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); Initialize(); + SwapCode = 1234; + Filetype = ExplicitVR; + // Load will set it to true if sucessfull + Group0002Parsed = false; + IsDocumentAlreadyLoaded = false; + LoadMode = 0x00000000; // default : load everything, later +} + +/** + * \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 will set it to true if sucessfull + IsDocumentAlreadyLoaded = false; + + Load(filename); +} +/** + * \brief Canonical destructor. + */ +Document::~Document () +{ + RefPubDict = NULL; + RefShaDict = NULL; +} + +//----------------------------------------------------------------------------- +// Public + +/** + * \brief Loader + * @param filename 'Document' (File or DicomDir) to be opened for parsing + */ +void Document::Load( std::string const &filename ) +{ + // We should clean out anything that already exists. + // Check IsDocumentAlreadyLoaded to be sure. + if( IsDocumentAlreadyLoaded ) + { + gdcmWarningMacro( "A file was already parsed inside this " << + "gdcm::Document (previous name was: " + << Filename.c_str() << ". New name is :" + << filename ); + // todo : clean out the 'Document' + // We should call ClearEntry() on the parent object ?!? + } + + Filename = filename; Fp = 0; if ( !OpenFile() ) { + // warning already performed in OpenFile() + //gdcmWarningMacro( "Unable to open as an ACR/DICOM file: " + // << Filename.c_str() ); + Filetype = Unknown; return; } @@ -67,17 +133,36 @@ Document::Document( std::string const &filename ) gdcmWarningMacro( "Starting parsing of file: " << Filename.c_str()); Fp->seekg(0, std::ios::end); - long lgt = Fp->tellg(); - + long lgt = Fp->tellg(); // total length of the file + Fp->seekg(0, std::ios::beg); - CheckSwap(); - long beg = Fp->tellg(); - lgt -= beg; - + // CheckSwap returns a boolean + // (false if no swap info of any kind was found) + if (! CheckSwap() ) + { + gdcmWarningMacro( "Neither a DICOM V3 nor an ACR-NEMA file: " + << Filename.c_str()); + CloseFile(); + return ; + } + + long beg = Fp->tellg(); // just after DICOM preamble (if any) + + lgt -= beg; // remaining length to parse + ParseDES( this, beg, lgt, false); // Loading is done during parsing - Fp->seekg( 0, std::ios::beg); + if ( IsEmpty() ) + { + gdcmWarningMacro( "No tag in internal hash table for: " + << Filename.c_str()); + CloseFile(); + return ; + } + IsDocumentAlreadyLoaded = true; + + Fp->seekg( 0, std::ios::beg); // Load 'non string' values @@ -120,8 +205,8 @@ Document::Document( std::string const &filename ) // Specific code to allow gdcm to read ACR-LibIDO formated images // Note: ACR-LibIDO is an extension of the ACR standard that was // used at CREATIS. For the time being (say a couple years) - // we keep this kludge to allow a smooth move to gdcm for - // CREATIS developpers (sorry folks). + // we keep this kludge to allow CREATIS users + // reading their old images. // // if recognition code tells us we deal with a LibIDO image // we switch lineNumber and columnNumber @@ -141,33 +226,6 @@ Document::Document( std::string const &filename ) // --- End of ACR-LibIDO kludge --- } -/** - * \brief This default constructor doesn't parse the file. You should - * then invoke \ref Document::SetFileName and then the parsing. - */ -Document::Document() - :ElementSet(-1) -{ - Fp = 0; - - SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); - Initialize(); - SwapCode = 1234; - Filetype = ExplicitVR; - Group0002Parsed = false; -} - -/** - * \brief Canonical destructor. - */ -Document::~Document () -{ - RefPubDict = NULL; - RefShaDict = NULL; -} - -//----------------------------------------------------------------------------- -// Public /** * \brief Get the public dictionary used */ @@ -205,12 +263,11 @@ 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() { @@ -236,7 +293,7 @@ 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; @@ -406,10 +463,12 @@ std::ifstream *Document::OpenFile() Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary); if( ! *Fp ) { - gdcmDebugMacro( "Cannot open file: " << Filename.c_str()); + gdcmErrorMacro( "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; @@ -420,20 +479,20 @@ std::ifstream *Document::OpenFile() return 0; } - //ACR -- or DICOM with no Preamble; may start with a Shadow Group -- + //-- ACR or DICOM with no Preamble; may start with a Shadow Group -- if( zero == 0x0001 || zero == 0x0100 || zero == 0x0002 || zero == 0x0200 || zero == 0x0003 || zero == 0x0300 || zero == 0x0004 || zero == 0x0400 || zero == 0x0005 || zero == 0x0500 || zero == 0x0006 || zero == 0x0600 || zero == 0x0007 || zero == 0x0700 || zero == 0x0008 || zero == 0x0800 ) { - std::string msg - = Util::Format("ACR/DICOM with no preamble: (%04x)\n", zero); + std::string msg = Util::Format( + "ACR/DICOM starting at the begining of the file:(%04x)\n", zero); gdcmWarningMacro( msg.c_str() ); return Fp; } - //DICOM + //-- DICOM -- Fp->seekg(126L, std::ios::cur); char dicm[4] = {' ',' ',' ',' '}; Fp->read(dicm, (size_t)4); @@ -447,10 +506,11 @@ std::ifstream *Document::OpenFile() HasDCMPreamble = true; return Fp; } - + + // -- Neither ACR/No Preamble Dicom nor DICOMV3 file CloseFile(); - gdcmWarningMacro( "Not DICOM/ACR (missing preamble)" << Filename.c_str()); - + gdcmWarningMacro( "Neither ACR/No Preamble Dicom nor DICOMV3 file: " + << Filename.c_str()); return 0; } @@ -766,7 +826,10 @@ void Document::Initialize() /** * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries) - * @return length of the parsed set. + * @param set DocEntrySet we are going to parse ('zero level' or a SQItem) + * @param offset start of parsing + * @param l_max length to parse + * @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) @@ -839,6 +902,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, 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 ) ) @@ -856,10 +920,32 @@ void Document::ParseDES(DocEntrySet *set, long offset, LoadDocEntry( newValEntry ); bool delimitor=newValEntry->IsItemDelimitor(); + + if ( LoadMode & NO_SHADOW ) // User asked to skip, if possible, + // shadow groups ( if possible : + // whether element 0x0000 exits) + { + if ( newValEntry->GetGroup()%2 != 0 ) + { + if ( newValEntry->GetElement() == 0x0000 ) + { + std::string strLgrGroup = newValEntry->GetValue(); + int lgrGroup; + if ( strLgrGroup != GDCM_UNFOUND) + { + lgrGroup = atoi(strLgrGroup.c_str()); + Fp->seekg(lgrGroup , std::ios::cur); + used = false; + continue; + } + } + } + } + if( !set->AddEntry( newValEntry ) ) { // If here expect big troubles - //delete newValEntry; //otherwise mem leak + // delete newValEntry; //otherwise mem leak used=false; } @@ -883,7 +969,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, else { // 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 ) @@ -895,6 +981,14 @@ void Document::ParseDES(DocEntrySet *set, long offset, delim_mode = false; } } + + if ( (LoadMode & NO_SEQ) && ! delim_mode ) // User asked to skip SQ + { + Fp->seekg( l, std::ios::cur); + used = false; + continue; + } + // no other way to create it ... newSeqEntry->SetDelimitorMode( delim_mode ); @@ -902,9 +996,8 @@ void Document::ParseDES(DocEntrySet *set, long offset, // 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 (/*Document *dummy =*/ dynamic_cast< Document* > ( set ) ) + if ( dynamic_cast< Document* > ( set ) ) { - //(void)dummy; newSeqEntry->SetDepthLevel( 1 ); newSeqEntry->SetKey( newSeqEntry->GetKey() ); } @@ -928,6 +1021,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, { used = false; } + if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) { if( !used ) @@ -1053,8 +1147,10 @@ void Document::LoadDocEntry(DocEntry *entry) { //s << "gdcm::NotLoaded (BinEntry)"; s << GDCM_NOTLOADED; - s << " Address:" << (long)entry->GetOffset(); - s << " Length:" << entry->GetLength(); + s << " Ad.:" << (long)entry->GetOffset(); + s << " x(" << std::hex << entry->GetOffset() << ")"; + s << std::dec; + s << " Lgt:" << entry->GetLength(); s << " x(" << std::hex << entry->GetLength() << ")"; binEntryPtr->SetValue(s.str()); } @@ -1237,72 +1333,6 @@ void Document::FindDocEntryLength( DocEntry *entry ) // Length is encoded on 2 bytes. length16 = ReadInt16(); - - // FIXME : This heuristic supposes that the first group following - // group 0002 *has* and element 0000. - // BUT ... Element 0000 is optionnal :-( - - - // Fixed using : HandleOutOfGroup0002() - // (first hereafter strategy ...) - - // We can tell the current file is encoded in big endian (like - // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag - // and it's value is the one of the encoding of a big endian file. - // In order to deal with such big endian encoded files, we have - // (at least) two strategies: - // * when we load the "Transfer Syntax" tag with value of big endian - // encoding, we raise the proper flags. Then we wait for the end - // of the META group (0x0002) among which is "Transfer Syntax", - // before switching the swap code to big endian. We have to postpone - // the switching of the swap code since the META group is fully encoded - // in little endian, and big endian coding only starts at the next - // group. The corresponding code can be hard to analyse and adds - // many additional unnecessary tests for regular tags. - // * the second strategy consists in waiting for trouble, that shall - // appear when we find the first group with big endian encoding. This - // is easy to detect since the length of a "Group Length" tag (the - // ones with zero as element number) has to be of 4 (0x0004). When we - // encounter 1024 (0x0400) chances are the encoding changed and we - // found a group with big endian encoding. - //---> Unfortunately, element 0000 is optional. - //---> This will not work when missing! - // We shall use this second strategy. In order to make sure that we - // can interpret the presence of an apparently big endian encoded - // length of a "Group Length" without committing a big mistake, we - // add an additional check: we look in the already parsed elements - // for the presence of a "Transfer Syntax" whose value has to be "big - // endian encoding". When this is the case, chances are we have got our - // hands on a big endian encoded file: we switch the swap code to - // big endian and proceed... - -// if ( element == 0x0000 && length16 == 0x0400 ) -// { -// std::string ts = GetTransferSyntax(); -// if ( Global::GetTS()->GetSpecialTransferSyntax(ts) -// != TS::ExplicitVRBigEndian ) -// { -// throw FormatError( "Document::FindDocEntryLength()", -// " not explicit VR." ); -// return; -// } -// length16 = 4; -// SwitchByteSwapCode(); -// -// // Restore the unproperly loaded values i.e. the group, the element -// // and the dictionary entry depending on them. -// uint16_t correctGroup = SwapShort( entry->GetGroup() ); -// uint16_t correctElem = SwapShort( entry->GetElement() ); -// DictEntry *newTag = GetDictEntry( correctGroup, correctElem ); -// if ( !newTag ) -// { -// // This correct tag is not in the dictionary. Create a new one. -// newTag = NewVirtualDictEntry(correctGroup, correctElem); -// } -// // FIXME this can create a memory leaks on the old entry that be -// // left unreferenced. -// entry->SetDictEntry( newTag ); -// } // 0xffff means that we deal with 'No Length' Sequence // or 'No Length' SQItem @@ -1362,12 +1392,14 @@ uint32_t Document::FindDocEntryLengthOBOrOW() if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) ) { long filePosition = Fp->tellg(); - gdcmWarningMacro( "Neither an Item tag nor a Sequence delimiter tag on :" + gdcmWarningMacro( + "Neither an Item tag nor a Sequence delimiter tag on :" << std::hex << group << " , " << elem << ") -before- position x(" << filePosition << ")" ); Fp->seekg(positionOnEntry, std::ios::beg); - throw FormatUnexpected( "Neither an Item tag nor a Sequence delimiter tag."); + throw FormatUnexpected( + "Neither an Item tag nor a Sequence delimiter tag."); } if ( elem == 0xe0dd ) { @@ -1665,10 +1697,10 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, */ bool Document::IsDocEntryAnInteger(DocEntry *entry) { - uint16_t elem = entry->GetElement(); - uint16_t group = entry->GetGroup(); - const std::string &vr = entry->GetVR(); - uint32_t length = entry->GetLength(); + uint16_t elem = entry->GetElement(); + uint16_t group = entry->GetGroup(); + const std::string &vr = entry->GetVR(); + uint32_t length = entry->GetLength(); // When we have some semantics on the element we just read, and if we // a priori know we are dealing with an integer, then we shall be @@ -1791,16 +1823,18 @@ bool Document::CheckSwap() Fp->seekg(0, std::ios::beg); 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"); + Fp->seekg(0, std::ios::beg); // Our next best chance would be to be considering a 'clean' ACR/NEMA file. - // By clean we mean that the length of the first tag is written down. + // 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. @@ -1809,7 +1843,6 @@ bool Document::CheckSwap() // representation of a 32 bits integer. Hence the following dirty // trick : s32 = *((uint32_t *)(entCur)); - switch( s32 ) { case 0x00040000 : @@ -1920,26 +1953,6 @@ void Document::SetMaxSizeLoadEntry(long newSize) MaxSizeLoadEntry = newSize; } -/** - * \brief Header Elements too long will not be printed - * See comments of \ref Document::MAX_SIZE_PRINT_ELEMENT_VALUE - * @param newSize new size - */ -void Document::SetMaxSizePrintEntry(long newSize) -{ - if ( newSize < 0 ) - { - return; - } - if ((uint32_t)newSize >= (uint32_t)0xffffffff ) - { - MaxSizePrintEntry = 0xffffffff; - return; - } - MaxSizePrintEntry = newSize; -} - - /** * \brief Read the next tag but WITHOUT loading it's value * (read the 'Group Number', the 'Element Number', @@ -1976,9 +1989,14 @@ DocEntry *Document::ReadNextDocEntry() if( vr == GDCM_UNKNOWN) { - DictEntry *dictEntry = GetDictEntry(group,elem); - if( dictEntry ) - realVR = dictEntry->GetVR(); + if ( elem == 0x0000 ) // Group Length + realVR = "UL"; // must be UL + else + { + DictEntry *dictEntry = GetDictEntry(group,elem); + if( dictEntry ) + realVR = dictEntry->GetVR(); + } } DocEntry *newEntry; @@ -2013,7 +2031,6 @@ DocEntry *Document::ReadNextDocEntry() catch ( FormatError e ) { // Call it quits - //std::cout << e; delete newEntry; return 0; }