X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=src%2FgdcmDocument.cxx;h=9c12f0ce8a8aa4f97e5210cd1c9dfec6311580ac;hb=25c9af365d9221bd45f3fbe27275ff0f66a265b1;hp=9557f569a883d2aa7af52b9a8662ca2b6e260be7;hpb=e045317452417dcb7d58ac263a3a6b2df3de849d;p=gdcm.git diff --git a/src/gdcmDocument.cxx b/src/gdcmDocument.cxx index 9557f569..9c12f0ce 100644 --- a/src/gdcmDocument.cxx +++ b/src/gdcmDocument.cxx @@ -3,8 +3,8 @@ Program: gdcm Module: $RCSfile: gdcmDocument.cxx,v $ Language: C++ - Date: $Date: 2005/01/11 22:40:31 $ - Version: $Revision: 1.178 $ + Date: $Date: 2005/02/07 09:51:03 $ + Version: $Revision: 1.225 $ Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de l'Image). All rights reserved. See Doc/License.txt or @@ -24,26 +24,25 @@ #include "gdcmUtil.h" #include "gdcmDebug.h" #include "gdcmTS.h" -#include "gdcmException.h" #include "gdcmDictSet.h" -#include "gdcmRLEFramesInfo.h" -#include "gdcmJPEGFragmentsInfo.h" #include "gdcmDocEntrySet.h" #include "gdcmSQItem.h" #include #include +#include // For nthos: -#if defined(_MSC_VER) || defined(__BORLANDC__) +#if defined(_MSC_VER) || defined(__BORLANDC__) || defined(__MINGW32__) #include -#else +#endif + +#ifdef CMAKE_HAVE_NETINET_IN_H #include #endif namespace gdcm { - //----------------------------------------------------------------------------- // Refer to Document::CheckSwap() //const unsigned int Document::HEADER_LENGTH_TO_READ = 256; @@ -54,16 +53,17 @@ 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 file to be opened for parsing + * @param filename 'Document' (File or DicomDir) to be opened for parsing */ -Document::Document( std::string const &filename ) : ElementSet(-1) +Document::Document( std::string const &filename ) + :ElementSet(-1) { SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); Filename = filename; - Initialise(); + Initialize(); Fp = 0; if ( !OpenFile() ) @@ -73,38 +73,39 @@ Document::Document( std::string const &filename ) : ElementSet(-1) Group0002Parsed = false; - gdcmVerboseMacro( "Starting parsing of file: " << Filename.c_str()); + gdcmWarningMacro( "Starting parsing of file: " << Filename.c_str()); // Fp->seekg( 0, std::ios::beg); - Fp->seekg(0, std::ios::end); + Fp->seekg(0, std::ios::end); long lgt = Fp->tellg(); - Fp->seekg( 0, std::ios::beg); + Fp->seekg(0, std::ios::beg); + CheckSwap(); long beg = Fp->tellg(); lgt -= beg; - ParseDES( this, beg, lgt, false); // le Load sera fait a la volee + ParseDES( this, beg, lgt, false); // Loading is done during parsing Fp->seekg( 0, std::ios::beg); // Load 'non string' values - std::string PhotometricInterpretation = GetEntry(0x0028,0x0004); + std::string PhotometricInterpretation = GetEntryValue(0x0028,0x0004); if( PhotometricInterpretation == "PALETTE COLOR " ) { LoadEntryBinArea(0x0028,0x1200); // gray LUT /// FIXME FIXME FIXME /// The tags refered by the three following lines used to be CORRECTLY /// defined as having an US Value Representation in the public - /// dictionnary. BUT the semantics implied by the three following + /// dictionary. BUT the semantics implied by the three following /// lines state that the corresponding tag contents are in fact /// the ones of a BinEntry. - /// In order to fix things "Quick and Dirty" the dictionnary was - /// altered on PURPOUS but now contains a WRONG value. + /// 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 - /// wether the following tags are either: + /// 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 @@ -125,7 +126,7 @@ Document::Document( std::string const &filename ) : ElementSet(-1) 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) @@ -136,30 +137,31 @@ Document::Document( std::string const &filename ) : ElementSet(-1) // we switch lineNumber and columnNumber // std::string RecCode; - RecCode = GetEntry(0x0008, 0x0010); // recognition code + RecCode = GetEntryValue(0x0008, 0x0010); // recognition code (RET) if (RecCode == "ACRNEMA_LIBIDO_1.1" || RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares // with "little-endian strings" { Filetype = ACR_LIBIDO; - std::string rows = GetEntry(0x0028, 0x0010); - std::string columns = GetEntry(0x0028, 0x0011); - SetEntry(columns, 0x0028, 0x0010); - SetEntry(rows , 0x0028, 0x0011); + std::string rows = GetEntryValue(0x0028, 0x0010); + std::string columns = GetEntryValue(0x0028, 0x0011); + SetValEntry(columns, 0x0028, 0x0010); + SetValEntry(rows , 0x0028, 0x0011); } - // ----------------- End of ACR-LibIDO kludge ------------------ + // --- 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) +Document::Document() + :ElementSet(-1) { Fp = 0; SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE); - Initialise(); + Initialize(); SwapCode = 1234; Filetype = ExplicitVR; Group0002Parsed = false; @@ -172,32 +174,6 @@ Document::~Document () { RefPubDict = NULL; RefShaDict = NULL; - - delete RLEInfo; - delete JPEGInfo; -} - -//----------------------------------------------------------------------------- -// Print - -/** - * \brief Prints The Dict Entries of THE public Dicom Dictionary - * @return - */ -void Document::PrintPubDict(std::ostream &os) -{ - RefPubDict->SetPrintLevel(PrintLevel); - RefPubDict->Print(os); -} - -/** - * \brief Prints The Dict Entries of THE shadow Dicom Dictionary - * @return - */ -void Document::PrintShaDict(std::ostream &os) -{ - RefShaDict->SetPrintLevel(PrintLevel); - RefShaDict->Print(os); } //----------------------------------------------------------------------------- @@ -250,19 +226,59 @@ bool Document::IsReadable() { if( Filetype == Unknown) { - gdcmVerboseMacro( "Wrong filetype"); + gdcmWarningMacro( "Wrong filetype"); return false; } - if( TagHT.empty() ) - { - gdcmVerboseMacro( "No tags in internal hash table."); + if ( IsEmpty() ) + { + gdcmWarningMacro( "No tag in internal hash table."); return false; } return true; } +/** + * \brief Predicate for dicom version 3 file. + * @return True when the file is a dicom version 3. + */ +bool Document::IsDicomV3() +{ + // Checking if Transfer Syntax exists is enough + // Anyway, it's to late check if the 'Preamble' was found ... + // And ... would it be a rich idea to check ? + // (some 'no Preamble' DICOM images exist !) + return GetDocEntry(0x0002, 0x0010) != NULL; +} + +/** + * \brief Predicate for Papyrus file + * Dedicated to whomsoever it may concern + * @return True when the file is a Papyrus file. + */ +bool Document::IsPapyrus() +{ + // check for Papyrus private Sequence + DocEntry *e = GetDocEntry(0x0041, 0x1050); + if ( !e ) + return false; + // check if it's actually a Sequence + if ( !dynamic_cast(e) ) + return false; + return true; +} + +/** + * \brief returns the File Type + * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown) + * @return the FileType code + */ +FileType Document::GetFileType() +{ + return Filetype; +} + /** * \brief Accessor to the Transfer Syntax (when present) of the * current document (it internally handles reading the @@ -301,28 +317,77 @@ std::string Document::GetTransferSyntax() } /** - * \brief Predicate for dicom version 3 file. - * @return True when the file is a dicom version 3. + * \brief Accesses the info from 0002,0010 : Transfer Syntax and TS + * @return The full Transfer Syntax Name (as opposed to Transfer Syntax UID) */ -bool Document::IsDicomV3() +std::string Document::GetTransferSyntaxName() { - // Checking if Transfer Syntax exists is enough - // Anyway, it's to late check if the 'Preamble' was found ... - // And ... would it be a rich idea to check ? - // (some 'no Preamble' DICOM images exist !) - return GetDocEntry(0x0002, 0x0010) != NULL; -} + // use the TS (TS : Transfer Syntax) + std::string transferSyntax = GetEntryValue(0x0002,0x0010); + + if ( (transferSyntax.find(GDCM_NOTLOADED) < transferSyntax.length()) ) + { + gdcmErrorMacro( "Transfer Syntax not loaded. " << std::endl + << "Better you increase MAX_SIZE_LOAD_ELEMENT_VALUE" ); + return "Uncompressed ACR-NEMA"; + } + if ( transferSyntax == GDCM_UNFOUND ) + { + gdcmWarningMacro( "Unfound Transfer Syntax (0002,0010)"); + return "Uncompressed ACR-NEMA"; + } + + // we do it only when we need it + const TSKey &tsName = Global::GetTS()->GetValue( transferSyntax ); + // Global::GetTS() is a global static you shall never try to delete it! + return tsName; +} +// +// --------------- Swap Code ------------------ /** - * \brief returns the File Type - * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown) - * @return the FileType code + * \brief Swaps the bytes so they agree with the processor order + * @return The properly swaped 16 bits integer. */ -FileType Document::GetFileType() +uint16_t Document::SwapShort(uint16_t a) { - return Filetype; + if ( SwapCode == 4321 || SwapCode == 2143 ) + { + a = ((( a << 8 ) & 0xff00 ) | (( a >> 8 ) & 0x00ff ) ); + } + return a; } +/** + * \brief Swaps back the bytes of 4-byte long integer accordingly to + * processor order. + * @return The properly swaped 32 bits integer. + */ +uint32_t Document::SwapLong(uint32_t a) +{ + switch (SwapCode) + { + case 1234 : + break; + case 4321 : + a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) | + ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) ); + break; + case 3412 : + a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) ); + break; + case 2143 : + a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) ); + break; + default : + gdcmErrorMacro( "Unset swap code:" << SwapCode ); + a = 0; + } + return a; +} + +// +// -----------------File I/O --------------- /** * \brief Tries to open the file \ref Document::Filename and * checks the preamble when existing. @@ -330,7 +395,6 @@ FileType Document::GetFileType() */ std::ifstream *Document::OpenFile() { - HasDCMPreamble = false; if (Filename.length() == 0) { @@ -339,19 +403,20 @@ std::ifstream *Document::OpenFile() if(Fp) { - gdcmVerboseMacro( "File already open: " << Filename.c_str()); + gdcmWarningMacro( "File already open: " << Filename.c_str()); + CloseFile(); } Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary); if( ! *Fp ) { - gdcmVerboseMacro( "Cannot open file: " << Filename.c_str()); + gdcmDebugMacro( "Cannot open file: " << Filename.c_str()); delete Fp; Fp = 0; return 0; } - uint16_t zero; + uint16_t zero = 0; Fp->read((char*)&zero, (size_t)2); if( Fp->eof() ) { @@ -368,13 +433,13 @@ std::ifstream *Document::OpenFile() { std::string msg = Util::Format("ACR/DICOM with no preamble: (%04x)\n", zero); - gdcmVerboseMacro( msg.c_str() ); + gdcmWarningMacro( msg.c_str() ); return Fp; } //DICOM Fp->seekg(126L, std::ios::cur); - char dicm[4]; + char dicm[4] = {' ',' ',' ',' '}; Fp->read(dicm, (size_t)4); if( Fp->eof() ) { @@ -388,7 +453,7 @@ std::ifstream *Document::OpenFile() } CloseFile(); - gdcmVerboseMacro( "Not DICOM/ACR (missing preamble)" << Filename.c_str()); + gdcmWarningMacro( "Not DICOM/ACR (missing preamble)" << Filename.c_str()); return 0; } @@ -405,23 +470,23 @@ bool Document::CloseFile() delete Fp; Fp = 0; } - - return true; //FIXME how do we detect a non-close ifstream ? + return true; //FIXME how do we detect a non-closed ifstream ? } /** * \brief Writes in a file all the Header Entries (Dicom Elements) - * @param fp file pointer on an already open file + * @param fp file pointer on an already open file (actually: Output File Stream) * @param filetype Type of the File to be written * (ACR-NEMA, ExplicitVR, ImplicitVR) - * \return Always true. + * @return Always true. */ void 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 + // \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 if ( filetype == ImplicitVR || filetype == ExplicitVR ) { @@ -432,640 +497,278 @@ void Document::WriteContent(std::ofstream *fp, FileType filetype) fp->write("DICM", 4); } -/** - * \todo rewrite later, if really usefull - * - 'Group Length' element is optional in DICOM - * - but un-updated odd groups lengthes can causes pb - * (xmedcon breaker) - * - * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) ) - * UpdateGroupLength(false,filetype); - * if ( filetype == ACR) - * UpdateGroupLength(true,ACR); - */ - + /* + * \todo rewrite later, if really usefull + * - 'Group Length' element is optional in DICOM + * - but un-updated odd groups lengthes can causes pb + * (xmedcon breaker) + * + * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) ) + * UpdateGroupLength(false,filetype); + * if ( filetype == ACR) + * UpdateGroupLength(true,ACR); + */ + ElementSet::WriteContent(fp, filetype); // This one is recursive } +// ----------------------------------------- +// Content entries /** - * \brief Modifies the value of a given Doc Entry (Dicom Element) - * when it exists. Create it with the given value when unexistant. - * @param value (string) Value to be set - * @param group Group number of the Entry - * @param elem Element number of the Entry - * @param vr V(alue) R(epresentation) of the Entry -if private Entry- - * \return pointer to the modified/created Header Entry (NULL when creation - * failed). - */ -ValEntry *Document::ReplaceOrCreate(std::string const &value, - uint16_t group, - uint16_t elem, - TagName const &vr ) -{ - ValEntry *valEntry = 0; - DocEntry *currentEntry = GetDocEntry( group, elem); - - if (currentEntry) - { - valEntry = dynamic_cast< ValEntry* >(currentEntry); - - // Verify the VR - if( valEntry ) - if( valEntry->GetVR()!=vr ) - valEntry=NULL; - - // if currentEntry doesn't correspond to the requested valEntry - if( !valEntry) - { - if (!RemoveEntry(currentEntry)) - { - gdcmVerboseMacro( "Removal of previous DocEntry failed."); - - return NULL; - } - } - } - - // Create a new valEntry if necessary - if (!valEntry) - { - valEntry = NewValEntry(group, elem, vr); - - if ( !AddEntry(valEntry)) - { - gdcmVerboseMacro("AddEntry failed although this is a creation."); - - delete valEntry; - return NULL; - } - } - - // Set the binEntry value - SetEntry(value, valEntry); - return valEntry; -} - -/* - * \brief Modifies the value of a given Header Entry (Dicom Element) - * when it exists. Create it with the given value when unexistant. - * A copy of the binArea is made to be kept in the Document. - * @param binArea (binary) value to be set - * @param Group Group number of the Entry - * @param Elem Element number of the Entry - * @param vr V(alue) R(epresentation) of the Entry -if private Entry- - * \return pointer to the modified/created Header Entry (NULL when creation - * failed). + * \brief Loads (from disk) the element content + * when a string is not suitable + * @param group group number of the Entry + * @param elem element number of the Entry */ -BinEntry *Document::ReplaceOrCreate(uint8_t *binArea, - int lgth, - uint16_t group, - uint16_t elem, - TagName const &vr ) +void Document::LoadEntryBinArea(uint16_t group, uint16_t elem) { - BinEntry *binEntry = 0; - DocEntry *currentEntry = GetDocEntry( group, elem); - - // Verify the currentEntry - if (currentEntry) - { - binEntry = dynamic_cast< BinEntry* >(currentEntry); - - // Verify the VR - if( binEntry ) - if( binEntry->GetVR()!=vr ) - binEntry=NULL; - - // if currentEntry doesn't correspond to the requested valEntry - if( !binEntry) - { - if (!RemoveEntry(currentEntry)) - { - gdcmVerboseMacro( "Removal of previous DocEntry failed."); - - return NULL; - } - } - } - - // Create a new binEntry if necessary - if (!binEntry) - { - binEntry = NewBinEntry(group, elem, vr); - - if ( !AddEntry(binEntry)) - { - gdcmVerboseMacro( "AddEntry failed allthough this is a creation."); - - delete binEntry; - return NULL; - } - } + // Search the corresponding DocEntry + DocEntry *docElement = GetDocEntry(group, elem); + if ( !docElement ) + return; - // Set the binEntry value - uint8_t *tmpArea; - if (lgth>0 && binArea) - { - tmpArea = new uint8_t[lgth]; - memcpy(tmpArea,binArea,lgth); - } - else - { - tmpArea = 0; - } - if (!SetEntry(tmpArea,lgth,binEntry)) - { - if (tmpArea) - { - delete[] tmpArea; - } - } + BinEntry *binElement = dynamic_cast(docElement); + if( !binElement ) + return; - return binEntry; -} + LoadEntryBinArea(binElement); +} -/* - * \brief Modifies the value of a given Header Entry (Dicom Element) - * when it exists. Create it when unexistant. - * @param Group Group number of the Entry - * @param Elem Element number of the Entry - * \return pointer to the modified/created SeqEntry (NULL when creation - * failed). +/** + * \brief Loads (from disk) the element content + * when a string is not suitable + * @param elem Entry whose binArea is going to be loaded */ -SeqEntry *Document::ReplaceOrCreate( uint16_t group, uint16_t elem) +void Document::LoadEntryBinArea(BinEntry *elem) { - SeqEntry *seqEntry = 0; - DocEntry *currentEntry = GetDocEntry( group, elem); - - // Verify the currentEntry - if (currentEntry) - { - seqEntry = dynamic_cast< SeqEntry* >(currentEntry); - - // Verify the VR - if( seqEntry ) - if( seqEntry->GetVR()!="SQ" ) - seqEntry=NULL; + if(elem->GetBinArea()) + return; - // if currentEntry doesn't correspond to the requested valEntry - if( !seqEntry) - { - if (!RemoveEntry(currentEntry)) - { - gdcmVerboseMacro( "Removal of previous DocEntry failed."); + bool openFile = !Fp; + if(openFile) + OpenFile(); - return NULL; - } - } - } + size_t o =(size_t)elem->GetOffset(); + Fp->seekg(o, std::ios::beg); - // Create a new seqEntry if necessary - if (!seqEntry) + size_t l = elem->GetLength(); + uint8_t *a = new uint8_t[l]; + if( !a ) { - seqEntry = NewSeqEntry(group, elem); - - if ( !AddEntry(seqEntry)) - { - gdcmVerboseMacro( "AddEntry failed allthough this is a creation."); - - delete seqEntry; - return NULL; - } + gdcmWarningMacro( "Cannot allocate BinEntry content"); + return; } - return seqEntry; -} - -/** - * \brief Set a new value if the invoked element exists - * Seems to be useless !!! - * @param value new element value - * @param group group number of the Entry - * @param elem element number of the Entry - * \return boolean - */ -bool Document::ReplaceIfExist(std::string const &value, - uint16_t group, uint16_t elem ) -{ - SetEntry(value, group, elem); - - return true; -} - -//----------------------------------------------------------------------------- -// Protected - -/** - * \brief Checks if a given Dicom Element exists within the H table - * @param group Group number of the searched Dicom Element - * @param element Element number of the searched Dicom Element - * @return true is found - */ -bool Document::CheckIfEntryExist(uint16_t group, uint16_t element ) -{ - const std::string &key = DictEntry::TranslateToKey(group, element ); - return TagHT.count(key) != 0; -} - - -/** - * \brief Searches within Header Entries (Dicom Elements) parsed with - * the public and private dictionaries - * for the element value representation of a given tag. - * @param group Group number of the searched tag. - * @param element Element number of the searched tag. - * @return Corresponding element value representation when it exists, - * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. - */ -std::string Document::GetEntry(uint16_t group, uint16_t element) -{ - TagKey key = DictEntry::TranslateToKey(group, element); - if ( !TagHT.count(key)) + /// \todo check the result + Fp->read((char*)a, l); + if( Fp->fail() || Fp->eof()) { - return GDCM_UNFOUND; + delete[] a; + return; } - return ((ValEntry *)TagHT.find(key)->second)->GetValue(); -} + elem->SetBinArea(a); -/** - * \brief Searches within Header Entries (Dicom Elements) parsed with - * the public and private dictionaries - * for the element value representation of a given tag.. - * - * Obtaining the VR (Value Representation) might be needed by caller - * to convert the string typed content to caller's native type - * (think of C++ vs Python). The VR is actually of a higher level - * of semantics than just the native C++ type. - * @param group Group number of the searched tag. - * @param element Element number of the searched tag. - * @return Corresponding element value representation when it exists, - * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. - */ -std::string Document::GetEntryVR(uint16_t group, uint16_t element) -{ - DocEntry *elem = GetDocEntry(group, element); - if ( !elem ) - { - return GDCM_UNFOUND; - } - return elem->GetVR(); + if(openFile) + CloseFile(); } /** - * \brief Searches within Header Entries (Dicom Elements) parsed with - * the public and private dictionaries - * for the value length of a given tag.. - * @param group Group number of the searched tag. - * @param element Element number of the searched tag. - * @return Corresponding element length; -2 if not found + * \brief Loads the element while preserving the current + * underlying file position indicator as opposed to + * LoadDocEntry that modifies it. + * @param entry DocEntry whose value will be loaded. */ -int Document::GetEntryLength(uint16_t group, uint16_t element) +void Document::LoadDocEntrySafe(DocEntry *entry) { - DocEntry *elem = GetDocEntry(group, element); - if ( !elem ) + if(Fp) { - return -2; //magic number + long PositionOnEntry = Fp->tellg(); + LoadDocEntry(entry); + Fp->seekg(PositionOnEntry, std::ios::beg); } - return elem->GetLength(); } /** - * \brief Accesses an existing DocEntry (i.e. a Dicom Element) - * through it's (group, element) and modifies it's content with - * the given value. - * @param content new value (string) to substitute with - * @param group group number of the Dicom Element to modify - * @param element element number of the Dicom Element to modify + * \brief Compares two documents, according to \ref DicomDir rules + * \warning Does NOT work with ACR-NEMA files + * \todo Find a trick to solve the pb (use RET fields ?) + * @param document to compare with current one + * @return true if 'smaller' */ -bool Document::SetEntry(std::string const& content, - uint16_t group, uint16_t element) +bool Document::operator<(Document &document) { - ValEntry *entry = GetValEntry(group, element); - if (!entry ) + // Patient Name + std::string s1 = GetEntryValue(0x0010,0x0010); + std::string s2 = document.GetEntryValue(0x0010,0x0010); + if(s1 < s2) { - gdcmVerboseMacro( "No corresponding ValEntry (try promotion first)."); - return false; + return true; } - return SetEntry(content,entry); -} - -/** - * \brief Accesses an existing DocEntry (i.e. a Dicom Element) - * through it's (group, element) and modifies it's content with - * the given value. - * @param content new value (void* -> uint8_t*) to substitute with - * @param lgth new value length - * @param group group number of the Dicom Element to modify - * @param element element number of the Dicom Element to modify - */ -bool Document::SetEntry(uint8_t*content, int lgth, - uint16_t group, uint16_t element) -{ - BinEntry *entry = GetBinEntry(group, element); - if (!entry ) + else if( s1 > s2 ) { - gdcmVerboseMacro( "No corresponding ValEntry (try promotion first)."); return false; } - - return SetEntry(content,lgth,entry); -} - -/** - * \brief Accesses an existing DocEntry (i.e. a Dicom Element) - * and modifies it's content with the given value. - * @param content new value (string) to substitute with - * @param entry Entry to be modified - */ -bool Document::SetEntry(std::string const &content,ValEntry *entry) -{ - if(entry) + else { - entry->SetValue(content); - return true; - } - return false; -} - -/** - * \brief Accesses an existing BinEntry (i.e. a Dicom Element) - * and modifies it's content with the given value. - * @param content new value (void* -> uint8_t*) to substitute with - * @param entry Entry to be modified - * @param lgth new value length - */ -bool Document::SetEntry(uint8_t *content, int lgth, BinEntry *entry) -{ - if(entry) - { - // Hope Binary field length is *never* wrong - /*if(lgth%2) // Non even length are padded with a space (020H). - { - lgth++; - //content = content + '\0'; // fing a trick to enlarge a binary field? - }*/ - - entry->SetBinArea(content); - entry->SetLength(lgth); - entry->SetValue(GDCM_BINLOADED); - return true; + // Patient ID + s1 = GetEntryValue(0x0010,0x0020); + s2 = document.GetEntryValue(0x0010,0x0020); + if ( s1 < s2 ) + { + return true; + } + else if ( s1 > s2 ) + { + return false; + } + else + { + // Study Instance UID + s1 = GetEntryValue(0x0020,0x000d); + s2 = document.GetEntryValue(0x0020,0x000d); + if ( s1 < s2 ) + { + return true; + } + else if( s1 > s2 ) + { + return false; + } + else + { + // Serie Instance UID + s1 = GetEntryValue(0x0020,0x000e); + s2 = document.GetEntryValue(0x0020,0x000e); + if ( s1 < s2 ) + { + return true; + } + else if( s1 > s2 ) + { + return false; + } + } + } } return false; } +//----------------------------------------------------------------------------- +// Protected /** - * \brief Gets (from Header) a 'non string' element value - * (LoadElementValues has already be executed) - * @param group group number of the Entry - * @param elem element number of the Entry - * @return Pointer to the 'non string' area - */ -void *Document::GetEntryBinArea(uint16_t group, uint16_t elem) -{ - DocEntry *entry = GetDocEntry(group, elem); - if (!entry) - { - gdcmVerboseMacro( "No entry"); - return 0; - } - if ( BinEntry *binEntry = dynamic_cast(entry) ) - { - return binEntry->GetBinArea(); - } - - return 0; -} - -/** - * \brief Loads (from disk) the element content - * when a string is not suitable - * @param group group number of the Entry - * @param elem element number of the Entry - */ -void Document::LoadEntryBinArea(uint16_t group, uint16_t elem) -{ - // Search the corresponding DocEntry - DocEntry *docElement = GetDocEntry(group, elem); - if ( !docElement ) - return; - - BinEntry *binElement = dynamic_cast(docElement); - if( !binElement ) - return; - - LoadEntryBinArea(binElement); -} - -/** - * \brief Loads (from disk) the element content - * when a string is not suitable - * @param element Entry whose binArea is going to be loaded + * \brief Reads a supposed to be 16 Bits integer + * (swaps it depending on processor endianness) + * @return read value */ -void Document::LoadEntryBinArea(BinEntry *element) +uint16_t Document::ReadInt16() + throw( FormatError ) { - if(element->GetBinArea()) - return; - - bool openFile = !Fp; - if(openFile) - OpenFile(); - - size_t o =(size_t)element->GetOffset(); - Fp->seekg(o, std::ios::beg); - - size_t l = element->GetLength(); - uint8_t *a = new uint8_t[l]; - if( !a ) + uint16_t g; + Fp->read ((char*)&g, (size_t)2); + if ( Fp->fail() ) { - gdcmVerboseMacro( "Cannot allocate a"); - return; + throw FormatError( "Document::ReadInt16()", " file error." ); } - - /// \todo check the result - Fp->read((char*)a, l); - if( Fp->fail() || Fp->eof()) + if( Fp->eof() ) { - delete[] a; - return; + throw FormatError( "Document::ReadInt16()", "EOF." ); } - - element->SetBinArea(a); - - if(openFile) - CloseFile(); + g = SwapShort(g); + return g; } /** - * \brief Sets a 'non string' value to a given Dicom Element - * @param area area containing the 'non string' value - * @param group Group number of the searched Dicom Element - * @param element Element number of the searched Dicom Element - * @return + * \brief Reads a supposed to be 32 Bits integer + * (swaps it depending on processor endianness) + * @return read value */ -/*bool Document::SetEntryBinArea(uint8_t *area, - uint16_t group, uint16_t element) +uint32_t Document::ReadInt32() + throw( FormatError ) { - DocEntry *currentEntry = GetDocEntry(group, element); - if ( !currentEntry ) - { - return false; - } - - if ( BinEntry *binEntry = dynamic_cast(currentEntry) ) + uint32_t g; + Fp->read ((char*)&g, (size_t)4); + if ( Fp->fail() ) { - binEntry->SetBinArea( area ); - return true; + throw FormatError( "Document::ReadInt32()", " file error." ); } - - return false; -}*/ - -/** - * \brief retrieves a Dicom Element (the first one) using (group, element) - * \warning (group, element) IS NOT an identifier inside the Dicom Header - * if you think it's NOT UNIQUE, check the count number - * and use iterators to retrieve ALL the Dicoms Elements within - * a given couple (group, element) - * @param group Group number of the searched Dicom Element - * @param element Element number of the searched Dicom Element - * @return - */ -DocEntry *Document::GetDocEntry(uint16_t group, uint16_t element) -{ - TagKey key = DictEntry::TranslateToKey(group, element); - if ( !TagHT.count(key)) + if( Fp->eof() ) { - return NULL; + throw FormatError( "Document::ReadInt32()", "EOF." ); } - return TagHT.find(key)->second; + g = SwapLong(g); + return g; } /** - * \brief Same as \ref Document::GetDocEntry except it only - * returns a result when the corresponding entry is of type - * ValEntry. - * @return When present, the corresponding ValEntry. + * \brief skips bytes inside the source file + * \warning NOT end user intended method ! + * @return */ -ValEntry *Document::GetValEntry(uint16_t group, uint16_t element) +void Document::SkipBytes(uint32_t nBytes) { - DocEntry *currentEntry = GetDocEntry(group, element); - if ( !currentEntry ) - { - return 0; - } - if ( ValEntry *entry = dynamic_cast(currentEntry) ) - { - return entry; - } - gdcmVerboseMacro( "Unfound ValEntry."); - - return 0; + //FIXME don't dump the returned value + Fp->seekg((long)nBytes, std::ios::cur); } /** - * \brief Same as \ref Document::GetDocEntry except it only - * returns a result when the corresponding entry is of type - * BinEntry. - * @return When present, the corresponding BinEntry. + * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader + * @param filetype Type of the File to be written */ -BinEntry *Document::GetBinEntry(uint16_t group, uint16_t element) +int Document::ComputeGroup0002Length( FileType filetype ) { - DocEntry *currentEntry = GetDocEntry(group, element); - if ( !currentEntry ) - { - return 0; - } - if ( BinEntry *entry = dynamic_cast(currentEntry) ) + uint16_t gr; + std::string vr; + + int groupLength = 0; + bool found0002 = false; + + // for each zero-level Tag in the DCM Header + DocEntry *entry = GetFirstEntry(); + while( entry ) { - return entry; - } - gdcmVerboseMacro( "Unfound BinEntry."); - - return 0; -} + gr = entry->GetGroup(); -/** - * \brief Loads the element while preserving the current - * underlying file position indicator as opposed to - * to LoadDocEntry that modifies it. - * @param entry Header Entry whose value shall be loaded. - * @return - */ -void Document::LoadDocEntrySafe(DocEntry *entry) -{ - if(Fp) - { - long PositionOnEntry = Fp->tellg(); - LoadDocEntry(entry); - Fp->seekg(PositionOnEntry, std::ios::beg); - } -} + if( gr == 0x0002 ) + { + found0002 = true; -/** - * \brief Swaps back the bytes of 4-byte long integer accordingly to - * processor order. - * @return The properly swaped 32 bits integer. - */ -uint32_t Document::SwapLong(uint32_t a) -{ - switch (SwapCode) - { - case 1234 : - break; - case 4321 : - a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) | - ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) ); - break; - - case 3412 : - a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) ); - break; - - case 2143 : - a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) ); + if( entry->GetElement() != 0x0000 ) + { + vr = entry->GetVR(); + + if( filetype == ExplicitVR ) + { + if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") ) + { + // explicit VR AND OB, OW, SQ : 4 more bytes + groupLength += 4; + } + } + groupLength += 2 + 2 + 4 + entry->GetLength(); + } + } + else if (found0002 ) break; - default : - gdcmErrorMacro( "Unset swap code:" << SwapCode ); - a = 0; - } - return a; -} - -/** - * \brief Unswaps back the bytes of 4-byte long integer accordingly to - * processor order. - * @return The properly unswaped 32 bits integer. - */ -uint32_t Document::UnswapLong(uint32_t a) -{ - return SwapLong(a); -} -/** - * \brief Swaps the bytes so they agree with the processor order - * @return The properly swaped 16 bits integer. - */ -uint16_t Document::SwapShort(uint16_t a) -{ - if ( SwapCode == 4321 || SwapCode == 2143 ) - { - a = ((( a << 8 ) & 0x0ff00 ) | (( a >> 8 ) & 0x00ff ) ); + entry = GetNextEntry(); } - return a; + return groupLength; } +//----------------------------------------------------------------------------- +// Private /** - * \brief Unswaps the bytes so they agree with the processor order - * @return The properly unswaped 16 bits integer. + * \brief Loads all the needed Dictionaries + * \warning NOT end user intended method ! */ -uint16_t Document::UnswapShort(uint16_t a) +void Document::Initialize() { - return SwapShort(a); + RefPubDict = Global::GetDicts()->GetDefaultPubDict(); + RefShaDict = NULL; + Filetype = Unknown; } -//----------------------------------------------------------------------------- -// Private - /** * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries) * @return length of the parsed set. @@ -1078,7 +781,7 @@ void Document::ParseDES(DocEntrySet *set, long offset, BinEntry *newBinEntry; SeqEntry *newSeqEntry; VRKey vr; - bool used=false; + bool used = false; while (true) { @@ -1087,8 +790,9 @@ void Document::ParseDES(DocEntrySet *set, long offset, break; } - used=true; + used = true; newDocEntry = ReadNextDocEntry( ); + if ( !newDocEntry ) { break; @@ -1103,23 +807,25 @@ void Document::ParseDES(DocEntrySet *set, long offset, { if ( newBinEntry ) { - if ( ! Global::GetVR()->IsVROfBinaryRepresentable(vr) ) + if ( Filetype == ExplicitVR && + !Global::GetVR()->IsVROfBinaryRepresentable(vr) ) { ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR - gdcmVerboseMacro( "Neither Valentry, nor BinEntry." + 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 (Document *dummy = dynamic_cast< Document* > ( set ) ) + if ( dynamic_cast< Document* > ( set ) ) { - (void)dummy; newBinEntry->SetKey( newBinEntry->GetKey() ); } // but when "this" is a SQItem, we are inserting this new - // valEntry in a sequence item, and the kay has the + // valEntry in a sequence item, and the key has the // generalized form (refer to \ref BaseTagKey): if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) ) { @@ -1140,9 +846,8 @@ void Document::ParseDES(DocEntrySet *set, long offset, /////////////////////// 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 (Document *dummy = dynamic_cast< Document* > ( set ) ) + if ( dynamic_cast< Document* > ( set ) ) { - (void)dummy; newValEntry->SetKey( newValEntry->GetKey() ); } // ...but when "set" is a SQItem, we are inserting this new @@ -1177,26 +882,6 @@ void Document::ParseDES(DocEntrySet *set, long offset, } } - if ( ( newDocEntry->GetGroup() == 0x7fe0 ) - && ( newDocEntry->GetElement() == 0x0010 ) ) - { - std::string ts = GetTransferSyntax(); - if ( Global::GetTS()->IsRLELossless(ts) ) - { - long positionOnEntry = Fp->tellg(); - Fp->seekg( newDocEntry->GetOffset(), std::ios::beg ); - ComputeRLEInfo(); - Fp->seekg( positionOnEntry, std::ios::beg ); - } - else if ( Global::GetTS()->IsJPEG(ts) ) - { - long positionOnEntry = Fp->tellg(); - Fp->seekg( newDocEntry->GetOffset(), std::ios::beg ); - ComputeJPEGFragmentInfo(); - Fp->seekg( positionOnEntry, std::ios::beg ); - } - } - // Just to make sure we are at the beginning of next entry. SkipToNextDocEntry(newDocEntry); } @@ -1222,13 +907,13 @@ 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 (/*Document *dummy =*/ dynamic_cast< Document* > ( set ) ) { - (void)dummy; + //(void)dummy; newSeqEntry->SetDepthLevel( 1 ); newSeqEntry->SetKey( newSeqEntry->GetKey() ); } - // But when "set" is allready a SQItem, we are building a nested + // 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 ) ) @@ -1244,14 +929,19 @@ void Document::ParseDES(DocEntrySet *set, long offset, newDocEntry->GetOffset(), l, delim_mode); } - set->AddEntry( newSeqEntry ); + if( !set->AddEntry( newSeqEntry ) ) + { + used = false; + } if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) { + if( !used ) + delete newDocEntry; break; } } - if(!used) + if( !used ) delete newDocEntry; } } @@ -1265,10 +955,13 @@ void Document::ParseSQ( SeqEntry *seqEntry, { int SQItemNumber = 0; bool dlm_mod; + long offsetStartCurrentSQItem = offset; while (true) { - DocEntry *newDocEntry = ReadNextDocEntry(); + // the first time, we read the fff0,e000 of the first SQItem + DocEntry *newDocEntry = ReadNextDocEntry(); + if ( !newDocEntry ) { // FIXME Should warn user @@ -1278,7 +971,7 @@ void Document::ParseSQ( SeqEntry *seqEntry, { if ( newDocEntry->IsSequenceDelimitor() ) { - seqEntry->SetSequenceDelimitationItem( newDocEntry ); + seqEntry->SetDelimitationItem( newDocEntry ); break; } } @@ -1287,7 +980,7 @@ void Document::ParseSQ( SeqEntry *seqEntry, delete newDocEntry; break; } - + // create the current SQItem SQItem *itemSQ = new SQItem( seqEntry->GetDepthLevel() ); std::ostringstream newBase; newBase << seqEntry->GetKey() @@ -1305,11 +998,23 @@ void Document::ParseSQ( SeqEntry *seqEntry, { dlm_mod = false; } - - ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod); + // FIXME, TODO + // when we're here, element fffe,e000 is already passed. + // it's lost for the SQItem we're going to process !! + + //ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod); + //delete newDocEntry; // FIXME well ... it's too late to use it ! + + // Let's try :------------ + // remove fff0,e000, created out of the SQItem delete newDocEntry; - - seqEntry->AddEntry( itemSQ, SQItemNumber ); + Fp->seekg(offsetStartCurrentSQItem, std::ios::beg); + // fill up the current SQItem, starting at the beginning of fff0,e000 + ParseDES(itemSQ, offsetStartCurrentSQItem, l+8, dlm_mod); + offsetStartCurrentSQItem = Fp->tellg(); + // end try ----------------- + + seqEntry->AddSQItem( itemSQ, SQItemNumber ); SQItemNumber++; if ( !delim_mode && ((long)(Fp->tellg())-offset ) >= l_max ) { @@ -1319,9 +1024,9 @@ void Document::ParseSQ( SeqEntry *seqEntry, } /** - * \brief Loads the element content if its length doesn't exceed - * the value specified with Document::SetMaxSizeLoadEntry() - * @param entry Header Entry (Dicom Element) to be dealt with + * \brief Loads the element content if its length doesn't exceed + * the value specified with Document::SetMaxSizeLoadEntry() + * @param entry Header Entry (Dicom Element) to be dealt with */ void Document::LoadDocEntry(DocEntry *entry) { @@ -1454,7 +1159,7 @@ void Document::LoadDocEntry(DocEntry *entry) if( length % 2 ) { newValue = Util::DicomString(str, length+1); - gdcmVerboseMacro("Warning: bad length: " << length << + gdcmWarningMacro("Warning: bad length: " << length << ",For string :" << newValue.c_str()); // Since we change the length of string update it length //entry->SetReadLength(length+1); @@ -1469,7 +1174,7 @@ void Document::LoadDocEntry(DocEntry *entry) { if ( Fp->fail() || Fp->eof()) { - gdcmVerboseMacro("Unread element value"); + gdcmWarningMacro("Unread element value"); valEntry->SetValue(GDCM_UNREAD); return; } @@ -1490,7 +1195,6 @@ void Document::LoadDocEntry(DocEntry *entry) } } - /** * \brief Find the value Length of the passed Header Entry * @param entry Header Entry whose length of the value shall be loaded. @@ -1498,7 +1202,6 @@ void Document::LoadDocEntry(DocEntry *entry) void Document::FindDocEntryLength( DocEntry *entry ) throw ( FormatError ) { - uint16_t element = entry->GetElement(); std::string vr = entry->GetVR(); uint16_t length16; @@ -1546,6 +1249,14 @@ 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 @@ -1566,6 +1277,8 @@ void Document::FindDocEntryLength( DocEntry *entry ) // 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 @@ -1574,38 +1287,39 @@ void Document::FindDocEntryLength( DocEntry *entry ) // 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 ); - } - - // Heuristic: well, some files are really ill-formed. +// 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 if ( length16 == 0xffff) - { - // 0xffff means that we deal with 'Unknown Length' Sequence + { length16 = 0; } FixDocEntryFoundLength( entry, (uint32_t)length16 ); @@ -1619,12 +1333,72 @@ void Document::FindDocEntryLength( DocEntry *entry ) // on Data elements "Implicit and Explicit VR Data Elements shall // not coexist in a Data Set and Data Sets nested within it".] // Length is on 4 bytes. + + // Well ... group 0002 is always coded in 'Explicit VR Litle Endian' + // even if Transfer Syntax is 'Implicit VR ...' FixDocEntryFoundLength( entry, ReadInt32() ); return; } } +/** + * \brief Find the Length till the next sequence delimiter + * \warning NOT end user intended method ! + * @return + */ +uint32_t Document::FindDocEntryLengthOBOrOW() + throw( FormatUnexpected ) +{ + // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data. + long positionOnEntry = Fp->tellg(); + bool foundSequenceDelimiter = false; + uint32_t totalLength = 0; + + while ( !foundSequenceDelimiter ) + { + uint16_t group; + uint16_t elem; + try + { + group = ReadInt16(); + elem = ReadInt16(); + } + catch ( FormatError ) + { + throw FormatError("Unexpected end of file encountered during ", + "Document::FindDocEntryLengthOBOrOW()"); + } + // We have to decount the group and element we just read + totalLength += 4; + if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) ) + { + long filePosition = Fp->tellg(); + gdcmWarningMacro( "Neither an Item tag nor a Sequence delimiter tag on :" + << std::hex << group << " , " << elem + << ") -before- position x(" << filePosition << ")" ); + + Fp->seekg(positionOnEntry, std::ios::beg); + throw FormatUnexpected( "Neither an Item tag nor a Sequence delimiter tag."); + } + if ( elem == 0xe0dd ) + { + foundSequenceDelimiter = true; + } + uint32_t itemLength = ReadInt32(); + // We add 4 bytes since we just read the ItemLength with ReadInt32 + totalLength += itemLength + 4; + SkipBytes(itemLength); + + if ( foundSequenceDelimiter ) + { + break; + } + } + Fp->seekg( positionOnEntry, std::ios::beg); + return totalLength; +} + /** * \brief Find the Value Representation of the current Dicom Element. * @return Value Representation of the current Entry @@ -1665,9 +1439,6 @@ std::string Document::FindDocEntryVR() */ bool Document::CheckDocEntryVR(VRKey vr) { - // CLEANME searching the dicom_vr at each occurence is expensive. - // PostPone this test in an optional integrity check at the end - // of parsing or only in debug mode. if ( !Global::GetVR()->IsValidVR(vr) ) return false; @@ -1745,7 +1516,6 @@ std::string Document::GetDocEntryValue(DocEntry *entry) #endif //GDCM_NO_ANSI_STRING_STREAM return s.str(); } - return ((ValEntry *)entry)->GetValue(); } @@ -1816,14 +1586,15 @@ void Document::SkipDocEntry(DocEntry *entry) } /** - * \brief Skips to the begining of the next Header Entry + * \brief Skips to the beginning of the next Header Entry * \warning NOT end user intended method ! * @param currentDocEntry entry to skip */ void Document::SkipToNextDocEntry(DocEntry *currentDocEntry) { Fp->seekg((long)(currentDocEntry->GetOffset()), std::ios::beg); - Fp->seekg( (long)(currentDocEntry->GetReadLength()),std::ios::cur); + if (currentDocEntry->GetGroup() != 0xfffe) // for fffe pb + Fp->seekg( (long)(currentDocEntry->GetReadLength()),std::ios::cur); } /** @@ -1847,7 +1618,7 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, if ( foundLength % 2) { - gdcmVerboseMacro( "Warning : Tag with uneven length " << foundLength + gdcmWarningMacro( "Warning : Tag with uneven length " << foundLength << " in x(" << std::hex << gr << "," << elem <<")"); } @@ -1894,8 +1665,7 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, { foundLength = 0; } - } - + } entry->SetLength(foundLength); } @@ -1907,10 +1677,10 @@ void Document::FixDocEntryFoundLength(DocEntry *entry, */ bool Document::IsDocEntryAnInteger(DocEntry *entry) { - uint16_t elem = entry->GetElement(); - uint16_t group = entry->GetGroup(); + uint16_t elem = entry->GetElement(); + uint16_t group = entry->GetGroup(); const std::string &vr = entry->GetVR(); - uint32_t length = entry->GetLength(); + 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 @@ -1931,7 +1701,7 @@ 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(); - gdcmVerboseMacro( "Erroneous Group Length element length on : (" + gdcmWarningMacro( "Erroneous Group Length element length on : (" << std::hex << group << " , " << elem << ") -before- position x(" << filePosition << ")" << "lgt : " << length ); @@ -1945,136 +1715,6 @@ bool Document::IsDocEntryAnInteger(DocEntry *entry) return false; } -/** - * \brief Find the Length till the next sequence delimiter - * \warning NOT end user intended method ! - * @return - */ - -uint32_t Document::FindDocEntryLengthOBOrOW() - throw( FormatUnexpected ) -{ - // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data. - long positionOnEntry = Fp->tellg(); - bool foundSequenceDelimiter = false; - uint32_t totalLength = 0; - - while ( !foundSequenceDelimiter ) - { - uint16_t group; - uint16_t elem; - try - { - group = ReadInt16(); - elem = ReadInt16(); - } - catch ( FormatError ) - { - throw FormatError("Unexpected end of file encountered during ", - "Document::FindDocEntryLengthOBOrOW()"); - } - - // We have to decount the group and element we just read - totalLength += 4; - - if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) ) - { - long filePosition = Fp->tellg(); - gdcmVerboseMacro( "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."); - } - - if ( elem == 0xe0dd ) - { - foundSequenceDelimiter = true; - } - - uint32_t itemLength = ReadInt32(); - // We add 4 bytes since we just read the ItemLength with ReadInt32 - totalLength += itemLength + 4; - SkipBytes(itemLength); - - if ( foundSequenceDelimiter ) - { - break; - } - } - Fp->seekg( positionOnEntry, std::ios::beg); - return totalLength; -} - -/** - * \brief Reads a supposed to be 16 Bits integer - * (swaps it depending on processor endianity) - * @return read value - */ -uint16_t Document::ReadInt16() - throw( FormatError ) -{ - uint16_t g; - Fp->read ((char*)&g, (size_t)2); - if ( Fp->fail() ) - { - throw FormatError( "Document::ReadInt16()", " file error." ); - } - if( Fp->eof() ) - { - throw FormatError( "Document::ReadInt16()", "EOF." ); - } - g = SwapShort(g); - return g; -} - -/** - * \brief Reads a supposed to be 32 Bits integer - * (swaps it depending on processor endianity) - * @return read value - */ -uint32_t Document::ReadInt32() - throw( FormatError ) -{ - uint32_t g; - Fp->read ((char*)&g, (size_t)4); - if ( Fp->fail() ) - { - throw FormatError( "Document::ReadInt32()", " file error." ); - } - if( Fp->eof() ) - { - throw FormatError( "Document::ReadInt32()", "EOF." ); - } - g = SwapLong(g); - 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 Loads all the needed Dictionaries - * \warning NOT end user intended method ! - */ -void Document::Initialise() -{ - RefPubDict = Global::GetDicts()->GetDefaultPubDict(); - RefShaDict = NULL; - RLEInfo = new RLEFramesInfo; - JPEGInfo = new JPEGFragmentsInfo; - Filetype = Unknown; -} - /** * \brief Discover what the swap code is (among little endian, big endian, * bad little endian, bad big endian). @@ -2108,25 +1748,24 @@ bool Document::CheckSwap() net2host = false; } - // The easiest case is the one of a DICOM header, since it possesses a - // file preamble where it suffice to look for the string "DICM". + // The easiest case is the one of a 'true' DICOM header, we just have + // to look for the string "DICM" inside the file preamble. Fp->read(deb, 256); char *entCur = deb + 128; if( memcmp(entCur, "DICM", (size_t)4) == 0 ) { - gdcmVerboseMacro( "Looks like DICOM Version3" ); + gdcmWarningMacro( "Looks like DICOM Version3 (preamble + DCM)" ); - // Next, determine the value representation (VR). Let's skip to the - // first element (0002, 0000) and check there if we find "UL" - // - or "OB" if the 1st one is (0002,0001) -, + // Group 0002 should always be VR, and the first element 0000 + // Let's be carefull (so many wrong headers ...) + // and determine the value representation (VR) : + // Let's skip to the first element (0002,0000) and check there if we find + // "UL" - or "OB" if the 1st one is (0002,0001) -, // in which case we (almost) know it is explicit VR. // WARNING: if it happens to be implicit VR then what we will read // is the length of the group. If this ascii representation of this // length happens to be "UL" then we shall believe it is explicit VR. - // FIXME: in order to fix the above warning, we could read the next - // element value (or a couple of elements values) in order to make - // sure we are not commiting a big mistake. // We need to skip : // * the 128 bytes of File Preamble (often padded with zeroes), // * the 4 bytes of "DICM" string, @@ -2134,42 +1773,42 @@ bool Document::CheckSwap() // i.e. a total of 136 bytes. entCur = deb + 136; - // FIXME : FIXME: - // Sometimes (see : gdcmData/icone.dcm) group 0x0002 *is* Explicit VR, - // but elem 0002,0010 (Transfer Syntax) tells us the file is - // *Implicit* VR. -and it is !- + // group 0x0002 *is always* Explicit VR Sometimes , + // even if elem 0002,0010 (Transfer Syntax) tells us the file is + // *Implicit* VR (see former 'gdcmData/icone.dcm') if( memcmp(entCur, "UL", (size_t)2) == 0 || memcmp(entCur, "OB", (size_t)2) == 0 || memcmp(entCur, "UI", (size_t)2) == 0 || memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later - // when Write DCM *adds* + // 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; - gdcmVerboseMacro( "Explicit Value Representation"); + gdcmWarningMacro( "Group 0002 : Explicit Value Representation"); } else { Filetype = ImplicitVR; - gdcmVerboseMacro( "Not an explicit Value Representation"); + gdcmWarningMacro( "Group 0002 :Not an explicit Value Representation;" + << "Looks like a bugged Header!"); } if ( net2host ) { SwapCode = 4321; - gdcmVerboseMacro( "HostByteOrder != NetworkByteOrder"); + gdcmWarningMacro( "HostByteOrder != NetworkByteOrder"); } else { SwapCode = 1234; - gdcmVerboseMacro( "HostByteOrder = NetworkByteOrder"); + gdcmWarningMacro( "HostByteOrder = NetworkByteOrder"); } - // Position the file position indicator at first tag (i.e. - // after the file preamble and the "DICM" string). + // Position the file position indicator at first tag + // (i.e. after the file preamble and the "DICM" string). Fp->seekg(0, std::ios::beg); Fp->seekg ( 132L, std::ios::beg); return true; @@ -2178,7 +1817,7 @@ bool Document::CheckSwap() // 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). - gdcmVerboseMacro( "Not a DICOM Version3 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. @@ -2253,27 +1892,19 @@ bool Document::CheckSwap() Filetype = ACR; return true; default : - gdcmVerboseMacro( "ACR/NEMA unfound swap info (Really hopeless !)"); + gdcmWarningMacro( "ACR/NEMA unfound swap info (Really hopeless !)"); Filetype = Unknown; return false; } - // Then the only info we have is the net2host one. - //if (! net2host ) - // SwapCode = 1234; - //else - // SwapCode = 4321; - //return; } } - - /** * \brief Change the Byte Swap code. */ void Document::SwitchByteSwapCode() { - gdcmVerboseMacro( "Switching Byte Swap code."); + gdcmWarningMacro( "Switching Byte Swap code from "<< SwapCode); if ( SwapCode == 1234 ) { SwapCode = 4321; @@ -2310,7 +1941,6 @@ void Document::SetMaxSizeLoadEntry(long newSize) MaxSizeLoadEntry = newSize; } - /** * \brief Header Elements too long will not be printed * \todo See comments of \ref Document::MAX_SIZE_PRINT_ELEMENT_VALUE @@ -2318,7 +1948,6 @@ void Document::SetMaxSizeLoadEntry(long newSize) */ void Document::SetMaxSizePrintEntry(long newSize) { - //DOH !! This is exactly SetMaxSizeLoadEntry FIXME FIXME if ( newSize < 0 ) { return; @@ -2332,68 +1961,6 @@ void Document::SetMaxSizePrintEntry(long newSize) } - -/** - * \brief Handle broken private tag from Philips NTSCAN - * where the endianess is being switch 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. - 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 - reversedEndian++; - SwitchByteSwapCode(); - // fix the tag - group = 0xfffe; - elem = 0xe000; - } - else if (group == 0xfffe && elem == 0xe00d && reversedEndian) - { - // end of reversed endian group - reversedEndian--; - SwitchByteSwapCode(); - } -} - -/** - * \brief Group 0002 is always coded Little Endian - * whatever Transfer Syntax is - * @return no return - */ -void Document::HandleOutOfGroup0002(uint16_t group) -{ - // 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 - - TagKey key = DictEntry::TranslateToKey(0x0002, 0x0010); - if ( !TagHT.count(key)) - { - gdcmVerboseMacro("True DICOM File, with NO Tansfer Syntax ?!?"); - return; - } - - // FIXME Strangely, this works with - //'Implicit VR Transfer Syntax (GE Private) - - if ( ((ValEntry *)TagHT.find(key)->second)->GetValue() - == "Explicit VR - Big Endian" ) - { - gdcmVerboseMacro("Tansfer Syntax = Explicit VR - Big Endian"); - SwitchByteSwapCode(); - } - } -} - /** * \brief Read the next tag but WITHOUT loading it's value * (read the 'Group Number', the 'Element Number', @@ -2415,16 +1982,15 @@ DocEntry *Document::ReadNextDocEntry() { // We reached the EOF (or an error occured) therefore // header parsing has to be considered as finished. - //std::cout << e; return 0; } // Sometimes file contains groups of tags with reversed endianess. HandleBrokenEndian(group, elem); -// In 'true DICOM' files Group 0002 is allways little endian + // In 'true DICOM' files Group 0002 is always little endian if ( HasDCMPreamble ) - HandleOutOfGroup0002(group); + HandleOutOfGroup0002(group, elem); std::string vr = FindDocEntryVR(); std::string realVR = vr; @@ -2449,11 +2015,14 @@ DocEntry *Document::ReadNextDocEntry() if( Filetype == ExplicitVR ) { // We thought this was explicit VR, but we end up with an - // implicit VR tag. Let's backtrack. - std::string msg; - msg = Util::Format("Falsely explicit vr file (%04x,%04x)\n", - newEntry->GetGroup(), newEntry->GetElement()); - gdcmVerboseMacro( msg.c_str() ); + // implicit VR tag. Let's backtrack. + if ( newEntry->GetGroup() != 0xfffe ) + { + std::string msg; + msg = Util::Format("Entry (%04x,%04x) should be Explicit VR\n", + newEntry->GetGroup(), newEntry->GetElement()); + gdcmWarningMacro( msg.c_str() ); + } } newEntry->SetImplicitVR(); } @@ -2475,484 +2044,77 @@ DocEntry *Document::ReadNextDocEntry() return newEntry; } - -/** - * \brief Generate a free TagKey i.e. a TagKey that is not present - * in the TagHt dictionary. - * @param group The generated tag must belong to this group. - * @return The element of tag with given group which is fee. - */ -uint32_t Document::GenerateFreeTagKeyInGroup(uint16_t group) -{ - for (uint32_t elem = 0; elem < UINT32_MAX; elem++) - { - TagKey key = DictEntry::TranslateToKey(group, elem); - if (TagHT.count(key) == 0) - { - return elem; - } - } - return UINT32_MAX; -} - -/** - * \brief Assuming the internal file pointer \ref Document::Fp - * is placed at the beginning of a tag check whether this - * tag is (TestGroup, TestElement). - * \warning On success the internal file pointer \ref Document::Fp - * is modified to point after the tag. - * On failure (i.e. when the tag wasn't the expected tag - * (TestGroup, TestElement) the internal file pointer - * \ref Document::Fp is restored to it's original position. - * @param testGroup The expected group of the tag. - * @param testElement The expected Element of the tag. - * @return True on success, false otherwise. - */ -bool Document::ReadTag(uint16_t testGroup, uint16_t testElement) -{ - long positionOnEntry = Fp->tellg(); - long currentPosition = Fp->tellg(); // On debugging purposes - - //// Read the Item Tag group and element, and make - // sure they are what we expected: - uint16_t itemTagGroup; - uint16_t itemTagElement; - try - { - itemTagGroup = ReadInt16(); - itemTagElement = ReadInt16(); - } - catch ( FormatError e ) - { - //std::cerr << e << std::endl; - return false; - } - if ( itemTagGroup != testGroup || itemTagElement != testElement ) - { - gdcmVerboseMacro( "Wrong Item Tag found:" - << " We should have found tag (" - << std::hex << testGroup << "," << testElement << ")" << std::endl - << " but instead we encountered tag (" - << std::hex << itemTagGroup << "," << itemTagElement << ")" - << " at address: " << " 0x(" << (unsigned int)currentPosition << ")" - ) ; - Fp->seekg(positionOnEntry, std::ios::beg); - - return false; - } - return true; -} - /** - * \brief Assuming the internal file pointer \ref Document::Fp - * is placed at the beginning of a tag (TestGroup, TestElement), - * read the length associated to the Tag. - * \warning On success the internal file pointer \ref Document::Fp - * is modified to point after the tag and it's length. - * On failure (i.e. when the tag wasn't the expected tag - * (TestGroup, TestElement) the internal file pointer - * \ref Document::Fp is restored to it's original position. - * @param testGroup The expected group of the tag. - * @param testElement The expected Element of the tag. - * @return On success returns the length associated to the tag. On failure - * returns 0. - */ -uint32_t Document::ReadTagLength(uint16_t testGroup, uint16_t testElement) -{ - long positionOnEntry = Fp->tellg(); - (void)positionOnEntry; - - if ( !ReadTag(testGroup, testElement) ) - { - return 0; - } - - //// Then read the associated Item Length - long currentPosition = Fp->tellg(); - uint32_t itemLength = ReadInt32(); - { - gdcmVerboseMacro( "Basic Item Length is: " - << itemLength << std::endl - << " at address: " << std::hex << (unsigned int)currentPosition); - } - return itemLength; -} - -/** - * \brief When parsing the Pixel Data of an encapsulated file, read - * the basic offset table (when present, and BTW dump it). - */ -void Document::ReadAndSkipEncapsulatedBasicOffsetTable() -{ - //// Read the Basic Offset Table Item Tag length... - uint32_t itemLength = ReadTagLength(0xfffe, 0xe000); - - // When present, read the basic offset table itself. - // Notes: - since the presence of this basic offset table is optional - // we can't rely on it for the implementation, and we will simply - // trash it's content (when present). - // - still, when present, we could add some further checks on the - // lengths, but we won't bother with such fuses for the time being. - if ( itemLength != 0 ) - { - char *basicOffsetTableItemValue = new char[itemLength + 1]; - Fp->read(basicOffsetTableItemValue, itemLength); - -#ifdef GDCM_DEBUG - for (unsigned int i=0; i < itemLength; i += 4 ) - { - uint32_t individualLength = str2num( &basicOffsetTableItemValue[i], - uint32_t); - gdcmVerboseMacro( "Read one length: " << - std::hex << individualLength ); - } -#endif //GDCM_DEBUG - - delete[] basicOffsetTableItemValue; - } -} - -/** - * \brief Parse pixel data from disk of [multi-]fragment RLE encoding. - * Compute the RLE extra information and store it in \ref RLEInfo - * for later pixel retrieval usage. + * \brief Handle broken private tag from Philips NTSCAN + * where the endianess is being switch to BigEndian for no + * apparent reason + * @return no return */ -void Document::ComputeRLEInfo() +void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem) { - std::string ts = GetTransferSyntax(); - if ( !Global::GetTS()->IsRLELossless(ts) ) + // Endian reversion. Some files contain groups of tags with reversed endianess. + static int reversedEndian = 0; + // try to fix endian switching in the middle of headers + if ((group == 0xfeff) && (elem == 0x00e0)) { - return; - } - - // Encoded pixel data: for the time being we are only concerned with - // Jpeg or RLE Pixel data encodings. - // As stated in PS 3.5-2003, section 8.2 p44: - // "If sent in Encapsulated Format (i.e. other than the Native Format) the - // value representation OB is used". - // Hence we expect an OB value representation. Concerning OB VR, - // the section PS 3.5-2003, section A.4.c p 58-59, states: - // "For the Value Representations OB and OW, the encoding shall meet the - // following specifications depending on the Data element tag:" - // [...snip...] - // - the first item in the sequence of items before the encoded pixel - // data stream shall be basic offset table item. The basic offset table - // item value, however, is not required to be present" - - ReadAndSkipEncapsulatedBasicOffsetTable(); - - // Encapsulated RLE Compressed Images (see PS 3.5-2003, Annex G) - // Loop on the individual frame[s] and store the information - // on the RLE fragments in a RLEFramesInfo. - // Note: - when only a single frame is present, this is a - // classical image. - // - when more than one frame are present, then we are in - // the case of a multi-frame image. - long frameLength; - while ( (frameLength = ReadTagLength(0xfffe, 0xe000)) ) - { - // Parse the RLE Header and store the corresponding RLE Segment - // Offset Table information on fragments of this current Frame. - // Note that the fragment pixels themselves are not loaded - // (but just skipped). - long frameOffset = Fp->tellg(); - - uint32_t nbRleSegments = ReadInt32(); - if ( nbRleSegments > 16 ) - { - // There should be at most 15 segments (refer to RLEFrame class) - gdcmVerboseMacro( "Too many segments."); - } - - uint32_t rleSegmentOffsetTable[16]; - for( int k = 1; k <= 15; k++ ) - { - rleSegmentOffsetTable[k] = ReadInt32(); - } - - // Deduce from both the RLE Header and the frameLength the - // fragment length, and again store this info in a - // RLEFramesInfo. - long rleSegmentLength[15]; - // skipping (not reading) RLE Segments - if ( nbRleSegments > 1) - { - for(unsigned int k = 1; k <= nbRleSegments-1; k++) - { - rleSegmentLength[k] = rleSegmentOffsetTable[k+1] - - rleSegmentOffsetTable[k]; - SkipBytes(rleSegmentLength[k]); - } - } - - rleSegmentLength[nbRleSegments] = frameLength - - rleSegmentOffsetTable[nbRleSegments]; - SkipBytes(rleSegmentLength[nbRleSegments]); - - // Store the collected info - RLEFrame *newFrameInfo = new RLEFrame; - newFrameInfo->NumberFragments = nbRleSegments; - for( unsigned int uk = 1; uk <= nbRleSegments; uk++ ) - { - newFrameInfo->Offset[uk] = frameOffset + rleSegmentOffsetTable[uk]; - newFrameInfo->Length[uk] = rleSegmentLength[uk]; - } - RLEInfo->Frames.push_back( newFrameInfo ); - } - - // Make sure that at the end of the item we encounter a 'Sequence - // Delimiter Item': - if ( !ReadTag(0xfffe, 0xe0dd) ) + // start endian swap mark for group found + reversedEndian++; + SwitchByteSwapCode(); + // fix the tag + group = 0xfffe; + elem = 0xe000; + } + else if (group == 0xfffe && elem == 0xe00d && reversedEndian) { - gdcmVerboseMacro( "No sequence delimiter item at end of RLE item sequence"); + // end of reversed endian group + reversedEndian--; + SwitchByteSwapCode(); } } /** - * \brief Parse pixel data from disk of [multi-]fragment Jpeg encoding. - * Compute the jpeg extra information (fragment[s] offset[s] and - * length) and store it[them] in \ref JPEGInfo for later pixel - * retrieval usage. + * \brief Group 0002 is always coded Little Endian + * whatever Transfer Syntax is + * @return no return */ -void Document::ComputeJPEGFragmentInfo() +void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem) { - // If you need to, look for comments of ComputeRLEInfo(). - std::string ts = GetTransferSyntax(); - if ( ! Global::GetTS()->IsJPEG(ts) ) - { - return; - } - - ReadAndSkipEncapsulatedBasicOffsetTable(); - - // Loop on the fragments[s] and store the parsed information in a - // JPEGInfo. - long fragmentLength; - while ( (fragmentLength = ReadTagLength(0xfffe, 0xe000)) ) - { - long fragmentOffset = Fp->tellg(); - - // Store the collected info - JPEGFragment *newFragment = new JPEGFragment; - newFragment->Offset = fragmentOffset; - newFragment->Length = fragmentLength; - JPEGInfo->Fragments.push_back( newFragment ); - - SkipBytes( fragmentLength ); - } - - // Make sure that at the end of the item we encounter a 'Sequence - // Delimiter Item': - if ( !ReadTag(0xfffe, 0xe0dd) ) + // Endian reversion. Some files contain groups of tags with reversed endianess. + if ( !Group0002Parsed && group != 0x0002) { - gdcmVerboseMacro( "No sequence delimiter item at end of JPEG item sequence"); - } -} - -/** - * \brief Walk recursively the given \ref DocEntrySet, and feed - * the given hash table (\ref TagDocEntryHT) with all the - * \ref DocEntry (Dicom entries) encountered. - * This method does the job for \ref BuildFlatHashTable. - * @param builtHT Where to collect all the \ref DocEntry encountered - * when recursively walking the given set. - * @param set The structure to be traversed (recursively). - */ -void Document::BuildFlatHashTableRecurse( TagDocEntryHT &builtHT, - DocEntrySet *set ) -{ - if (ElementSet *elementSet = dynamic_cast< ElementSet* > ( set ) ) - { - TagDocEntryHT const ¤tHT = elementSet->GetTagHT(); - for( TagDocEntryHT::const_iterator i = currentHT.begin(); - i != currentHT.end(); - ++i) - { - DocEntry *entry = i->second; - if ( SeqEntry *seqEntry = dynamic_cast(entry) ) - { - const ListSQItem& items = seqEntry->GetSQItems(); - for( ListSQItem::const_iterator item = items.begin(); - item != items.end(); - ++item) - { - BuildFlatHashTableRecurse( builtHT, *item ); - } - continue; - } - builtHT[entry->GetKey()] = entry; - } - return; - } + Group0002Parsed = true; + // we just came out of group 0002 + // if Transfer syntax is Big Endian we have to change CheckSwap - if (SQItem *SQItemSet = dynamic_cast< SQItem* > ( set ) ) - { - const ListDocEntry& currentList = SQItemSet->GetDocEntries(); - for (ListDocEntry::const_iterator i = currentList.begin(); - i != currentList.end(); - ++i) + std::string ts = GetTransferSyntax(); + if ( !Global::GetTS()->IsTransferSyntax(ts) ) { - DocEntry *entry = *i; - if ( SeqEntry *seqEntry = dynamic_cast(entry) ) - { - const ListSQItem& items = seqEntry->GetSQItems(); - for( ListSQItem::const_iterator item = items.begin(); - item != items.end(); - ++item) - { - BuildFlatHashTableRecurse( builtHT, *item ); - } - continue; - } - builtHT[entry->GetKey()] = entry; + gdcmWarningMacro("True DICOM File, with NO Tansfer Syntax: " << ts ); + return; } - } -} - -/** - * \brief Build a \ref TagDocEntryHT (i.e. a std::map<>) from the current - * Document. - * - * The structure used by a Document (through \ref ElementSet), - * in order to hold the parsed entries of a Dicom header, is a recursive - * one. This is due to the fact that the sequences (when present) - * can be nested. Additionaly, the sequence items (represented in - * gdcm as \ref SQItem) add an extra complexity to the data - * structure. Hence, a gdcm user whishing to visit all the entries of - * a Dicom header will need to dig in the gdcm internals (which - * implies exposing all the internal data structures to the API). - * In order to avoid this burden to the user, \ref BuildFlatHashTable - * recursively builds a temporary hash table, which holds all the - * Dicom entries in a flat structure (a \ref TagDocEntryHT i.e. a - * std::map<>). - * \warning Of course there is NO integrity constrain between the - * returned \ref TagDocEntryHT and the \ref ElementSet used - * to build it. Hence if the underlying \ref ElementSet is - * altered, then it is the caller responsability to invoke - * \ref BuildFlatHashTable again... - * @return The flat std::map<> we juste build. - */ -TagDocEntryHT *Document::BuildFlatHashTable() -{ - TagDocEntryHT *FlatHT = new TagDocEntryHT; - BuildFlatHashTableRecurse( *FlatHT, this ); - return FlatHT; -} - + // Group 0002 is always 'Explicit ...' enven when Transfer Syntax says 'Implicit ..." - -/** - * \brief Compares two documents, according to \ref DicomDir rules - * \warning Does NOT work with ACR-NEMA files - * \todo Find a trick to solve the pb (use RET fields ?) - * @param document - * @return true if 'smaller' - */ -bool Document::operator<(Document &document) -{ - // Patient Name - std::string s1 = GetEntry(0x0010,0x0010); - std::string s2 = document.GetEntry(0x0010,0x0010); - if(s1 < s2) - { - return true; - } - else if( s1 > s2 ) - { - return false; - } - else - { - // Patient ID - s1 = GetEntry(0x0010,0x0020); - s2 = document.GetEntry(0x0010,0x0020); - if ( s1 < s2 ) - { - return true; - } - else if ( s1 > s2 ) - { - return false; - } - else - { - // Study Instance UID - s1 = GetEntry(0x0020,0x000d); - s2 = document.GetEntry(0x0020,0x000d); - if ( s1 < s2 ) + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ImplicitVRLittleEndian ) { - return true; + Filetype = ImplicitVR; } - else if( s1 > s2 ) - { - return false; - } - else - { - // Serie Instance UID - s1 = GetEntry(0x0020,0x000e); - s2 = document.GetEntry(0x0020,0x000e); - if ( s1 < s2 ) - { - return true; - } - else if( s1 > s2 ) - { - return false; - } - } - } - } - return false; -} - - -/** - * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader - * @param filetype Type of the File to be written - */ -int Document::ComputeGroup0002Length( FileType filetype ) -{ - uint16_t gr, el; - std::string vr; - - int groupLength = 0; - bool found0002 = false; - - // for each zero-level Tag in the DCM Header - DocEntry *entry; - - Initialize(); - entry = GetNextEntry(); - while(entry) - { - gr = entry->GetGroup(); - - if (gr == 0x0002) + + // FIXME Strangely, this works with + //'Implicit VR Transfer Syntax (GE Private) + if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ExplicitVRBigEndian ) { - found0002 = true; - - el = entry->GetElement(); - vr = entry->GetVR(); - - if (filetype == ExplicitVR) - { - if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") ) - { - groupLength += 4; // explicit VR AND OB, OW, SQ : 4 more bytes - } - } - groupLength += 2 + 2 + 4 + entry->GetLength(); + gdcmWarningMacro("Transfer Syntax Name = [" + << GetTransferSyntaxName() << "]" ); + SwitchByteSwapCode(); + group = SwapShort(group); + elem = SwapShort(elem); } - else if (found0002 ) - break; - - entry = GetNextEntry(); } - return groupLength; } -} // end namespace gdcm +//----------------------------------------------------------------------------- +// Print //----------------------------------------------------------------------------- +} // end namespace gdcm