-// gdcmDocument.cxx
-//-----------------------------------------------------------------------------
+/*=========================================================================
+
+ Program: gdcm
+ Module: $RCSfile: gdcmDocument.cxx,v $
+ Language: C++
+ Date: $Date: 2008/09/15 15:49:22 $
+ Version: $Revision: 1.381 $
+
+ Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de
+ l'Image). All rights reserved. See Doc/License.txt or
+ http://www.creatis.insa-lyon.fr/Public/Gdcm/License.html for details.
+
+ This software is distributed WITHOUT ANY WARRANTY; without even
+ the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
+ PURPOSE. See the above copyright notices for more information.
+
+=========================================================================*/
#include "gdcmDocument.h"
-#include "gdcmValEntry.h"
-#include "gdcmBinEntry.h"
#include "gdcmSeqEntry.h"
-
#include "gdcmGlobal.h"
#include "gdcmUtil.h"
#include "gdcmDebug.h"
+#include "gdcmTS.h"
+#include "gdcmDictSet.h"
+#include "gdcmDocEntrySet.h"
+#include "gdcmSQItem.h"
+#include "gdcmDataEntry.h"
-#include <errno.h>
#include <vector>
+#include <iomanip>
+#include <fstream>
+#include <ctype.h> // for isdigit
+#include <stdlib.h> // for atoi
-// For nthos:
-#ifdef _MSC_VER
- #include <winsock.h>
-#else
- #include <netinet/in.h>
-#endif
-
-# include <iomanip>
-
-#define UI1_2_840_10008_1_2 "1.2.840.10008.1.2"
-#define UI1_2_840_10008_1_2_1 "1.2.840.10008.1.2.1"
-#define UI1_2_840_10008_1_2_2 "1.2.840.10008.1.2.2"
-#define UI1_2_840_10008_1_2_1_99 "1.2.840.10008.1.2.1.99"
-
-typedef struct {
- guint32 totalSQlength;
- guint32 alreadyParsedlength;
-} pileElem;
+#if defined(__BORLANDC__)
+ #include <mem.h> // for memset
+#endif
+namespace GDCM_NAME_SPACE
+{
//-----------------------------------------------------------------------------
-// Refer to gdcmDocument::CheckSwap()
-const unsigned int gdcmDocument::HEADER_LENGTH_TO_READ = 256;
-// Refer to gdcmDocument::SetMaxSizeLoadEntry()
-const unsigned int gdcmDocument::MAX_SIZE_LOAD_ELEMENT_VALUE = 4096;
-
-const unsigned int gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE = 64;
+// Refer to Document::SetMaxSizeLoadEntry()
+const unsigned int Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096
//-----------------------------------------------------------------------------
// Constructor / Destructor
+// Constructors and destructors are protected to avoid user to invoke directly
/**
- * \brief constructor
- * @param inFilename file to be opened for parsing
- * @param exception_on_error whether we throw an exception or not
- * @param enable_sequences = true to allow the header
- * to be parsed *inside* the SeQuences,
- * when they have an actual length
- * \warning enable_sequences *has to be* true for reading PAPYRUS 3.0 files
- * @param ignore_shadow to allow skipping the shadow elements,
- * to save memory space.
- * \warning The TRUE value for this param has to be used
- * with a FALSE value for the 'enable_sequence' param.
- * ('public elements' may be embedded in 'shadow Sequences')
- */
-gdcmDocument::gdcmDocument(const char *inFilename,
- bool exception_on_error,
- bool enable_sequences,
- bool ignore_shadow) {
- enableSequences=enable_sequences;
- ignoreShadow =ignore_shadow;
-
- SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
- filename = inFilename;
- Initialise();
-
- if ( !OpenFile(exception_on_error))
- return;
-
- rewind(fp);
- //if (!CheckSwap())
- // return false; // to go on compiling
-
- fseek(fp,0L,SEEK_END);
- long lgt = ftell(fp);
-
- rewind(fp);
- CheckSwap();
- long beg = ftell(fp);
- lgt -= beg;
-
- //std::cout << "beg, lgt : " << beg << " " << lgt << std::endl;
-
- long l=ParseDES( this, beg, lgt, false); // le Load sera fait a la volee
- CloseFile();
-
- printLevel = 1; // 'Medium' print level by default
-}
-
-/**
- * \brief constructor
- * @param exception_on_error
+ * \brief This default constructor neither loads nor parses the file.
+ * You should then invoke Document::Load.
+ *
*/
-gdcmDocument::gdcmDocument(bool exception_on_error) {
- (void)exception_on_error;
- enableSequences=0;
+Document::Document()
+ :ElementSet()
+{
+ Fp = 0;
SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
- Initialise();
-
- printLevel = 1; // 'Medium' print level by default
+ Initialize();
+ SwapCode = 1234;
+ Filetype = ExplicitVR;
+ CurrentOffsetPosition = 0;
+ OffsetOfPreviousParseDES =0;
+ // Load will set it to true if sucessfull
+ Group0002Parsed = false;
+ IsDocumentAlreadyLoaded = false;
+ IsDocumentModified = true;
+ LoadMode = LD_ALL; // default : load everything, later
+
+ SetFileName("");
+ changeFromUN=false;
+ UnexpectedEOF=false;
}
/**
* \brief Canonical destructor.
*/
-gdcmDocument::~gdcmDocument (void) {
- RefPubDict = NULL;
- RefShaDict = NULL;
-}
-
-//-----------------------------------------------------------------------------
-// Print
-/**
-
-
-/**
- * \brief Prints The Dict Entries of THE public Dicom Dictionary
- * @return
- */
-void gdcmDocument::PrintPubDict(std::ostream & os) {
- RefPubDict->Print(os);
-}
-
-/**
- * \brief Prints The Dict Entries of THE shadow Dicom Dictionary
- * @return
- */
-void gdcmDocument::PrintShaDict(std::ostream & os) {
- RefShaDict->Print(os);
+Document::~Document ()
+{
+ CloseFile();
}
//-----------------------------------------------------------------------------
// Public
/**
- * \brief Get the public dictionary used
- */
-gdcmDict *gdcmDocument::GetPubDict(void) {
- return(RefPubDict);
-}
-
-/**
- * \brief Get the shadow dictionary used
- */
-gdcmDict *gdcmDocument::GetShaDict(void) {
- return(RefShaDict);
-}
-
-/**
- * \brief Set the shadow dictionary used
- * \param dict dictionary to use in shadow
+ * \brief Loader. use SetLoadMode(), SetFileName() before !
+ * @return false if file cannot be open or no swap info was found,
+ * or no tag was found.
*/
-bool gdcmDocument::SetShaDict(gdcmDict *dict){
- RefShaDict=dict;
- return(!RefShaDict);
+bool Document::Load( )
+{
+ if ( GetFileName() == "" )
+ {
+ gdcmWarningMacro( "Use SetFileName, before !" );
+ return false;
+ }
+ return DoTheLoadingDocumentJob( );
}
-/**
- * \brief Set the shadow dictionary used
- * \param dictName name of the dictionary to use in shadow
- */
-bool gdcmDocument::SetShaDict(DictKey dictName){
- RefShaDict=gdcmGlobal::GetDicts()->GetDict(dictName);
- return(!RefShaDict);
-}
+//#ifndef GDCM_LEGACY_REMOVE
/**
- * \brief This predicate, based on hopefully reasonable heuristics,
- * decides whether or not the current gdcmDocument was properly parsed
- * and contains the mandatory information for being considered as
- * a well formed and usable Dicom/Acr File.
- * @return true when gdcmDocument is the one of a reasonable Dicom/Acr file,
- * false otherwise.
+ * \brief Loader. (DEPRECATED : not to break the API)
+ * @param fileName 'Document' (File or DicomDir) to be open for parsing
+ * @return false if file cannot be open or no swap info was found,
+ * or no tag was found.
*/
-bool gdcmDocument::IsReadable(void) {
- if(filetype==gdcmUnknown) {
- return(false);
- }
- if(listEntries.size()<=0) {
- return(false);
- }
-
- return(true);
+ /*
+bool Document::Load( std::string const &fileName )
+{
+ Filename = fileName;
+ return DoTheLoadingDocumentJob( );
}
+*/
+//#endif
/**
- * \brief Determines if the Transfer Syntax was already encountered
- * and if it corresponds to a ImplicitVRLittleEndian one.
- * @return True when ImplicitVRLittleEndian found. False in all other cases.
+ * \brief Performs the Loading Job (internal use only)
+ * @return false if file cannot be open or no swap info was found,
+ * or no tag was found.
*/
-bool gdcmDocument::IsImplicitVRLittleEndianTransferSyntax(void) {
- gdcmDocEntry *Element = GetDocEntryByNumber(0x0002, 0x0010);
- if ( !Element )
- return false;
- LoadDocEntrySafe(Element);
-
- std::string Transfer = Element->GetValue();
- if ( Transfer == UI1_2_840_10008_1_2 )
+bool Document::DoTheLoadingDocumentJob( )
+{
+ if ( ! IsDocumentModified ) // Nothing to do !
return true;
- return false;
-}
-/**
- * \brief Determines if the Transfer Syntax was already encountered
- * and if it corresponds to a ExplicitVRLittleEndian one.
- * @return True when ExplicitVRLittleEndian found. False in all other cases.
- */
-bool gdcmDocument::IsExplicitVRLittleEndianTransferSyntax(void) {
- gdcmDocEntry* Element = GetDocEntryByNumber(0x0002, 0x0010);
- if ( !Element )
+ ClearEntry();
+
+ Fp = 0;
+ if ( !OpenFile() )
+ {
+ // warning already performed in OpenFile()
+ Filetype = Unknown;
return false;
- LoadDocEntrySafe(Element);
+ }
- std::string Transfer = Element->GetValue();
- if ( Transfer == UI1_2_840_10008_1_2_1 )
- return true;
- return false;
-}
+ Group0002Parsed = false;
-/**
- * \brief Determines if the Transfer Syntax was already encountered
- * and if it corresponds to a DeflatedExplicitVRLittleEndian one.
- * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases.
- */
-bool gdcmDocument::IsDeflatedExplicitVRLittleEndianTransferSyntax(void) {
- gdcmDocEntry* Element = GetDocEntryByNumber(0x0002, 0x0010);
- if ( !Element )
- return false;
- LoadDocEntrySafe(Element);
+ gdcmDebugMacro( "Starting parsing of file: " << Filename.c_str());
- std::string Transfer = Element->GetValue();
- if ( Transfer == UI1_2_840_10008_1_2_1_99 )
- return true;
- return false;
-}
+ // Computes the total length of the file
+ Fp->seekg(0, std::ios::end); // Once per Document !
+ long lgt = Fp->tellg(); // Once per Document !
+ Fp->seekg(0, std::ios::beg); // Once per Document !
-/**
- * \brief Determines if the Transfer Syntax was already encountered
- * and if it corresponds to a Explicit VR Big Endian one.
- * @return True when big endian found. False in all other cases.
- */
-bool gdcmDocument::IsExplicitVRBigEndianTransferSyntax(void) {
- gdcmDocEntry* Element = GetDocEntryByNumber(0x0002, 0x0010);
- if ( !Element )
- return false;
- LoadDocEntrySafe(Element);
+ // CheckSwap returns a boolean
+ // (false if no swap info of any kind was found)
+ if (! CheckSwap() )
+ {
+ gdcmWarningMacro( "Neither a DICOM V3 nor an ACR-NEMA file: "
+ << Filename.c_str());
+ CloseFile();
+ return false;
+ }
- std::string Transfer = Element->GetValue();
- if ( Transfer == UI1_2_840_10008_1_2_2 ) //1.2.2 ??? A verifier !
- return true;
- return false;
-}
+ long beg = Fp->tellg(); // just after DICOM preamble (if any)
-/**
- * \brief returns the File Type
- * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
- * @return the FileType code
- */
-FileType gdcmDocument::GetFileType(void) {
- return(filetype);
-}
+ lgt -= beg; // remaining length to parse
-/**
- * \brief opens the file
- * @param exception_on_error
- * @return
- */
-FILE *gdcmDocument::OpenFile(bool exception_on_error)
- throw(gdcmFileError)
-{
- fp=fopen(filename.c_str(),"rb");
- if(exception_on_error)
- {
- if(!fp)
- throw gdcmFileError("gdcmDocument::gdcmDocument(const char *, bool)");
- }
-
- if ( fp )
- {
- guint16 zero;
- fread(&zero, (size_t)2, (size_t)1, fp);
-
- //ACR -- or DICOM with no Preamble --
- if( zero == 0x0008 || zero == 0x0800 || zero == 0x0002 || zero == 0x0200)
- return(fp);
-
- //DICOM
- fseek(fp, 126L, SEEK_CUR);
- char dicm[4];
- fread(dicm, (size_t)4, (size_t)1, fp);
- if( memcmp(dicm, "DICM", 4) == 0 )
- return(fp);
-
- fclose(fp);
- dbg.Verbose(0, "gdcmDocument::OpenFile not DICOM/ACR", filename.c_str());
- }
- else {
- dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file", filename.c_str());
- }
- return(NULL);
-}
+ // Recursive call.
+ // Loading is done during parsing
+ OffsetOfPreviousParseDES = beg;
+ ParseDES( this, beg, lgt, false); // delim_mode is first defaulted to false
-/**
- * \brief closes the file
- * @return TRUE if the close was successfull
- */
-bool gdcmDocument::CloseFile(void) {
- int closed = fclose(fp);
- fp = (FILE *)0;
- if (! closed)
- return false;
- return true;
-}
+ if ( IsEmpty() )
+ {
+ gdcmErrorMacro( "No tag in internal hash table for: "
+ << Filename.c_str());
+ CloseFile();
+ return false;
+ }
+ IsDocumentAlreadyLoaded = true;
-/**
- * \brief Writes in a file all the Header Entries (Dicom Elements)
- * of the Chained List
- * @param fp file pointer on an already open file
- * @param type Type of the File to be written
- * (ACR-NEMA, ExplicitVR, ImplicitVR)
- * \return Always true.
- */
-bool gdcmDocument::Write(FILE *fp, FileType type) {
-/// \todo
-/// ==============
-/// The stuff will have to be rewritten using the SeQuence based
-/// tree-like stucture instead of the chained list .
-/// (so we shall remove the GroupHT from the gdcmDocument)
-/// To be checked
-/// =============
-
- /// \todo move the following lines (and a lot of others, to be written)
- /// to a future function CheckAndCorrectHeader
+ //Fp->seekg(0, std::ios::beg); // Once per Document!
- /// \todo
- /// Question :
- /// Comment pourrait-on savoir si le DcmHeader vient d'un fichier
- /// DicomV3 ou non (FileType est un champ de gdcmDocument ...)
- /// WARNING : Si on veut ecrire du DICOM V3 a partir d'un DcmHeader ACR-NEMA
- /// no way
- /// a moins de se livrer a un tres complique ajout des champs manquants.
- /// faire un CheckAndCorrectHeader (?)
-
- if (type == gdcmImplicitVR)
- {
- std::string implicitVRTransfertSyntax = UI1_2_840_10008_1_2;
- ReplaceOrCreateByNumber(implicitVRTransfertSyntax,0x0002, 0x0010);
-
- /// \todo Refer to standards on page 21, chapter 6.2
- /// "Value representation": values with a VR of UI shall be
- /// padded with a single trailing null
- /// Dans le cas suivant on doit pader manuellement avec un 0
+ // Load 'non string' values
- SetEntryLengthByNumber(18, 0x0002, 0x0010);
- }
-
- if (type == gdcmExplicitVR)
+ std::string PhotometricInterpretation = GetEntryString(0x0028,0x0004);
+ if ( PhotometricInterpretation == "PALETTE COLOR " )
{
- std::string explicitVRTransfertSyntax = UI1_2_840_10008_1_2_1;
- ReplaceOrCreateByNumber(explicitVRTransfertSyntax,0x0002, 0x0010);
+ // FIXME
+ // Probabely this line should be outside the 'if'
+ // Try to find an image sample holding a 'gray LUT'
+ LoadEntryBinArea(0x0028,0x1200); // gray LUT
+
+ /// FIXME
+ /// --> FIXME : The difference between BinEntry and DataEntry
+ /// --> no longer exists, but the alteration of Dicom Dictionary remains.
+ /// --> Old comment restored on purpose.
+ /// --> New one (replacing both BinEntry and ValEntry by DataEntry)
+ /// --> had absolutely no meaning.
+ /// --> The whole comment will be removed when the stuff is cleaned !
+ /// -->
+ /// The tags refered by the three following lines used to be CORRECTLY
+ /// defined as having an US Value Representation in the public
+ /// dictionary. BUT the semantics implied by the three following
+ /// lines state that the corresponding tag contents are in fact
+ /// the ones of a BinEntry.
+ /// 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 decide of the semantics by deciding
+ /// whether the following tags are either :
+ /// - multivaluated US, and hence loaded as ValEntry, but afterwards
+ /// also used as BinEntry, which requires the proper conversion,
+ /// - OW, and hence loaded as BinEntry, but afterwards also used
+ /// as ValEntry, which requires the proper conversion.
- /// \todo Refer to standards on page 21, chapter 6.2
- /// "Value representation": values with a VR of UI shall be
- /// padded with a single trailing null
- /// Dans le cas suivant on doit pader manuellement avec un 0
+ // --> OB (byte aray) or OW (short int aray)
+ // The actual VR has to be deduced from other entries.
+ // Our way of loading them may fail in some cases :
+ // We must or not SwapByte depending on other field values.
+
+ LoadEntryBinArea(0x0028,0x1201); // R LUT
+ LoadEntryBinArea(0x0028,0x1202); // G LUT
+ LoadEntryBinArea(0x0028,0x1203); // B LUT
- SetEntryLengthByNumber(20, 0x0002, 0x0010);
+ // Segmented Red Palette Color LUT Data
+ LoadEntryBinArea(0x0028,0x1221);
+ // Segmented Green Palette Color LUT Data
+ LoadEntryBinArea(0x0028,0x1222);
+ // Segmented Blue Palette Color LUT Data
+ LoadEntryBinArea(0x0028,0x1223);
+ }
+
+ //FIXME later : how to use it?
+ SeqEntry *modLutSeq = GetSeqEntry(0x0028,0x3000); // Modality LUT Sequence
+ if ( modLutSeq !=0 )
+ {
+ SQItem *sqi= modLutSeq->GetFirstSQItem();
+ if ( sqi != 0 )
+ {
+ DataEntry *dataEntry = sqi->GetDataEntry(0x0028,0x3006); // LUT Data
+ if ( dataEntry != 0 )
+ {
+ if ( dataEntry->GetLength() != 0 )
+ {
+ // FIXME : CTX dependent means : contexted dependant.
+ // see upper comment.
+ LoadEntryBinArea(dataEntry); //LUT Data (CTX dependent)
+ }
+ }
+ }
}
-/**
- * \todo rewrite later, if really usefull
- *
- * --> Warning : un-updated odd groups lengths can causes pb
- * --> (xmedcon breaks)
- * --> to be re- written with future org.
- *
- * if ( (type == ImplicitVR) || (type == ExplicitVR) )
- * UpdateGroupLength(false,type);
- * if ( type == ACR)
- * UpdateGroupLength(true,ACR);
- */
+ // Force Loading some more elements if user asked to.
- WriteEntries(fp,type);
- return(true);
-}
+ GDCM_NAME_SPACE::DocEntry *d;
+ for (ListElements::iterator it = UserForceLoadList.begin();
+ it != UserForceLoadList.end();
+ ++it)
+ {
+ gdcmDebugMacro( "Force Load " << std::hex
+ << (*it).Group << "|" <<(*it).Elem );
+
+ d = GetDocEntry( (*it).Group, (*it).Elem);
+
+ if ( d == NULL)
+ {
+ gdcmWarningMacro( "You asked to ForceLoad " << std::hex
+ << (*it).Group <<"|"<< (*it).Elem
+ << " that doesn't exist" );
+ continue;
+ }
-/**
- * \brief Modifies the value of a given Header Entry (Dicom Element)
- * when it exists. Create it with the given value when unexistant.
- * \warning Adds the Header Entry to the HTable, NOT to the chained List
- * @param Value Value to be set
- * @param Group Group of the Entry
- * @param Elem Element of the Entry
- * \return pointer to the modified/created Header Entry (NULL when creation
- * failed).
- */
-
-
-gdcmDocEntry * gdcmDocument::ReplaceOrCreateByNumber(
- std::string Value,
- guint16 Group,
- guint16 Elem ){
- gdcmDocEntry* a;
- a = GetDocEntryByNumber( Group, Elem);
- if (a == NULL) {
- a =NewDocEntryByNumber(Group, Elem);
- if (a == NULL)
- return NULL;
- AddEntry(a);
- }
- SetEntryByNumber(Value, Group, Elem);
- //a->SetValue(Value);
- return(a);
-}
+ LoadDocEntry(d, true);
+ }
-/**
- * \brief Set a new value if the invoked element exists
- * Seems to be useless !!!
- * @param Value new element value
- * @param Group group of the Entry
- * @param Elem element of the Entry
- * \return boolean
- */
-bool gdcmDocument::ReplaceIfExistByNumber(char* Value, guint16 Group, guint16 Elem )
-{
- std::string v = Value;
- SetEntryByNumber(v, Group, Elem);
+ 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 of years)
+ // 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
+ //
+ std::string RecCode;
+ RecCode = GetEntryString(0x0008, 0x0010); // recognition code (RET)
+
+
+ if(RecCode.find("ACRNEMA_LIBIDO") == 0 || // any version
+ RecCode.find("CANRME_AILIBOD") == 0) // for brain-damaged softwares
+ // with "little-endian strings"
+ {
+
+ Filetype = ACR_LIBIDO;
+ std::string rows = GetEntryString(0x0028, 0x0010);
+ std::string columns = GetEntryString(0x0028, 0x0011);
+ SetEntryString(columns, 0x0028, 0x0010);
+ SetEntryString(rows , 0x0028, 0x0011);
+ }
+ // --- End of ACR-LibIDO kludge ---
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 number of occurences
- */
-int gdcmDocument::CheckIfEntryExistByNumber(guint16 group, guint16 element ) {
- std::string key = gdcmDictEntry::TranslateToKey(group, element );
- return (tagHT.count(key));
}
-/**
- * \brief Searches within Header Entries (Dicom Elements) parsed with
- * the public and private dictionaries
- * for the element value of a given tag.
- * \warning Don't use any longer : use GetPubEntryByName
- * @param tagName name of the searched element.
- * @return Corresponding element value when it exists,
- * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
- */
-std::string gdcmDocument::GetEntryByName(std::string tagName) {
- gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
- if( dictEntry == NULL)
- return GDCM_UNFOUND;
-
- return(GetEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement()));
-}
/**
- * \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 tagName name of the searched element.
- * @return Corresponding element value representation when it exists,
- * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
+ * \brief Adds a new element we want to load anyway
+ * @param group Group number of the target tag.
+ * @param elem Element number of the target tag.
*/
-std::string gdcmDocument::GetEntryVRByName(std::string tagName) {
- gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
- if( dictEntry == NULL)
- return GDCM_UNFOUND;
-
- gdcmDocEntry* elem = GetDocEntryByNumber(dictEntry->GetGroup(),
- dictEntry->GetElement());
- return elem->GetVR();
+void Document::AddForceLoadElement (uint16_t group, uint16_t elem)
+{
+ DicomElement el;
+ el.Group = group;
+ el.Elem = elem;
+ UserForceLoadList.push_back(el);
}
-
-
/**
- * \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 of the searched tag.
- * @param element Element of the searched tag.
- * @return Corresponding element value representation when it exists,
- * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
+ * \brief Get the public dictionary used
*/
-std::string gdcmDocument::GetEntryByNumber(guint16 group, guint16 element){
- TagKey key = gdcmDictEntry::TranslateToKey(group, element);
- if ( ! tagHT.count(key))
- return GDCM_UNFOUND;
- return tagHT.find(key)->second->GetValue();
+Dict *Document::GetPubDict()
+{
+ return RefPubDict;
}
/**
- * \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 of the searched tag.
- * @param element Element of the searched tag.
- * @return Corresponding element value representation when it exists,
- * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
+ * \brief Get the shadow dictionary used
*/
-std::string gdcmDocument::GetEntryVRByNumber(guint16 group, guint16 element) {
- gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
- if ( !elem )
- return GDCM_UNFOUND;
- return elem->GetVR();
+Dict *Document::GetShaDict()
+{
+ return RefShaDict;
}
/**
- * \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 of the searched tag.
- * @param element Element of the searched tag.
- * @return Corresponding element length; -2 if not found
+ * \brief Set the shadow dictionary used
+ * @param dict dictionary to use in shadow
*/
-int gdcmDocument::GetEntryLengthByNumber(guint16 group, guint16 element) {
- gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
- if ( !elem )
- return -2;
- return elem->GetLength();
+bool Document::SetShaDict(Dict *dict)
+{
+ RefShaDict = dict;
+ return !RefShaDict;
}
+
/**
- * \brief Sets the value (string) of the Header Entry (Dicom Element)
- * @param content string value of the Dicom Element
- * @param tagName name of the searched Dicom Element.
- * @return true when found
+ * \brief Set the shadow dictionary used
+ * @param dictName name of the dictionary to use in shadow
*/
-bool gdcmDocument::SetEntryByName(std::string content,std::string tagName) {
- gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
- if( dictEntry == NULL)
- return false;
-
- return(SetEntryByNumber(content,dictEntry->GetGroup(),
- dictEntry->GetElement()));
+bool Document::SetShaDict(DictKey const &dictName)
+{
+ RefShaDict = Global::GetDicts()->GetDict(dictName);
+ return !RefShaDict;
}
/**
- * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
- * through it's (group, element) and modifies it's content with
- * the given value.
- * @param content new value to substitute with
- * @param group group of the Dicom Element to modify
- * @param element element of the Dicom Element to modify
+ * \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 gdcmDocument::SetEntryByNumber(std::string content,
- guint16 group,
- guint16 element)
+bool Document::IsParsable()
{
- TagKey key = gdcmDictEntry::TranslateToKey(group, element);
- if ( ! tagHT.count(key))
+ if ( Filetype == Unknown )
+ {
+ gdcmWarningMacro( "Wrong filetype for " << GetFileName());
return false;
- int l = content.length();
- if(l%2) // Non even length are padded with a space (020H).
- {
- l++;
- content = content + '\0';
}
-
- gdcmDocEntry * a;
- IterHT p;
- TagDocEntryHT::iterator p2;
- // DO NOT remove the following lines : they explain the stuff
- //p= tagHT.equal_range(key); // get a pair of iterators first-last synonym
- //p2=p.first; // iterator on the first synonym
- //a=p2->second; // H Table target column (2-nd col)
-
- // or, easier :
-
- // Oops ! this was for multimap, right ?
- a = ((tagHT.equal_range(key)).first)->second;
-
- ((gdcmValEntry*)a)->SetValue(content);
-
- std::string vr = a->GetVR();
-
- guint32 lgr;
- if( (vr == "US") || (vr == "SS") )
- lgr = 2;
- else if( (vr == "UL") || (vr == "SL") )
- lgr = 4;
- else
- lgr = l;
-
- a->SetLength(lgr);
- return true;
-}
-/**
- * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
- * in the PubDocEntrySet of this instance
- * through it's (group, element) and modifies it's length with
- * the given value.
- * \warning Use with extreme caution.
- * @param l new length to substitute with
- * @param group group of the Entry to modify
- * @param element element of the Entry to modify
- * @return true on success, false otherwise.
- */
-bool gdcmDocument::SetEntryLengthByNumber(guint32 l,
- guint16 group,
- guint16 element)
-{
- TagKey key = gdcmDictEntry::TranslateToKey(group, element);
- if ( ! tagHT.count(key))
+ if ( IsEmpty() )
+ {
+ gdcmWarningMacro( "No tag in internal hash table.");
return false;
- if (l%2) l++; // length must be even
- ( ((tagHT.equal_range(key)).first)->second )->SetLength(l);
+ }
- return true ;
+ return true;
}
-
/**
- * \brief Gets (from Header) the offset of a 'non string' element value
- * (LoadElementValues has already be executed)
- * @param Group group of the Entry
- * @param Elem element of the Entry
- * @return File Offset of the Element Value
+ * \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.
*/
-size_t gdcmDocument::GetEntryOffsetByNumber(guint16 Group, guint16 Elem)
+bool Document::IsReadable()
{
- gdcmDocEntry* Entry = GetDocEntryByNumber(Group, Elem);
- if (!Entry)
- {
- dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber",
- "failed to Locate gdcmDocEntry");
- return (size_t)0;
- }
- return Entry->GetOffset();
+ return IsParsable();
}
/**
- * \brief Gets (from Header) a 'non string' element value
- * (LoadElementValues has already be executed)
- * @param Group group of the Entry
- * @param Elem element of the Entry
- * @return Pointer to the 'non string' area
+ * \brief Predicate for dicom version 3 file.
+ * @return True when the file is a dicom version 3.
*/
-void * gdcmDocument::GetEntryVoidAreaByNumber(guint16 Group, guint16 Elem)
+bool Document::IsDicomV3()
{
- gdcmDocEntry* Entry = GetDocEntryByNumber(Group, Elem);
- if (!Entry)
- {
- dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber",
- "failed to Locate gdcmDocEntry");
- return (NULL);
- }
- return ((gdcmBinEntry *)Entry)->GetVoidArea();
+ // Checking if Transfer Syntax exists is enough
+ // 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;
}
/**
- * \brief Loads (from disk) the element content
- * when a string is not suitable
- * @param Group group of the Entry
- * @param Elem element of the Entry
+ * \brief Predicate for Papyrus file
+ * Dedicated to whomsoever it may concern
+ * @return True when the file is a Papyrus file.
*/
-void *gdcmDocument::LoadEntryVoidArea(guint16 Group, guint16 Elem)
+bool Document::IsPapyrus()
{
- gdcmDocEntry * Element= GetDocEntryByNumber(Group, Elem);
- if ( !Element )
- return NULL;
- size_t o =(size_t)Element->GetOffset();
- fseek(fp, o, SEEK_SET);
- size_t l=Element->GetLength();
- char* a = new char[l];
- if(!a)
- return NULL;
-
- SetEntryVoidAreaByNumber(a, Group, Elem);
- /// \todo check the result
- size_t l2 = fread(a, 1, l ,fp);
- if(l != l2)
- {
- delete[] a;
- return NULL;
- }
-
- return a;
+ // check for Papyrus private Sequence
+ DocEntry *e = GetDocEntry(0x0041, 0x1050);
+ if ( !e )
+ return false;
+ // check if it's actually a Sequence
+ if ( !dynamic_cast<SeqEntry*>(e) )
+ return false;
+ return true;
}
/**
- * \brief Sets a 'non string' value to a given Dicom Element
- * @param area
- * @param group Group number of the searched Dicom Element
- * @param element Element number of the searched Dicom Element
- * @return
+ * \brief returns the File Type
+ * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
+ * @return the FileType code
*/
-bool gdcmDocument::SetEntryVoidAreaByNumber(void * area,
- guint16 group,
- guint16 element)
+FileType Document::GetFileType()
{
- TagKey key = gdcmDictEntry::TranslateToKey(group, element);
- if ( ! tagHT.count(key))
- return false;
- // This was for multimap ?
- (( gdcmBinEntry *)( ((tagHT.equal_range(key)).first)->second ))->SetVoidArea(area);
-
- return true;
+ return Filetype;
}
/**
- * \brief Update the entries with the shadow dictionary.
- * Only non even entries are analyzed
+ * \brief Accessor to the Transfer Syntax (when present) of the
+ * current document (it internally handles reading the
+ * value from disk when only parsing occured).
+ * @return The encountered Transfer Syntax of the current document, if DICOM.
+ * GDCM_UNKNOWN for ACR-NEMA files (or broken headers ...)
*/
-void gdcmDocument::UpdateShaEntries(void) {
- gdcmDictEntry *entry;
- std::string vr;
-
- for(ListTag::iterator it=listEntries.begin();
- it!=listEntries.end();
- ++it)
+std::string Document::GetTransferSyntax()
+{
+ DocEntry *entry = GetDocEntry(0x0002, 0x0010);
+ if ( !entry )
{
- // Odd group => from public dictionary
- if((*it)->GetGroup()%2==0)
- continue;
-
- // Peer group => search the corresponding dict entry
- if(RefShaDict)
- entry=RefShaDict->GetDictEntryByNumber((*it)->GetGroup(),(*it)->GetElement());
- else
- entry=NULL;
+ return GDCM_UNKNOWN;
+ }
- if((*it)->IsImplicitVR())
- vr="Implicit";
- else
- vr=(*it)->GetVR();
-
- //(*it)->SetValue(GetDocEntryUnvalue(*it));
- // to go on compiling
- if(entry){
- // Set the new entry and the new value
- (*it)->SetDictEntry(entry);
- CheckDocEntryVR(*it,vr);
-
- //(*it)->SetValue(GetDocEntryValue(*it));
- // to go on compiling
-
+ // The entry might be present but not loaded (parsing and loading
+ // happen at different stages): try loading and proceed with check...
+
+ // Well ...
+ // (parsing and loading happen at the very same stage!)
+ //LoadDocEntrySafe(entry); //JPRx
+ if (DataEntry *dataEntry = dynamic_cast<DataEntry *>(entry) )
+ {
+ std::string transfer = dataEntry->GetString();
+ // The actual transfer (as read from disk) might be padded. We
+ // first need to remove the potential padding. We can make the
+ // weak assumption that padding was not executed with digits...
+ if ( transfer.length() == 0 )
+ {
+ // for brain damaged headers
+ gdcmWarningMacro( "Transfer Syntax has length = 0.");
+ return GDCM_UNKNOWN;
}
- else
+ while ( !isdigit((unsigned char)transfer[transfer.length()-1]) )
{
- // Remove precedent value transformation
- (*it)->SetDictEntry(NewVirtualDictEntry((*it)->GetGroup(),(*it)->GetElement(),vr));
+ transfer.erase(transfer.length()-1, 1);
+ if ( transfer.length() == 0 )
+ {
+ // for brain damaged headers
+ gdcmWarningMacro( "Transfer Syntax contains no valid character.");
+ return GDCM_UNKNOWN;
+ }
}
+ return transfer;
}
+ return GDCM_UNKNOWN;
}
/**
- * \brief Searches within the Header Entries for a Dicom Element of
- * a given tag.
- * @param tagName name of the searched Dicom Element.
- * @return Corresponding Dicom Element when it exists, and NULL
- * otherwise.
+ * \brief Accesses the info from 0002,0010 : Transfer Syntax and TS
+ * @return The full Transfer Syntax Name (as opposed to Transfer Syntax UID)
*/
- gdcmDocEntry *gdcmDocument::GetDocEntryByName(std::string tagName) {
- gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
- if( dictEntry == NULL)
- return NULL;
+std::string Document::GetTransferSyntaxName()
+{
+ // use the TS (TS : Transfer Syntax)
+ std::string transferSyntax = GetEntryString(0x0002,0x0010);
- return(GetDocEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement()));
-}
+ 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 )
+ {
+ gdcmDebugMacro( "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 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
+ * \brief Swaps the bytes so they agree with the processor order
+ * @return The properly swaped 16 bits integer.
*/
-gdcmDocEntry* gdcmDocument::GetDocEntryByNumber(guint16 group, guint16 element)
+uint16_t Document::SwapShort(uint16_t a)
{
- TagKey key = gdcmDictEntry::TranslateToKey(group, element);
- if ( ! tagHT.count(key))
- return NULL;
- return tagHT.find(key)->second;
+ if ( SwapCode == 4321 || SwapCode == 2143 )
+ {
+ //a = ((( a << 8 ) & 0xff00 ) | (( a >> 8 ) & 0x00ff ) );
+ // Save CPU time
+ a = ( a << 8 ) | ( a >> 8 );
+ }
+ return a;
}
/**
- * \brief retrieves the Dicom Elements (all of them) using (group, element)
- * @param group Group number of the searched Dicom Element.
- * @param element Element number of the searched Dicom Element.
- * @return a range (i.e.pair<,>) containing all elements whose key is group|element)
+ * \brief Swaps back the bytes of 4-byte long integer accordingly to
+ * processor order.
+ * @return The properly swaped 32 bits integer.
*/
-
-IterHT gdcmDocument::GetDocEntrySameNumber(guint16 group, guint16 element){
- TagKey key = gdcmDictEntry::TranslateToKey(group, element);
- return (tagHT.equal_range(key));
-}
+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) );
+// save CPU time
+ a=( ( a<<24) | ((a<<8) & 0x00ff0000) |
+ ((a>>8) & 0x0000ff00) | (a>>24) );
+ break;
+ case 3412 :
+// a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
+ a=( (a<<16) | (a>>16) );
+ break;
+ case 2143 :
+ a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
+ break;
+ default :
+ gdcmErrorMacro( "Unexpected swap code:" << SwapCode );
+ a = 0;
+ }
+ return a;
+}
/**
- * \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
+ * \brief Swaps back the bytes of 8-byte long 'double' accordingly to
+ * processor order.
+ * @return The properly swaped 64 bits double.
*/
-void gdcmDocument::LoadDocEntrySafe(gdcmDocEntry * entry) {
- long PositionOnEntry = ftell(fp);
- LoadDocEntry(entry);
- fseek(fp, PositionOnEntry, SEEK_SET);
-}
+double Document::SwapDouble(double a)
+{
+ switch (SwapCode)
+ {
+ // There were no 'double' at ACR-NEMA time.
+ // We just have to deal with 'straight Little Endian' and
+ // 'straight Big Endian'
+ case 1234 :
+ break;
+ case 4321 :
+ {
+ char *beg = (char *)&a;
+ char *end = beg + 7;
+ char t;
+ for (unsigned int i = 0; i<7; i++)
+ {
+ t = *beg;
+ *beg = *end;
+ *end = t;
+ beg++,
+ end--;
+ }
+ }
+ break;
+ default :
+ gdcmErrorMacro( "Unexpected swap code:" << SwapCode );
+ a = 0.;
+ }
+ return a;
+}
+//
+// -----------------File I/O ---------------
/**
- * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader
- * \warning : to be re-written using the chained list instead of the H table.
- * \warning : DO NOT use (doesn't work any longer because of the multimap)
- * \todo : to be re-written using the chained list instead of the H table
- * @param SkipSequence TRUE if we don't want to write Sequences (ACR-NEMA Files)
- * @param type Type of the File (ExplicitVR,ImplicitVR, ACR, ...)
+ * \brief Tries to open the file Document::Filename and
+ * checks the preamble when existing,
+ * or if the file starts with an ACR-NEMA look-like element.
+ * @return The FILE pointer on success, 0 on failure.
*/
-
- /* Probabely useless, now !
-
-void gdcmDocument::UpdateGroupLength(bool SkipSequence, FileType type) {
- guint16 gr, el;
- std::string vr;
-
- gdcmDocEntry *elem;
- char trash[10];
- std::string str_trash;
+std::ifstream *Document::OpenFile()
+{
+ HasDCMPreamble = false;
+ if (Filename.length() == 0)
+ {
+ return 0;
+ }
+
+ if ( Fp )
+ {
+ gdcmDebugMacro( "File already open: " << Filename.c_str());
+ CloseFile();
+ }
+
+ Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary);
+ if ( ! *Fp )
+ {
+ // Don't user gdcmErrorMacro :
+ // a spurious message will appear when you use, for instance
+ // gdcm::FileHelper *fh = new gdcm::FileHelper( outputFileName );
+ // to create outputFileName.
- GroupKey key;
- GroupHT groupHt; // to hold the length of each group
- TagKey tk;
- // remember :
- // typedef std::map<GroupKey, int> GroupHT;
+ // FIXME : if the upper comment is still usefull
+ // --> the constructor is not so good ...
- gdcmDocEntry *elemZ;
-
- // for each Tag in the DCM Header
+ gdcmWarningMacro( "Cannot open file: " << Filename.c_str());
+ delete Fp;
+ Fp = 0;
+ return 0;
+ //exit(1); // No function is allowed to leave the application instead
+ // of warning the caller
+ }
+
+ uint16_t zero = 0;
+ Fp->read((char*)&zero, (size_t)2);
+ if ( Fp->eof() )
+ {
+ CloseFile();
+ return 0;
+ }
+
+ //-- DICOM --
+ Fp->seekg(126L, std::ios::cur); // Once per Document
+ char dicm[4]; // = {' ',' ',' ',' '};
+ Fp->read(dicm, (size_t)4);
+ if ( Fp->eof() )
+ {
+ CloseFile();
+ return 0;
+ }
- for (TagDocEntryHT::iterator tag2 = tagHT.begin();
- tag2 != tagHT.end();
- ++tag2)
- {
- elem = tag2->second;
- gr = elem->GetGroup();
- el = elem->GetElement();
- vr = elem->GetVR();
-
- sprintf(trash, "%04x", gr);
- key = trash; // generate 'group tag'
-
- // if the caller decided not to take SEQUENCEs into account
- // e.g : he wants to write an ACR-NEMA File
-
- if (SkipSequence && vr == "SQ")
- continue;
-
- // Still unsolved problem :
- // we cannot find the 'Sequence Delimitation Item'
- // since it's at the end of the Hash Table
- // (fffe,e0dd)
-
- // there is SEQUENCE in ACR-NEMA
- // WARNING :
- // --> la descente a l'interieur' des SQ
- // devrait etre faite avec une liste chainee, pas avec une HTable...
-
- if ( groupHt.count(key) == 0) // we just read the first elem of a given group
- {
- if (el == 0x0000) // the first elem is 0x0000
- {
- groupHt[key] = 0; // initialize group length
- }
- else
- {
- groupHt[key] = 2 + 2 + 4 + elem->GetLength(); // non 0x0000 first group elem
- }
- }
- else // any elem but the first
- {
- if (type == ExplicitVR)
- {
- if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") )
- {
- groupHt[key] += 4; // explicit VR AND OB, OW, SQ : 4 more bytes
- }
- }
- groupHt[key] += 2 + 2 + 4 + elem->GetLength();
- }
+ if ( memcmp(dicm, "DICM", 4) == 0 )
+ {
+ HasDCMPreamble = true;
+ return Fp;
}
- unsigned short int gr_bid;
-
- for (GroupHT::iterator g = groupHt.begin(); // for each group we found
- g != groupHt.end();
- ++g)
- {
- // FIXME: g++ -Wall -Wstrict-prototypes reports on following line:
- // warning: unsigned int format, different type arg
- sscanf(g->first.c_str(),"%x",&gr_bid); //FIXME
- tk = g->first + "|0000"; // generate the element full tag
-
- if ( tagHT.count(tk) == 0) // if element 0x0000 not found
- {
- gdcmDictEntry * tagZ = new gdcmDictEntry(gr_bid, 0x0000, "UL");
- elemZ = new gdcmDocEntry(tagZ);
- elemZ->SetLength(4);
- AddEntry(elemZ); // create it
- }
- else
- {
- elemZ=GetDocEntryByNumber(gr_bid, 0x0000);
- }
- sprintf(trash ,"%d",g->second);
- str_trash=trash;
- //elemZ->SetValue(str_trash); // to go on compiling
- }
+ //-- Broken ACR or DICOM (?) with no Preamble; may start with a Shadow Group --
+ // FIXME : We cannot be sure the preable is only zeroes..
+ // (see ACUSON-24-YBR_FULL-RLE.dcm )
+ if (
+ zero == 0x0001 || zero == 0x0100 || zero == 0x0002 || zero == 0x0200 ||
+ zero == 0x0003 || zero == 0x0300 || zero == 0x0004 || zero == 0x0400 ||
+ zero == 0x0005 || zero == 0x0500 || zero == 0x0006 || zero == 0x0600 ||
+ zero == 0x0007 || zero == 0x0700 || zero == 0x0008 || zero == 0x0800 ||
+ zero == 0x0028 || 0x2800 // worse : some ACR-NEMA like files
+ // start 00028 group ?!?
+ )
+ {
+ std::string msg = Util::Format(
+ "ACR/DICOM starting by 0x(%04x) at the beginning of the file\n", zero);
+ // FIXME : is it a Warning message, or a Debug message?
+ gdcmWarningMacro( msg.c_str() );
+ return Fp;
+ }
+
+ // -- Neither ACR/No Preamble Dicom nor DICOMV3 file
+ CloseFile();
+ // Don't user Warning nor Error, not to pollute the output
+ // while directory recursive parsing ...
+ gdcmDebugMacro( "Neither ACR/No Preamble Dicom nor DICOMV3 file: "
+ << Filename.c_str());
+ return 0;
}
-*/
-
/**
- * \brief Writes in a file (according to the requested format)
- * the group, the element, the value representation and the length
- * of a single gdcmDocEntry passed as argument.
- * @param tag pointer on the gdcmDocEntry to be written
- * @param _fp already open file pointer
- * @param type type of the File to be written
+ * \brief closes the file
+ * @return TRUE if the close was successfull
*/
-void gdcmDocument::WriteEntryTagVRLength(gdcmDocEntry *tag,
- FILE *_fp,
- FileType type)
+bool Document::CloseFile()
{
- guint16 group = tag->GetGroup();
- std::string vr = tag->GetVR();
- guint16 el = tag->GetElement();
- guint32 lgr = tag->GetReadLength();
-
- if ( (group == 0xfffe) && (el == 0x0000) )
- // Fix in order to make some MR PHILIPS images e-film readable
- // see gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm:
- // we just *always* ignore spurious fffe|0000 tag !
- return;
-
- fwrite ( &group,(size_t)2 ,(size_t)1 ,_fp); //group
- fwrite ( &el,(size_t)2 ,(size_t)1 ,_fp); //element
-
- if ( type == gdcmExplicitVR ) {
-
- // Special case of delimiters:
- if (group == 0xfffe) {
- // Delimiters have NO Value Representation and have NO length.
- // Hence we skip writing the VR and length and we pad by writing
- // 0xffffffff
-
- int ff=0xffffffff;
- fwrite (&ff,(size_t)4 ,(size_t)1 ,_fp);
- return;
- }
-
- guint16 z=0;
- guint16 shortLgr = lgr;
- if (vr == "unkn") { // Unknown was 'written'
- // deal with Little Endian
- fwrite ( &shortLgr,(size_t)2 ,(size_t)1 ,_fp);
- fwrite ( &z, (size_t)2 ,(size_t)1 ,_fp);
- } else {
- fwrite (vr.c_str(),(size_t)2 ,(size_t)1 ,_fp);
- if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") )
- {
- fwrite ( &z, (size_t)2 ,(size_t)1 ,_fp);
- fwrite ( &lgr,(size_t)4 ,(size_t)1 ,_fp);
- } else {
- fwrite ( &shortLgr,(size_t)2 ,(size_t)1 ,_fp);
- }
- }
- }
- else // IMPLICIT VR
- {
- fwrite ( &lgr,(size_t)4 ,(size_t)1 ,_fp);
+ if ( Fp )
+ {
+ Fp->close();
+ delete Fp;
+ Fp = 0;
}
+ return true;
}
-
+
/**
- * \brief Writes in a file (according to the requested format)
- * the value of a single gdcmDocEntry passed as argument.
- * @param tag Pointer on the gdcmDocEntry to be written
- * @param _fp Already open file pointer
- * @param type type of the File to be written
+ * \brief Writes in a file all the Entries (Dicom Elements)
+ * @param fp file pointer on an already open file (actually: Output File Stream)
+ * @param filetype Type of the File to be written
+ * (ACR-NEMA, ExplicitVR, ImplicitVR)
*/
-
- // TODO : to be re -written recursively !
-
-void gdcmDocument::WriteEntryValue(gdcmDocEntry *tag, FILE *_fp,FileType type)
+void Document::WriteContent(std::ofstream *fp, FileType filetype, bool, bool)
{
- (void)type;
- guint16 group = tag->GetGroup();
- std::string vr = tag->GetVR();
- guint32 lgr = tag->GetReadLength();
+ // Skip if user wants to write an ACR-NEMA file
- if (vr == "SQ")
- // SeQuences have no value:
+ if ( filetype == ImplicitVR || filetype == ExplicitVR ||
+ filetype == JPEG || filetype == JPEG2000 )
+ {
+ // writing Dicom File Preamble
+ char filePreamble[128];
+ memset(filePreamble, 0, 128);
+ fp->write(filePreamble, 128);
+ 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);
+ *
+ * --> Computing group length for groups with embeded Sequences
+ * --> was too much tricky / we were [in a hurry / too lazy]
+ * --> We don't write the element 0x0000 (group length)
+ */
+ // This one is recursive
+ // false : outside MetaElements
+ // false : outside Sequence
+ ElementSet::WriteContent(fp, filetype, false, false);
+}
+
+// -----------------------------------------
+// Content entries
+/**
+ * \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 *docEntry = GetDocEntry(group, elem);
+ if ( !docEntry )
+ {
+ gdcmDebugMacro(std::hex << group << "|" << elem
+ << " doesn't exist" );
return;
- if (group == 0xfffe)
- // Delimiters have no associated value:
+ }
+ DataEntry *dataEntry = dynamic_cast<DataEntry *>(docEntry);
+ if ( !dataEntry )
+ {
+ gdcmWarningMacro(std::hex << group << "|" << elem
+ << " is NOT a DataEntry");
return;
-
- void *voidArea;
- // voidArea = tag->GetVoidArea();
- // to go on compiling
- if (voidArea != NULL)
- { // there is a 'non string' LUT, overlay, etc
- fwrite ( voidArea,(size_t)lgr ,(size_t)1 ,_fp); // Elem value
- return;
}
-
- if (vr == "US" || vr == "SS")
- {
- // some 'Short integer' fields may be mulivaluated
- // each single value is separated from the next one by '\'
- // we split the string and write each value as a short int
- std::vector<std::string> tokens;
- tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
- Tokenize (tag->GetValue(), tokens, "\\");
- for (unsigned int i=0; i<tokens.size();i++)
- {
- guint16 val_uint16 = atoi(tokens[i].c_str());
- void *ptr = &val_uint16;
- fwrite ( ptr,(size_t)2 ,(size_t)1 ,_fp);
- }
- tokens.clear();
+ LoadEntryBinArea(dataEntry);
+}
+
+/**
+ * \brief Loads (from disk) the element content
+ * when a string is not suitable
+ * @param entry Entry whose binArea is going to be loaded
+ */
+void Document::LoadEntryBinArea(DataEntry *entry)
+{
+ if( entry->GetBinArea() )
+ return;
+
+ bool openFile = !Fp;
+ if ( openFile )
+ OpenFile();
+
+ //size_t o =(size_t)entry->GetOffset();
+ Fp->seekg((size_t)entry->GetOffset(), std::ios::beg); // FIXME : for each DataEntry !
+
+ size_t l = entry->GetLength();
+ uint8_t *data = new uint8_t[l];
+ if ( !data )
+ {
+ gdcmWarningMacro( "Cannot allocate DataEntry content for : "
+ << std::hex << entry->GetGroup()
+ << "|" << entry->GetElement() );
return;
}
- // some 'Integer' fields may be mulivaluated
- // each single value is separated from the next one by '\'
- // we split the string and write each value as an int
- if (vr == "UL" || vr == "SL")
+
+ // Read the data
+ Fp->read((char*)data, l);
+ if ( Fp->fail() || Fp->eof() )
{
- std::vector<std::string> tokens;
- tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
- Tokenize (tag->GetValue(), tokens, "\\");
- for (unsigned int i=0; i<tokens.size();i++)
+ delete[] data;
+ entry->SetState(DataEntry::STATE_UNREAD);
+ return;
+ }
+
+ // Swap the data content if necessary
+ uint32_t i;
+ unsigned short vrLgth =
+ Global::GetVR()->GetAtomicElementLength(entry->GetVR());
+
+// FIXME : trouble expected if we read an ... OW Entry (LUT, etc ..)
+// if( entry->GetVR() == "OW" )
+// vrLgth = 1;
+
+ switch(vrLgth)
+ {
+ case 1:
+ {
+ break;
+ }
+ case 2:
+ {
+ uint16_t *data16 = (uint16_t *)data;
+ for(i=0;i<l/vrLgth;i++)
+ data16[i] = SwapShort(data16[i]);
+ break;
+ }
+ case 4:
{
- guint32 val_uint32 = atoi(tokens[i].c_str());
- void *ptr = &val_uint32;
- fwrite ( ptr,(size_t)4 ,(size_t)1 ,_fp);
+ uint32_t *data32 = (uint32_t *)data;
+ for(i=0;i<l/vrLgth;i++)
+ data32[i] = SwapLong(data32[i]);
+ break;
}
- tokens.clear();
- return;
- }
- fwrite (tag->GetValue().c_str(), (size_t)lgr ,(size_t)1, _fp); // Elem value
+ case 8:
+ {
+ double *data64 = (double *)data;
+ for(i=0;i<l/vrLgth;i++)
+ data64[i] = SwapDouble(data64[i]);
+ break;
+ }
+ }
+
+ entry->SetBinArea(data);
+
+ if ( openFile ) // The file is left in the state (open/close) it was at entrance
+ CloseFile();
}
/**
- * \brief Writes in a file (according to the requested format)
- * a single gdcmDocEntry passed as argument.
- * \sa WriteEntryValue, WriteEntryTagVRLength.
- * @param tag Pointer on the gdcmDocEntry to be written
- * @param _fp Already open file pointer
- * @param type type of the File to be written
+ * \brief Loads the element while preserving the current
+ * underlying file position indicator as opposed to
+ * LoadDocEntry that modifies it
+ * \note seems to be unused!.
+ * @param entry DocEntry whose value will be loaded.
*/
+//void Document::LoadDocEntrySafe(DocEntry *entry)
+//{
+// if ( Fp )
+// {
+// long PositionOnEntry = Fp->tellg(); // LoadDocEntrySafe is not used
+// LoadDocEntry(entry);
+// Fp->seekg(PositionOnEntry, std::ios::beg); // LoadDocEntrySafe is not used
+// }
+//}
-bool gdcmDocument::WriteEntry(gdcmDocEntry *tag, FILE *_fp,FileType type)
+/**
+ * \brief Compares two documents, according to DicomDir rules
+ * \warning Does NOT work with ACR-NEMA files
+ * \todo Find a trick to solve the pb (use RET fields ?)
+ * @param document to compare with current one
+ * @return true if 'smaller'
+ */
+bool Document::operator<(Document &document)
{
- guint32 length = tag->GetLength();
-
- // The value of a tag MUST (see the DICOM norm) be an odd number of
- // bytes. When this is not the case, pad with an additional byte:
- if(length%2==1)
- {
-// tag->SetValue(tag->GetValue()+"\0");
-// to go on compiling
- tag->SetLength(tag->GetReadLength()+1);
+ // Patient Name
+ std::string s1 = GetEntryString(0x0010,0x0010);
+ std::string s2 = document.GetEntryString(0x0010,0x0010);
+ if (s1 < s2)
+ {
+ return true;
}
-
- WriteEntryTagVRLength(tag, _fp, type);
- WriteEntryValue(tag, _fp, type);
- return true;
+ else if ( s1 > s2 )
+ {
+ return false;
+ }
+ else
+ {
+ // Patient ID
+ s1 = GetEntryString(0x0010,0x0020);
+ s2 = document.GetEntryString(0x0010,0x0020);
+ if ( s1 < s2 )
+ {
+ return true;
+ }
+ else if ( s1 > s2 )
+ {
+ return false;
+ }
+ else
+ {
+ // Study Instance UID
+ s1 = GetEntryString(0x0020,0x000d);
+ s2 = document.GetEntryString(0x0020,0x000d);
+ if ( s1 < s2 )
+ {
+ return true;
+ }
+ else if ( s1 > s2 )
+ {
+ return false;
+ }
+ else
+ {
+ // Serie Instance UID
+ s1 = GetEntryString(0x0020,0x000e);
+ s2 = document.GetEntryString(0x0020,0x000e);
+ if ( s1 < s2 )
+ {
+ return true;
+ }
+ else if ( s1 > s2 )
+ {
+ return false;
+ }
+ }
+ }
+ }
+ return false;
}
+//-----------------------------------------------------------------------------
+// Protected
+
/**
- * \brief writes on disc according to the requested format
- * (ACR-NEMA, ExplicitVR, ImplicitVR) the image
- * using the Chained List
- * \warning does NOT add the missing elements in the header :
- * it's up to the user doing it !
- * (function CheckHeaderCoherence to be written)
- * \warning DON'T try, right now, to write a DICOM image
- * from an ACR Header (meta elements will be missing!)
- * \sa WriteEntriesDeprecated (Special temporary method for Theralys)
- * @param type type of the File to be written
- * (ACR-NEMA, ExplicitVR, ImplicitVR)
- * @param _fp already open file pointer
+ * \brief Reads a given length of bytes
+ * (in order to avoid to many CPU time-consuming fread-s)
+ * @param l length to read
*/
-
-bool gdcmDocument::WriteEntries(FILE *_fp,FileType type)
+void Document::ReadBegBuffer(size_t l)
+ throw( FormatError )
{
- /// \todo (?) check write failures (after *each* fwrite)
-
- for (ListTag::iterator tag2=listEntries.begin();
- tag2 != listEntries.end();
- ++tag2)
+ Fp->read (BegBuffer, (size_t)l);
+ if ( Fp->fail() )
{
- if ( type == gdcmACR ){
- if ((*tag2)->GetGroup() < 0x0008)
- // Ignore pure DICOM V3 groups
- continue;
- if ((*tag2)->GetElement() %2)
- // Ignore the "shadow" groups
- continue;
- if ((*tag2)->GetVR() == "SQ" ) // ignore Sequences
- continue;
- if ((*tag2)->GetSQDepthLevel() != 0) // Not only ignore the SQ element
- continue;
- }
- if (! WriteEntry(*tag2,_fp,type) )
- return false;
+ throw FormatError( "Document::ReadBegBuffer()", " file error." );
}
- return true;
-}
+ if ( Fp->eof() )
+ {
+ throw FormatError( "Document::ReadBegBuffer()", "EOF." );
+ }
+ PtrBegBuffer = BegBuffer;
+ CurrentOffsetPosition+=l;
+}
+/**
+ * \brief Reads a supposed to be 16 Bits integer
+ * (swaps it depending on processor endianness)
+ * @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 writes on disc according to the requested format
- * (ACR-NEMA, ExplicitVR, ImplicitVR) the image,
- * using only the last synonym of each mutimap H Table post.
- * \warning Uses the H Table, instead of the Chained List
- * in order to be compliant with the old way to proceed
- * (added elements taken in to account)
- * Only THERALYS, during a transitory phase is supposed
- * to use this method !!!
- * \warning DON'T try, right now, to write a DICOM image
- * from an ACR Header (meta elements will be missing!)
- * \sa WriteEntries
- * @param _fp already open file pointer
- * @param type type of the File to be written
- * (ACR-NEMA, ExplicitVR, ImplicitVR)
+ * \brief Gets from BegBuffer a supposed to be 16 Bits integer
+ * (swaps it depending on processor endianness)
+ * @return read value
*/
-void gdcmDocument::WriteEntriesDeprecated(FILE *_fp,FileType type) {
-
- // restent a tester les echecs en ecriture (apres chaque fwrite)
-
- for (TagDocEntryHT::iterator tag2=tagHT.begin();
- tag2 != tagHT.end();
- ++tag2){
- if ( type == gdcmACR ){
- if ((*tag2->second).GetGroup() < 0x0008) continue; // ignore pure DICOM V3 groups
- if ((*tag2->second).GetElement() %2) continue; // ignore shadow groups
- if ((*tag2->second).GetVR() == "SQ" ) continue; // ignore Sequences
- if ((*tag2->second).GetSQDepthLevel() != 0) continue; // Not only ignore the SQ element
- }
- if ( ! WriteEntry(tag2->second,_fp,type))
- break;
+uint16_t Document::GetInt16()
+{
+ uint16_t g = *((uint16_t*)PtrBegBuffer);
+ g = SwapShort(g);
+ PtrBegBuffer+=2;
+ return g;
+}
+/**
+ * \brief Reads a supposed to be 32 Bits integer
+ * (swaps it depending on processor endianness)
+ * @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 Swaps back the bytes of 4-byte long integer accordingly to
- * processor order.
- * @return The properly swaped 32 bits integer.
+ * \brief Gets from BegBuffer a supposed to be 32 Bits integer
+ * (swaps it depending on processor endianness)
+ * @return read value
*/
-guint32 gdcmDocument::SwapLong(guint32 a) {
- switch (sw) {
- case 0 :
- 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;
+uint32_t Document::GetInt32()
+{
+ uint32_t g = *((uint32_t*)PtrBegBuffer);
+ g = SwapLong(g);
+ PtrBegBuffer+=4;
+ return g;
+}
+
+/**
+ * \brief Re-computes the length of the Dicom group 0002.
+ */
+int Document::ComputeGroup0002Length( )
+{
+ uint16_t gr;
+ VRKey vr;
- case 2143 :
- a=( ((a<<8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
+ int groupLength = 0;
+ bool found0002 = false;
+
+ // for each zero-level Tag in the DCM Header
+ DocEntry *entry = GetFirstEntry();
+ while( entry )
+ {
+ gr = entry->GetGroup();
+
+ if ( gr == 0x0002 )
+ {
+ found0002 = true;
+
+ if ( entry->GetElement() != 0x0000 )
+ {
+ vr = entry->GetVR();
+
+ //if ( (vr == "OB")||(vr == "OW")||(vr == "UT")||(vr == "SQ"))
+ // (no SQ, OW, OL, UT in group 0x0002;)
+ if ( vr == "OB" )
+ {
+ // explicit VR AND (OB, OW, OL, SQ, UT, UN) : 4 more bytes
+ groupLength += 4;
+ }
+ groupLength += 2 + 2 + 4 + entry->GetLength();
+ }
+ }
+ else if (found0002 )
break;
- default :
- std::cout << "swapCode= " << sw << std::endl;
- dbg.Error(" gdcmDocument::SwapLong : unset swap code");
- a=0;
+
+ entry = GetNextEntry();
}
- return(a);
+ return groupLength;
}
/**
- * \brief Unswaps back the bytes of 4-byte long integer accordingly to
- * processor order.
- * @return The properly unswaped 32 bits integer.
+ * \brief CallStartMethod
*/
-guint32 gdcmDocument::UnswapLong(guint32 a) {
- return (SwapLong(a));
+void Document::CallStartMethod()
+{
+ Progress = 0.0f;
+ Abort = false;
+ CommandManager::ExecuteCommand(this,CMD_STARTPROGRESS);
}
/**
- * \brief Swaps the bytes so they agree with the processor order
- * @return The properly swaped 16 bits integer.
+ * \brief CallProgressMethod
*/
-guint16 gdcmDocument::SwapShort(guint16 a) {
- if ( (sw==4321) || (sw==2143) )
- a =(((a<<8) & 0x0ff00) | ((a>>8)&0x00ff));
- return (a);
+void Document::CallProgressMethod()
+{
+ CommandManager::ExecuteCommand(this,CMD_PROGRESS);
}
/**
- * \brief Unswaps the bytes so they agree with the processor order
- * @return The properly unswaped 16 bits integer.
+ * \brief CallEndMethod
*/
-guint16 gdcmDocument::UnswapShort(guint16 a) {
- return (SwapShort(a));
+void Document::CallEndMethod()
+{
+ Progress = 1.0f;
+ CommandManager::ExecuteCommand(this,CMD_ENDPROGRESS);
}
//-----------------------------------------------------------------------------
// Private
-
-
+/**
+ * \brief Loads all the needed Dictionaries
+ * \warning NOT end user intended method !
+ */
+void Document::Initialize()
+{
+ RefPubDict = Global::GetDicts()->GetDefaultPubDict();
+ RefShaDict = NULL;
+ Filetype = Unknown;
+}
/**
* \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries)
- * @return false if file is not ACR-NEMA / PAPYRUS / DICOM
+ * @param set DocEntrySet we are going to parse ('zero level' or a SQItem)
+ * @param offset start of parsing
+ * @param l_max length to parse (meaningless when we are in 'delimitor mode')
+ * @param delim_mode : whether we are in 'delimitor mode' (l=0xffffff) or not
*/
+void Document::ParseDES(DocEntrySet *set, long offset,
+ long l_max, bool delim_mode)
+{
+ DocEntry *newDocEntry;
+ DataEntry *newDataEntry;
+ SeqEntry *newSeqEntry;
+ //VRKey vr;
+ bool used; // will be set to false when something wrong happens to an Entry.
+ // (Entry will then be deleted)
+ bool delim_mode_intern = delim_mode;
+ bool first = true;
+ gdcmDebugMacro( "Enter in ParseDES, delim-mode " << delim_mode
+ << " at offset " << std::hex << "0x(" << offset << ")" );
+ while (true)
+ {
+
+ ///\todo FIXME : On 64 bits processors, tellg gives unexpected results after a while ?
+ /// Probabely a bug in gdcm code somewhere (some memory erased ?)
-long gdcmDocument::ParseDES(gdcmDocEntrySet *set, long offset, long l_max, bool delim_mode) {
-
- gdcmDocEntry *NewDocEntry = (gdcmDocEntry *)0;
- gdcmValEntry *vl;
- gdcmBinEntry *bn;
- gdcmSeqEntry *sq;
- string vr;
- long l;
- while (true) {
- NewDocEntry = ReadNextDocEntry( );
- if (!NewDocEntry)
+// Uncomment to track the bug
+
+ if( Debug::GetDebugFlag() )
+ std::cout << std::dec <<"(long)(Fp->tellg()) " << (long)(Fp->tellg()) // in Debug mode
+ << std::hex << " 0x(" <<(long)(Fp->tellg()) << ")" << std::endl;
+
+
+ // if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) // Once per DocEntry
+ if ( !delim_mode ) // 'and then' doesn't exist in C++ :-(
+ if ( ((long)(Fp->tellg())-offset) >= l_max) // Once per DocEntry, when no delim mode
+ {
+ break;
+ }
+
+ newDocEntry = ReadNextDocEntry( );
+
+ if ( !newDocEntry )
+ {
break;
-
- std::cout << hex << NewDocEntry->GetGroup()
- << " "
- << NewDocEntry->GetElement()
- << " "
- << NewDocEntry->GetVR()
- << " "
- << NewDocEntry->GetReadLength()
- << " "
- << NewDocEntry->GetOffset()
- << std::endl;
-
- vr = NewDocEntry->GetVR();
- if (vr!="SQ") {
-
- if (vr == "AE" || vr == "AS" || vr == "DA" || vr == "PN" ||
- vr == "UI" || vr == "TM" || vr == "SH" || vr == "LO" ||
- vr == "CS" || vr == "IS" || vr == "LO" || vr == "LT" ||
- vr == "SH" || vr == "ST" ||
- vr == "SL" || vr == "SS" || vr == "UL" || vr == "US"
- ) {
- // --- ValEntry
- vl= new gdcmValEntry(NewDocEntry->GetDictEntry());
- vl->Copy(NewDocEntry);
- set->AddEntry(vl);
- LoadDocEntry(vl);
- std::cout << "value [" << vl->GetValue() << "]" << std::endl;
- if (!delim_mode && vl->isItemDelimitor())
- break;
- } else { // BinEntry
-
- // Hope the following VR *do* correspond to a BinEntry
-
- //AT Attribute Tag; // 2 16-bit unsigned short integers
- //FL Floating Point Single; // 32-bit IEEE 754:1985 float
- //FD Floating Point Double; // 64-bit IEEE 754:1985 double
- //UN Unknown; // Any length of bytes
- //UT Unlimited Text; // At most 2^32 -1 chars
- //OB Other Byte String; // String of bytes (VR independant)
- //OW Other Word String; // String of 16-bit words (VR dependant)
-
- bn = new gdcmBinEntry(NewDocEntry->GetDictEntry());
- bn->Copy(NewDocEntry);
- LoadDocEntry(bn);
- std::cout << "value [" << "Bin Entry, in voidArea" << "]" << std::endl;
- }
-
- SkipToNextDocEntry(NewDocEntry); // to be sure to be at the beginning
- l = NewDocEntry->GetFullLength();
-
- } else { // VR = "SQ"
-
- //SkipDocEntry(NewDocEntry);
+ }
- std::cout << "gdcmDocument::ParseDES : SQ found " << std::endl;
- l=NewDocEntry->GetReadLength();
- if (l == 0xffffffff)
- delim_mode = true;
+ // Uncoment this cerr line to be able to 'follow' the DocEntries
+ // when something *very* strange happens
+ if( Debug::GetDebugFlag() )
+ std::cerr<<newDocEntry->GetKey()<<" "<<newDocEntry->GetVR()<<std::endl;
+
+ // an Item Starter found elsewhere but in the first position
+ // of a SeqEntry means previous entry was a Sequence
+ // but we didn't get it (private Sequence + Implicit VR)
+ // we have to backtrack.
+ if ( !first && newDocEntry->IsItemStarter() )
+ {
+ // Debug message within the method !
+ newDocEntry = Backtrack(newDocEntry, set);
+ }
+ else
+ {
+ PreviousDocEntry = newDocEntry;
+ }
+
+ used = true;
+ newDataEntry = dynamic_cast<DataEntry*>(newDocEntry);
+
+ if ( newDataEntry )
+ {
+ //////////////////////////// DataEntry
+
+ //vr = newDocEntry->GetVR(); // useless ?
+
+ if ( !set->AddEntry( newDataEntry ) )
+ {
+ gdcmDebugMacro( "in ParseDES : cannot add a DataEntry "
+ << newDataEntry->GetKey()
+ << " (at offset : 0x("
+ << newDataEntry->GetOffset() << ") )" );
+ used=false;
+ }
else
- delim_mode = false;
- sq = new gdcmSeqEntry(NewDocEntry->GetDictEntry());
- sq->Copy(NewDocEntry);
- sq->SetDelimitorMode(delim_mode);
- long lgt = ParseSQ((gdcmDocEntrySet *)sq, offset, l, delim_mode);
- // FIXME : on en fait quoi, de lgt ?
- set->AddEntry(sq);
- if ( !delim_mode && ftell(fp)-offset >= l_max)
+ {
+ newDataEntry->Delete();
+ // Load only if we can add (not a duplicate key)
+ LoadDocEntry( newDataEntry );
+ }
+ if ( newDataEntry->GetElement() == 0x0000 ) // if on group length
+ {
+ if ( newDataEntry->GetGroup()%2 != 0 ) // if Shadow Group
+ {
+ if ( LoadMode & LD_NOSHADOW ) // if user asked to skip shad.gr
+ {
+ std::string strLgrGroup = newDataEntry->GetString();
+
+ int lgrGroup;
+ //if ( newDataEntry->IsUnfound() ) /?!? JPR
+ {
+ lgrGroup = atoi(strLgrGroup.c_str());
+ Fp->seekg(lgrGroup, std::ios::cur); // Once per Shadow group, when NOSHADOW
+ RemoveEntry( newDocEntry ); // Remove and delete
+ continue;
+ }
+ }
+ }
+ }
+
+ bool delimitor = newDataEntry->IsItemDelimitor();
+ bool outOfBounds = false;
+ if (!delim_mode )
+ if ( ((long)(Fp->tellg())-offset) >= l_max ) //Once per DataEntry when no delim mode
+ outOfBounds = true;
+
+ // 'and then', 'or else' don't exist in C++ :-(
+ // if ( (delimitor) ||
+ // (!delim_mode && ((long)(Fp->tellg())-offset) >= l_max) ) // Once per DataEntry
+
+ if ( delimitor || outOfBounds )
+ {
+ if ( !used )
+ newDocEntry->Delete();
break;
- }
- }
- cout << endl;
- delete NewDocEntry;
- return l; // ??
-}
+ }
+
+ // Just to make sure we are at the beginning of next entry.
+ SkipToNextDocEntry(newDocEntry); // FIXME : once per DocEntry, segfault if commented out
+ }
+ else
+ {
+ /////////////////////// SeqEntry : VR = "SQ"
+
+ unsigned long l = newDocEntry->GetReadLength();
+ if ( l != 0 ) // don't mess the delim_mode for 'zero-length sequence'
+ {
+ if ( l == 0xffffffff )
+ {
+ delim_mode_intern = true;
+ }
+ else
+ {
+ delim_mode_intern = false;
+ }
+ }
+ if ( (LoadMode & LD_NOSHADOWSEQ) && ! delim_mode_intern )
+ {
+ // User asked to skip SeQuences *only* if they belong to Shadow Group
+ if ( newDocEntry->GetGroup()%2 != 0 )
+ {
+ Fp->seekg( l, std::ios::cur); // once per SQITEM, when NOSHADOWSEQ
+ newDocEntry->Delete(); // Delete, not in the set
+ continue;
+ }
+ }
+ if ( (LoadMode & LD_NOSEQ) && ! delim_mode_intern )
+ {
+ // User asked to skip *any* SeQuence
+ Fp->seekg( l, std::ios::cur); // Once per SQ, when NOSEQ
+ newDocEntry->Delete(); // Delete, not in the set
+ continue;
+ }
+ // delay the dynamic cast as late as possible
+ newSeqEntry = dynamic_cast<SeqEntry*>(newDocEntry);
+
+ // no other way to create the Delimitor ...
+ newSeqEntry->SetDelimitorMode( delim_mode_intern );
+
+ // At the top of the hierarchy, stands a Document. When "set"
+ // is a Document, then we are building the first depth level.
+ // Hence the SeqEntry we are building simply has a depth
+ // level of one:
+ if ( set == this ) // ( dynamic_cast< Document* > ( set ) )
+ {
+ newSeqEntry->SetDepthLevel( 1 );
+ }
+ // 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:
+ // time waste hunting
+ else if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) )
+ {
+ newSeqEntry->SetDepthLevel( parentSQItem->GetDepthLevel() + 1 );
+ }
-/**
- * \brief Parses a Sequence ( SeqEntry after SeqEntry)
- * @return parsed length for this level
- */
+ if ( l != 0 )
+ { // Don't try to parse zero-length sequences
+ gdcmDebugMacro( "Entry in ParseSQ, delim " << delim_mode_intern
+ << " at offset 0x(" << std::hex
+ << newDocEntry->GetOffset() << ")");
-long gdcmDocument::ParseSQ(gdcmDocEntrySet *set, long offset, long l_max, bool delim_mode) {
+ bool res = ParseSQ( newSeqEntry,
+ newDocEntry->GetOffset(),
+ l, delim_mode_intern);
- std::cout << "Entree ds gdcmDocument::ParseSQ" << std::endl;
+ gdcmDebugMacro( "Exit from ParseSQ, delim " << delim_mode_intern << " -->return : " << res);
+ }
+ if ( !set->AddEntry( newSeqEntry ) )
+ {
+ gdcmWarningMacro( "in ParseDES : cannot add a SeqEntry "
+ << newSeqEntry->GetKey()
+ << " (at offset : 0x("
+ << newSeqEntry->GetOffset() << ") )" );
+ used = false;
+ }
+ else
+ {
+ newDocEntry->Delete();
+ }
- gdcmDocEntry *NewDocEntry = (gdcmDocEntry *)0;
- gdcmSQItem *itemSQ;
- bool dlm_mod;
- int lgr, l, lgth;
- cout << "=============== on entre ds une Sequence" <<endl;
-
- while (true) {
- std::cout << "gdcmDocument::ParseSQ on itere" << std::endl;
-
- if(delim_mode) {
- NewDocEntry = ReadNextDocEntry();
- if (NewDocEntry->isSequenceDelimitor()) {
- // add the Sequence Delimitor // TODO : find the trick to put it proprerly !
- // ((gdcmSeqEntry *)set)->SetSequenceDelimitationItem(NewDocEntry);
- break;
- }
- }
- if (!delim_mode && (ftell(fp)-offset) >= l_max){
- break;
- }
- itemSQ = new gdcmSQItem();
- itemSQ->AddEntry(NewDocEntry); // no value, no voidArea. Think of it while printing !
- l= NewDocEntry->GetLength();
- if (l ==0xffffffff)
- dlm_mod = true;
- else
- dlm_mod=false;
+ // if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) // Once per SeqEntry
+
+ if ( !delim_mode ) // 'and then' doesn't exist in C++ :-(
+ if ( ((long)(Fp->tellg())-offset) >= l_max) // Once per SeqEntry when no delim mode
+
+ {
+ if ( !used )
+ newDocEntry->Delete();
+ break;
+ }
+ } // end SeqEntry : VR = "SQ"
+
+ if ( !used )
+ {
+ newDocEntry->Delete();
+ }
+ first = false;
- lgr=ParseDES(itemSQ, offset, l, dlm_mod);
- ((gdcmSeqEntry *)set)->AddEntry(itemSQ);
- }
- //Update(lgth);
- lgth = ftell(fp) - offset;
- cout << "=============== lgr Sequence :" << lgth <<endl;
- return(lgth);
+ if (UnexpectedEOF) // some terminator was missing
+ {
+ break;
+ }
+ } // end While
+ gdcmDebugMacro( "Exit from ParseDES, delim-mode " << delim_mode );
}
-
/**
- * \brief Parses the header of the file and load element values.
- * @return false if file is not ACR-NEMA / PAPYRUS / DICOM
- */
+ * \brief Parses a Sequence ( SeqEntry after SeqEntry)
+ * @return false if expected fff0,e000 not found
+ */
+bool Document::ParseSQ( SeqEntry *seqEntry,
+ long offset, long l_max, bool delim_mode)
+{
+ int SQItemNumber = 0;
+ bool dlm_mod;
+ long offsetStartCurrentSQItem = offset;
-/* =======================================================================
-// Replaced by ParseSQ
-
-bool gdcmDocument::LoadHeaderEntries(bool exception_on_error) throw(gdcmFormatError) {
- (void)exception_on_error;
- rewind(fp);
- if (!CheckSwap())
- return false;
+ while (true)
+ {
+ // the first time, we read the fff0,e000 of the first SQItem
+ DocEntry *newDocEntry = ReadNextDocEntry();
+
+ if ( !newDocEntry )
+ {
+ // The most frequent is when a SQ terminator is missing (?!?)
+ gdcmWarningMacro("in ParseSQ : should never get here!");
+ UnexpectedEOF = true;
+ return false;
+ }
+ if ( delim_mode )
+ {
+ if ( newDocEntry->IsSequenceDelimitor() )
+ {
+ seqEntry->SetDelimitationItem( newDocEntry );
+ newDocEntry->Delete();
+ break;
+ }
+ }
+ else // ! delim_mode
+ {
+ if ( ((long)(Fp->tellg())-offset) >= l_max) // Once per SQItem when no delim mode
+ {
+ newDocEntry->Delete();
+ break;
+ }
+ }
+ // create the current SQItem
+ SQItem *itemSQ = SQItem::New( seqEntry->GetDepthLevel() );
+ unsigned int l = newDocEntry->GetReadLength();
- gdcmDocEntry *newDocEntry = (gdcmDocEntry *)0;
- while ( (newDocEntry = ReadNextDocEntry()) ) {
- SkipDocEntry(newDocEntry);
- if ( (ignoreShadow==0) || (newDocEntry->GetGroup()%2) == 0) {
- AddEntry(newDocEntry);
- }
- }
- rewind(fp);
- // Be carefull : merging this two loops may cause troubles ...
- for (ListTag::iterator i = GetListEntry().begin();
- i != GetListEntry().end();
- ++i)
- {
- LoadDocEntry(*i);
- }
- rewind(fp);
-
- // Load 'non string' values
+ if ( l == 0xffffffff )
+ {
+ dlm_mod = true;
+ }
+ else
+ {
+ dlm_mod = false;
+ }
+
+ // avoid infinite loop when Bad assumption was made on illegal 'unknown length' UN //JPRx
+
+ if (offsetStartCurrentSQItem <= OffsetOfPreviousParseDES)
+ {
+ gdcmWarningMacro("Bad assumption was made on illegal 'unknown length' UN!" << std::endl <<
+ "OffsetOfPreviousParseDES " << std::hex << OffsetOfPreviousParseDES
+ << " offsetStartCurrentSQItem " << offsetStartCurrentSQItem);
+ /// \todo when "Bad assumption (SQ) on illegal 'unknown length' UN", Backtrack again + try OB
+ return false;
+ }
+ else
+ {
+ OffsetOfPreviousParseDES = offsetStartCurrentSQItem;
+ }
+
+ // fill up the current SQItem, starting at the beginning of fff0,e000
+ Fp->seekg(offsetStartCurrentSQItem, std::ios::beg); // Once per SQItem
+ ParseDES(itemSQ, offsetStartCurrentSQItem, l+8, dlm_mod);
+ offsetStartCurrentSQItem = Fp->tellg(); // Once per SQItem
-
- // --------------------------------------------------------------
- // Special Patch to allow gdcm to read ACR-LibIDO formated images
- //
- // if recognition code tells us we deal with a LibIDO image
- // we switch lineNumber and columnNumber
- //
- std::string RecCode;
- RecCode = GetEntryByNumber(0x0008, 0x0010); // recognition code
- if (RecCode == "ACRNEMA_LIBIDO_1.1" ||
- RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares
- // with "little-endian strings"
- {
- filetype = ACR_LIBIDO;
- std::string rows = GetEntryByNumber(0x0028, 0x0010);
- std::string columns = GetEntryByNumber(0x0028, 0x0011);
- SetEntryByNumber(columns, 0x0028, 0x0010);
- SetEntryByNumber(rows , 0x0028, 0x0011);
- }
- // ----------------- End of Special Patch ----------------
+ seqEntry->AddSQItem( itemSQ, SQItemNumber );
+ itemSQ->Delete();
+ newDocEntry->Delete();
+ SQItemNumber++;
+ //if ( !delim_mode && ((long)(Fp->tellg())-offset ) >= l_max ) //JPRx
+ if ( !delim_mode && (offsetStartCurrentSQItem-offset ) >= l_max )
+ {
+ break;
+ }
+ }
return true;
}
-======================================================================= */
+/**
+ * \brief When a private Sequence + Implicit VR is encountered
+ * we cannot guess it's a Sequence till we find the first
+ * Item Starter. We then backtrack to do the job.
+ * @param docEntry Item Starter that warned us
+ * @param set DocEntrySet (ElementSet/SQItem) the DocEntry will belong
+ */
+DocEntry *Document::Backtrack(DocEntry *docEntry, DocEntrySet *set)
+{
+ // delete the Item Starter, built erroneously out of any Sequence
+ // it's not yet in the HTable/chained list
+ docEntry->Delete();
+
+ // Get all info we can from PreviousDocEntry
+ uint16_t group = PreviousDocEntry->GetGroup();
+ uint16_t elem = PreviousDocEntry->GetElement();
+ uint32_t lgt = PreviousDocEntry->GetLength();
+ long offset = PreviousDocEntry->GetOffset();
+
+ gdcmDebugMacro( "Backtrack :" << std::hex << group
+ << "|" << elem
+ << " at offset 0x(" <<offset << ")" );
+
+ set->RemoveEntry( PreviousDocEntry );
+
+ // forge the Seq Entry
+ DocEntry *newEntry = NewSeqEntry(group, elem);
+ newEntry->SetLength(lgt);
+ newEntry->SetOffset(offset);
+
+ // Move back to the beginning of the Sequence
+
+ Fp->seekg(offset, std::ios::beg); // Only for Shadow Implicit VR SQ
+ return newEntry; // It will added where it has to be!
+}
/**
- * \brief Loads the element content if its length doesn't exceed
- * the value specified with gdcmDocument::SetMaxSizeLoadEntry()
- * @param Entry Header Entry (Dicom Element) to be dealt with
+ * \brief Loads (or not) the element content depending if its length exceeds
+ * or not the value specified with Document::SetMaxSizeLoadEntry()
+ * @param entry Header Entry (Dicom Element) to be dealt with
+ * @param forceLoad whether you want to force loading of 'long' elements
*/
-void gdcmDocument::LoadDocEntry(gdcmDocEntry *Entry) {
- size_t item_read;
- guint16 group = Entry->GetGroup();
- std::string vr= Entry->GetVR();
- guint32 length = Entry->GetLength();
+void Document::LoadDocEntry(DocEntry *entry, bool forceLoad)
+{
+ uint16_t group = entry->GetGroup();
+ uint16_t elem = entry->GetElement();
+ const VRKey &vr = entry->GetVR();
+ uint32_t length = entry->GetLength();
- fseek(fp, (long)Entry->GetOffset(), SEEK_SET);
+ // Fp->seekg((long)entry->GetOffset(), std::ios::beg); // JPRx
// A SeQuence "contains" a set of Elements.
// (fffe e000) tells us an Element is beginning
// (fffe e00d) tells us an Element just ended
// (fffe e0dd) tells us the current SeQuence just ended
- if( group == 0xfffe ) {
+ // (fffe 0000) is an 'impossible' tag value,
+ // found in MR-PHILIPS-16-Multi-Seq.dcm
+
+ if ( (group == 0xfffe && elem != 0x0000 ) || vr == "SQ" )
+ {
// NO more value field for SQ !
- //Entry->SetValue("gdcm::Skipped");
- // appel recursif de Load Value
- // (meme pb que pour le parsing)
return;
}
- // When the length is zero things are easy:
- if ( length == 0 ) {
- ((gdcmValEntry *)Entry)->SetValue("");
+ DataEntry *dataEntryPtr = dynamic_cast< DataEntry* >(entry);
+ if( !dataEntryPtr )
+ {
return;
}
- // The elements whose length is bigger than the specified upper bound
- // are not loaded. Instead we leave a short notice of the offset of
- // the element content and it's length.
- if (length > MaxSizeLoadEntry) {
- std::ostringstream s;
- s << "gdcm::NotLoaded.";
- s << " Address:" << (long)Entry->GetOffset();
- s << " Length:" << Entry->GetLength();
- s << " x(" << std::hex << Entry->GetLength() << ")";
- ((gdcmValEntry *)Entry)->SetValue(s.str());
+ // When the length is zero things are easy:
+ if ( length == 0 )
+ {
+ dataEntryPtr->SetBinArea(NULL,true);
return;
}
-
- // Any compacter code suggested (?)
- if ( IsDocEntryAnInteger(Entry) ) {
- guint32 NewInt;
- std::ostringstream s;
- int nbInt;
- // When short integer(s) are expected, read and convert the following
- // n *two characters properly i.e. as short integers as opposed to strings.
- // Elements with Value Multiplicity > 1
- // contain a set of integers (not a single one)
- if (vr == "US" || vr == "SS") {
- nbInt = length / 2;
- NewInt = ReadInt16();
- s << NewInt;
- if (nbInt > 1){
- for (int i=1; i < nbInt; i++) {
- s << '\\';
- NewInt = ReadInt16();
- s << NewInt;
- }
- }
- }
- // When integer(s) are expected, read and convert the following
- // n * four characters properly i.e. as integers as opposed to strings.
- // Elements with Value Multiplicity > 1
- // contain a set of integers (not a single one)
- else if (vr == "UL" || vr == "SL") {
- nbInt = length / 4;
- NewInt = ReadInt32();
- s << NewInt;
- if (nbInt > 1) {
- for (int i=1; i < nbInt; i++) {
- s << '\\';
- NewInt = ReadInt32();
- s << NewInt;
- }
- }
- }
-#ifdef GDCM_NO_ANSI_STRING_STREAM
- s << std::ends; // to avoid oddities on Solaris
-#endif //GDCM_NO_ANSI_STRING_STREAM
- ((gdcmValEntry *)Entry)->SetValue(s.str());
- return;
+ // The elements whose length is bigger than the specified upper bound
+ // are not loaded.
+
+ if (!forceLoad)
+ {
+ if (length > MaxSizeLoadEntry)
+ {
+ dataEntryPtr->SetBinArea(NULL,true);
+ dataEntryPtr->SetState(DataEntry::STATE_NOTLOADED);
+
+ // to be sure we are at the end of the value ...
+ // Fp->seekg((long)entry->GetOffset()+(long)entry->GetLength(),
+ // std::ios::beg); //JPRx
+ return;
+ }
}
- // We need an additional byte for storing \0 that is not on disk
- std::string NewValue(length,0);
- item_read = fread(&(NewValue[0]), (size_t)length, (size_t)1, fp);
- if ( item_read != 1 ) {
- dbg.Verbose(1, "gdcmDocument::LoadElementValue","unread element value");
- ((gdcmValEntry *)Entry)->SetValue("gdcm::UnRead");
- return;
- }
-
- if( (vr == "UI") ) // Because of correspondance with the VR dic
- ((gdcmValEntry *)Entry)->SetValue(NewValue.c_str());
- else
- ((gdcmValEntry *)Entry)->SetValue(NewValue);
+ /// \todo: a method that *doesn't* load anything (maybe with MaxSizeLoadEntry=0 ?)
+ /// + a ForceLoad call on the +/- 20 'usefull' fields
+ /// Allow user to tell the fields he wants to ForceLoad
+ /// during initial stage.
+ /// Later, a GetString or GetBinArea will load the value from disk, if not loaded
+ /// + a method that load *everything* that's not yet loaded
+
+ LoadEntryBinArea(dataEntryPtr); // last one, not to erase length !
}
-
/**
- * \brief Find the value Length of the passed Header Entry
- * @param Entry Header Entry whose length of the value shall be loaded.
+ * \brief Find the value Length of the passed Doc Entry
+ * @param entry Header Entry whose length of the value shall be loaded.
*/
- void gdcmDocument::FindDocEntryLength (gdcmDocEntry *Entry) {
- guint16 element = Entry->GetElement();
- //guint16 group = Entry->GetGroup(); //FIXME
- std::string vr = Entry->GetVR();
- guint16 length16;
-
-
- if ( (filetype == gdcmExplicitVR) && (! Entry->IsImplicitVR()) )
+void Document::FindDocEntryLength( DocEntry *entry )
+ throw ( FormatError )
+{
+ const VRKey &vr = entry->GetVR();
+ uint16_t length16;
+ if ( Filetype == ExplicitVR && !entry->IsImplicitVR() )
{
- if ( (vr=="OB") || (vr=="OW") || (vr=="SQ") || (vr=="UN") )
+
+ // WARNING :
+ //
+ // For some images, length of UN elements is coded on 2 bytes (instead of 4)
+ // There are *not* readable !
+ // You can make a quick and dirty patch, commenting out
+ //| vr == "UN"
+ // in the following line.
+ // (the 'straight' images will no longer be readable ...)
+
+ if ( vr == "OB" || vr == "OW" || vr == "OL" || vr == "SQ" || vr == "UT"
+ || vr == "UN" || changeFromUN == true)
{
- // The following reserved two bytes (see PS 3.5-2001, section
- // 7.1.2 Data element structure with explicit vr p27) must be
+ changeFromUN = false;
+ // The following reserved two bytes (see PS 3.5-2003, section
+ // "7.1.2 Data element structure with explicit vr", p 27) must be
// skipped before proceeding on reading the length on 4 bytes.
- fseek(fp, 2L, SEEK_CUR);
- guint32 length32 = ReadInt32();
- if ( (vr == "OB") && (length32 == 0xffffffff) )
+ //Fp->seekg( 2L, std::ios::cur); // Once per OB,OW,OL,UT,UN,SQ DocEntry
+ uint32_t length32 = ReadInt32(); // Once per OB,OW,OL,UT,UN,SQ DocEntry
+ CurrentOffsetPosition+=4;
+ if ( (vr == "OB" || vr == "OW" || vr == "OL") && length32 == 0xffffffff )
{
- Entry->SetLength(FindDocEntryLengthOB());
+ uint32_t lengthOB;
+ try
+ {
+ lengthOB = FindDocEntryLengthOBOrOW();// for encapsulation of encoded pixel
+ }
+ catch ( FormatUnexpected )
+ {
+ // Computing the length failed (this happens with broken
+ // files like gdcm-JPEG-LossLess3a.dcm). We still have a
+ // chance to get the pixels by deciding the element goes
+ // until the end of the file. Hence we artificially fix the
+ // the length and proceed.
+ gdcmWarningMacro( " Computing the length failed for " <<
+ entry->GetKey() <<" in " <<GetFileName());
+
+ long currentPosition = Fp->tellg(); // Only for gdcm-JPEG-LossLess3a.dcm-like
+ Fp->seekg(0L,std::ios::end); // Only for gdcm-JPEG-LossLess3a.dcm-like
+
+ long lengthUntilEOF = (long)(Fp->tellg())-currentPosition; // Only for gdcm-JPEG-LossLess3a.dcm-like
+ Fp->seekg(currentPosition, std::ios::beg); // Only for gdcm-JPEG-LossLess3a.dcm-like
+
+ entry->SetReadLength(lengthUntilEOF);
+ entry->SetLength(lengthUntilEOF);
+ return;
+ }
+ entry->SetReadLength(lengthOB);
+ entry->SetLength(lengthOB);
return;
}
- FixDocEntryFoundLength(Entry, length32);
+ FixDocEntryFoundLength(entry, length32);
return;
}
-
// Length is encoded on 2 bytes.
- length16 = ReadInt16();
-
- // 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.
- // 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) )
- {
- if ( ! IsExplicitVRBigEndianTransferSyntax() )
- {
- dbg.Verbose(0, "gdcmDocument::FindLength", "not explicit VR");
- errno = 1;
- return;
- }
- length16 = 4;
- SwitchSwapToBigEndian();
- // Restore the unproperly loaded values i.e. the group, the element
- // and the dictionary entry depending on them.
- guint16 CorrectGroup = SwapShort(Entry->GetGroup());
- guint16 CorrectElem = SwapShort(Entry->GetElement());
- gdcmDictEntry * NewTag = GetDictEntryByNumber(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.
+ //length16 = ReadInt16();
+ length16 = GetInt16();
+ // 0xffff means that we deal with 'No Length' Sequence
+ // or 'No Length' SQItem
if ( length16 == 0xffff)
- {
+ {
length16 = 0;
- //dbg.Verbose(0, "gdcmDocument::FindLength",
- // "Erroneous element length fixed.");
- // Actually, length= 0xffff means that we deal with
- // Unknown Sequence Length
}
- FixDocEntryFoundLength(Entry, (guint32)length16);
+ FixDocEntryFoundLength( entry, (uint32_t)length16 );
return;
}
else
// 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.
-
- FixDocEntryFoundLength(Entry, ReadInt32());
+
+ // Well ... group 0002 is always coded in 'Explicit VR Litle Endian'
+ // even if Transfer Syntax is 'Implicit VR ...'
+ // --> Except for 'Implicit VR Big Endian Transfer Syntax GE Private'
+ // where Group 0x0002 is *also* encoded in Implicit VR !
+
+ FixDocEntryFoundLength( entry, GetInt32() /*ReadInt32()*/ );
return;
}
}
/**
- * \brief Find the Value Representation of the current Dicom Element.
- * @param Entry
+ * \brief Find the Length till the next sequence delimiter
+ * \warning NOT end user intended method !
+ * @return
*/
-void gdcmDocument::FindDocEntryVR( gdcmDocEntry *Entry)
+uint32_t Document::FindDocEntryLengthOBOrOW()
+ throw( FormatUnexpected )
{
- if (filetype != gdcmExplicitVR)
- return;
-
- char VR[3];
-
- long PositionOnEntry = ftell(fp);
- // Warning: we believe this is explicit VR (Value Representation) because
- // we used a heuristic that found "UL" in the first tag. Alas this
- // doesn't guarantee that all the tags will be in explicit VR. In some
- // cases (see e-film filtered files) one finds implicit VR tags mixed
- // within an explicit VR file. Hence we make sure the present tag
- // is in explicit VR and try to fix things if it happens not to be
- // the case.
+ // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
- (void)fread (&VR, (size_t)2,(size_t)1, fp);
- VR[2]=0;
- if(!CheckDocEntryVR(Entry,VR))
- {
- fseek(fp, PositionOnEntry, SEEK_SET);
- // When this element is known in the dictionary we shall use, e.g. for
- // the semantics (see the usage of IsAnInteger), the VR proposed by the
- // dictionary entry. Still we have to flag the element as implicit since
- // we know now our assumption on expliciteness is not furfilled.
- // avoid .
- if ( Entry->IsVRUnknown() )
- Entry->SetVR("Implicit");
- Entry->SetImplicitVR();
- }
-}
+ long positionOnEntry = Fp->tellg(); // Only for OB,OW DataElements
-/**
- * \brief Check the correspondance between the VR of the header entry
- * and the taken VR. If they are different, the header entry is
- * updated with the new VR.
- * @param Entry Header Entry to check
- * @param vr Dicom Value Representation
- * @return false if the VR is incorrect of if the VR isn't referenced
- * otherwise, it returns true
-*/
-bool gdcmDocument::CheckDocEntryVR(gdcmDocEntry *Entry, VRKey vr)
-{
- char msg[100]; // for sprintf
- bool RealExplicit = true;
-
- // Assume we are reading a falsely explicit VR file i.e. we reached
- // a tag where we expect reading a VR but are in fact we read the
- // first to bytes of the length. Then we will interogate (through find)
- // the dicom_vr dictionary with oddities like "\004\0" which crashes
- // both GCC and VC++ implementations of the STL map. Hence when the
- // expected VR read happens to be non-ascii characters we consider
- // we hit falsely explicit VR tag.
-
- if ( (!isalpha(vr[0])) && (!isalpha(vr[1])) )
- RealExplicit = false;
-
- // 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 ( RealExplicit && !gdcmGlobal::GetVR()->Count(vr) )
- RealExplicit= false;
-
- if ( !RealExplicit )
- {
- // We thought this was explicit VR, but we end up with an
- // implicit VR tag. Let's backtrack.
- sprintf(msg,"Falsely explicit vr file (%04x,%04x)\n",
- Entry->GetGroup(),Entry->GetElement());
- dbg.Verbose(1, "gdcmDocument::FindVR: ",msg);
- if (Entry->GetGroup()%2 && Entry->GetElement() == 0x0000) { // Group length is UL !
- gdcmDictEntry* NewEntry = NewVirtualDictEntry(
- Entry->GetGroup(),Entry->GetElement(),
- "UL","FIXME","Group Length");
- Entry->SetDictEntry(NewEntry);
- }
- return(false);
- }
+ bool foundSequenceDelimiter = false;
+ uint32_t totalLength = 0;
- if ( Entry->IsVRUnknown() )
- {
- // When not a dictionary entry, we can safely overwrite the VR.
- if (Entry->GetElement() == 0x0000) { // Group length is UL !
- Entry->SetVR("UL");
- } else {
- Entry->SetVR(vr);
- }
- }
- else if ( Entry->GetVR() != vr )
+ while ( !foundSequenceDelimiter )
{
- // The VR present in the file and the dictionary disagree. We assume
- // the file writer knew best and use the VR of the file. Since it would
- // be unwise to overwrite the VR of a dictionary (since it would
- // compromise it's next user), we need to clone the actual DictEntry
- // and change the VR for the read one.
- gdcmDictEntry* NewEntry = NewVirtualDictEntry(
- Entry->GetGroup(),Entry->GetElement(),
- vr,"FIXME",Entry->GetName());
- Entry->SetDictEntry(NewEntry);
- }
- return(true);
-}
+ uint16_t group;
+ uint16_t elem;
-/**
- * \brief Get the transformed value of the header entry. The VR value
- * is used to define the transformation to operate on the value
- * \warning NOT end user intended method !
- * @param Entry
- * @return Transformed entry value
- */
-std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *Entry)
-{
- if ( (IsDocEntryAnInteger(Entry)) && (Entry->IsImplicitVR()) )
- {
- std::string val=Entry->GetValue();
- std::string vr=Entry->GetVR();
- guint32 length = Entry->GetLength();
- std::ostringstream s;
- int nbInt;
-
- // When short integer(s) are expected, read and convert the following
- // n * 2 bytes properly i.e. as a multivaluated strings
- // (each single value is separated fromthe next one by '\'
- // as usual for standard multivaluated filels
- // Elements with Value Multiplicity > 1
- // contain a set of short integers (not a single one)
-
- if (vr == "US" || vr == "SS")
+ try
+ { ///\todo make sure there is never OL encoded pixel data!
+
+ //group = ReadInt16(); // Once per fragment (if any) of OB,OW DataElements
+ //elem = ReadInt16(); // Once per fragment (if any) of OB,OW DataElements
+ ReadBegBuffer(4); // Once per fragment (if any) of OB,OW DataElements
+ }
+ catch ( FormatError )
{
- guint16 NewInt16;
-
- nbInt = length / 2;
- for (int i=0; i < nbInt; i++)
- {
- if(i!=0)
- s << '\\';
- NewInt16 = (val[2*i+0]&0xFF)+((val[2*i+1]&0xFF)<<8);
- NewInt16 = SwapShort(NewInt16);
- s << NewInt16;
- }
+ throw FormatError("Unexpected end of file encountered during ",
+ "Document::FindDocEntryLengthOBOrOW()");
}
+ group = GetInt16();
+ elem = GetInt16();
- // When integer(s) are expected, read and convert the following
- // n * 4 bytes properly i.e. as a multivaluated strings
- // (each single value is separated fromthe next one by '\'
- // as usual for standard multivaluated filels
- // Elements with Value Multiplicity > 1
- // contain a set of integers (not a single one)
- else if (vr == "UL" || vr == "SL")
+ // We have to decount the group and element we just read
+ totalLength += 4;
+ if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) )
{
- guint32 NewInt32;
-
- nbInt = length / 4;
- for (int i=0; i < nbInt; i++)
- {
- if(i!=0)
- s << '\\';
- NewInt32= (val[4*i+0]&0xFF)+((val[4*i+1]&0xFF)<<8)+
- ((val[4*i+2]&0xFF)<<16)+((val[4*i+3]&0xFF)<<24);
- NewInt32=SwapLong(NewInt32);
- s << NewInt32;
- }
+ gdcmWarningMacro(
+ "Neither an Item tag nor a Sequence delimiter tag on :"
+ << std::hex << group << "|" << elem << ") Pos. on entry was 0x(" <<positionOnEntry<< ") "
+ );
+
+ Fp->seekg(positionOnEntry, std::ios::beg); // Once per fragment (if any) of OB,OW DataElements
+ throw FormatUnexpected(
+ "Neither an Item tag nor a Sequence delimiter tag.");
+ }
+ if ( elem == 0xe0dd )
+ {
+ foundSequenceDelimiter = true;
+ }
+ uint32_t itemLength = ReadInt32(); // Once per fragment (if any) of OB,OW DataElements
+ // We add 4 bytes since we just read the ItemLength with ReadInt32
+ totalLength += itemLength + 4;
+ SkipBytes(itemLength);
+
+ if ( foundSequenceDelimiter )
+ {
+ break;
}
-#ifdef GDCM_NO_ANSI_STRING_STREAM
- s << std::ends; // to avoid oddities on Solaris
-#endif //GDCM_NO_ANSI_STRING_STREAM
- return(s.str());
}
-
- return(Entry->GetValue());
+ Fp->seekg( positionOnEntry, std::ios::beg); // Only once for OB,OW DataElements
+ return totalLength;
}
/**
- * \brief Get the reverse transformed value of the header entry. The VR
- * value is used to define the reverse transformation to operate on
- * the value
- * \warning NOT end user intended method !
- * @param Entry
- * @return Reverse transformed entry value
+ * \brief Find the Value Representation of the current Dicom Element.
+ * @return Value Representation of the current Entry
*/
-std::string gdcmDocument::GetDocEntryUnvalue(gdcmDocEntry *Entry)
+VRKey Document::FindDocEntryVR()
{
- if ( (IsDocEntryAnInteger(Entry)) && (Entry->IsImplicitVR()) )
+ if ( Filetype != ExplicitVR )
{
- std::string vr=Entry->GetVR();
- std::ostringstream s;
- std::vector<std::string> tokens;
-
- if (vr == "US" || vr == "SS")
- {
- guint16 NewInt16;
+ return GDCM_VRUNKNOWN;
+ }
- tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
- Tokenize (Entry->GetValue(), tokens, "\\");
- for (unsigned int i=0; i<tokens.size();i++)
- {
- NewInt16 = atoi(tokens[i].c_str());
- s<<(NewInt16&0xFF)<<((NewInt16>>8)&0xFF);
- }
- tokens.clear();
- }
- if (vr == "UL" || vr == "SL")
- {
- guint32 NewInt32;
+ // Delimiters (0xfffe), are not explicit VR ...
+ if ( CurrentGroup == 0xfffe )
+ return GDCM_VRUNKNOWN;
+
+ //long positionOnEntry;
+ //if( Debug::GetWarningFlag() )
+ // positionOnEntry = Fp->tellg(); // Only in Warning Mode
+
+ // Warning: we believe this is explicit VR (Value Representation) because
+ // we used a heuristic that found "UL" in the first tag and/or
+ // 'Transfer Syntax' told us it is.
+ // Alas this doesn't guarantee that all the tags will be in explicit VR.
+ // In some cases one finds implicit VR tags mixed within an explicit VR file
+ // Well...
+ // 'Normaly' the only case is : group 0002 Explicit, and other groups Implicit
+ //
+ // Hence we make sure the present tag is in explicit VR and try to fix things
+ // if it happens not to be the case.
- tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
- Tokenize (Entry->GetValue(), tokens, "\\");
- for (unsigned int i=0; i<tokens.size();i++)
- {
- NewInt32 = atoi(tokens[i].c_str());
- s<<(char)(NewInt32&0xFF)<<(char)((NewInt32>>8)&0xFF)
- <<(char)((NewInt32>>16)&0xFF)<<(char)((NewInt32>>24)&0xFF);
- }
- tokens.clear();
- }
+ VRKey vr;
+ //Fp->read(&(vr[0]),(size_t)2);
+ vr[0] = *PtrBegBuffer++;
+ vr[1] = *PtrBegBuffer++;
+
+ //if ( !CheckDocEntryVR(vr) ) // avoid useless function call
+ if ( !Global::GetVR()->IsValidVR(vr) )
+ {
-#ifdef GDCM_NO_ANSI_STRING_STREAM
- s << std::ends; // to avoid oddities on Solaris
-#endif //GDCM_NO_ANSI_STRING_STREAM
- return(s.str());
+ gdcmWarningMacro( "Unknown VR " << vr.GetHexaRepresentation() << std::hex
+ << " at offset : 0x(" << CurrentOffsetPosition-4
+ << ") for group " << std::hex << CurrentGroup );
+
+ //Fp->seekg(positionOnEntry, std::ios::beg); //JPRx
+ //Fp->seekg((long)-2, std::ios::cur);// only for unrecognized VR (?!?)
+ //see :MR_Philips_Intera_PrivateSequenceExplicitVR.dcm
+ PtrBegBuffer-=2;
+ return GDCM_VRUNKNOWN;
}
+ return vr;
+}
- return(Entry->GetValue());
+/**
+ * \brief Check the correspondance between the VR of the header entry
+ * and the taken VR. If they are different, the header entry is
+ * updated with the new VR.
+ * @param vr Dicom Value Representation
+ * @return false if the VR is incorrect or if the VR isn't referenced
+ * otherwise, it returns true
+*/
+bool Document::CheckDocEntryVR(const VRKey &vr)
+{
+ return Global::GetVR()->IsValidVR(vr);
}
/**
* \brief Skip a given Header Entry
- * \warning NOT end user intended method !
- * @param entry
+ * @param entry entry to skip
*/
-void gdcmDocument::SkipDocEntry(gdcmDocEntry *entry)
+void Document::SkipDocEntry(DocEntry *entry)
{
SkipBytes(entry->GetLength());
}
/**
- * \brief Skips to the begining of the next Header Entry
- * \warning NOT end user intended method !
- * @param entry
+ * \brief Skips to the beginning of the next Header Entry
+ * @param currentDocEntry entry to skip
*/
-void gdcmDocument::SkipToNextDocEntry(gdcmDocEntry *entry)
+void Document::SkipToNextDocEntry(DocEntry *currentDocEntry)
{
- (void)fseek(fp, (long)(entry->GetOffset()), SEEK_SET);
- (void)fseek(fp, (long)(entry->GetReadLength()), SEEK_CUR);
+ long l = currentDocEntry->GetReadLength();
+ if ( (uint32_t) l == (uint32_t)-1 ) // length = 0xffff shouldn't appear here ...
+ // ... but PMS imagers happen !
+ return;
+ Fp->seekg((size_t)(currentDocEntry->GetOffset()), std::ios::beg); //FIXME :each DocEntry
+ if (currentDocEntry->GetGroup() != 0xfffe) // for fffe pb
+ {
+ Fp->seekg( l,std::ios::cur); //FIXME :each DocEntry
+ }
}
-/**
- * \brief Loads the value for a a given VLEntry
- * \warning NOT end user intended method !
- * @param entry
- */
-void gdcmDocument::LoadVLEntry(gdcmDocEntry *entry)
-{
- //SkipBytes(entry->GetLength());
- LoadDocEntry(entry);
-}
/**
* \brief When the length of an element value is obviously wrong (because
* the parser went Jabberwocky) one can hope improving things by
- * applying this heuristic.
+ * applying some heuristics.
+ * @param entry entry to check
+ * @param foundLength first assumption about length (before bug fix, or set to zero if =0xffffffff)
*/
-void gdcmDocument::FixDocEntryFoundLength(gdcmDocEntry *Entry, guint32 FoundLength)
+void Document::FixDocEntryFoundLength(DocEntry *entry,
+ uint32_t foundLength)
{
- Entry->SetReadLength(FoundLength); // will be updated only if a bug is found
- if ( FoundLength == 0xffffffff) {
- FoundLength = 0;
- }
+ entry->SetReadLength( foundLength );// will be updated only if a bug is found
- guint16 gr =Entry->GetGroup();
- guint16 el =Entry->GetElement();
-
- if (FoundLength%2) {
- std::ostringstream s;
- s << "Warning : Tag with uneven length " << FoundLength
- << " in x(" << std::hex << gr << "," << el <<")" << std::dec;
- dbg.Verbose(0,s.str().c_str());
+ if ( foundLength == 0xffffffff)
+ {
+ //foundLength = 0;
+ //entry->SetLength(foundLength);
+ entry->SetLength(0);
+ return; // return ASAP; don't waist time on useless tests
}
- // Sorry for the patch!
- // XMedCom did the trick to read some naughty GE images ...
- if (FoundLength == 13) {
- // The following 'if' will be removed when there is no more
- // images on Creatis HDs with a 13 length for Manufacturer...
- if ( (Entry->GetGroup() != 0x0008) ||
- ( (Entry->GetElement() != 0x0070) && (Entry->GetElement() != 0x0080) ) ){
- // end of remove area
- FoundLength =10;
- Entry->SetReadLength(10); // a bug is to be fixed
+ uint16_t gr = entry->GetGroup();
+ uint16_t elem = entry->GetElement();
+
+ if ( foundLength % 2)
+ {
+ gdcmWarningMacro( "Warning : Tag (" << std::hex << gr << "|" << elem << ") with uneven length "
+ << std::dec << foundLength << " 0x(" << std::hex << foundLength << ") "
+ //<< " at offset x(" << offset << ")"
+ );
+ }
+
+ //////// Fix for some naughty General Electric images.
+ // Allthough not recent many such GE corrupted images are still present
+ // on Creatis hard disks. Hence this fix shall remain when such images
+ // are no longer in use (we are talking a few years, here)...
+ // Note: XMedCon probably uses such a trick since it is able to read
+ // those pesky GE images ...
+ if ( foundLength == 13)
+ {
+ // Only happens for this length !
+ if ( gr != 0x0008 || ( elem != 0x0070 && elem != 0x0080 ) )
+ {
+ foundLength = 10;
+ entry->SetReadLength(10); // a bug is to be fixed !?
}
}
- // to fix some garbage 'Leonardo' Siemens images
- // May be commented out to avoid overhead
- else if ( (Entry->GetGroup() == 0x0009) &&
- ( (Entry->GetElement() == 0x1113) || (Entry->GetElement() == 0x1114) ) ){
- FoundLength =4;
- Entry->SetReadLength(4); // a bug is to be fixed
- }
- // end of fix
-
- // to try to 'go inside' SeQuences (with length), and not to skip them
- else if ( Entry->GetVR() == "SQ")
+ //////// Fix for some brain-dead 'Leonardo' Siemens images.
+ // Occurence of such images is quite low (unless one leaves close to a
+ // 'Leonardo' source. Hence, one might consider commenting out the
+ // following fix on efficiency reasons.
+ else if ( gr == 0x0009 && ( elem == 0x1113 || elem == 0x1114 ) )
{
- if (enableSequences) // only if the user does want to !
- FoundLength =0; // ReadLength is unchanged
- }
-
- // we found a 'delimiter' element
- // fffe|xxxx is just a marker, we don't take its length into account
- else if(Entry->GetGroup() == 0xfffe)
- {
- // *normally, fffe|0000 doesn't exist !
- if( Entry->GetElement() != 0x0000 ) // gdcm-MR-PHILIPS-16-Multi-Seq.dcm
- // causes extra troubles :-(
- FoundLength =0;
- }
-
- Entry->SetUsableLength(FoundLength);
+ // Ideally we should check we are in Explicit and double check
+ // that VR=UL... this is done properly in gdcm2
+ if( foundLength == 6 )
+ {
+ gdcmWarningMacro( "Replacing Length from 6 into 4" );
+ foundLength = 4;
+ entry->SetReadLength(4); // a bug is to be fixed !
+ }
+ else if ( foundLength%4 )
+ {
+ gdcmErrorMacro( "This looks like to a buggy Siemens DICOM file."
+ "The length of this tag seems to be wrong" );
+ }
+ }
+
+ else if ( entry->GetVR() == "SQ" )
+ {
+ foundLength = 0; // ReadLength is unchanged
+ }
+
+ //////// We encountered a 'delimiter' element i.e. a tag of the form
+ // "fffe|xxxx" which is just a marker. Delimiters length should not be
+ // taken into account.
+ else if ( gr == 0xfffe )
+ {
+ // According to the norm, fffe|0000 shouldn't exist. BUT the Philips
+ // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to
+ // causes extra troubles...
+ if ( elem != 0x0000 )
+ {
+ foundLength = 0;
+ }
+ else
+ {
+ foundLength=12; // to skip the mess that follows this bugged Tag !
+ }
+ }
+ entry->SetLength(foundLength);
}
/**
* \brief Apply some heuristics to predict whether the considered
* element value contains/represents an integer or not.
- * @param Entry The element value on which to apply the predicate.
+ * @param entry The element value on which to apply the predicate.
* @return The result of the heuristical predicate.
*/
-bool gdcmDocument::IsDocEntryAnInteger(gdcmDocEntry *Entry) {
- guint16 element = Entry->GetElement();
- guint16 group = Entry->GetGroup();
- std::string vr = Entry->GetVR();
- guint32 length = Entry->GetLength();
+bool Document::IsDocEntryAnInteger(DocEntry *entry)
+{
+ uint16_t elem = entry->GetElement();
+ uint16_t group = entry->GetGroup();
+ const VRKey &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
- // able to swap it's element value properly.
- if ( element == 0 ) // This is the group length of the group
+ // able to swap its element value properly.
+ if ( elem == 0 ) // This is the group length of the group
{
- if (length == 4)
+ if ( length == 4 )
+ {
return true;
+ }
else
{
- std::ostringstream s;
- int filePosition = ftell(fp);
- s << "Erroneous Group Length element length on : (" \
- << std::hex << group << " , " << element
+ // Although this should never happen, still some images have a
+ // corrupted group length [e.g. have a glance at offset x(8336) of
+ // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm.
+ // Since for dicom compliant and well behaved headers, the present
+ // test is useless (and might even look a bit paranoid), when we
+ // encounter such an ill-formed image, we simply display a warning
+ // message and proceed on parsing (while crossing fingers).
+ long filePosition = Fp->tellg(); // Only when elem 0x0000 length is not 4 (?!?)
+ (void)filePosition;
+ gdcmWarningMacro( "Erroneous Group Length element length on : ("
+ << std::hex << group << " , " << elem
<< ") -before- position x(" << filePosition << ")"
- << "lgt : " << length;
- // These 2 lines commented out : a *very dirty* patch
- // to go on PrintHeader'ing gdcm-MR-PHILIPS-16-Multi-Seq.dcm.
- // have a glance at offset x(8336) ...
- // For *regular* headers, the test is useless..
- // lets's print a warning message and go on,
- // instead of giving up with an error message
-
- //std::cout << s.str().c_str() << std::endl;
- // dbg.Error("gdcmDocument::IsDocEntryAnInteger",
- // s.str().c_str());
+ << "lgt : " << length );
}
}
- if ( (vr == "UL") || (vr == "US") || (vr == "SL") || (vr == "SS") )
+
+ if ( vr == "UL" || vr == "US" || vr == "SL" || vr == "SS" )
+ {
return true;
-
+ }
return false;
}
-/**
- * \brief Find the Length till the next sequence delimiter
- * \warning NOT end user intended method !
- * @return
- */
-
- guint32 gdcmDocument::FindDocEntryLengthOB(void) {
- // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
- guint16 g;
- guint16 n;
- long PositionOnEntry = ftell(fp);
- bool FoundSequenceDelimiter = false;
- guint32 TotalLength = 0;
- guint32 ItemLength;
-
- while ( ! FoundSequenceDelimiter)
- {
- g = ReadInt16();
- n = ReadInt16();
- if (errno == 1)
- return 0;
- TotalLength += 4; // We even have to decount the group and element
-
- if ( g != 0xfffe && g!=0xb00c ) //for bogus header
- {
- char msg[100]; // for sprintf. Sorry
- sprintf(msg,"wrong group (%04x) for an item sequence (%04x,%04x)\n",g, g,n);
- dbg.Verbose(1, "gdcmDocument::FindLengthOB: ",msg);
- errno = 1;
- return 0;
- }
- if ( n == 0xe0dd || ( g==0xb00c && n==0x0eb6 ) ) // for bogus header
- FoundSequenceDelimiter = true;
- else if ( n != 0xe000 )
- {
- char msg[100]; // for sprintf. Sorry
- sprintf(msg,"wrong element (%04x) for an item sequence (%04x,%04x)\n",
- n, g,n);
- dbg.Verbose(1, "gdcmDocument::FindLengthOB: ",msg);
- errno = 1;
- return 0;
- }
- ItemLength = ReadInt32();
- TotalLength += ItemLength + 4; // We add 4 bytes since we just read
- // the ItemLength with ReadInt32
- SkipBytes(ItemLength);
- }
- fseek(fp, PositionOnEntry, SEEK_SET);
- return TotalLength;
-}
-
-/**
- * \brief Reads a supposed to be 16 Bits integer
- * (swaps it depending on processor endianity)
- * @return read value
- */
-guint16 gdcmDocument::ReadInt16(void) {
- guint16 g;
- size_t item_read;
- item_read = fread (&g, (size_t)2,(size_t)1, fp);
- if ( item_read != 1 ) {
- if(ferror(fp))
- dbg.Verbose(0, "gdcmDocument::ReadInt16", " File Error");
- errno = 1;
- return 0;
- }
- errno = 0;
- g = SwapShort(g);
- return g;
-}
-
-/**
- * \brief Reads a supposed to be 32 Bits integer
- * (swaps it depending on processor endianity)
- * @return read value
- */
-guint32 gdcmDocument::ReadInt32(void) {
- guint32 g;
- size_t item_read;
- item_read = fread (&g, (size_t)4,(size_t)1, fp);
- if ( item_read != 1 ) {
- if(ferror(fp))
- dbg.Verbose(0, "gdcmDocument::ReadInt32", " File Error");
- errno = 1;
- return 0;
- }
- errno = 0;
- g = SwapLong(g);
- return g;
-}
-
-/**
- * \brief skips bytes inside the source file
- * \warning NOT end user intended method !
- * @return
- */
-void gdcmDocument::SkipBytes(guint32 NBytes) {
- //FIXME don't dump the returned value
- (void)fseek(fp, (long)NBytes, SEEK_CUR);
-}
-
-/**
- * \brief Loads all the needed Dictionaries
- * \warning NOT end user intended method !
- */
-void gdcmDocument::Initialise(void)
-{
- RefPubDict = gdcmGlobal::GetDicts()->GetDefaultPubDict();
- RefShaDict = (gdcmDict*)0;
-}
/**
* \brief Discover what the swap code is (among little endian, big endian,
* it's neither ACR-NEMA nor DICOM
* true when we hope ours assuptions are OK
*/
-bool gdcmDocument::CheckSwap() {
-
- // The only guaranted way of finding the swap code is to find a
- // group tag since we know it's length has to be of four bytes i.e.
- // 0x00000004. Finding the swap code in then straigthforward. Trouble
- // occurs when we can't find such group...
-
- guint32 x=4; // x : for ntohs
- bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
- guint32 s32;
- guint16 s16;
+bool Document::CheckSwap()
+{
+ uint32_t s32;
+ uint16_t s16;
- int lgrLue;
- char *entCur;
- char deb[HEADER_LENGTH_TO_READ];
+ char deb[256];
// First, compare HostByteOrder and NetworkByteOrder in order to
// determine if we shall need to swap bytes (i.e. the Endian type).
- if (x==ntohs(x))
- net2host = true;
- else
- net2host = false;
+ bool net2host = Util::IsCurrentProcessorBigEndian();
- // 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".
- lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, fp);
+ // 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);
- entCur = deb + 128;
- if(memcmp(entCur, "DICM", (size_t)4) == 0) {
- dbg.Verbose(1, "gdcmDocument::CheckSwap:", "looks like DICOM Version3");
+ char *entCur = deb + 128;
+ if ( memcmp(entCur, "DICM", (size_t)4) == 0 )
+ {
+ gdcmDebugMacro( "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,
// 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 (Transfert 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*
+ if ( memcmp(entCur, "UL", (size_t)2) == 0 ||
+ memcmp(entCur, "OB", (size_t)2) == 0 ||
+ memcmp(entCur, "UI", (size_t)2) == 0 ||
+ memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later
+ // when Write DCM *adds*
// FIXME
- // Use gdcmDocument::dicom_vr to test all the possibilities
+ // Use Document::dicom_vr to test all the possibilities
// instead of just checking for UL, OB and UI !? group 0000
{
- filetype = gdcmExplicitVR;
- dbg.Verbose(1, "gdcmDocument::CheckSwap:",
- "explicit Value Representation");
+ Filetype = ExplicitVR;
+ gdcmDebugMacro( "Group 0002 : Explicit Value Representation");
}
else
{
- filetype = gdcmImplicitVR;
- dbg.Verbose(1, "gdcmDocument::CheckSwap:",
- "not an explicit Value Representation");
+ Filetype = ImplicitVR;
+ gdcmWarningMacro( "Group 0002 :Not an explicit Value Representation;"
+ << "Looks like a bugged Header!");
}
- if (net2host)
+ // Here, we assume that the file IS kosher Dicom !
+ // (The meta elements - group 0x0002 - ARE little endian !)
+ if ( net2host )
{
- sw = 4321;
- dbg.Verbose(1, "gdcmDocument::CheckSwap:",
- "HostByteOrder != NetworkByteOrder");
- }
+ SwapCode = 4321;
+ gdcmDebugMacro( "HostByteOrder != NetworkByteOrder, SwapCode = 4321");
+ }
else
{
- sw = 0;
- dbg.Verbose(1, "gdcmDocument::CheckSwap:",
- "HostByteOrder = NetworkByteOrder");
+ SwapCode = 1234;
+ gdcmDebugMacro( "HostByteOrder = NetworkByteOrder, SwapCode = 1234");
}
- // Position the file position indicator at first tag (i.e.
- // after the file preamble and the "DICM" string).
- rewind(fp);
- fseek (fp, 132L, SEEK_SET);
+ // Position the file position indicator at first tag
+ // (i.e. after the file preamble and the "DICM" string).
+
+ Fp->seekg ( 132L, std::ios::beg); // Once per Document
+ CurrentOffsetPosition = 132;
return true;
- } // End of DicomV3
+ } // ------------------------------- 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).
- dbg.Verbose(1, "gdcmDocument::CheckSwap:", "not a DICOM Version3 file");
- rewind(fp);
+
+ gdcmWarningMacro( "Not a Kosher DICOM Version3 file (no preamble)");
+
+ Fp->seekg(0, std::ios::beg); // Once per ACR-NEMA Document
+ CurrentOffsetPosition = 0;
+ // Let's check 'No Preamble Dicom File' :
+ // Should start with group 0x0002
+ // and be Explicit Value Representation
+
+ s16 = *((uint16_t *)(deb));
+ SwapCode = 0;
+ switch ( s16 )
+ {
+ case 0x0002 :
+ SwapCode = 1234;
+ entCur = deb + 4;
+ break;
+ case 0x0200 :
+ SwapCode = 4321;
+ entCur = deb + 6;
+ }
+
+ if ( SwapCode != 0 )
+ {
+ if ( memcmp(entCur, "UL", (size_t)2) == 0 ||
+ memcmp(entCur, "OB", (size_t)2) == 0 ||
+ memcmp(entCur, "UI", (size_t)2) == 0 ||
+ memcmp(entCur, "SH", (size_t)2) == 0 ||
+ memcmp(entCur, "AE", (size_t)2) == 0 ||
+ memcmp(entCur, "OB", (size_t)2) == 0 )
+ {
+ Filetype = ExplicitVR; // FIXME : not enough to say it's Explicit
+ // Wait untill reading Transfer Syntax
+ gdcmDebugMacro( "Group 0002 : Explicit Value Representation");
+ return true;
+ }
+ }
+// ------------------------------- End of 'No Preamble' DicomV3 -------------
// Our next best chance would be to be considering a 'clean' ACR/NEMA file.
- // By clean we mean that the length of the first tag is written down.
+ // By clean we mean that the length of the first group is written down.
// If this is the case and since the length of the first group HAS to be
// four (bytes), then determining the proper swap code is straightforward.
// We assume the array of char we are considering contains the binary
// representation of a 32 bits integer. Hence the following dirty
// trick :
- s32 = *((guint32 *)(entCur));
-
- switch (s32) {
+ s32 = *((uint32_t *)(entCur));
+ switch( s32 )
+ {
case 0x00040000 :
- sw = 3412;
- filetype = gdcmACR;
+ SwapCode = 3412;
+ Filetype = ACR;
return true;
case 0x04000000 :
- sw = 4321;
- filetype = gdcmACR;
+ SwapCode = 4321;
+ Filetype = ACR;
return true;
case 0x00000400 :
- sw = 2143;
- filetype = gdcmACR;
+ SwapCode = 2143;
+ Filetype = ACR;
return true;
case 0x00000004 :
- sw = 0;
- filetype = gdcmACR;
+ SwapCode = 1234;
+ Filetype = ACR;
return true;
default :
-
- // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
- // It is time for despaired wild guesses.
- // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA,
- // i.e. the 'group length' element is not present :
-
- // check the supposed to be 'group number'
- // 0x0002 or 0x0004 or 0x0008
- // to determine ' sw' value .
- // Only 0 or 4321 will be possible
- // (no oportunity to check for the formerly well known
- // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian'
- // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -4, 8-)
- // the file IS NOT ACR-NEMA nor DICOM V3
- // Find a trick to tell it the caller...
-
- s16 = *((guint16 *)(deb));
-
- switch (s16) {
- case 0x0002 :
- case 0x0004 :
- case 0x0008 :
- sw = 0;
- filetype = gdcmACR;
- return true;
- case 0x0200 :
- case 0x0400 :
- case 0x0800 :
- sw = 4321;
- filetype = gdcmACR;
- return true;
- default :
- dbg.Verbose(0, "gdcmDocument::CheckSwap:",
- "ACR/NEMA unfound swap info (Really hopeless !)");
- filetype = gdcmUnknown;
- return false;
- }
+ // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
+ // It is time for despaired wild guesses.
+ // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA,
+ // i.e. the 'group length' element is not present :
+
+ // check the supposed-to-be 'group number'
+ // in ( 0x0001 .. 0x0008 )
+ // to determine ' SwapCode' value .
+ // Only 0 or 4321 will be possible
+ // (no oportunity to check for the formerly well known
+ // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian'
+ // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc-3, 4, ..., 8-)
+ // the file IS NOT ACR-NEMA nor DICOM V3
+ // Find a trick to tell it the caller...
- // Then the only info we have is the net2host one.
- //if (! net2host )
- // sw = 0;
- //else
- // sw = 4321;
- //return;
+ s16 = *((uint16_t *)(deb));
+
+ gdcmDebugMacro("not a DicomV3 nor a 'clean' ACR/NEMA;"
+ << " (->despaired wild guesses !)");
+ switch ( s16 )
+ {
+ case 0x0001 :
+ case 0x0002 :
+ case 0x0003 :
+ case 0x0004 :
+ case 0x0005 :
+ case 0x0006 :
+ case 0x0007 :
+ case 0x0008 :
+ case 0x0028 :
+ SwapCode = 1234;
+ Filetype = ACR;
+ return true;
+ case 0x0100 :
+ case 0x0200 :
+ case 0x0300 :
+ case 0x0400 :
+ case 0x0500 :
+ case 0x0600 :
+ case 0x0700 :
+ case 0x0800 :
+ case 0x2800 :
+ SwapCode = 4321;
+ Filetype = ACR;
+ return true;
+ default :
+
+ s16 = *((uint16_t *)(deb));
+ if (s16 != 0x0000)
+ return false;
+ s16 = *((uint16_t *)(deb+2));
+
+ Fp->seekg ( 0L, std::ios::beg); // Once per Document
+ CurrentOffsetPosition = 0;
+ switch(s16) // try an other trick!
+ // -> to be able to decode 0029|1010 DataElement
+ // -> and be not less cleaver than dcmdump ;-)
+ {
+ case 0x0004 :
+ SwapCode = 1234;
+ break;
+ case 0x0400 :
+ SwapCode = 3412;
+ break;
+ default:
+ gdcmWarningMacro("ACR/NEMA unfound swap info (Hopeless !)");
+ Filetype = Unknown;
+ return false;
+ }
+ // Check if next 2 bytes are a VR
+ // Probabely something more time-consuming exists with std::string
+ const char VRvalues[] = "AEASATCSDADTFLFDISLOLTPNSHSLSSSTTMUIULUSUTOBOWOLOFATUNSQRT";
+ int nbVal = 29;
+ const char *pt = VRvalues;
+ for (int i=0;i<nbVal;i++)
+ {
+ if(*(deb+4) == *pt++) {
+ if(*(deb+5) == *pt++) {
+ Filetype = ExplicitVR;
+ return true;
+ }
+ else {
+ pt++;
+ }
+ }
+ }
+ Filetype = ImplicitVR;
+ return true;
+ }
}
}
/**
- * \brief Restore the unproperly loaded values i.e. the group, the element
- * and the dictionary entry depending on them.
+ * \brief Change the Byte Swap code.
*/
-void gdcmDocument::SwitchSwapToBigEndian(void)
+void Document::SwitchByteSwapCode()
{
- dbg.Verbose(1, "gdcmDocument::SwitchSwapToBigEndian",
- "Switching to BigEndian mode.");
- if ( sw == 0 )
+ gdcmDebugMacro( "Switching Byte Swap code from "<< SwapCode
+ << " at: 0x" << std::hex << Fp->tellg() ); // Only when DEBUG
+ if ( SwapCode == 1234 )
{
- sw = 4321;
- return;
+ SwapCode = 4321;
}
- if ( sw == 4321 )
+ else if ( SwapCode == 4321 )
{
- sw = 0;
- return;
+ SwapCode = 1234;
}
- if ( sw == 3412 )
+ else if ( SwapCode == 3412 )
{
- sw = 2143;
- return;
+ SwapCode = 2143;
+ }
+ else if ( SwapCode == 2143 )
+ {
+ SwapCode = 3412;
}
- if ( sw == 2143 )
- sw = 3412;
+ gdcmDebugMacro( " Into: "<< SwapCode );
}
/**
- * \brief during parsing, Header Elements too long are not loaded in memory
- * @param NewSize
+ * \brief during parsing, Header Elements too long are not loaded in memory
+ * @param newSize new size
*/
-void gdcmDocument::SetMaxSizeLoadEntry(long NewSize)
+void Document::SetMaxSizeLoadEntry(long newSize)
{
- if (NewSize < 0)
+ if ( newSize < 0 )
+ {
return;
- if ((guint32)NewSize >= (guint32)0xffffffff)
+ }
+ if ((uint32_t)newSize >= (uint32_t)0xffffffff )
{
MaxSizeLoadEntry = 0xffffffff;
return;
}
- MaxSizeLoadEntry = NewSize;
+ MaxSizeLoadEntry = newSize;
}
-
/**
- * \brief Header Elements too long will not be printed
- * \todo See comments of \ref gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE
- * @param NewSize
+ * \brief Read the next tag WITHOUT loading it's value
+ * (read the 'Group Number', the 'Element Number',
+ * gets the Dict Entry
+ * gets the VR, gets the length, gets the offset value)
+ * @return On succes : the newly created DocEntry, NULL on failure.
*/
-void gdcmDocument::SetMaxSizePrintEntry(long NewSize)
+DocEntry *Document::ReadNextDocEntry()
{
- if (NewSize < 0)
- return;
- if ((guint32)NewSize >= (guint32)0xffffffff)
+ try
{
- MaxSizePrintEntry = 0xffffffff;
- return;
+ ReadBegBuffer(8); // Avoid to many time consuming freads
+ //CurrentGroup = ReadInt16();
+ //CurrentElem = ReadInt16();
}
- MaxSizePrintEntry = NewSize;
-}
+ catch ( FormatError )
+ {
+ // We reached the EOF (or an error occured) therefore
+ // header parsing has to be considered as finished.
+ return 0;
+ }
+
+ changeFromUN = false;
+ CurrentGroup = GetInt16();
+ CurrentElem = GetInt16();
+
+ // In 'true DICOM' files Group 0002 is always little endian
+ if ( HasDCMPreamble )
+ {
+ if ( !Group0002Parsed && CurrentGroup != 0x0002) // avoid calling a function when useless
+ HandleOutOfGroup0002(CurrentGroup, CurrentElem);
+ else
+ // Sometimes file contains groups of tags with reversed endianess.
+ HandleBrokenEndian(CurrentGroup, CurrentElem);
+ }
+ VRKey vr = FindDocEntryVR();
+ VRKey realVR = vr;
+ if ( vr == GDCM_VRUNKNOWN )
+ {
+ if ( CurrentElem == 0x0000 ) // Group Length
+ {
+ realVR = "UL"; // must be UL
+ }
+ else if (CurrentGroup == 0xfffe) // Don't get DictEntry for Delimitors
+ {
+ realVR = "UL";
+ }
-/**
- * \brief Read the next tag but WITHOUT loading it's value
- * (read the 'Group Number', the 'Element Number',
- * gets the Dict Entry
- * gets the VR, gets the length, gets the offset value)
- * @return On succes the newly created DocEntry, NULL on failure.
- */
-gdcmDocEntry *gdcmDocument::ReadNextDocEntry(void) {
- guint16 g,n;
- gdcmDocEntry *NewEntry;
- g = ReadInt16();
- n = ReadInt16();
+ // Was commented out in order not to generate 'Shadow Groups' where some
+ // Data Elements are Explicit VR and some other ones Implicit VR
+ // -> Better we fix the problem at Write time
+
+ else if (CurrentGroup%2 == 1 )
+ {
+ if (CurrentElem >= 0x0010 && CurrentElem <=0x00ff )
+ {
+ // DICOM PS 3-5 7.8.1 a) states that :
+ // Private Creator Data Elements numbered (gggg,0010-00FF) (gggg is odd)
+ // attributes have to be LO (Long String) and the VM shall be equal to 1
+ realVR = "LO";
+
+ // Seems not to be true
+ // Still in gdcmtk, David Clunnie disagrees, Marco Eichelberg says it's OK ...
+ // We let it for a while?
+ //(We should check length==4, for more security, but we don't have it yet !)
+ }
+ else if ( CurrentElem == 0x0001)
+ {
+ realVR = "UL"; // Private Group Length To End
+ }
+ else // check the private dictionary for shadow elements when Implicit VR!
+ {
+ DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem);
+ if ( dictEntry )
+ {
+ realVR = dictEntry->GetVR();
+ dictEntry->Unregister(); // GetDictEntry registered it
+ }
+ }
+ }
- if (errno == 1)
- // We reached the EOF (or an error occured) therefore
- // header parsing has to be considered as finished.
- return (gdcmDocEntry *)0;
-
-// Pb : how to propagate the element length (used in SkipDocEntry)
-// direct call to SkipBytes ?
-
-// if (ignoreShadow == 1 && g%2 ==1)
- // if user wants to skip shadow groups
- // and current element *is* a shadow element
- // we don't create anything
-// return (gdcmDocEntry *)1; // to tell caller it's NOT finished
-
- NewEntry = NewDocEntryByNumber(g, n);
- FindDocEntryVR(NewEntry);
- FindDocEntryLength(NewEntry);
-
- if (errno == 1) {
- // Call it quits
- delete NewEntry;
- return NULL;
+ else
+ {
+ DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem);//only when ImplicitVR
+ if ( dictEntry )
+ {
+ realVR = dictEntry->GetVR();
+ dictEntry->Unregister(); // GetDictEntry registered it
+ }
+ }
}
- NewEntry->SetOffset(ftell(fp));
- return NewEntry;
-}
-
-/**
- * \brief Build a new Element Value from all the low level arguments.
- * Check for existence of dictionary entry, and build
- * a default one when absent.
- * @param Name Name of the underlying DictEntry
- */
-gdcmDocEntry *gdcmDocument::NewDocEntryByName(std::string Name)
-{
- gdcmDictEntry *NewTag = GetDictEntryByName(Name);
- if (!NewTag)
- NewTag = NewVirtualDictEntry(0xffff, 0xffff, "LO", "unkn", Name);
- gdcmDocEntry* NewEntry = new gdcmDocEntry(NewTag);
- if (!NewEntry)
+ // if UN found, let's check the dictionary, and trust it!
+ // (maybe a private dictionary exists?)
+ else if (vr == "UN")
{
- dbg.Verbose(1, "gdcmDocument::ObtainDocEntryByName",
- "failed to allocate gdcmDocEntry");
- return (gdcmDocEntry *)0;
+ DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem);
+ if ( dictEntry )
+ {
+ realVR = dictEntry->GetVR();
+ dictEntry->Unregister(); // GetDictEntry registered it
+
+ // for VR = "UN", length is always stored on 4 bytes.
+ // remember this info, in order not to crash later
+ changeFromUN=true;
+ /// \todo : fixme If inside a vr = "UN" DataElement (but SQ according to a private dictionnary)
+ /// there is some more vr = "UN" DataElements, it will probabely fail.
+ /// --> find a -non time consuming- trick to store changeFromUN info at DataElement level,
+ /// not at the Document level.
+ /// --> ?!? JPR
+
+ }
}
- return NewEntry;
-}
-
-/**
- * \brief Request a new virtual dict entry to the dict set
- * @param group group of the underlying DictEntry
- * @param element element of the underlying DictEntry
- * @param vr VR of the underlying DictEntry
- * @param fourth owner group
- * @param name english name
- */
-gdcmDictEntry *gdcmDocument::NewVirtualDictEntry(guint16 group, guint16 element,
- std::string vr,
- std::string fourth,
- std::string name)
-{
- return gdcmGlobal::GetDicts()->NewVirtualDictEntry(group,element,vr,fourth,name);
-}
-
-/**
- * \brief Build a new Element Value from all the low level arguments.
- * Check for existence of dictionary entry, and build
- * a default one when absent.
- * @param Group group of the underlying DictEntry
- * @param Elem element of the underlying DictEntry
- */
-gdcmDocEntry *gdcmDocument::NewDocEntryByNumber(guint16 Group, guint16 Elem)
-{
- // Find out if the tag we encountered is in the dictionaries:
- gdcmDictEntry *DictEntry = GetDictEntryByNumber(Group, Elem);
- if (!DictEntry)
- DictEntry = NewVirtualDictEntry(Group, Elem);
- gdcmDocEntry *NewEntry = new gdcmDocEntry(DictEntry);
- if (!NewEntry)
+ DocEntry *newEntry;
+ //if ( Global::GetVR()->IsVROfSequence(realVR) )
+ if (realVR == "SQ")
{
- dbg.Verbose(1, "gdcmDocument::NewDocEntryByNumber",
- "failed to allocate gdcmDocEntry");
- return NULL;
+ newEntry = NewSeqEntry(CurrentGroup, CurrentElem);
}
- return NewEntry;
-}
-
-/// \todo Never used; commented out, waiting for removal.
-/**
- * \brief Small utility function that creates a new manually crafted
- * (as opposed as read from the file) gdcmDocEntry with user
- * specified name and adds it to the public tag hash table.
- * \note A fake TagKey is generated so the PubDict can keep it's coherence.
- * @param NewTagName The name to be given to this new tag.
- * @param VR The Value Representation to be given to this new tag.
- * @return The newly hand crafted Element Value.
- */
-//gdcmDocEntry *gdcmDocument::NewManualDocEntryToPubDict(std::string NewTagName,
-// std::string VR)
-//{
-// gdcmDocEntry *NewEntry = NULL;
-// guint32 StuffGroup = 0xffff; // Group to be stuffed with additional info
-// guint32 FreeElem = 0;
-// gdcmDictEntry *DictEntry = NULL;
-//
-// FreeElem = GenerateFreeTagKeyInGroup(StuffGroup);
-// if (FreeElem == UINT32_MAX)
-// {
-// dbg.Verbose(1, "gdcmHeader::NewManualDocEntryToPubDict",
-// "Group 0xffff in Public Dict is full");
-// return NULL;
-// }
-//
-// DictEntry = NewVirtualDictEntry(StuffGroup, FreeElem,
-// VR, "GDCM", NewTagName);
-// NewEntry = new gdcmDocEntry(DictEntry);
-// AddEntry(NewEntry);
-// 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.
- */
-guint32 gdcmDocument::GenerateFreeTagKeyInGroup(guint16 group)
-{
- for (guint32 elem = 0; elem < UINT32_MAX; elem++)
+ else
{
- TagKey key = gdcmDictEntry::TranslateToKey(group, elem);
- if (tagHT.count(key) == 0)
- return elem;
+ newEntry = NewDataEntry(CurrentGroup, CurrentElem, realVR);
+ static_cast<DataEntry *>(newEntry)->SetState(DataEntry::STATE_NOTLOADED);
}
- return UINT32_MAX;
-}
-
-/**
- * \brief Searches both the public and the shadow dictionary (when they
- * exist) for the presence of the DictEntry with given name.
- * The public dictionary has precedence on the shadow one.
- * @param Name name of the searched DictEntry
- * @return Corresponding DictEntry when it exists, NULL otherwise.
- */
-gdcmDictEntry *gdcmDocument::GetDictEntryByName(std::string Name)
-{
- gdcmDictEntry *found = (gdcmDictEntry *)0;
- if (!RefPubDict && !RefShaDict)
+ if ( vr == GDCM_VRUNKNOWN )
{
- dbg.Verbose(0, "gdcmDocument::GetDictEntry",
- "we SHOULD have a default dictionary");
+ if ( Filetype == ExplicitVR )
+ {
+ // We thought this was explicit VR, but we end up with an
+ // implicit VR tag. Let's backtrack.
+
+ //if ( newEntry->GetGroup() != 0xfffe )
+ if (CurrentGroup != 0xfffe)
+ {
+ int offset = Fp->tellg();//Only when heuristic for Explicit/Implicit was wrong
+
+ gdcmWarningMacro("Entry (" << newEntry->GetKey() << ") at x("
+ << std::hex << offset << ") should be Explicit VR");
+ }
+ }
+ newEntry->SetImplicitVR();
}
- if (RefPubDict)
+
+ try
{
- found = RefPubDict->GetDictEntryByName(Name);
- if (found)
- return found;
+ FindDocEntryLength(newEntry);
}
- if (RefShaDict)
+ catch ( FormatError )
{
- found = RefShaDict->GetDictEntryByName(Name);
- if (found)
- return found;
+ // Call it quits
+ newEntry->Delete();
+ return 0;
}
- return found;
+
+ newEntry->SetOffset(Fp->tellg()); // for each DocEntry
+ return newEntry;
}
/**
- * \brief Searches both the public and the shadow dictionary (when they
- * exist) for the presence of the DictEntry with given
- * group and element. The public dictionary has precedence on the
- * shadow one.
- * @param group group of the searched DictEntry
- * @param element element of the searched DictEntry
- * @return Corresponding DictEntry when it exists, NULL otherwise.
+ * \brief Handle broken private tag from Philips NTSCAN
+ * where the endianess is being switched to BigEndian
+ * for no apparent reason
+ * @return no return
*/
-gdcmDictEntry *gdcmDocument::GetDictEntryByNumber(guint16 group,guint16 element)
+void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem)
{
- gdcmDictEntry *found = (gdcmDictEntry *)0;
- if (!RefPubDict && !RefShaDict)
+ // for strange PMS Gyroscan Intera images
+ // Item 'starter' has a tag : 0x3f3f,0x3f00, for no apparent reason
+
+ // --- Feel free to remove this test *on your own coy of gdcm*
+ // if you are sure you'll never face this problem.
+
+ if ((group == 0x3f3f) && (elem == 0x3f00))
{
- dbg.Verbose(0, "gdcmDocument::GetDictEntry",
- "we SHOULD have a default dictionary");
+ // start endian swap mark for group found
+ gdcmDebugMacro( " delimiter 0x3f3f found." );
+ // fix the tag
+ group = 0xfffe;
+ elem = 0xe000;
+ return;
}
- if (RefPubDict)
+ // --- End of removable code
+
+ // Endian reversion.
+ // Some files contain groups of tags with reversed endianess.
+ static int reversedEndian = 0;
+ // try to fix endian switching in the middle of headers
+ if ((group == 0xfeff) && (elem == 0x00e0))
+ {
+ // start endian swap mark for group found
+ gdcmDebugMacro( "Start endian swap mark found." );
+ reversedEndian++;
+ SwitchByteSwapCode();
+ // fix the tag
+ group = 0xfffe;
+ elem = 0xe000;
+ }
+ else if (group == 0xfffe && elem == 0xe00d && reversedEndian)
+ {
+ // end of reversed endian group
+ gdcmDebugMacro( "End of reversed endian." );
+ reversedEndian--;
+ SwitchByteSwapCode();
+ }
+ else if (group == 0xfeff && elem == 0xdde0)
{
- found = RefPubDict->GetDictEntryByNumber(group, element);
- if (found)
- return found;
+ // reversed Sequence Terminator found
+ // probabely a bug in the header !
+ // Do what you want, it breaks !
+ //reversedEndian--;
+ //SwitchByteSwapCode();
+ gdcmWarningMacro( "Should never get here! reversed Sequence Terminator!" );
+ // fix the tag
+ group = 0xfffe;
+ elem = 0xe0dd;
}
- if (RefShaDict)
+ else if (group == 0xfffe && elem == 0xe0dd)
{
- found = RefShaDict->GetDictEntryByNumber(group, element);
- if (found)
- return found;
+ gdcmDebugMacro( "Straight Sequence Terminator." );
}
- return found;
}
+/**
+ * \brief Group 0002 is always coded Little Endian
+ * whatever Transfer Syntax is
+ * @return no return
+ */
+void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem)
+{
+ // Endian reversion.
+ // Some files contain groups of tags with reversed endianess.
+
+ Group0002Parsed = true;
+ // we just came out of group 0002
+ // if Transfer Syntax is Big Endian we have to change CheckSwap
+
+ std::string ts = GetTransferSyntax();
+ TS::SpecialType s = Global::GetTS()->GetSpecialTransferSyntax(ts);
+
+ // Group 0002 is always 'Explicit ...'
+ // even when Transfer Syntax says 'Implicit ..."
+
+ if ( s == TS::ImplicitVRLittleEndian
+ ||
+ s == TS::ImplicitVRBigEndianPrivateGE
+ )
+ {
+ Filetype = ImplicitVR;
+ }
+
+ // FIXME Strangely, this works with
+ //'Implicit VR BigEndian Transfer Syntax' (GE Private)
+ //
+ // --> Probabely normal, since we considered we never have
+ // to trust manufacturers.
+ // (we often find 'Implicit VR' tag,
+ // even when Transfer Syntax tells us it's Explicit ...
+
+ // NEVER trust the meta elements!
+ // (see what ezDICOM does ...)
+
+ /*
+ if ( s == TS::ExplicitVRBigEndian )
+ {
+ gdcmDebugMacro("Transfer Syntax Name = ["
+ << GetTransferSyntaxName() << "]" );
+ SwitchByteSwapCode();
+ group = SwapShort(group);
+ elem = SwapShort(elem);
+ }
+ */
+ //-- Broken ACR may start with a Shadow Group --
+ // worse : some ACR-NEMA like files start 00028 group ?!?
+ if ( !( (group >= 0x0001 && group <= 0x0008) || group == 0x0028 ) )
+ {
+ // We trust what we see.
+ SwitchByteSwapCode();
+ group = SwapShort(group);
+ elem = SwapShort(elem);
+ // not what we where told (by meta elements) !
+ gdcmDebugMacro("Transfer Syntax Name = ["
+ << GetTransferSyntaxName() << "]" );
+ }
+
+ /// \todo find a trick to warn user and stop processing
+
+ if ( s == TS::DeflatedExplicitVRLittleEndian)
+ {
+ gdcmWarningMacro("Transfer Syntax ["
+ << GetTransferSyntaxName() << "] :"
+ << " not yet dealt with ");
+ return;
+ }
+
+ // The following shouldn't occur very often
+ // Let's check at the very end.
+
+ if ( ts == GDCM_UNKNOWN )
+ {
+ gdcmDebugMacro("True DICOM File, with NO Transfer Syntax (?!) " );
+ return;
+ }
+
+ if ( !Global::GetTS()->IsTransferSyntax(ts) )
+ {
+ gdcmWarningMacro("True DICOM File, with illegal Transfer Syntax: ["
+ << ts << "]");
+ return;
+ }
+}
+//-----------------------------------------------------------------------------
+// Print
//-----------------------------------------------------------------------------
+} // end namespace gdcm
git://git.creatis.insa-lyon.fr / gdcm.git / blobdiff