1 /*=========================================================================
4 Module: $RCSfile: gdcmDocument.cxx,v $
6 Date: $Date: 2005/01/19 10:30:51 $
7 Version: $Revision: 1.199 $
9 Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de
10 l'Image). All rights reserved. See Doc/License.txt or
11 http://www.creatis.insa-lyon.fr/Public/Gdcm/License.html for details.
13 This software is distributed WITHOUT ANY WARRANTY; without even
14 the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 PURPOSE. See the above copyright notices for more information.
17 =========================================================================*/
19 #include "gdcmDocument.h"
20 #include "gdcmValEntry.h"
21 #include "gdcmBinEntry.h"
22 #include "gdcmSeqEntry.h"
23 #include "gdcmGlobal.h"
25 #include "gdcmDebug.h"
27 #include "gdcmException.h"
28 #include "gdcmDictSet.h"
29 #include "gdcmRLEFramesInfo.h"
30 #include "gdcmJPEGFragmentsInfo.h"
31 #include "gdcmDocEntrySet.h"
32 #include "gdcmSQItem.h"
38 #if defined(_MSC_VER) || defined(__BORLANDC__) || defined(__MINGW32__)
42 #ifdef CMAKE_HAVE_NETINET_IN_H
43 #include <netinet/in.h>
49 //-----------------------------------------------------------------------------
50 // Refer to Document::CheckSwap()
51 //const unsigned int Document::HEADER_LENGTH_TO_READ = 256;
53 // Refer to Document::SetMaxSizeLoadEntry()
54 const unsigned int Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096
55 const unsigned int Document::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff;
57 //-----------------------------------------------------------------------------
58 // Constructor / Destructor
62 * @param filename file to be opened for parsing
64 Document::Document( std::string const &filename ) : ElementSet(-1)
66 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
76 Group0002Parsed = false;
78 gdcmVerboseMacro( "Starting parsing of file: " << Filename.c_str());
79 // Fp->seekg( 0, std::ios::beg);
81 Fp->seekg(0, std::ios::end);
82 long lgt = Fp->tellg();
84 Fp->seekg( 0, std::ios::beg);
87 long beg = Fp->tellg();
90 ParseDES( this, beg, lgt, false); // Loading is done during parsing
92 Fp->seekg( 0, std::ios::beg);
94 // Load 'non string' values
96 std::string PhotometricInterpretation = GetEntry(0x0028,0x0004);
97 if( PhotometricInterpretation == "PALETTE COLOR " )
99 LoadEntryBinArea(0x0028,0x1200); // gray LUT
100 /// FIXME FIXME FIXME
101 /// The tags refered by the three following lines used to be CORRECTLY
102 /// defined as having an US Value Representation in the public
103 /// dictionary. BUT the semantics implied by the three following
104 /// lines state that the corresponding tag contents are in fact
105 /// the ones of a BinEntry.
106 /// In order to fix things "Quick and Dirty" the dictionary was
107 /// altered on PURPOSE but now contains a WRONG value.
108 /// In order to fix things and restore the dictionary to its
109 /// correct value, one needs to decided of the semantics by deciding
110 /// whether the following tags are either:
111 /// - multivaluated US, and hence loaded as ValEntry, but afterwards
112 /// also used as BinEntry, which requires the proper conversion,
113 /// - OW, and hence loaded as BinEntry, but afterwards also used
114 /// as ValEntry, which requires the proper conversion.
115 LoadEntryBinArea(0x0028,0x1201); // R LUT
116 LoadEntryBinArea(0x0028,0x1202); // G LUT
117 LoadEntryBinArea(0x0028,0x1203); // B LUT
119 // Segmented Red Palette Color LUT Data
120 LoadEntryBinArea(0x0028,0x1221);
121 // Segmented Green Palette Color LUT Data
122 LoadEntryBinArea(0x0028,0x1222);
123 // Segmented Blue Palette Color LUT Data
124 LoadEntryBinArea(0x0028,0x1223);
126 //FIXME later : how to use it?
127 LoadEntryBinArea(0x0028,0x3006); //LUT Data (CTX dependent)
131 // --------------------------------------------------------------
132 // Specific code to allow gdcm to read ACR-LibIDO formated images
133 // Note: ACR-LibIDO is an extension of the ACR standard that was
134 // used at CREATIS. For the time being (say a couple years)
135 // we keep this kludge to allow a smooth move to gdcm for
136 // CREATIS developpers (sorry folks).
138 // if recognition code tells us we deal with a LibIDO image
139 // we switch lineNumber and columnNumber
142 RecCode = GetEntry(0x0008, 0x0010); // recognition code (RET)
143 if (RecCode == "ACRNEMA_LIBIDO_1.1" ||
144 RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares
145 // with "little-endian strings"
147 Filetype = ACR_LIBIDO;
148 std::string rows = GetEntry(0x0028, 0x0010);
149 std::string columns = GetEntry(0x0028, 0x0011);
150 SetEntry(columns, 0x0028, 0x0010);
151 SetEntry(rows , 0x0028, 0x0011);
153 // ----------------- End of ACR-LibIDO kludge ------------------
157 * \brief This default constructor doesn't parse the file. You should
158 * then invoke \ref Document::SetFileName and then the parsing.
160 Document::Document() : ElementSet(-1)
164 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
167 Filetype = ExplicitVR;
168 Group0002Parsed = false;
172 * \brief Canonical destructor.
174 Document::~Document ()
183 //-----------------------------------------------------------------------------
187 * \brief Prints The Dict Entries of THE public Dicom Dictionary
190 void Document::PrintPubDict(std::ostream &os)
192 RefPubDict->SetPrintLevel(PrintLevel);
193 RefPubDict->Print(os);
197 * \brief Prints The Dict Entries of THE shadow Dicom Dictionary
200 void Document::PrintShaDict(std::ostream &os)
202 RefShaDict->SetPrintLevel(PrintLevel);
203 RefShaDict->Print(os);
206 //-----------------------------------------------------------------------------
209 * \brief Get the public dictionary used
211 Dict *Document::GetPubDict()
217 * \brief Get the shadow dictionary used
219 Dict *Document::GetShaDict()
225 * \brief Set the shadow dictionary used
226 * @param dict dictionary to use in shadow
228 bool Document::SetShaDict(Dict *dict)
235 * \brief Set the shadow dictionary used
236 * @param dictName name of the dictionary to use in shadow
238 bool Document::SetShaDict(DictKey const &dictName)
240 RefShaDict = Global::GetDicts()->GetDict(dictName);
245 * \brief This predicate, based on hopefully reasonable heuristics,
246 * decides whether or not the current Document was properly parsed
247 * and contains the mandatory information for being considered as
248 * a well formed and usable Dicom/Acr File.
249 * @return true when Document is the one of a reasonable Dicom/Acr file,
252 bool Document::IsReadable()
254 if( Filetype == Unknown)
256 gdcmVerboseMacro( "Wrong filetype");
262 gdcmVerboseMacro( "No tag in internal hash table.");
270 * \brief Accessor to the Transfer Syntax (when present) of the
271 * current document (it internally handles reading the
272 * value from disk when only parsing occured).
273 * @return The encountered Transfer Syntax of the current document.
275 std::string Document::GetTransferSyntax()
277 DocEntry *entry = GetDocEntry(0x0002, 0x0010);
283 // The entry might be present but not loaded (parsing and loading
284 // happen at different stages): try loading and proceed with check...
285 LoadDocEntrySafe(entry);
286 if (ValEntry *valEntry = dynamic_cast< ValEntry* >(entry) )
288 std::string transfer = valEntry->GetValue();
289 // The actual transfer (as read from disk) might be padded. We
290 // first need to remove the potential padding. We can make the
291 // weak assumption that padding was not executed with digits...
292 if ( transfer.length() == 0 )
294 // for brain damaged headers
297 while ( !isdigit((unsigned char)transfer[transfer.length()-1]) )
299 transfer.erase(transfer.length()-1, 1);
307 * \brief Predicate for dicom version 3 file.
308 * @return True when the file is a dicom version 3.
310 bool Document::IsDicomV3()
312 // Checking if Transfer Syntax exists is enough
313 // Anyway, it's to late check if the 'Preamble' was found ...
314 // And ... would it be a rich idea to check ?
315 // (some 'no Preamble' DICOM images exist !)
316 return GetDocEntry(0x0002, 0x0010) != NULL;
320 * \brief returns the File Type
321 * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
322 * @return the FileType code
324 FileType Document::GetFileType()
330 * \brief Tries to open the file \ref Document::Filename and
331 * checks the preamble when existing.
332 * @return The FILE pointer on success.
334 std::ifstream *Document::OpenFile()
337 HasDCMPreamble = false;
338 if (Filename.length() == 0)
345 gdcmVerboseMacro( "File already open: " << Filename.c_str());
348 Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary);
351 gdcmDebugMacro( "Cannot open file: " << Filename.c_str());
358 Fp->read((char*)&zero, (size_t)2);
365 //ACR -- or DICOM with no Preamble; may start with a Shadow Group --
367 zero == 0x0001 || zero == 0x0100 || zero == 0x0002 || zero == 0x0200 ||
368 zero == 0x0003 || zero == 0x0300 || zero == 0x0004 || zero == 0x0400 ||
369 zero == 0x0005 || zero == 0x0500 || zero == 0x0006 || zero == 0x0600 ||
370 zero == 0x0007 || zero == 0x0700 || zero == 0x0008 || zero == 0x0800 )
373 = Util::Format("ACR/DICOM with no preamble: (%04x)\n", zero);
374 gdcmVerboseMacro( msg.c_str() );
379 Fp->seekg(126L, std::ios::cur);
381 Fp->read(dicm, (size_t)4);
387 if( memcmp(dicm, "DICM", 4) == 0 )
389 HasDCMPreamble = true;
394 gdcmVerboseMacro( "Not DICOM/ACR (missing preamble)" << Filename.c_str());
400 * \brief closes the file
401 * @return TRUE if the close was successfull
403 bool Document::CloseFile()
412 return true; //FIXME how do we detect a non-closed ifstream ?
416 * \brief Writes in a file all the Header Entries (Dicom Elements)
417 * @param fp file pointer on an already open file (actually: Output File Stream)
418 * @param filetype Type of the File to be written
419 * (ACR-NEMA, ExplicitVR, ImplicitVR)
420 * \return Always true.
422 void Document::WriteContent(std::ofstream *fp, FileType filetype)
424 /// \todo move the following lines (and a lot of others, to be written)
425 /// to a future function CheckAndCorrectHeader
426 /// (necessary if user wants to write a DICOM V3 file
427 /// starting from an ACR-NEMA (V2) Header
429 if ( filetype == ImplicitVR || filetype == ExplicitVR )
431 // writing Dicom File Preamble
432 char filePreamble[128];
433 memset(filePreamble, 0, 128);
434 fp->write(filePreamble, 128);
435 fp->write("DICM", 4);
439 * \todo rewrite later, if really usefull
440 * - 'Group Length' element is optional in DICOM
441 * - but un-updated odd groups lengthes can causes pb
444 * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) )
445 * UpdateGroupLength(false,filetype);
446 * if ( filetype == ACR)
447 * UpdateGroupLength(true,ACR);
450 ElementSet::WriteContent(fp, filetype); // This one is recursive
454 * \brief Modifies the value of a given Doc Entry (Dicom Element)
455 * when it exists. Create it with the given value when unexistant.
456 * @param value (string) Value to be set
457 * @param group Group number of the Entry
458 * @param elem Element number of the Entry
459 * @param vr V(alue) R(epresentation) of the Entry -if private Entry-
460 * \return pointer to the modified/created Header Entry (NULL when creation
463 ValEntry *Document::ReplaceOrCreate(std::string const &value,
468 ValEntry *valEntry = 0;
469 DocEntry *currentEntry = GetDocEntry( group, elem);
473 valEntry = dynamic_cast< ValEntry* >(currentEntry);
477 if( valEntry->GetVR()!=vr )
480 // if currentEntry doesn't correspond to the requested valEntry
483 if (!RemoveEntry(currentEntry))
485 gdcmVerboseMacro( "Removal of previous DocEntry failed.");
492 // Create a new valEntry if necessary
495 valEntry = NewValEntry(group, elem, vr);
497 if ( !AddEntry(valEntry))
499 gdcmVerboseMacro("AddEntry failed although this is a creation.");
506 // Set the binEntry value
507 SetEntry(value, valEntry);
512 * \brief Modifies the value of a given Header Entry (Dicom Element)
513 * when it exists. Create it with the given value when unexistant.
514 * A copy of the binArea is made to be kept in the Document.
515 * @param binArea (binary) value to be set
516 * @param group Group number of the Entry
517 * @param elem Element number of the Entry
518 * @param vr V(alue) R(epresentation) of the Entry -if private Entry-
519 * \return pointer to the modified/created Header Entry (NULL when creation
522 BinEntry *Document::ReplaceOrCreate(uint8_t *binArea,
528 BinEntry *binEntry = 0;
529 DocEntry *currentEntry = GetDocEntry( group, elem);
531 // Verify the currentEntry
534 binEntry = dynamic_cast< BinEntry* >(currentEntry);
538 if( binEntry->GetVR()!=vr )
541 // if currentEntry doesn't correspond to the requested valEntry
544 if (!RemoveEntry(currentEntry))
546 gdcmVerboseMacro( "Removal of previous DocEntry failed.");
553 // Create a new binEntry if necessary
556 binEntry = NewBinEntry(group, elem, vr);
558 if ( !AddEntry(binEntry))
560 gdcmVerboseMacro( "AddEntry failed allthough this is a creation.");
567 // Set the binEntry value
569 if (lgth>0 && binArea)
571 tmpArea = new uint8_t[lgth];
572 memcpy(tmpArea,binArea,lgth);
578 if (!SetEntry(tmpArea,lgth,binEntry))
590 * \brief Modifies the value of a given Header Entry (Dicom Element)
591 * when it exists. Create it when unexistant.
592 * @param group Group number of the Entry
593 * @param elem Element number of the Entry
594 * \return pointer to the modified/created SeqEntry (NULL when creation
597 SeqEntry *Document::ReplaceOrCreate( uint16_t group, uint16_t elem)
599 SeqEntry *seqEntry = 0;
600 DocEntry *currentEntry = GetDocEntry( group, elem);
602 // Verify the currentEntry
605 seqEntry = dynamic_cast< SeqEntry* >(currentEntry);
609 if( seqEntry->GetVR()!="SQ" )
612 // if currentEntry doesn't correspond to the requested valEntry
615 if (!RemoveEntry(currentEntry))
617 gdcmVerboseMacro( "Removal of previous DocEntry failed.");
623 // Create a new seqEntry if necessary
626 seqEntry = NewSeqEntry(group, elem);
628 if ( !AddEntry(seqEntry))
630 gdcmVerboseMacro( "AddEntry failed allthough this is a creation.");
640 * \brief Set a new value if the invoked element exists
641 * Seems to be useless !!!
642 * @param value new element value
643 * @param group group number of the Entry
644 * @param elem element number of the Entry
647 bool Document::ReplaceIfExist(std::string const &value,
648 uint16_t group, uint16_t elem )
650 SetEntry(value, group, elem);
655 //-----------------------------------------------------------------------------
659 * \brief Checks if a given Dicom Element exists within the H table
660 * @param group Group number of the searched Dicom Element
661 * @param elem Element number of the searched Dicom Element
662 * @return true is found
664 bool Document::CheckIfEntryExist(uint16_t group, uint16_t elem )
666 const std::string &key = DictEntry::TranslateToKey(group, elem );
667 return TagHT.count(key) != 0;
672 * \brief Searches within Header Entries (Dicom Elements) parsed with
673 * the public and private dictionaries
674 * for the element value representation of a given tag.
675 * @param group Group number of the searched tag.
676 * @param elem Element number of the searched tag.
677 * @return Corresponding element value representation when it exists,
678 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
680 std::string Document::GetEntry(uint16_t group, uint16_t elem)
682 TagKey key = DictEntry::TranslateToKey(group, elem);
683 if ( !TagHT.count(key))
688 return ((ValEntry *)TagHT.find(key)->second)->GetValue();
692 * \brief Searches within Header Entries (Dicom Elements) parsed with
693 * the public and private dictionaries
694 * for the element value representation of a given tag..
695 * Obtaining the VR (Value Representation) might be needed by caller
696 * to convert the string typed content to caller's native type
697 * (think of C++ vs Python). The VR is actually of a higher level
698 * of semantics than just the native C++ type.
699 * @param group Group number of the searched tag.
700 * @param elem Element number of the searched tag.
701 * @return Corresponding element value representation when it exists,
702 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
704 std::string Document::GetEntryVR(uint16_t group, uint16_t elem)
706 DocEntry *element = GetDocEntry(group, elem);
711 return element->GetVR();
715 * \brief Searches within Header Entries (Dicom Elements) parsed with
716 * the public and private dictionaries
717 * for the value length of a given tag..
718 * @param group Group number of the searched tag.
719 * @param elem Element number of the searched tag.
720 * @return Corresponding element length; -2 if not found
722 int Document::GetEntryLength(uint16_t group, uint16_t elem)
724 DocEntry *element = GetDocEntry(group, elem);
727 return -2; //magic number
729 return element->GetLength();
733 * \brief Accesses an existing DocEntry (i.e. a Dicom Element)
734 * through it's (group, element) and modifies it's content with
736 * @param content new value (string) to substitute with
737 * @param group group number of the Dicom Element to modify
738 * @param elem element number of the Dicom Element to modify
740 bool Document::SetEntry(std::string const& content,
741 uint16_t group, uint16_t elem)
743 ValEntry *entry = GetValEntry(group, elem);
746 gdcmVerboseMacro( "No corresponding ValEntry (try promotion first).");
749 return SetEntry(content,entry);
753 * \brief Accesses an existing DocEntry (i.e. a Dicom Element)
754 * through it's (group, element) and modifies it's content with
756 * @param content new value (void* -> uint8_t*) to substitute with
757 * @param lgth new value length
758 * @param group group number of the Dicom Element to modify
759 * @param elem element number of the Dicom Element to modify
761 bool Document::SetEntry(uint8_t*content, int lgth,
762 uint16_t group, uint16_t elem)
764 BinEntry *entry = GetBinEntry(group, elem);
767 gdcmVerboseMacro( "No corresponding ValEntry (try promotion first).");
771 return SetEntry(content,lgth,entry);
775 * \brief Accesses an existing DocEntry (i.e. a Dicom Element)
776 * and modifies it's content with the given value.
777 * @param content new value (string) to substitute with
778 * @param entry Entry to be modified
780 bool Document::SetEntry(std::string const &content, ValEntry *entry)
784 entry->SetValue(content);
791 * \brief Accesses an existing BinEntry (i.e. a Dicom Element)
792 * and modifies it's content with the given value.
793 * @param content new value (void* -> uint8_t*) to substitute with
794 * @param entry Entry to be modified
795 * @param lgth new value length
797 bool Document::SetEntry(uint8_t *content, int lgth, BinEntry *entry)
801 // Hope Binary field length is *never* wrong
802 /*if(lgth%2) // Non even length are padded with a space (020H).
805 //content = content + '\0'; // fing a trick to enlarge a binary field?
808 entry->SetBinArea(content);
809 entry->SetLength(lgth);
810 entry->SetValue(GDCM_BINLOADED);
817 * \brief Gets (from Header) a 'non string' element value
818 * (LoadElementValues has already be executed)
819 * @param group group number of the Entry
820 * @param elem element number of the Entry
821 * @return Pointer to the 'non string' area
823 void *Document::GetEntryBinArea(uint16_t group, uint16_t elem)
825 DocEntry *entry = GetDocEntry(group, elem);
828 gdcmVerboseMacro( "No entry");
831 if ( BinEntry *binEntry = dynamic_cast<BinEntry*>(entry) )
833 return binEntry->GetBinArea();
840 * \brief Loads (from disk) the element content
841 * when a string is not suitable
842 * @param group group number of the Entry
843 * @param elem element number of the Entry
845 void Document::LoadEntryBinArea(uint16_t group, uint16_t elem)
847 // Search the corresponding DocEntry
848 DocEntry *docElement = GetDocEntry(group, elem);
852 BinEntry *binElement = dynamic_cast<BinEntry *>(docElement);
856 LoadEntryBinArea(binElement);
860 * \brief Loads (from disk) the element content
861 * when a string is not suitable
862 * @param elem Entry whose binArea is going to be loaded
864 void Document::LoadEntryBinArea(BinEntry *elem)
866 if(elem->GetBinArea())
873 size_t o =(size_t)elem->GetOffset();
874 Fp->seekg(o, std::ios::beg);
876 size_t l = elem->GetLength();
877 uint8_t *a = new uint8_t[l];
880 gdcmVerboseMacro( "Cannot allocate BinEntry content");
884 /// \todo check the result
885 Fp->read((char*)a, l);
886 if( Fp->fail() || Fp->eof())
899 * \brief Sets a 'non string' value to a given Dicom Element
900 * @param area area containing the 'non string' value
901 * @param group Group number of the searched Dicom Element
902 * @param elem Element number of the searched Dicom Element
905 /*bool Document::SetEntryBinArea(uint8_t *area,
906 uint16_t group, uint16_t elem)
908 DocEntry *currentEntry = GetDocEntry(group, elem);
914 if ( BinEntry *binEntry = dynamic_cast<BinEntry*>(currentEntry) )
916 binEntry->SetBinArea( area );
924 * \brief retrieves a Dicom Element using (group, element)
925 * @param group Group number of the searched Dicom Element
926 * @param elem Element number of the searched Dicom Element
929 DocEntry *Document::GetDocEntry(uint16_t group, uint16_t elem)
931 TagKey key = DictEntry::TranslateToKey(group, elem);
932 if ( !TagHT.count(key))
936 return TagHT.find(key)->second;
940 * \brief Same as \ref Document::GetDocEntry except it only
941 * returns a result when the corresponding entry is of type
943 * @param group Group number of the searched Dicom Element
944 * @param elem Element number of the searched Dicom Element
945 * @return When present, the corresponding ValEntry.
947 ValEntry *Document::GetValEntry(uint16_t group, uint16_t elem)
949 DocEntry *currentEntry = GetDocEntry(group, elem);
954 if ( ValEntry *entry = dynamic_cast<ValEntry*>(currentEntry) )
958 gdcmVerboseMacro( "Unfound ValEntry.");
964 * \brief Same as \ref Document::GetDocEntry except it only
965 * returns a result when the corresponding entry is of type
967 * @param group Group number of the searched Dicom Element
968 * @param elem Element number of the searched Dicom Element
969 * @return When present, the corresponding BinEntry.
971 BinEntry *Document::GetBinEntry(uint16_t group, uint16_t elem)
973 DocEntry *currentEntry = GetDocEntry(group, elem);
978 if ( BinEntry *entry = dynamic_cast<BinEntry*>(currentEntry) )
982 gdcmVerboseMacro( "Unfound BinEntry.");
988 * \brief Same as \ref Document::GetDocEntry except it only
989 * returns a result when the corresponding entry is of type
991 * @param group Group number of the searched Dicom Element
992 * @param elem Element number of the searched Dicom Element
993 * @return When present, the corresponding SeqEntry.
995 SeqEntry *Document::GetSeqEntry(uint16_t group, uint16_t elem)
997 DocEntry *currentEntry = GetDocEntry(group, elem);
1002 if ( SeqEntry *entry = dynamic_cast<SeqEntry*>(currentEntry) )
1006 gdcmVerboseMacro( "Unfound SeqEntry.");
1013 * \brief Loads the element while preserving the current
1014 * underlying file position indicator as opposed to
1015 * LoadDocEntry that modifies it.
1016 * @param entry Header Entry whose value will be loaded.
1019 void Document::LoadDocEntrySafe(DocEntry *entry)
1023 long PositionOnEntry = Fp->tellg();
1024 LoadDocEntry(entry);
1025 Fp->seekg(PositionOnEntry, std::ios::beg);
1030 * \brief Swaps back the bytes of 4-byte long integer accordingly to
1032 * @return The properly swaped 32 bits integer.
1034 uint32_t Document::SwapLong(uint32_t a)
1041 a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) |
1042 ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) );
1045 a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
1048 a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
1051 gdcmErrorMacro( "Unset swap code:" << SwapCode );
1058 * \brief Unswaps back the bytes of 4-byte long integer accordingly to
1060 * @return The properly unswaped 32 bits integer.
1062 uint32_t Document::UnswapLong(uint32_t a)
1068 * \brief Swaps the bytes so they agree with the processor order
1069 * @return The properly swaped 16 bits integer.
1071 uint16_t Document::SwapShort(uint16_t a)
1073 if ( SwapCode == 4321 || SwapCode == 2143 )
1075 a = ((( a << 8 ) & 0x0ff00 ) | (( a >> 8 ) & 0x00ff ) );
1081 * \brief Unswaps the bytes so they agree with the processor order
1082 * @return The properly unswaped 16 bits integer.
1084 uint16_t Document::UnswapShort(uint16_t a)
1086 return SwapShort(a);
1089 //-----------------------------------------------------------------------------
1093 * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries)
1094 * @return length of the parsed set.
1096 void Document::ParseDES(DocEntrySet *set, long offset,
1097 long l_max, bool delim_mode)
1099 DocEntry *newDocEntry = 0;
1100 ValEntry *newValEntry;
1101 BinEntry *newBinEntry;
1102 SeqEntry *newSeqEntry;
1108 if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max)
1114 newDocEntry = ReadNextDocEntry( );
1121 vr = newDocEntry->GetVR();
1122 newValEntry = dynamic_cast<ValEntry*>(newDocEntry);
1123 newBinEntry = dynamic_cast<BinEntry*>(newDocEntry);
1124 newSeqEntry = dynamic_cast<SeqEntry*>(newDocEntry);
1126 if ( newValEntry || newBinEntry )
1130 if ( Filetype == ExplicitVR && ! Global::GetVR()->IsVROfBinaryRepresentable(vr) )
1132 ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR
1133 gdcmVerboseMacro( std::hex << newDocEntry->GetGroup()
1134 << "|" << newDocEntry->GetElement()
1135 << " : Neither Valentry, nor BinEntry."
1136 "Probably unknown VR.");
1139 //////////////////// BinEntry or UNKOWN VR:
1140 // When "this" is a Document the Key is simply of the
1141 // form ( group, elem )...
1142 if ( dynamic_cast< Document* > ( set ) )
1144 newBinEntry->SetKey( newBinEntry->GetKey() );
1146 // but when "this" is a SQItem, we are inserting this new
1147 // valEntry in a sequence item, and the key has the
1148 // generalized form (refer to \ref BaseTagKey):
1149 if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) )
1151 newBinEntry->SetKey( parentSQItem->GetBaseTagKey()
1152 + newBinEntry->GetKey() );
1155 LoadDocEntry( newBinEntry );
1156 if( !set->AddEntry( newBinEntry ) )
1158 //Expect big troubles if here
1159 //delete newBinEntry;
1165 /////////////////////// ValEntry
1166 // When "set" is a Document, then we are at the top of the
1167 // hierarchy and the Key is simply of the form ( group, elem )...
1168 if ( dynamic_cast< Document* > ( set ) )
1170 newValEntry->SetKey( newValEntry->GetKey() );
1172 // ...but when "set" is a SQItem, we are inserting this new
1173 // valEntry in a sequence item. Hence the key has the
1174 // generalized form (refer to \ref BaseTagKey):
1175 if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) )
1177 newValEntry->SetKey( parentSQItem->GetBaseTagKey()
1178 + newValEntry->GetKey() );
1181 LoadDocEntry( newValEntry );
1182 bool delimitor=newValEntry->IsItemDelimitor();
1183 if( !set->AddEntry( newValEntry ) )
1185 // If here expect big troubles
1186 //delete newValEntry; //otherwise mem leak
1196 if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max)
1204 if ( ( newDocEntry->GetGroup() == 0x7fe0 )
1205 && ( newDocEntry->GetElement() == 0x0010 ) )
1207 std::string ts = GetTransferSyntax();
1208 if ( Global::GetTS()->IsRLELossless(ts) )
1210 long positionOnEntry = Fp->tellg();
1211 Fp->seekg( newDocEntry->GetOffset(), std::ios::beg );
1213 Fp->seekg( positionOnEntry, std::ios::beg );
1215 else if ( Global::GetTS()->IsJPEG(ts) )
1217 long positionOnEntry = Fp->tellg();
1218 Fp->seekg( newDocEntry->GetOffset(), std::ios::beg );
1219 ComputeJPEGFragmentInfo();
1220 Fp->seekg( positionOnEntry, std::ios::beg );
1224 // Just to make sure we are at the beginning of next entry.
1225 SkipToNextDocEntry(newDocEntry);
1230 unsigned long l = newDocEntry->GetReadLength();
1231 if ( l != 0 ) // don't mess the delim_mode for zero-length sequence
1233 if ( l == 0xffffffff )
1242 // no other way to create it ...
1243 newSeqEntry->SetDelimitorMode( delim_mode );
1245 // At the top of the hierarchy, stands a Document. When "set"
1246 // is a Document, then we are building the first depth level.
1247 // Hence the SeqEntry we are building simply has a depth
1249 if (/*Document *dummy =*/ dynamic_cast< Document* > ( set ) )
1252 newSeqEntry->SetDepthLevel( 1 );
1253 newSeqEntry->SetKey( newSeqEntry->GetKey() );
1255 // But when "set" is already a SQItem, we are building a nested
1256 // sequence, and hence the depth level of the new SeqEntry
1257 // we are building, is one level deeper:
1258 if (SQItem *parentSQItem = dynamic_cast< SQItem* > ( set ) )
1260 newSeqEntry->SetDepthLevel( parentSQItem->GetDepthLevel() + 1 );
1261 newSeqEntry->SetKey( parentSQItem->GetBaseTagKey()
1262 + newSeqEntry->GetKey() );
1266 { // Don't try to parse zero-length sequences
1267 ParseSQ( newSeqEntry,
1268 newDocEntry->GetOffset(),
1271 set->AddEntry( newSeqEntry );
1272 if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max)
1284 * \brief Parses a Sequence ( SeqEntry after SeqEntry)
1285 * @return parsed length for this level
1287 void Document::ParseSQ( SeqEntry *seqEntry,
1288 long offset, long l_max, bool delim_mode)
1290 int SQItemNumber = 0;
1292 long offsetStartCurrentSQItem = offset;
1296 // the first time, we read the fff0,e000 of the first SQItem
1297 DocEntry *newDocEntry = ReadNextDocEntry();
1301 // FIXME Should warn user
1306 if ( newDocEntry->IsSequenceDelimitor() )
1308 seqEntry->SetSequenceDelimitationItem( newDocEntry );
1312 if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max)
1317 // create the current SQItem
1318 SQItem *itemSQ = new SQItem( seqEntry->GetDepthLevel() );
1319 std::ostringstream newBase;
1320 newBase << seqEntry->GetKey()
1324 itemSQ->SetBaseTagKey( newBase.str() );
1325 unsigned int l = newDocEntry->GetReadLength();
1327 if ( l == 0xffffffff )
1336 // when we're here, element fffe,e000 is already passed.
1337 // it's lost for the SQItem we're going to process !!
1339 //ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod);
1340 //delete newDocEntry; // FIXME well ... it's too late to use it !
1342 // Let's try :------------
1343 // remove fff0,e000, created out of the SQItem
1345 Fp->seekg(offsetStartCurrentSQItem, std::ios::beg);
1346 // fill up the current SQItem, starting at the beginning of fff0,e000
1347 ParseDES(itemSQ, offsetStartCurrentSQItem, l+8, dlm_mod);
1348 offsetStartCurrentSQItem = Fp->tellg();
1349 // end try -----------------
1351 seqEntry->AddEntry( itemSQ, SQItemNumber );
1353 if ( !delim_mode && ((long)(Fp->tellg())-offset ) >= l_max )
1361 * \brief Loads the element content if its length doesn't exceed
1362 * the value specified with Document::SetMaxSizeLoadEntry()
1363 * @param entry Header Entry (Dicom Element) to be dealt with
1365 void Document::LoadDocEntry(DocEntry *entry)
1367 uint16_t group = entry->GetGroup();
1368 std::string vr = entry->GetVR();
1369 uint32_t length = entry->GetLength();
1371 Fp->seekg((long)entry->GetOffset(), std::ios::beg);
1373 // A SeQuence "contains" a set of Elements.
1374 // (fffe e000) tells us an Element is beginning
1375 // (fffe e00d) tells us an Element just ended
1376 // (fffe e0dd) tells us the current SeQuence just ended
1377 if( group == 0xfffe )
1379 // NO more value field for SQ !
1383 // When the length is zero things are easy:
1386 ((ValEntry *)entry)->SetValue("");
1390 // The elements whose length is bigger than the specified upper bound
1391 // are not loaded. Instead we leave a short notice of the offset of
1392 // the element content and it's length.
1394 std::ostringstream s;
1395 if (length > MaxSizeLoadEntry)
1397 if (BinEntry *binEntryPtr = dynamic_cast< BinEntry* >(entry) )
1399 //s << "gdcm::NotLoaded (BinEntry)";
1400 s << GDCM_NOTLOADED;
1401 s << " Address:" << (long)entry->GetOffset();
1402 s << " Length:" << entry->GetLength();
1403 s << " x(" << std::hex << entry->GetLength() << ")";
1404 binEntryPtr->SetValue(s.str());
1406 // Be carefull : a BinEntry IS_A ValEntry ...
1407 else if (ValEntry *valEntryPtr = dynamic_cast< ValEntry* >(entry) )
1409 // s << "gdcm::NotLoaded. (ValEntry)";
1410 s << GDCM_NOTLOADED;
1411 s << " Address:" << (long)entry->GetOffset();
1412 s << " Length:" << entry->GetLength();
1413 s << " x(" << std::hex << entry->GetLength() << ")";
1414 valEntryPtr->SetValue(s.str());
1419 gdcmErrorMacro( "MaxSizeLoadEntry exceeded, neither a BinEntry "
1420 << "nor a ValEntry ?! Should never print that !" );
1423 // to be sure we are at the end of the value ...
1424 Fp->seekg((long)entry->GetOffset()+(long)entry->GetLength(),
1429 // When we find a BinEntry not very much can be done :
1430 if (BinEntry *binEntryPtr = dynamic_cast< BinEntry* >(entry) )
1432 s << GDCM_BINLOADED;
1433 binEntryPtr->SetValue(s.str());
1434 LoadEntryBinArea(binEntryPtr); // last one, not to erase length !
1438 /// \todo Any compacter code suggested (?)
1439 if ( IsDocEntryAnInteger(entry) )
1443 // When short integer(s) are expected, read and convert the following
1444 // n *two characters properly i.e. consider them as short integers as
1445 // opposed to strings.
1446 // Elements with Value Multiplicity > 1
1447 // contain a set of integers (not a single one)
1448 if (vr == "US" || vr == "SS")
1451 NewInt = ReadInt16();
1455 for (int i=1; i < nbInt; i++)
1458 NewInt = ReadInt16();
1463 // See above comment on multiple integers (mutatis mutandis).
1464 else if (vr == "UL" || vr == "SL")
1467 NewInt = ReadInt32();
1471 for (int i=1; i < nbInt; i++)
1474 NewInt = ReadInt32();
1479 #ifdef GDCM_NO_ANSI_STRING_STREAM
1480 s << std::ends; // to avoid oddities on Solaris
1481 #endif //GDCM_NO_ANSI_STRING_STREAM
1483 ((ValEntry *)entry)->SetValue(s.str());
1487 // FIXME: We need an additional byte for storing \0 that is not on disk
1488 char *str = new char[length+1];
1489 Fp->read(str, (size_t)length);
1490 str[length] = '\0'; //this is only useful when length is odd
1491 // Special DicomString call to properly handle \0 and even length
1492 std::string newValue;
1495 newValue = Util::DicomString(str, length+1);
1496 gdcmVerboseMacro("Warning: bad length: " << length <<
1497 ",For string :" << newValue.c_str());
1498 // Since we change the length of string update it length
1499 //entry->SetReadLength(length+1);
1503 newValue = Util::DicomString(str, length);
1507 if ( ValEntry *valEntry = dynamic_cast<ValEntry* >(entry) )
1509 if ( Fp->fail() || Fp->eof())
1511 gdcmVerboseMacro("Unread element value");
1512 valEntry->SetValue(GDCM_UNREAD);
1518 // Because of correspondance with the VR dic
1519 valEntry->SetValue(newValue);
1523 valEntry->SetValue(newValue);
1528 gdcmErrorMacro( "Should have a ValEntry, here !");
1534 * \brief Find the value Length of the passed Header Entry
1535 * @param entry Header Entry whose length of the value shall be loaded.
1537 void Document::FindDocEntryLength( DocEntry *entry )
1538 throw ( FormatError )
1540 std::string vr = entry->GetVR();
1543 if ( Filetype == ExplicitVR && !entry->IsImplicitVR() )
1545 if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UN" )
1547 // The following reserved two bytes (see PS 3.5-2003, section
1548 // "7.1.2 Data element structure with explicit vr", p 27) must be
1549 // skipped before proceeding on reading the length on 4 bytes.
1550 Fp->seekg( 2L, std::ios::cur);
1551 uint32_t length32 = ReadInt32();
1553 if ( (vr == "OB" || vr == "OW") && length32 == 0xffffffff )
1558 lengthOB = FindDocEntryLengthOBOrOW();
1560 catch ( FormatUnexpected )
1562 // Computing the length failed (this happens with broken
1563 // files like gdcm-JPEG-LossLess3a.dcm). We still have a
1564 // chance to get the pixels by deciding the element goes
1565 // until the end of the file. Hence we artificially fix the
1566 // the length and proceed.
1567 long currentPosition = Fp->tellg();
1568 Fp->seekg(0L,std::ios::end);
1570 long lengthUntilEOF = (long)(Fp->tellg())-currentPosition;
1571 Fp->seekg(currentPosition, std::ios::beg);
1573 entry->SetReadLength(lengthUntilEOF);
1574 entry->SetLength(lengthUntilEOF);
1577 entry->SetReadLength(lengthOB);
1578 entry->SetLength(lengthOB);
1581 FixDocEntryFoundLength(entry, length32);
1585 // Length is encoded on 2 bytes.
1586 length16 = ReadInt16();
1588 // FIXME : This heuristic supposes that the first group following
1589 // group 0002 *has* and element 0000.
1590 // BUT ... Element 0000 is optionnal :-(
1593 // Fixed using : HandleOutOfGroup0002()
1594 // (first hereafter strategy ...)
1596 // We can tell the current file is encoded in big endian (like
1597 // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag
1598 // and it's value is the one of the encoding of a big endian file.
1599 // In order to deal with such big endian encoded files, we have
1600 // (at least) two strategies:
1601 // * when we load the "Transfer Syntax" tag with value of big endian
1602 // encoding, we raise the proper flags. Then we wait for the end
1603 // of the META group (0x0002) among which is "Transfer Syntax",
1604 // before switching the swap code to big endian. We have to postpone
1605 // the switching of the swap code since the META group is fully encoded
1606 // in little endian, and big endian coding only starts at the next
1607 // group. The corresponding code can be hard to analyse and adds
1608 // many additional unnecessary tests for regular tags.
1609 // * the second strategy consists in waiting for trouble, that shall
1610 // appear when we find the first group with big endian encoding. This
1611 // is easy to detect since the length of a "Group Length" tag (the
1612 // ones with zero as element number) has to be of 4 (0x0004). When we
1613 // encounter 1024 (0x0400) chances are the encoding changed and we
1614 // found a group with big endian encoding.
1615 //---> Unfortunately, element 0000 is optional.
1616 //---> This will not work when missing!
1617 // We shall use this second strategy. In order to make sure that we
1618 // can interpret the presence of an apparently big endian encoded
1619 // length of a "Group Length" without committing a big mistake, we
1620 // add an additional check: we look in the already parsed elements
1621 // for the presence of a "Transfer Syntax" whose value has to be "big
1622 // endian encoding". When this is the case, chances are we have got our
1623 // hands on a big endian encoded file: we switch the swap code to
1624 // big endian and proceed...
1627 // if ( element == 0x0000 && length16 == 0x0400 )
1629 // std::string ts = GetTransferSyntax();
1630 // if ( Global::GetTS()->GetSpecialTransferSyntax(ts)
1631 // != TS::ExplicitVRBigEndian )
1633 // throw FormatError( "Document::FindDocEntryLength()",
1634 // " not explicit VR." );
1638 // SwitchByteSwapCode();
1640 // Restore the unproperly loaded values i.e. the group, the element
1641 // and the dictionary entry depending on them.
1642 // uint16_t correctGroup = SwapShort( entry->GetGroup() );
1643 // uint16_t correctElem = SwapShort( entry->GetElement() );
1644 // DictEntry *newTag = GetDictEntry( correctGroup, correctElem ); if ( !newTag )
1646 // This correct tag is not in the dictionary. Create a new one.
1647 // newTag = NewVirtualDictEntry(correctGroup, correctElem);
1649 // FIXME this can create a memory leaks on the old entry that be
1650 // left unreferenced.
1651 // entry->SetDictEntry( newTag );
1655 // 0xffff means that we deal with 'No Length' Sequence
1656 // or 'No Length' SQItem
1657 if ( length16 == 0xffff)
1661 FixDocEntryFoundLength( entry, (uint32_t)length16 );
1666 // Either implicit VR or a non DICOM conformal (see note below) explicit
1667 // VR that ommited the VR of (at least) this element. Farts happen.
1668 // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25
1669 // on Data elements "Implicit and Explicit VR Data Elements shall
1670 // not coexist in a Data Set and Data Sets nested within it".]
1671 // Length is on 4 bytes.
1673 // Well ... group 0002 is always coded in 'Explicit VR Litle Endian'
1674 // even if Transfer Syntax is 'Implicit VR ...'
1676 FixDocEntryFoundLength( entry, ReadInt32() );
1682 * \brief Find the Value Representation of the current Dicom Element.
1683 * @return Value Representation of the current Entry
1685 std::string Document::FindDocEntryVR()
1687 if ( Filetype != ExplicitVR )
1688 return GDCM_UNKNOWN;
1690 long positionOnEntry = Fp->tellg();
1691 // Warning: we believe this is explicit VR (Value Representation) because
1692 // we used a heuristic that found "UL" in the first tag. Alas this
1693 // doesn't guarantee that all the tags will be in explicit VR. In some
1694 // cases (see e-film filtered files) one finds implicit VR tags mixed
1695 // within an explicit VR file. Hence we make sure the present tag
1696 // is in explicit VR and try to fix things if it happens not to be
1700 Fp->read (vr, (size_t)2);
1703 if( !CheckDocEntryVR(vr) )
1705 Fp->seekg(positionOnEntry, std::ios::beg);
1706 return GDCM_UNKNOWN;
1712 * \brief Check the correspondance between the VR of the header entry
1713 * and the taken VR. If they are different, the header entry is
1714 * updated with the new VR.
1715 * @param vr Dicom Value Representation
1716 * @return false if the VR is incorrect of if the VR isn't referenced
1717 * otherwise, it returns true
1719 bool Document::CheckDocEntryVR(VRKey vr)
1721 // CLEANME searching the dicom_vr at each occurence is expensive.
1722 // PostPone this test in an optional integrity check at the end
1723 // of parsing or only in debug mode.
1724 if ( !Global::GetVR()->IsValidVR(vr) )
1731 * \brief Get the transformed value of the header entry. The VR value
1732 * is used to define the transformation to operate on the value
1733 * \warning NOT end user intended method !
1734 * @param entry entry to tranform
1735 * @return Transformed entry value
1737 std::string Document::GetDocEntryValue(DocEntry *entry)
1739 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1741 std::string val = ((ValEntry *)entry)->GetValue();
1742 std::string vr = entry->GetVR();
1743 uint32_t length = entry->GetLength();
1744 std::ostringstream s;
1747 // When short integer(s) are expected, read and convert the following
1748 // n * 2 bytes properly i.e. as a multivaluated strings
1749 // (each single value is separated fromthe next one by '\'
1750 // as usual for standard multivaluated filels
1751 // Elements with Value Multiplicity > 1
1752 // contain a set of short integers (not a single one)
1754 if( vr == "US" || vr == "SS" )
1759 for (int i=0; i < nbInt; i++)
1765 newInt16 = ( val[2*i+0] & 0xFF ) + ( ( val[2*i+1] & 0xFF ) << 8);
1766 newInt16 = SwapShort( newInt16 );
1771 // When integer(s) are expected, read and convert the following
1772 // n * 4 bytes properly i.e. as a multivaluated strings
1773 // (each single value is separated fromthe next one by '\'
1774 // as usual for standard multivaluated filels
1775 // Elements with Value Multiplicity > 1
1776 // contain a set of integers (not a single one)
1777 else if( vr == "UL" || vr == "SL" )
1782 for (int i=0; i < nbInt; i++)
1788 newInt32 = ( val[4*i+0] & 0xFF )
1789 + (( val[4*i+1] & 0xFF ) << 8 )
1790 + (( val[4*i+2] & 0xFF ) << 16 )
1791 + (( val[4*i+3] & 0xFF ) << 24 );
1792 newInt32 = SwapLong( newInt32 );
1796 #ifdef GDCM_NO_ANSI_STRING_STREAM
1797 s << std::ends; // to avoid oddities on Solaris
1798 #endif //GDCM_NO_ANSI_STRING_STREAM
1802 return ((ValEntry *)entry)->GetValue();
1806 * \brief Get the reverse transformed value of the header entry. The VR
1807 * value is used to define the reverse transformation to operate on
1809 * \warning NOT end user intended method !
1810 * @param entry Entry to reverse transform
1811 * @return Reverse transformed entry value
1813 std::string Document::GetDocEntryUnvalue(DocEntry *entry)
1815 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1817 std::string vr = entry->GetVR();
1818 std::vector<std::string> tokens;
1819 std::ostringstream s;
1821 if ( vr == "US" || vr == "SS" )
1825 tokens.erase( tokens.begin(), tokens.end()); // clean any previous value
1826 Util::Tokenize (((ValEntry *)entry)->GetValue(), tokens, "\\");
1827 for (unsigned int i=0; i<tokens.size(); i++)
1829 newInt16 = atoi(tokens[i].c_str());
1830 s << ( newInt16 & 0xFF )
1831 << (( newInt16 >> 8 ) & 0xFF );
1835 if ( vr == "UL" || vr == "SL")
1839 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1840 Util::Tokenize (((ValEntry *)entry)->GetValue(), tokens, "\\");
1841 for (unsigned int i=0; i<tokens.size();i++)
1843 newInt32 = atoi(tokens[i].c_str());
1844 s << (char)( newInt32 & 0xFF )
1845 << (char)(( newInt32 >> 8 ) & 0xFF )
1846 << (char)(( newInt32 >> 16 ) & 0xFF )
1847 << (char)(( newInt32 >> 24 ) & 0xFF );
1852 #ifdef GDCM_NO_ANSI_STRING_STREAM
1853 s << std::ends; // to avoid oddities on Solaris
1854 #endif //GDCM_NO_ANSI_STRING_STREAM
1858 return ((ValEntry *)entry)->GetValue();
1862 * \brief Skip a given Header Entry
1863 * \warning NOT end user intended method !
1864 * @param entry entry to skip
1866 void Document::SkipDocEntry(DocEntry *entry)
1868 SkipBytes(entry->GetLength());
1872 * \brief Skips to the begining of the next Header Entry
1873 * \warning NOT end user intended method !
1874 * @param currentDocEntry entry to skip
1876 void Document::SkipToNextDocEntry(DocEntry *currentDocEntry)
1878 Fp->seekg((long)(currentDocEntry->GetOffset()), std::ios::beg);
1879 if (currentDocEntry->GetGroup() != 0xfffe) // for fffe pb
1880 Fp->seekg( (long)(currentDocEntry->GetReadLength()),std::ios::cur);
1884 * \brief When the length of an element value is obviously wrong (because
1885 * the parser went Jabberwocky) one can hope improving things by
1886 * applying some heuristics.
1887 * @param entry entry to check
1888 * @param foundLength first assumption about length
1890 void Document::FixDocEntryFoundLength(DocEntry *entry,
1891 uint32_t foundLength)
1893 entry->SetReadLength( foundLength ); // will be updated only if a bug is found
1894 if ( foundLength == 0xffffffff)
1899 uint16_t gr = entry->GetGroup();
1900 uint16_t elem = entry->GetElement();
1902 if ( foundLength % 2)
1904 gdcmVerboseMacro( "Warning : Tag with uneven length " << foundLength
1905 << " in x(" << std::hex << gr << "," << elem <<")");
1908 //////// Fix for some naughty General Electric images.
1909 // Allthough not recent many such GE corrupted images are still present
1910 // on Creatis hard disks. Hence this fix shall remain when such images
1911 // are no longer in use (we are talking a few years, here)...
1912 // Note: XMedCom probably uses such a trick since it is able to read
1913 // those pesky GE images ...
1914 if ( foundLength == 13)
1916 // Only happens for this length !
1917 if ( gr != 0x0008 || ( elem != 0x0070 && elem != 0x0080 ) )
1920 entry->SetReadLength(10); /// \todo a bug is to be fixed !?
1924 //////// Fix for some brain-dead 'Leonardo' Siemens images.
1925 // Occurence of such images is quite low (unless one leaves close to a
1926 // 'Leonardo' source. Hence, one might consider commenting out the
1927 // following fix on efficiency reasons.
1928 else if ( gr == 0x0009 && ( elem == 0x1113 || elem == 0x1114 ) )
1931 entry->SetReadLength(4); /// \todo a bug is to be fixed !?
1934 else if ( entry->GetVR() == "SQ" )
1936 foundLength = 0; // ReadLength is unchanged
1939 //////// We encountered a 'delimiter' element i.e. a tag of the form
1940 // "fffe|xxxx" which is just a marker. Delimiters length should not be
1941 // taken into account.
1942 else if( gr == 0xfffe )
1944 // According to the norm, fffe|0000 shouldn't exist. BUT the Philips
1945 // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to
1946 // causes extra troubles...
1947 if( entry->GetElement() != 0x0000 )
1953 entry->SetLength(foundLength);
1957 * \brief Apply some heuristics to predict whether the considered
1958 * element value contains/represents an integer or not.
1959 * @param entry The element value on which to apply the predicate.
1960 * @return The result of the heuristical predicate.
1962 bool Document::IsDocEntryAnInteger(DocEntry *entry)
1964 uint16_t elem = entry->GetElement();
1965 uint16_t group = entry->GetGroup();
1966 const std::string &vr = entry->GetVR();
1967 uint32_t length = entry->GetLength();
1969 // When we have some semantics on the element we just read, and if we
1970 // a priori know we are dealing with an integer, then we shall be
1971 // able to swap it's element value properly.
1972 if ( elem == 0 ) // This is the group length of the group
1980 // Allthough this should never happen, still some images have a
1981 // corrupted group length [e.g. have a glance at offset x(8336) of
1982 // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm].
1983 // Since for dicom compliant and well behaved headers, the present
1984 // test is useless (and might even look a bit paranoid), when we
1985 // encounter such an ill-formed image, we simply display a warning
1986 // message and proceed on parsing (while crossing fingers).
1987 long filePosition = Fp->tellg();
1988 gdcmVerboseMacro( "Erroneous Group Length element length on : ("
1989 << std::hex << group << " , " << elem
1990 << ") -before- position x(" << filePosition << ")"
1991 << "lgt : " << length );
1995 if ( vr == "UL" || vr == "US" || vr == "SL" || vr == "SS" )
2003 * \brief Find the Length till the next sequence delimiter
2004 * \warning NOT end user intended method !
2008 uint32_t Document::FindDocEntryLengthOBOrOW()
2009 throw( FormatUnexpected )
2011 // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
2012 long positionOnEntry = Fp->tellg();
2013 bool foundSequenceDelimiter = false;
2014 uint32_t totalLength = 0;
2016 while ( !foundSequenceDelimiter )
2022 group = ReadInt16();
2025 catch ( FormatError )
2027 throw FormatError("Unexpected end of file encountered during ",
2028 "Document::FindDocEntryLengthOBOrOW()");
2031 // We have to decount the group and element we just read
2034 if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) )
2036 long filePosition = Fp->tellg();
2037 gdcmVerboseMacro( "Neither an Item tag nor a Sequence delimiter tag on :"
2038 << std::hex << group << " , " << elem
2039 << ") -before- position x(" << filePosition << ")" );
2041 Fp->seekg(positionOnEntry, std::ios::beg);
2042 throw FormatUnexpected( "Neither an Item tag nor a Sequence delimiter tag.");
2045 if ( elem == 0xe0dd )
2047 foundSequenceDelimiter = true;
2050 uint32_t itemLength = ReadInt32();
2051 // We add 4 bytes since we just read the ItemLength with ReadInt32
2052 totalLength += itemLength + 4;
2053 SkipBytes(itemLength);
2055 if ( foundSequenceDelimiter )
2060 Fp->seekg( positionOnEntry, std::ios::beg);
2065 * \brief Reads a supposed to be 16 Bits integer
2066 * (swaps it depending on processor endianity)
2067 * @return read value
2069 uint16_t Document::ReadInt16()
2070 throw( FormatError )
2073 Fp->read ((char*)&g, (size_t)2);
2076 throw FormatError( "Document::ReadInt16()", " file error." );
2080 throw FormatError( "Document::ReadInt16()", "EOF." );
2087 * \brief Reads a supposed to be 32 Bits integer
2088 * (swaps it depending on processor endianity)
2089 * @return read value
2091 uint32_t Document::ReadInt32()
2092 throw( FormatError )
2095 Fp->read ((char*)&g, (size_t)4);
2098 throw FormatError( "Document::ReadInt32()", " file error." );
2102 throw FormatError( "Document::ReadInt32()", "EOF." );
2109 * \brief skips bytes inside the source file
2110 * \warning NOT end user intended method !
2113 void Document::SkipBytes(uint32_t nBytes)
2115 //FIXME don't dump the returned value
2116 Fp->seekg((long)nBytes, std::ios::cur);
2120 * \brief Loads all the needed Dictionaries
2121 * \warning NOT end user intended method !
2123 void Document::Initialise()
2125 RefPubDict = Global::GetDicts()->GetDefaultPubDict();
2127 RLEInfo = new RLEFramesInfo;
2128 JPEGInfo = new JPEGFragmentsInfo;
2133 * \brief Discover what the swap code is (among little endian, big endian,
2134 * bad little endian, bad big endian).
2136 * @return false when we are absolutely sure
2137 * it's neither ACR-NEMA nor DICOM
2138 * true when we hope ours assuptions are OK
2140 bool Document::CheckSwap()
2142 // The only guaranted way of finding the swap code is to find a
2143 // group tag since we know it's length has to be of four bytes i.e.
2144 // 0x00000004. Finding the swap code in then straigthforward. Trouble
2145 // occurs when we can't find such group...
2147 uint32_t x = 4; // x : for ntohs
2148 bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
2154 // First, compare HostByteOrder and NetworkByteOrder in order to
2155 // determine if we shall need to swap bytes (i.e. the Endian type).
2156 if ( x == ntohs(x) )
2165 // The easiest case is the one of a 'true' DICOM header, we just have
2166 // to look for the string "DICM" inside the file preamble.
2169 char *entCur = deb + 128;
2170 if( memcmp(entCur, "DICM", (size_t)4) == 0 )
2172 gdcmVerboseMacro( "Looks like DICOM Version3 (preamble + DCM)" );
2174 // Group 0002 should always be VR, and the first element 0000
2175 // Let's be carefull (so many wrong headers ...)
2176 // and determine the value representation (VR) :
2177 // Let's skip to the first element (0002,0000) and check there if we find
2178 // "UL" - or "OB" if the 1st one is (0002,0001) -,
2179 // in which case we (almost) know it is explicit VR.
2180 // WARNING: if it happens to be implicit VR then what we will read
2181 // is the length of the group. If this ascii representation of this
2182 // length happens to be "UL" then we shall believe it is explicit VR.
2183 // We need to skip :
2184 // * the 128 bytes of File Preamble (often padded with zeroes),
2185 // * the 4 bytes of "DICM" string,
2186 // * the 4 bytes of the first tag (0002, 0000),or (0002, 0001)
2187 // i.e. a total of 136 bytes.
2190 // group 0x0002 *is always* Explicit VR Sometimes ,
2191 // even if elem 0002,0010 (Transfer Syntax) tells us the file is
2192 // *Implicit* VR (see former 'gdcmData/icone.dcm')
2194 if( memcmp(entCur, "UL", (size_t)2) == 0 ||
2195 memcmp(entCur, "OB", (size_t)2) == 0 ||
2196 memcmp(entCur, "UI", (size_t)2) == 0 ||
2197 memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later
2198 // when Write DCM *adds*
2200 // Use Document::dicom_vr to test all the possibilities
2201 // instead of just checking for UL, OB and UI !? group 0000
2203 Filetype = ExplicitVR;
2204 gdcmVerboseMacro( "Group 0002 : Explicit Value Representation");
2208 Filetype = ImplicitVR;
2209 gdcmVerboseMacro( "Group 0002 :Not an explicit Value Representation;"
2210 << "Looks like a bugged Header!");
2216 gdcmVerboseMacro( "HostByteOrder != NetworkByteOrder");
2221 gdcmVerboseMacro( "HostByteOrder = NetworkByteOrder");
2224 // Position the file position indicator at first tag
2225 // (i.e. after the file preamble and the "DICM" string).
2226 Fp->seekg(0, std::ios::beg);
2227 Fp->seekg ( 132L, std::ios::beg);
2231 // Alas, this is not a DicomV3 file and whatever happens there is no file
2232 // preamble. We can reset the file position indicator to where the data
2233 // is (i.e. the beginning of the file).
2234 gdcmVerboseMacro( "Not a DICOM Version3 file");
2235 Fp->seekg(0, std::ios::beg);
2237 // Our next best chance would be to be considering a 'clean' ACR/NEMA file.
2238 // By clean we mean that the length of the first tag is written down.
2239 // If this is the case and since the length of the first group HAS to be
2240 // four (bytes), then determining the proper swap code is straightforward.
2243 // We assume the array of char we are considering contains the binary
2244 // representation of a 32 bits integer. Hence the following dirty
2246 s32 = *((uint32_t *)(entCur));
2267 // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
2268 // It is time for despaired wild guesses.
2269 // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA,
2270 // i.e. the 'group length' element is not present :
2272 // check the supposed-to-be 'group number'
2273 // in ( 0x0001 .. 0x0008 )
2274 // to determine ' SwapCode' value .
2275 // Only 0 or 4321 will be possible
2276 // (no oportunity to check for the formerly well known
2277 // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian'
2278 // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -3, 4, ..., 8-)
2279 // the file IS NOT ACR-NEMA nor DICOM V3
2280 // Find a trick to tell it the caller...
2282 s16 = *((uint16_t *)(deb));
2309 gdcmVerboseMacro( "ACR/NEMA unfound swap info (Really hopeless !)");
2317 * \brief Change the Byte Swap code.
2319 void Document::SwitchByteSwapCode()
2321 gdcmVerboseMacro( "Switching Byte Swap code from "<< SwapCode);
2322 if ( SwapCode == 1234 )
2326 else if ( SwapCode == 4321 )
2330 else if ( SwapCode == 3412 )
2334 else if ( SwapCode == 2143 )
2341 * \brief during parsing, Header Elements too long are not loaded in memory
2344 void Document::SetMaxSizeLoadEntry(long newSize)
2350 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2352 MaxSizeLoadEntry = 0xffffffff;
2355 MaxSizeLoadEntry = newSize;
2360 * \brief Header Elements too long will not be printed
2361 * \todo See comments of \ref Document::MAX_SIZE_PRINT_ELEMENT_VALUE
2364 void Document::SetMaxSizePrintEntry(long newSize)
2370 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2372 MaxSizePrintEntry = 0xffffffff;
2375 MaxSizePrintEntry = newSize;
2381 * \brief Handle broken private tag from Philips NTSCAN
2382 * where the endianess is being switch to BigEndian for no
2386 void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem)
2388 // Endian reversion. Some files contain groups of tags with reversed endianess.
2389 static int reversedEndian = 0;
2390 // try to fix endian switching in the middle of headers
2391 if ((group == 0xfeff) && (elem == 0x00e0))
2393 // start endian swap mark for group found
2395 SwitchByteSwapCode();
2400 else if (group == 0xfffe && elem == 0xe00d && reversedEndian)
2402 // end of reversed endian group
2404 SwitchByteSwapCode();
2409 * \brief Accesses the info from 0002,0010 : Transfer Syntax and TS
2410 * @return The full Transfer Syntax Name (as opposed to Transfer Syntax UID)
2412 std::string Document::GetTransferSyntaxName()
2414 // use the TS (TS : Transfer Syntax)
2415 std::string transferSyntax = GetEntry(0x0002,0x0010);
2417 if ( (transferSyntax.find(GDCM_NOTLOADED) < transferSyntax.length()) )
2419 gdcmErrorMacro( "Transfer Syntax not loaded. " << std::endl
2420 << "Better you increase MAX_SIZE_LOAD_ELEMENT_VALUE" );
2421 return "Uncompressed ACR-NEMA";
2423 if ( transferSyntax == GDCM_UNFOUND )
2425 gdcmVerboseMacro( "Unfound Transfer Syntax (0002,0010)");
2426 return "Uncompressed ACR-NEMA";
2429 // we do it only when we need it
2430 const TSKey &tsName = Global::GetTS()->GetValue( transferSyntax );
2432 // Global::GetTS() is a global static you shall never try to delete it!
2437 * \brief Group 0002 is always coded Little Endian
2438 * whatever Transfer Syntax is
2441 void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem)
2443 // Endian reversion. Some files contain groups of tags with reversed endianess.
2444 if ( !Group0002Parsed && group != 0x0002)
2446 Group0002Parsed = true;
2447 // we just came out of group 0002
2448 // if Transfer syntax is Big Endian we have to change CheckSwap
2450 std::string ts = GetTransferSyntax();
2451 if ( !Global::GetTS()->IsTransferSyntax(ts) )
2453 gdcmVerboseMacro("True DICOM File, with NO Tansfer Syntax: " << ts );
2457 // Group 0002 is always 'Explicit ...' enven when Transfer Syntax says 'Implicit ..."
2459 if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ImplicitVRLittleEndian )
2461 Filetype = ImplicitVR;
2464 // FIXME Strangely, this works with
2465 //'Implicit VR Transfer Syntax (GE Private)
2466 if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ExplicitVRBigEndian )
2468 gdcmVerboseMacro("Transfer Syntax Name = ["
2469 << GetTransferSyntaxName() << "]" );
2470 SwitchByteSwapCode();
2471 group = SwapShort(group);
2472 elem = SwapShort(elem);
2478 * \brief Read the next tag but WITHOUT loading it's value
2479 * (read the 'Group Number', the 'Element Number',
2480 * gets the Dict Entry
2481 * gets the VR, gets the length, gets the offset value)
2482 * @return On succes the newly created DocEntry, NULL on failure.
2484 DocEntry *Document::ReadNextDocEntry()
2491 group = ReadInt16();
2494 catch ( FormatError e )
2496 // We reached the EOF (or an error occured) therefore
2497 // header parsing has to be considered as finished.
2502 // Sometimes file contains groups of tags with reversed endianess.
2503 HandleBrokenEndian(group, elem);
2505 // In 'true DICOM' files Group 0002 is always little endian
2506 if ( HasDCMPreamble )
2507 HandleOutOfGroup0002(group, elem);
2509 std::string vr = FindDocEntryVR();
2510 std::string realVR = vr;
2512 if( vr == GDCM_UNKNOWN)
2514 DictEntry *dictEntry = GetDictEntry(group,elem);
2516 realVR = dictEntry->GetVR();
2520 if( Global::GetVR()->IsVROfSequence(realVR) )
2521 newEntry = NewSeqEntry(group, elem);
2522 else if( Global::GetVR()->IsVROfStringRepresentable(realVR) )
2523 newEntry = NewValEntry(group, elem,vr);
2525 newEntry = NewBinEntry(group, elem,vr);
2527 if( vr == GDCM_UNKNOWN )
2529 if( Filetype == ExplicitVR )
2531 // We thought this was explicit VR, but we end up with an
2532 // implicit VR tag. Let's backtrack.
2533 if ( newEntry->GetGroup() != 0xfffe )
2536 msg = Util::Format("Entry (%04x,%04x) should be Explicit VR\n",
2537 newEntry->GetGroup(), newEntry->GetElement());
2538 gdcmVerboseMacro( msg.c_str() );
2541 newEntry->SetImplicitVR();
2546 FindDocEntryLength(newEntry);
2548 catch ( FormatError e )
2556 newEntry->SetOffset(Fp->tellg());
2563 * \brief Generate a free TagKey i.e. a TagKey that is not present
2564 * in the TagHt dictionary.
2565 * @param group The generated tag must belong to this group.
2566 * @return The element of tag with given group which is fee.
2568 uint32_t Document::GenerateFreeTagKeyInGroup(uint16_t group)
2570 for (uint32_t elem = 0; elem < UINT32_MAX; elem++)
2572 TagKey key = DictEntry::TranslateToKey(group, elem);
2573 if (TagHT.count(key) == 0)
2582 * \brief Assuming the internal file pointer \ref Document::Fp
2583 * is placed at the beginning of a tag check whether this
2584 * tag is (TestGroup, TestElement).
2585 * \warning On success the internal file pointer \ref Document::Fp
2586 * is modified to point after the tag.
2587 * On failure (i.e. when the tag wasn't the expected tag
2588 * (TestGroup, TestElement) the internal file pointer
2589 * \ref Document::Fp is restored to it's original position.
2590 * @param testGroup The expected group of the tag.
2591 * @param testElement The expected Element of the tag.
2592 * @return True on success, false otherwise.
2594 bool Document::ReadTag(uint16_t testGroup, uint16_t testElement)
2596 long positionOnEntry = Fp->tellg();
2597 long currentPosition = Fp->tellg(); // On debugging purposes
2599 //// Read the Item Tag group and element, and make
2600 // sure they are what we expected:
2601 uint16_t itemTagGroup;
2602 uint16_t itemTagElement;
2605 itemTagGroup = ReadInt16();
2606 itemTagElement = ReadInt16();
2608 catch ( FormatError e )
2610 //std::cerr << e << std::endl;
2613 if ( itemTagGroup != testGroup || itemTagElement != testElement )
2615 gdcmVerboseMacro( "Wrong Item Tag found:"
2616 << " We should have found tag ("
2617 << std::hex << testGroup << "," << testElement << ")" << std::endl
2618 << " but instead we encountered tag ("
2619 << std::hex << itemTagGroup << "," << itemTagElement << ")"
2620 << " at address: " << " 0x(" << (unsigned int)currentPosition << ")"
2622 Fp->seekg(positionOnEntry, std::ios::beg);
2630 * \brief Assuming the internal file pointer \ref Document::Fp
2631 * is placed at the beginning of a tag (TestGroup, TestElement),
2632 * read the length associated to the Tag.
2633 * \warning On success the internal file pointer \ref Document::Fp
2634 * is modified to point after the tag and it's length.
2635 * On failure (i.e. when the tag wasn't the expected tag
2636 * (TestGroup, TestElement) the internal file pointer
2637 * \ref Document::Fp is restored to it's original position.
2638 * @param testGroup The expected group of the tag.
2639 * @param testElement The expected Element of the tag.
2640 * @return On success returns the length associated to the tag. On failure
2643 uint32_t Document::ReadTagLength(uint16_t testGroup, uint16_t testElement)
2646 if ( !ReadTag(testGroup, testElement) )
2651 //// Then read the associated Item Length
2652 long currentPosition = Fp->tellg();
2653 uint32_t itemLength = ReadInt32();
2655 gdcmVerboseMacro( "Basic Item Length is: "
2656 << itemLength << std::endl
2657 << " at address: " << std::hex << (unsigned int)currentPosition);
2663 * \brief When parsing the Pixel Data of an encapsulated file, read
2664 * the basic offset table (when present, and BTW dump it).
2666 void Document::ReadAndSkipEncapsulatedBasicOffsetTable()
2668 //// Read the Basic Offset Table Item Tag length...
2669 uint32_t itemLength = ReadTagLength(0xfffe, 0xe000);
2671 // When present, read the basic offset table itself.
2672 // Notes: - since the presence of this basic offset table is optional
2673 // we can't rely on it for the implementation, and we will simply
2674 // trash it's content (when present).
2675 // - still, when present, we could add some further checks on the
2676 // lengths, but we won't bother with such fuses for the time being.
2677 if ( itemLength != 0 )
2679 char *basicOffsetTableItemValue = new char[itemLength + 1];
2680 Fp->read(basicOffsetTableItemValue, itemLength);
2683 for (unsigned int i=0; i < itemLength; i += 4 )
2685 uint32_t individualLength = str2num( &basicOffsetTableItemValue[i],
2687 gdcmVerboseMacro( "Read one length: " <<
2688 std::hex << individualLength );
2692 delete[] basicOffsetTableItemValue;
2697 * \brief Parse pixel data from disk of [multi-]fragment RLE encoding.
2698 * Compute the RLE extra information and store it in \ref RLEInfo
2699 * for later pixel retrieval usage.
2701 void Document::ComputeRLEInfo()
2703 std::string ts = GetTransferSyntax();
2704 if ( !Global::GetTS()->IsRLELossless(ts) )
2709 // Encoded pixel data: for the time being we are only concerned with
2710 // Jpeg or RLE Pixel data encodings.
2711 // As stated in PS 3.5-2003, section 8.2 p44:
2712 // "If sent in Encapsulated Format (i.e. other than the Native Format) the
2713 // value representation OB is used".
2714 // Hence we expect an OB value representation. Concerning OB VR,
2715 // the section PS 3.5-2003, section A.4.c p 58-59, states:
2716 // "For the Value Representations OB and OW, the encoding shall meet the
2717 // following specifications depending on the Data element tag:"
2719 // - the first item in the sequence of items before the encoded pixel
2720 // data stream shall be basic offset table item. The basic offset table
2721 // item value, however, is not required to be present"
2723 ReadAndSkipEncapsulatedBasicOffsetTable();
2725 // Encapsulated RLE Compressed Images (see PS 3.5-2003, Annex G)
2726 // Loop on the individual frame[s] and store the information
2727 // on the RLE fragments in a RLEFramesInfo.
2728 // Note: - when only a single frame is present, this is a
2730 // - when more than one frame are present, then we are in
2731 // the case of a multi-frame image.
2733 while ( (frameLength = ReadTagLength(0xfffe, 0xe000)) )
2735 // Parse the RLE Header and store the corresponding RLE Segment
2736 // Offset Table information on fragments of this current Frame.
2737 // Note that the fragment pixels themselves are not loaded
2738 // (but just skipped).
2739 long frameOffset = Fp->tellg();
2741 uint32_t nbRleSegments = ReadInt32();
2742 if ( nbRleSegments > 16 )
2744 // There should be at most 15 segments (refer to RLEFrame class)
2745 gdcmVerboseMacro( "Too many segments.");
2748 uint32_t rleSegmentOffsetTable[16];
2749 for( int k = 1; k <= 15; k++ )
2751 rleSegmentOffsetTable[k] = ReadInt32();
2754 // Deduce from both the RLE Header and the frameLength the
2755 // fragment length, and again store this info in a
2757 long rleSegmentLength[15];
2758 // skipping (not reading) RLE Segments
2759 if ( nbRleSegments > 1)
2761 for(unsigned int k = 1; k <= nbRleSegments-1; k++)
2763 rleSegmentLength[k] = rleSegmentOffsetTable[k+1]
2764 - rleSegmentOffsetTable[k];
2765 SkipBytes(rleSegmentLength[k]);
2769 rleSegmentLength[nbRleSegments] = frameLength
2770 - rleSegmentOffsetTable[nbRleSegments];
2771 SkipBytes(rleSegmentLength[nbRleSegments]);
2773 // Store the collected info
2774 RLEFrame *newFrameInfo = new RLEFrame;
2775 newFrameInfo->NumberFragments = nbRleSegments;
2776 for( unsigned int uk = 1; uk <= nbRleSegments; uk++ )
2778 newFrameInfo->Offset[uk] = frameOffset + rleSegmentOffsetTable[uk];
2779 newFrameInfo->Length[uk] = rleSegmentLength[uk];
2781 RLEInfo->Frames.push_back( newFrameInfo );
2784 // Make sure that at the end of the item we encounter a 'Sequence
2786 if ( !ReadTag(0xfffe, 0xe0dd) )
2788 gdcmVerboseMacro( "No sequence delimiter item at end of RLE item sequence");
2793 * \brief Parse pixel data from disk of [multi-]fragment Jpeg encoding.
2794 * Compute the jpeg extra information (fragment[s] offset[s] and
2795 * length) and store it[them] in \ref JPEGInfo for later pixel
2798 void Document::ComputeJPEGFragmentInfo()
2800 // If you need to, look for comments of ComputeRLEInfo().
2801 std::string ts = GetTransferSyntax();
2802 if ( ! Global::GetTS()->IsJPEG(ts) )
2807 ReadAndSkipEncapsulatedBasicOffsetTable();
2809 // Loop on the fragments[s] and store the parsed information in a
2811 long fragmentLength;
2812 while ( (fragmentLength = ReadTagLength(0xfffe, 0xe000)) )
2814 long fragmentOffset = Fp->tellg();
2816 // Store the collected info
2817 JPEGFragment *newFragment = new JPEGFragment;
2818 newFragment->Offset = fragmentOffset;
2819 newFragment->Length = fragmentLength;
2820 JPEGInfo->Fragments.push_back( newFragment );
2822 SkipBytes( fragmentLength );
2825 // Make sure that at the end of the item we encounter a 'Sequence
2827 if ( !ReadTag(0xfffe, 0xe0dd) )
2829 gdcmVerboseMacro( "No sequence delimiter item at end of JPEG item sequence");
2834 * \brief Walk recursively the given \ref DocEntrySet, and feed
2835 * the given hash table (\ref TagDocEntryHT) with all the
2836 * \ref DocEntry (Dicom entries) encountered.
2837 * This method does the job for \ref BuildFlatHashTable.
2838 * @param builtHT Where to collect all the \ref DocEntry encountered
2839 * when recursively walking the given set.
2840 * @param set The structure to be traversed (recursively).
2842 /*void Document::BuildFlatHashTableRecurse( TagDocEntryHT &builtHT,
2845 if (ElementSet *elementSet = dynamic_cast< ElementSet* > ( set ) )
2847 TagDocEntryHT const ¤tHT = elementSet->GetTagHT();
2848 for( TagDocEntryHT::const_iterator i = currentHT.begin();
2849 i != currentHT.end();
2852 DocEntry *entry = i->second;
2853 if ( SeqEntry *seqEntry = dynamic_cast<SeqEntry*>(entry) )
2855 const ListSQItem& items = seqEntry->GetSQItems();
2856 for( ListSQItem::const_iterator item = items.begin();
2857 item != items.end();
2860 BuildFlatHashTableRecurse( builtHT, *item );
2864 builtHT[entry->GetKey()] = entry;
2869 if (SQItem *SQItemSet = dynamic_cast< SQItem* > ( set ) )
2871 const ListDocEntry& currentList = SQItemSet->GetDocEntries();
2872 for (ListDocEntry::const_iterator i = currentList.begin();
2873 i != currentList.end();
2876 DocEntry *entry = *i;
2877 if ( SeqEntry *seqEntry = dynamic_cast<SeqEntry*>(entry) )
2879 const ListSQItem& items = seqEntry->GetSQItems();
2880 for( ListSQItem::const_iterator item = items.begin();
2881 item != items.end();
2884 BuildFlatHashTableRecurse( builtHT, *item );
2888 builtHT[entry->GetKey()] = entry;
2895 * \brief Build a \ref TagDocEntryHT (i.e. a std::map<>) from the current
2898 * The structure used by a Document (through \ref ElementSet),
2899 * in order to hold the parsed entries of a Dicom header, is a recursive
2900 * one. This is due to the fact that the sequences (when present)
2901 * can be nested. Additionaly, the sequence items (represented in
2902 * gdcm as \ref SQItem) add an extra complexity to the data
2903 * structure. Hence, a gdcm user whishing to visit all the entries of
2904 * a Dicom header will need to dig in the gdcm internals (which
2905 * implies exposing all the internal data structures to the API).
2906 * In order to avoid this burden to the user, \ref BuildFlatHashTable
2907 * recursively builds a temporary hash table, which holds all the
2908 * Dicom entries in a flat structure (a \ref TagDocEntryHT i.e. a
2910 * \warning Of course there is NO integrity constrain between the
2911 * returned \ref TagDocEntryHT and the \ref ElementSet used
2912 * to build it. Hence if the underlying \ref ElementSet is
2913 * altered, then it is the caller responsability to invoke
2914 * \ref BuildFlatHashTable again...
2915 * @return The flat std::map<> we juste build.
2917 /*TagDocEntryHT *Document::BuildFlatHashTable()
2919 TagDocEntryHT *FlatHT = new TagDocEntryHT;
2920 BuildFlatHashTableRecurse( *FlatHT, this );
2927 * \brief Compares two documents, according to \ref DicomDir rules
2928 * \warning Does NOT work with ACR-NEMA files
2929 * \todo Find a trick to solve the pb (use RET fields ?)
2931 * @return true if 'smaller'
2933 bool Document::operator<(Document &document)
2936 std::string s1 = GetEntry(0x0010,0x0010);
2937 std::string s2 = document.GetEntry(0x0010,0x0010);
2949 s1 = GetEntry(0x0010,0x0020);
2950 s2 = document.GetEntry(0x0010,0x0020);
2961 // Study Instance UID
2962 s1 = GetEntry(0x0020,0x000d);
2963 s2 = document.GetEntry(0x0020,0x000d);
2974 // Serie Instance UID
2975 s1 = GetEntry(0x0020,0x000e);
2976 s2 = document.GetEntry(0x0020,0x000e);
2993 * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader
2994 * @param filetype Type of the File to be written
2996 int Document::ComputeGroup0002Length( FileType filetype )
3001 int groupLength = 0;
3002 bool found0002 = false;
3004 // for each zero-level Tag in the DCM Header
3005 DocEntry *entry = GetFirstEntry();
3008 gr = entry->GetGroup();
3014 el = entry->GetElement();
3015 vr = entry->GetVR();
3017 if (filetype == ExplicitVR)
3019 if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") )
3021 groupLength += 4; // explicit VR AND OB, OW, SQ : 4 more bytes
3024 groupLength += 2 + 2 + 4 + entry->GetLength();
3026 else if (found0002 )
3029 entry = GetNextEntry();
3034 } // end namespace gdcm
3036 //-----------------------------------------------------------------------------