1 /*=========================================================================
4 Module: $RCSfile: gdcmDocument.cxx,v $
6 Date: $Date: 2004/07/19 15:16:18 $
7 Version: $Revision: 1.55 $
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.htm 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"
24 #include "gdcmGlobal.h"
26 #include "gdcmDebug.h"
35 #include <netinet/in.h>
40 // Implicit VR Little Endian
41 #define UI1_2_840_10008_1_2 "1.2.840.10008.1.2"
42 // Explicit VR Little Endian
43 #define UI1_2_840_10008_1_2_1 "1.2.840.10008.1.2.1"
44 // Deflated Explicit VR Little Endian
45 #define UI1_2_840_10008_1_2_1_99 "1.2.840.10008.1.2.1.99"
46 // Explicit VR Big Endian
47 #define UI1_2_840_10008_1_2_2 "1.2.840.10008.1.2.2"
48 // JPEG Baseline (Process 1)
49 #define UI1_2_840_10008_1_2_4_50 "1.2.840.10008.1.2.4.50"
50 // JPEG Extended (Process 2 & 4)
51 #define UI1_2_840_10008_1_2_4_51 "1.2.840.10008.1.2.4.51"
52 // JPEG Extended (Process 3 & 5)
53 #define UI1_2_840_10008_1_2_4_52 "1.2.840.10008.1.2.4.52"
54 // JPEG Spectral Selection, Non-Hierarchical (Process 6 & 8)
55 #define UI1_2_840_10008_1_2_4_53 "1.2.840.10008.1.2.4.53"
56 // JPEG Full Progression, Non-Hierarchical (Process 10 & 12)
57 #define UI1_2_840_10008_1_2_4_55 "1.2.840.10008.1.2.4.55"
58 // JPEG Lossless, Non-Hierarchical (Process 14)
59 #define UI1_2_840_10008_1_2_4_57 "1.2.840.10008.1.2.4.57"
60 // JPEG Lossless, Hierarchical, First-Order Prediction (Process 14,
61 // [Selection Value 1])
62 #define UI1_2_840_10008_1_2_4_70 "1.2.840.10008.1.2.4.70"
64 #define UI1_2_840_10008_1_2_4_90 "1.2.840.10008.1.2.4.90"
66 #define UI1_2_840_10008_1_2_4_91 "1.2.840.10008.1.2.4.91"
68 #define UI1_2_840_10008_1_2_5 "1.2.840.10008.1.2.5"
69 // UI1_1_2_840_10008_1_2_5
70 #define str2num(str, typeNum) *((typeNum *)(str))
72 //-----------------------------------------------------------------------------
73 // Refer to gdcmDocument::CheckSwap()
74 const unsigned int gdcmDocument::HEADER_LENGTH_TO_READ = 256;
76 // Refer to gdcmDocument::SetMaxSizeLoadEntry()
77 const unsigned int gdcmDocument::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096
78 const unsigned int gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff;
80 //-----------------------------------------------------------------------------
81 // Constructor / Destructor
85 * @param inFilename file to be opened for parsing
86 * @param exception_on_error whether we throw an exception or not
87 * @param enable_sequences = true to allow the header
88 * to be parsed *inside* the SeQuences,
89 * when they have an actual length
90 * \warning enable_sequences *has to be* true for reading PAPYRUS 3.0 files
91 * @param ignore_shadow to allow skipping the shadow elements,
92 * to save memory space.
93 * \warning The TRUE value for this param has to be used
94 * with a FALSE value for the 'enable_sequence' param.
95 * ('public elements' may be embedded in 'shadow Sequences')
97 gdcmDocument::gdcmDocument( std::string const & filename,
98 bool exception_on_error,
99 bool enable_sequences,
103 IgnoreShadow = ignore_shadow;
104 //EnableSequences=enable_sequences;
105 (void)enable_sequences;
106 EnableSequences = true; // JPR // TODO : remove params out of the constructor
107 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
111 if ( !OpenFile(exception_on_error))
116 dbg.Verbose(0, "gdcmDocument::gdcmDocument: starting parsing of file: ",
120 fseek(Fp,0L,SEEK_END);
121 long lgt = ftell(Fp);
125 long beg = ftell(Fp);
130 long l = ParseDES( this, beg, lgt, false); // le Load sera fait a la volee
131 (void)l; //is l used anywhere ?
135 // Load 'non string' values
137 std::string PhotometricInterpretation = GetEntryByNumber(0x0028,0x0004);
138 if( PhotometricInterpretation == "PALETTE COLOR " )
140 LoadEntryVoidArea(0x0028,0x1200); // gray LUT
141 LoadEntryVoidArea(0x0028,0x1201); // R LUT
142 LoadEntryVoidArea(0x0028,0x1202); // G LUT
143 LoadEntryVoidArea(0x0028,0x1203); // B LUT
145 LoadEntryVoidArea(0x0028,0x1221); // Segmented Red Palette Color LUT Data
146 LoadEntryVoidArea(0x0028,0x1222); // Segmented Green Palette Color LUT Data
147 LoadEntryVoidArea(0x0028,0x1223); // Segmented Blue Palette Color LUT Data
149 //FIXME later : how to use it?
150 LoadEntryVoidArea(0x0028,0x3006); //LUT Data (CTX dependent)
154 // --------------------------------------------------------------
155 // Special Patch to allow gdcm to read ACR-LibIDO formated images
157 // if recognition code tells us we deal with a LibIDO image
158 // we switch lineNumber and columnNumber
161 RecCode = GetEntryByNumber(0x0008, 0x0010); // recognition code
162 if (RecCode == "ACRNEMA_LIBIDO_1.1" ||
163 RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares
164 // with "little-endian strings"
166 Filetype = gdcmACR_LIBIDO;
167 std::string rows = GetEntryByNumber(0x0028, 0x0010);
168 std::string columns = GetEntryByNumber(0x0028, 0x0011);
169 SetEntryByNumber(columns, 0x0028, 0x0010);
170 SetEntryByNumber(rows , 0x0028, 0x0011);
172 // ----------------- End of Special Patch ----------------
174 PrintLevel = 1; // 'Medium' print level by default
179 * @param exception_on_error
181 gdcmDocument::gdcmDocument(bool exception_on_error)
184 (void)exception_on_error;
185 //EnableSequences=0; // ?!? JPR
187 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
190 PrintLevel = 1; // 'Medium' print level by default
194 * \brief Canonical destructor.
196 gdcmDocument::~gdcmDocument ()
201 // Recursive clean up of sequences
202 for (TagDocEntryHT::const_iterator it = TagHT.begin();
203 it != TagHT.end(); ++it )
210 //-----------------------------------------------------------------------------
214 * \brief Prints The Dict Entries of THE public Dicom Dictionary
217 void gdcmDocument::PrintPubDict(std::ostream & os)
219 RefPubDict->Print(os);
223 * \brief Prints The Dict Entries of THE shadow Dicom Dictionary
226 void gdcmDocument::PrintShaDict(std::ostream & os)
228 RefShaDict->Print(os);
231 //-----------------------------------------------------------------------------
234 * \brief Get the public dictionary used
236 gdcmDict *gdcmDocument::GetPubDict()
242 * \brief Get the shadow dictionary used
244 gdcmDict *gdcmDocument::GetShaDict()
250 * \brief Set the shadow dictionary used
251 * \param dict dictionary to use in shadow
253 bool gdcmDocument::SetShaDict(gdcmDict *dict)
260 * \brief Set the shadow dictionary used
261 * \param dictName name of the dictionary to use in shadow
263 bool gdcmDocument::SetShaDict(DictKey dictName)
265 RefShaDict = gdcmGlobal::GetDicts()->GetDict(dictName);
270 * \brief This predicate, based on hopefully reasonable heuristics,
271 * decides whether or not the current gdcmDocument was properly parsed
272 * and contains the mandatory information for being considered as
273 * a well formed and usable Dicom/Acr File.
274 * @return true when gdcmDocument is the one of a reasonable Dicom/Acr file,
277 bool gdcmDocument::IsReadable()
279 if( Filetype == gdcmUnknown)
281 std::cout << " gdcmDocument::IsReadable: Filetype " << Filetype
282 << " " << "gdcmUnknown " << gdcmUnknown << std::endl; //JPR
283 dbg.Verbose(0, "gdcmDocument::IsReadable: wrong filetype");
289 dbg.Verbose(0, "gdcmDocument::IsReadable: no tags in internal"
299 * \brief Internal function that checks whether the Transfer Syntax given
300 * as argument is the one present in the current document.
301 * @param SyntaxToCheck The transfert syntax we need to check against.
302 * @return True when SyntaxToCheck corresponds to the Transfer Syntax of
303 * the current document. False either when the document contains
304 * no Transfer Syntax, or when the Tranfer Syntaxes doesn't match.
306 bool gdcmDocument::IsGivenTransferSyntax(std::string const & syntaxToCheck)
308 gdcmDocEntry *entry = GetDocEntryByNumber(0x0002, 0x0010);
314 // The entry might be present but not loaded (parsing and loading
315 // happen at different stages): try loading and proceed with check...
316 LoadDocEntrySafe(entry);
317 if (gdcmValEntry* valEntry = dynamic_cast< gdcmValEntry* >(entry) )
319 std::string transfer = valEntry->GetValue();
321 // The actual transfer (as read from disk) might be padded. We
322 // first need to remove the potential padding. We can make the
323 // weak assumption that padding was not executed with digits...
324 while ( ! isdigit(transfer[transfer.length()-1]) )
326 transfer.erase(transfer.length()-1, 1);
328 if ( transfer == syntaxToCheck )
337 * \brief Determines if the Transfer Syntax of the present document
338 * corresponds to a Implicit Value Representation of
340 * \sa \ref gdcmDocument::IsGivenTransferSyntax.
341 * @return True when ImplicitVRLittleEndian found. False in all other cases.
343 bool gdcmDocument::IsImplicitVRLittleEndianTransferSyntax()
345 return IsGivenTransferSyntax(UI1_2_840_10008_1_2);
349 * \brief Determines if the Transfer Syntax was already encountered
350 * and if it corresponds to a ExplicitVRLittleEndian one.
351 * @return True when ExplicitVRLittleEndian found. False in all other cases.
353 bool gdcmDocument::IsExplicitVRLittleEndianTransferSyntax()
355 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1);
359 * \brief Determines if the Transfer Syntax was already encountered
360 * and if it corresponds to a DeflatedExplicitVRLittleEndian one.
361 * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases.
363 bool gdcmDocument::IsDeflatedExplicitVRLittleEndianTransferSyntax()
365 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1_99);
369 * \brief Determines if the Transfer Syntax was already encountered
370 * and if it corresponds to a Explicit VR Big Endian one.
371 * @return True when big endian found. False in all other cases.
373 bool gdcmDocument::IsExplicitVRBigEndianTransferSyntax()
375 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_2);
379 * \brief Determines if the Transfer Syntax was already encountered
380 * and if it corresponds to a JPEGBaseLineProcess1 one.
381 * @return True when JPEGBaseLineProcess1found. False in all other cases.
383 bool gdcmDocument::IsJPEGBaseLineProcess1TransferSyntax()
385 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_50);
389 * \brief Determines if the Transfer Syntax was already encountered
390 * and if it corresponds to a JPEGExtendedProcess2-4 one.
391 * @return True when JPEGExtendedProcess2-4 found. False in all other cases.
393 bool gdcmDocument::IsJPEGExtendedProcess2_4TransferSyntax()
395 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_51);
399 * \brief Determines if the Transfer Syntax was already encountered
400 * and if it corresponds to a JPEGExtendeProcess3-5 one.
401 * @return True when JPEGExtendedProcess3-5 found. False in all other cases.
403 bool gdcmDocument::IsJPEGExtendedProcess3_5TransferSyntax()
405 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_52);
409 * \brief Determines if the Transfer Syntax was already encountered
410 * and if it corresponds to a JPEGSpectralSelectionProcess6-8 one.
411 * @return True when JPEGSpectralSelectionProcess6-8 found. False in all
414 bool gdcmDocument::IsJPEGSpectralSelectionProcess6_8TransferSyntax()
416 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_53);
420 * \brief Determines if the Transfer Syntax was already encountered
421 * and if it corresponds to a RLE Lossless one.
422 * @return True when RLE Lossless found. False in all
425 bool gdcmDocument::IsRLELossLessTransferSyntax()
427 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_5);
431 * \brief Determines if Transfer Syntax was already encountered
432 * and if it corresponds to a JPEG Lossless one.
433 * @return True when RLE Lossless found. False in all
437 bool gdcmDocument::IsJPEGLossless()
439 return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_55)
440 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_57)
441 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_70) );
445 * \brief Determines if the Transfer Syntax was already encountered
446 * and if it corresponds to a JPEG2000 one
447 * @return True when JPEG2000 (Lossly or LossLess) found. False in all
450 bool gdcmDocument::IsJPEG2000()
452 return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_90)
453 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_91) );
457 * \brief Predicate for dicom version 3 file.
458 * @return True when the file is a dicom version 3.
460 bool gdcmDocument::IsDicomV3()
462 // Checking if Transfert Syntax exists is enough
463 // Anyway, it's to late check if the 'Preamble' was found ...
464 // And ... would it be a rich idea to check ?
465 // (some 'no Preamble' DICOM images exist !)
466 return GetDocEntryByNumber(0x0002, 0x0010) != NULL;
470 * \brief returns the File Type
471 * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
472 * @return the FileType code
474 FileType gdcmDocument::GetFileType()
480 * \brief opens the file
481 * @param exception_on_error
484 FILE *gdcmDocument::OpenFile(bool exception_on_error)
487 Fp = fopen(Filename.c_str(),"rb");
491 if(exception_on_error)
493 throw gdcmFileError("gdcmDocument::gdcmDocument(const char *, bool)");
497 dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file: ",
506 fread(&zero, (size_t)2, (size_t)1, Fp);
508 //ACR -- or DICOM with no Preamble --
509 if( zero == 0x0008 || zero == 0x0800 || zero == 0x0002 || zero == 0x0200 )
515 fseek(Fp, 126L, SEEK_CUR);
517 fread(dicm, (size_t)4, (size_t)1, Fp);
518 if( memcmp(dicm, "DICM", 4) == 0 )
524 dbg.Verbose(0, "gdcmDocument::OpenFile not DICOM/ACR", Filename.c_str());
528 dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file", Filename.c_str());
535 * \brief closes the file
536 * @return TRUE if the close was successfull
538 bool gdcmDocument::CloseFile()
540 int closed = fclose(Fp);
547 * \brief Writes in a file all the Header Entries (Dicom Elements)
548 * @param fp file pointer on an already open file
549 * @param filetype Type of the File to be written
550 * (ACR-NEMA, ExplicitVR, ImplicitVR)
551 * \return Always true.
553 void gdcmDocument::Write(FILE* fp,FileType filetype)
555 /// \todo move the following lines (and a lot of others, to be written)
556 /// to a future function CheckAndCorrectHeader
558 /// WARNING : Si on veut ecrire du DICOM V3 a partir d'un DcmHeader ACR-NEMA
559 /// no way (check : FileType est un champ de gdcmDocument ...)
560 /// a moins de se livrer a un tres complique ajout des champs manquants.
561 /// faire un CheckAndCorrectHeader (?)
563 if (filetype == gdcmImplicitVR)
565 std::string implicitVRTransfertSyntax = UI1_2_840_10008_1_2;
566 ReplaceOrCreateByNumber(implicitVRTransfertSyntax,0x0002, 0x0010);
568 /// \todo Refer to standards on page 21, chapter 6.2
569 /// "Value representation": values with a VR of UI shall be
570 /// padded with a single trailing null
571 /// in the following case we have to padd manually with a 0
573 SetEntryLengthByNumber(18, 0x0002, 0x0010);
576 if (filetype == gdcmExplicitVR)
578 std::string explicitVRTransfertSyntax = UI1_2_840_10008_1_2_1;
579 ReplaceOrCreateByNumber(explicitVRTransfertSyntax,0x0002, 0x0010);
581 /// \todo Refer to standards on page 21, chapter 6.2
582 /// "Value representation": values with a VR of UI shall be
583 /// padded with a single trailing null
584 /// Dans le cas suivant on doit pader manuellement avec un 0
586 SetEntryLengthByNumber(20, 0x0002, 0x0010);
590 * \todo rewrite later, if really usefull
591 * - 'Group Length' element is optional in DICOM
592 * - but un-updated odd groups lengthes can causes pb
595 * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) )
596 * UpdateGroupLength(false,filetype);
597 * if ( filetype == ACR)
598 * UpdateGroupLength(true,ACR);
601 gdcmElementSet::Write(fp, filetype); // This one is recursive
606 * \brief Modifies the value of a given Header Entry (Dicom Element)
607 * when it exists. Create it with the given value when unexistant.
608 * @param Value (string) Value to be set
609 * @param Group Group number of the Entry
610 * @param Elem Element number of the Entry
611 * \return pointer to the modified/created Header Entry (NULL when creation
615 gdcmValEntry * gdcmDocument::ReplaceOrCreateByNumber(
620 gdcmValEntry* valEntry = 0;
622 gdcmDocEntry* currentEntry = GetDocEntryByNumber( group, elem);
625 // The entry wasn't present and we simply create the required ValEntry:
626 currentEntry = NewDocEntryByNumber(group, elem);
629 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: call to"
630 " NewDocEntryByNumber failed.");
633 valEntry = new gdcmValEntry(currentEntry);
634 if ( !AddEntry(valEntry))
636 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: AddEntry"
637 " failed allthough this is a creation.");
642 valEntry = dynamic_cast< gdcmValEntry* >(currentEntry);
645 // We need to promote the gdcmDocEntry to a gdcmValEntry:
646 valEntry = new gdcmValEntry(currentEntry);
647 if (!RemoveEntry(currentEntry))
649 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: removal"
650 " of previous DocEntry failed.");
653 if ( !AddEntry(valEntry))
655 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: adding"
656 " promoted ValEntry failed.");
662 SetEntryByNumber(value, group, elem);
668 * \brief Modifies the value of a given Header Entry (Dicom Element)
669 * when it exists. Create it with the given value when unexistant.
670 * @param voidArea (binary) value to be set
671 * @param Group Group number of the Entry
672 * @param Elem Element number of the Entry
673 * \return pointer to the modified/created Header Entry (NULL when creation
676 gdcmBinEntry * gdcmDocument::ReplaceOrCreateByNumber(
683 gdcmDocEntry* a = GetDocEntryByNumber( group, elem);
686 a = NewBinEntryByNumber(group, elem);
692 b = new gdcmBinEntry(a);
694 b->SetVoidArea(voidArea);
696 SetEntryByNumber(voidArea, lgth, group, elem);
697 //b->SetVoidArea(voidArea); //what if b == 0 !!
703 * \brief Set a new value if the invoked element exists
704 * Seems to be useless !!!
705 * @param Value new element value
706 * @param Group group number of the Entry
707 * @param Elem element number of the Entry
710 bool gdcmDocument::ReplaceIfExistByNumber(const char* value, uint16_t group,
713 const std::string v = value;
714 SetEntryByNumber(v, group, elem);
719 //-----------------------------------------------------------------------------
723 * \brief Checks if a given Dicom Element exists within the H table
724 * @param group Group number of the searched Dicom Element
725 * @param element Element number of the searched Dicom Element
726 * @return number of occurences
728 int gdcmDocument::CheckIfEntryExistByNumber(uint16_t group, uint16_t element )
730 std::string key = gdcmDictEntry::TranslateToKey(group, element );
731 return TagHT.count(key);
735 * \brief Searches within Header Entries (Dicom Elements) parsed with
736 * the public and private dictionaries
737 * for the element value of a given tag.
738 * \warning Don't use any longer : use GetPubEntryByName
739 * @param tagName name of the searched element.
740 * @return Corresponding element value when it exists,
741 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
743 std::string gdcmDocument::GetEntryByName(TagName tagName)
745 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
751 return GetEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement());
755 * \brief Searches within Header Entries (Dicom Elements) parsed with
756 * the public and private dictionaries
757 * for the element value representation of a given tag.
759 * Obtaining the VR (Value Representation) might be needed by caller
760 * to convert the string typed content to caller's native type
761 * (think of C++ vs Python). The VR is actually of a higher level
762 * of semantics than just the native C++ type.
763 * @param tagName name of the searched element.
764 * @return Corresponding element value representation when it exists,
765 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
767 std::string gdcmDocument::GetEntryVRByName(TagName tagName)
769 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
770 if( dictEntry == NULL)
775 gdcmDocEntry* elem = GetDocEntryByNumber(dictEntry->GetGroup(),
776 dictEntry->GetElement());
777 return elem->GetVR();
782 * \brief Searches within Header Entries (Dicom Elements) parsed with
783 * the public and private dictionaries
784 * for the element value representation of a given tag.
785 * @param group Group number of the searched tag.
786 * @param element Element number of the searched tag.
787 * @return Corresponding element value representation when it exists,
788 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
790 std::string gdcmDocument::GetEntryByNumber(uint16_t group, uint16_t element)
792 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
793 /// \todo use map methods, instead of multimap JPR
794 if ( !TagHT.count(key))
799 return ((gdcmValEntry *)TagHT.find(key)->second)->GetValue();
803 * \brief Searches within Header Entries (Dicom Elements) parsed with
804 * the public and private dictionaries
805 * for the element value representation of a given tag..
807 * Obtaining the VR (Value Representation) might be needed by caller
808 * to convert the string typed content to caller's native type
809 * (think of C++ vs Python). The VR is actually of a higher level
810 * of semantics than just the native C++ type.
811 * @param group Group number of the searched tag.
812 * @param element Element number of the searched tag.
813 * @return Corresponding element value representation when it exists,
814 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
816 std::string gdcmDocument::GetEntryVRByNumber(uint16_t group, uint16_t element)
818 gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
823 return elem->GetVR();
827 * \brief Searches within Header Entries (Dicom Elements) parsed with
828 * the public and private dictionaries
829 * for the value length of a given tag..
830 * @param group Group number of the searched tag.
831 * @param element Element number of the searched tag.
832 * @return Corresponding element length; -2 if not found
834 int gdcmDocument::GetEntryLengthByNumber(uint16_t group, uint16_t element)
836 gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
839 return -2; //magic number
841 return elem->GetLength();
844 * \brief Sets the value (string) of the Header Entry (Dicom Element)
845 * @param content string value of the Dicom Element
846 * @param tagName name of the searched Dicom Element.
847 * @return true when found
849 bool gdcmDocument::SetEntryByName(std::string content,std::string tagName)
851 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
857 return SetEntryByNumber(content,dictEntry->GetGroup(),
858 dictEntry->GetElement());
862 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
863 * through it's (group, element) and modifies it's content with
865 * @param content new value (string) to substitute with
866 * @param group group number of the Dicom Element to modify
867 * @param element element number of the Dicom Element to modify
869 bool gdcmDocument::SetEntryByNumber(std::string content,
873 gdcmValEntry* valEntry = GetValEntryByNumber(group, element);
876 dbg.Verbose(0, "gdcmDocument::SetEntryByNumber: no corresponding",
877 " ValEntry (try promotion first).");
880 // Non even content must be padded with a space (020H)...
881 if( content.length() % 2 )
883 content += '\0'; // ... therefore we padd with (000H) .!?!
885 valEntry->SetValue(content);
887 // Integers have a special treatement for their length:
888 gdcmVRKey vr = valEntry->GetVR();
889 if( vr == "US" || vr == "SS" )
891 valEntry->SetLength(2);
893 else if( vr == "UL" || vr == "SL" )
895 valEntry->SetLength(4);
899 valEntry->SetLength(content.length());
906 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
907 * through it's (group, element) and modifies it's content with
909 * @param content new value (void *) to substitute with
910 * @param lgth new value length
911 * @param group group number of the Dicom Element to modify
912 * @param element element number of the Dicom Element to modify
914 bool gdcmDocument::SetEntryByNumber(void *content,
919 (void)lgth; //not used
920 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
921 if ( !TagHT.count(key))
926 /* Hope Binary field length is *never* wrong
927 if(lgth%2) // Non even length are padded with a space (020H).
930 //content = content + '\0'; // fing a trick to enlarge a binary field?
934 a = (gdcmBinEntry *)TagHT[key];
935 a->SetVoidArea(content);
936 //a->SetLength(lgth); // ???
941 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
942 * in the PubDocEntrySet of this instance
943 * through it's (group, element) and modifies it's length with
945 * \warning Use with extreme caution.
946 * @param l new length to substitute with
947 * @param group group number of the Entry to modify
948 * @param element element number of the Entry to modify
949 * @return true on success, false otherwise.
951 bool gdcmDocument::SetEntryLengthByNumber(uint32_t l,
955 /// \todo use map methods, instead of multimap JPR
956 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
957 if ( !TagHT.count(key) )
963 l++; // length must be even
965 ( ((TagHT.equal_range(key)).first)->second )->SetLength(l);
971 * \brief Gets (from Header) the offset of a 'non string' element value
972 * (LoadElementValues has already be executed)
973 * @param Group group number of the Entry
974 * @param Elem element number of the Entry
975 * @return File Offset of the Element Value
977 size_t gdcmDocument::GetEntryOffsetByNumber(uint16_t group, uint16_t elem)
979 gdcmDocEntry* entry = GetDocEntryByNumber(group, elem);
982 dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry present.");
985 return entry->GetOffset();
989 * \brief Gets (from Header) a 'non string' element value
990 * (LoadElementValues has already be executed)
991 * @param Group group number of the Entry
992 * @param Elem element number of the Entry
993 * @return Pointer to the 'non string' area
995 void * gdcmDocument::GetEntryVoidAreaByNumber(uint16_t group, uint16_t elem)
997 gdcmDocEntry* entry = GetDocEntryByNumber(group, elem);
1000 dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry");
1003 return ((gdcmBinEntry *)entry)->GetVoidArea();
1007 * \brief Loads (from disk) the element content
1008 * when a string is not suitable
1009 * @param Group group number of the Entry
1010 * @param Elem element number of the Entry
1012 void *gdcmDocument::LoadEntryVoidArea(uint16_t group, uint16_t elem)
1014 gdcmDocEntry *docElement = GetDocEntryByNumber(group, elem);
1019 size_t o =(size_t)docElement->GetOffset();
1020 fseek(Fp, o, SEEK_SET);
1021 size_t l = docElement->GetLength();
1022 char* a = new char[l];
1025 dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a");
1028 size_t l2 = fread(a, 1, l , Fp);
1034 /// \todo Drop any already existing void area! JPR
1035 SetEntryVoidAreaByNumber(a, group, elem);
1040 * \brief Loads (from disk) the element content
1041 * when a string is not suitable
1042 * @param Element Entry whose voidArea is going to be loaded
1044 void *gdcmDocument::LoadEntryVoidArea(gdcmBinEntry *element)
1046 size_t o =(size_t)element->GetOffset();
1047 fseek(Fp, o, SEEK_SET);
1048 size_t l = element->GetLength();
1049 char* a = new char[l];
1052 dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a");
1055 element->SetVoidArea((void *)a);
1056 /// \todo check the result
1057 size_t l2 = fread(a, 1, l , Fp);
1068 * \brief Sets a 'non string' value to a given Dicom Element
1069 * @param area area containing the 'non string' value
1070 * @param group Group number of the searched Dicom Element
1071 * @param element Element number of the searched Dicom Element
1074 bool gdcmDocument::SetEntryVoidAreaByNumber(void * area,
1078 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
1079 if ( !TagHT.count(key))
1083 // This was for multimap ?
1084 (( gdcmBinEntry *)( ((TagHT.equal_range(key)).first)->second ))->SetVoidArea(area);
1090 * \brief Update the entries with the shadow dictionary.
1091 * Only non even entries are analyzed
1093 void gdcmDocument::UpdateShaEntries()
1095 //gdcmDictEntry *entry;
1098 /// \todo TODO : still any use to explore recursively the whole structure?
1100 for(ListTag::iterator it=listEntries.begin();
1101 it!=listEntries.end();
1104 // Odd group => from public dictionary
1105 if((*it)->GetGroup()%2==0)
1108 // Peer group => search the corresponding dict entry
1110 entry=RefShaDict->GetDictEntryByNumber((*it)->GetGroup(),(*it)->GetElement());
1114 if((*it)->IsImplicitVR())
1119 (*it)->SetValue(GetDocEntryUnvalue(*it)); // to go on compiling
1121 // Set the new entry and the new value
1122 (*it)->SetDictEntry(entry);
1123 CheckDocEntryVR(*it,vr);
1125 (*it)->SetValue(GetDocEntryValue(*it)); // to go on compiling
1130 // Remove precedent value transformation
1131 (*it)->SetDictEntry(NewVirtualDictEntry((*it)->GetGroup(),(*it)->GetElement(),vr));
1138 * \brief Searches within the Header Entries for a Dicom Element of
1140 * @param tagName name of the searched Dicom Element.
1141 * @return Corresponding Dicom Element when it exists, and NULL
1144 gdcmDocEntry* gdcmDocument::GetDocEntryByName(std::string const & tagName)
1146 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
1152 return GetDocEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement());
1156 * \brief retrieves a Dicom Element (the first one) using (group, element)
1157 * \warning (group, element) IS NOT an identifier inside the Dicom Header
1158 * if you think it's NOT UNIQUE, check the count number
1159 * and use iterators to retrieve ALL the Dicoms Elements within
1160 * a given couple (group, element)
1161 * @param group Group number of the searched Dicom Element
1162 * @param element Element number of the searched Dicom Element
1165 gdcmDocEntry* gdcmDocument::GetDocEntryByNumber(uint16_t group,
1168 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
1169 if ( !TagHT.count(key))
1173 return TagHT.find(key)->second;
1177 * \brief Same as \ref gdcmDocument::GetDocEntryByNumber except it only
1178 * returns a result when the corresponding entry is of type
1180 * @return When present, the corresponding ValEntry.
1182 gdcmValEntry* gdcmDocument::GetValEntryByNumber(uint16_t group,
1185 gdcmDocEntry* currentEntry = GetDocEntryByNumber(group, element);
1186 if ( !currentEntry )
1190 if ( gdcmValEntry* valEntry = dynamic_cast<gdcmValEntry*>(currentEntry) )
1194 dbg.Verbose(0, "gdcmDocument::GetValEntryByNumber: unfound ValEntry.");
1200 * \brief Loads the element while preserving the current
1201 * underlying file position indicator as opposed to
1202 * to LoadDocEntry that modifies it.
1203 * @param entry Header Entry whose value shall be loaded.
1206 void gdcmDocument::LoadDocEntrySafe(gdcmDocEntry * entry)
1208 long PositionOnEntry = ftell(Fp);
1209 LoadDocEntry(entry);
1210 fseek(Fp, PositionOnEntry, SEEK_SET);
1214 * \brief Swaps back the bytes of 4-byte long integer accordingly to
1216 * @return The properly swaped 32 bits integer.
1218 uint32_t gdcmDocument::SwapLong(uint32_t a)
1225 a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) |
1226 ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) );
1230 a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
1234 a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
1237 std::cout << "swapCode= " << SwapCode << std::endl;
1238 dbg.Error(" gdcmDocument::SwapLong : unset swap code");
1245 * \brief Unswaps back the bytes of 4-byte long integer accordingly to
1247 * @return The properly unswaped 32 bits integer.
1249 uint32_t gdcmDocument::UnswapLong(uint32_t a)
1255 * \brief Swaps the bytes so they agree with the processor order
1256 * @return The properly swaped 16 bits integer.
1258 uint16_t gdcmDocument::SwapShort(uint16_t a)
1260 if ( SwapCode == 4321 || SwapCode == 2143 )
1262 a =((( a << 8 ) & 0x0ff00 ) | (( a >> 8 ) & 0x00ff ) );
1268 * \brief Unswaps the bytes so they agree with the processor order
1269 * @return The properly unswaped 16 bits integer.
1271 uint16_t gdcmDocument::UnswapShort(uint16_t a)
1273 return SwapShort(a);
1276 //-----------------------------------------------------------------------------
1280 * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries)
1281 * @return length of the parsed set.
1284 long gdcmDocument::ParseDES(gdcmDocEntrySet *set,
1289 gdcmDocEntry *newDocEntry = 0;
1290 gdcmValEntry *newValEntry = 0;
1291 unsigned long l = 0;
1293 int depth = set->GetDepthLevel();
1296 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1300 newDocEntry = ReadNextDocEntry( );
1306 gdcmVRKey vr = newDocEntry->GetVR();
1310 if ( gdcmGlobal::GetVR()->IsVROfGdcmStringRepresentable(vr) )
1313 newValEntry = new gdcmValEntry(newDocEntry->GetDictEntry());
1314 newValEntry->Copy(newDocEntry);
1315 newValEntry->SetDepthLevel(depth);
1316 set->AddEntry(newValEntry);
1317 LoadDocEntry(newValEntry);
1318 if (newValEntry->isItemDelimitor())
1322 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1329 if ( ! gdcmGlobal::GetVR()->IsVROfGdcmBinaryRepresentable(vr) )
1331 ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR
1332 dbg.Verbose(0, "gdcmDocument::ParseDES: neither Valentry, "
1333 "nor BinEntry. Probably unknown VR.");
1336 ////// BinEntry or UNKOWN VR:
1337 gdcmBinEntry *bn = new gdcmBinEntry(newDocEntry->GetDictEntry());
1338 bn->Copy(newDocEntry);
1343 if (newDocEntry->GetGroup() == 0x7fe0 &&
1344 newDocEntry->GetElement() == 0x0010 )
1346 if (newDocEntry->GetReadLength()==0xffffffff)
1348 // Broken US.3405.1.dcm
1349 Parse7FE0(); // to skip the pixels
1350 // (multipart JPEG/RLE are trouble makers)
1354 SkipToNextDocEntry(newDocEntry);
1355 l = newDocEntry->GetFullLength();
1360 // to be sure we are at the beginning
1361 SkipToNextDocEntry(newDocEntry);
1362 l = newDocEntry->GetFullLength();
1368 l = newDocEntry->GetReadLength();
1369 if ( l != 0 ) // don't mess the delim_mode for zero-length sequence
1371 if ( l == 0xffffffff )
1380 // no other way to create it ...
1381 gdcmSeqEntry *sq = new gdcmSeqEntry(newDocEntry->GetDictEntry(),
1382 set->GetDepthLevel());
1383 sq->Copy(newDocEntry);
1384 sq->SetDelimitorMode(delim_mode);
1385 sq->SetDepthLevel(depth);
1388 { // Don't try to parse zero-length sequences
1389 long lgt = ParseSQ( sq,
1390 newDocEntry->GetOffset(),
1392 (void)lgt; //not used...
1395 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1403 return l; // Probably useless
1407 * \brief Parses a Sequence ( SeqEntry after SeqEntry)
1408 * @return parsed length for this level
1410 long gdcmDocument::ParseSQ(gdcmSeqEntry *set,
1411 long offset, long l_max, bool delim_mode)
1413 int SQItemNumber = 0;
1415 //int depth = set->GetDepthLevel();
1416 //(void)depth; //not used
1420 gdcmDocEntry *newDocEntry = ReadNextDocEntry();
1423 // FIXME Should warn user
1428 if ( newDocEntry->isSequenceDelimitor() )
1430 set->SetSequenceDelimitationItem( newDocEntry );
1434 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1439 gdcmSQItem *itemSQ = new gdcmSQItem(set->GetDepthLevel());
1440 itemSQ->AddEntry(newDocEntry);
1441 unsigned int l = newDocEntry->GetReadLength();
1443 if ( l == 0xffffffff )
1452 int lgr = ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod);
1453 (void)lgr; //FIXME not used
1455 set->AddEntry(itemSQ, SQItemNumber);
1457 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1463 int lgth = ftell(Fp) - offset;
1468 * \brief Loads the element content if its length doesn't exceed
1469 * the value specified with gdcmDocument::SetMaxSizeLoadEntry()
1470 * @param Entry Header Entry (Dicom Element) to be dealt with
1472 void gdcmDocument::LoadDocEntry(gdcmDocEntry* entry)
1475 uint16_t group = entry->GetGroup();
1476 std::string vr = entry->GetVR();
1477 uint32_t length = entry->GetLength();
1479 fseek(Fp, (long)entry->GetOffset(), SEEK_SET);
1481 // A SeQuence "contains" a set of Elements.
1482 // (fffe e000) tells us an Element is beginning
1483 // (fffe e00d) tells us an Element just ended
1484 // (fffe e0dd) tells us the current SeQuence just ended
1485 if( group == 0xfffe )
1487 // NO more value field for SQ !
1491 // When the length is zero things are easy:
1494 ((gdcmValEntry *)entry)->SetValue("");
1498 // The elements whose length is bigger than the specified upper bound
1499 // are not loaded. Instead we leave a short notice of the offset of
1500 // the element content and it's length.
1502 std::ostringstream s;
1503 if (length > MaxSizeLoadEntry)
1505 if (gdcmBinEntry* binEntryPtr = dynamic_cast< gdcmBinEntry* >(entry) )
1507 s << "gdcm::NotLoaded (BinEntry)";
1508 s << " Address:" << (long)entry->GetOffset();
1509 s << " Length:" << entry->GetLength();
1510 s << " x(" << std::hex << entry->GetLength() << ")";
1511 binEntryPtr->SetValue(s.str());
1513 // to be sure we are at the end of the value ...
1514 fseek(Fp, (long)entry->GetOffset()+(long)entry->GetLength(), SEEK_SET);
1516 return; //FIXME FIXME FIXME FIXME ????
1518 // Be carefull : a BinEntry IS_A ValEntry ...
1519 if (gdcmValEntry* valEntryPtr = dynamic_cast< gdcmValEntry* >(entry) )
1521 s << "gdcm::NotLoaded. (ValEntry)";
1522 s << " Address:" << (long)entry->GetOffset();
1523 s << " Length:" << entry->GetLength();
1524 s << " x(" << std::hex << entry->GetLength() << ")";
1525 valEntryPtr->SetValue(s.str());
1527 // to be sure we are at the end of the value ...
1528 fseek(Fp,(long)entry->GetOffset()+(long)entry->GetLength(),SEEK_SET);
1533 // When we find a BinEntry not very much can be done :
1534 if (gdcmBinEntry* binEntryPtr = dynamic_cast< gdcmBinEntry* >(entry) )
1537 LoadEntryVoidArea(binEntryPtr);
1538 s << "gdcm::Loaded (BinEntry)";
1539 binEntryPtr->SetValue(s.str());
1543 /// \todo Any compacter code suggested (?)
1544 if ( IsDocEntryAnInteger(entry) )
1547 std::ostringstream s;
1549 // When short integer(s) are expected, read and convert the following
1550 // n *two characters properly i.e. consider them as short integers as
1551 // opposed to strings.
1552 // Elements with Value Multiplicity > 1
1553 // contain a set of integers (not a single one)
1554 if (vr == "US" || vr == "SS")
1557 NewInt = ReadInt16();
1561 for (int i=1; i < nbInt; i++)
1564 NewInt = ReadInt16();
1569 // See above comment on multiple integers (mutatis mutandis).
1570 else if (vr == "UL" || vr == "SL")
1573 NewInt = ReadInt32();
1577 for (int i=1; i < nbInt; i++)
1580 NewInt = ReadInt32();
1585 #ifdef GDCM_NO_ANSI_STRING_STREAM
1586 s << std::ends; // to avoid oddities on Solaris
1587 #endif //GDCM_NO_ANSI_STRING_STREAM
1589 ((gdcmValEntry *)entry)->SetValue(s.str());
1593 // We need an additional byte for storing \0 that is not on disk
1594 //std::string newValue(length,0);
1595 //item_read = fread(&(newValue[0]), (size_t)length, (size_t)1, Fp);
1596 //rah !! I can't believe it could work, normally this is a const char* !!!
1597 char *str = new char[length+1];
1598 item_read = fread(str, (size_t)length, (size_t)1, Fp);
1600 std::string newValue = str;
1602 if ( gdcmValEntry* valEntry = dynamic_cast<gdcmValEntry* >(entry) )
1604 if ( item_read != 1 )
1606 dbg.Verbose(1, "gdcmDocument::LoadDocEntry",
1607 "unread element value");
1608 valEntry->SetValue("gdcm::UnRead");
1614 // Because of correspondance with the VR dic
1615 valEntry->SetValue(newValue);
1619 valEntry->SetValue(newValue);
1624 dbg.Error(true, "gdcmDocument::LoadDocEntry"
1625 "Should have a ValEntry, here !");
1631 * \brief Find the value Length of the passed Header Entry
1632 * @param Entry Header Entry whose length of the value shall be loaded.
1634 void gdcmDocument::FindDocEntryLength (gdcmDocEntry *entry)
1636 uint16_t element = entry->GetElement();
1637 std::string vr = entry->GetVR();
1641 if ( Filetype == gdcmExplicitVR && !entry->IsImplicitVR() )
1643 if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UN" )
1645 // The following reserved two bytes (see PS 3.5-2001, section
1646 // 7.1.2 Data element structure with explicit vr p27) must be
1647 // skipped before proceeding on reading the length on 4 bytes.
1648 fseek(Fp, 2L, SEEK_CUR);
1649 uint32_t length32 = ReadInt32();
1651 if ( vr == "OB" && length32 == 0xffffffff )
1653 uint32_t lengthOB = FindDocEntryLengthOB();
1656 // Computing the length failed (this happens with broken
1657 // files like gdcm-JPEG-LossLess3a.dcm). We still have a
1658 // chance to get the pixels by deciding the element goes
1659 // until the end of the file. Hence we artificially fix the
1660 // the length and proceed.
1661 long currentPosition = ftell(Fp);
1662 fseek(Fp,0L,SEEK_END);
1663 long lengthUntilEOF = ftell(Fp) - currentPosition;
1664 fseek(Fp, currentPosition, SEEK_SET);
1665 entry->SetLength(lengthUntilEOF);
1669 entry->SetLength(lengthOB);
1672 FixDocEntryFoundLength(entry, length32);
1676 // Length is encoded on 2 bytes.
1677 length16 = ReadInt16();
1679 // We can tell the current file is encoded in big endian (like
1680 // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag
1681 // and it's value is the one of the encoding of a big endian file.
1682 // In order to deal with such big endian encoded files, we have
1683 // (at least) two strategies:
1684 // * when we load the "Transfer Syntax" tag with value of big endian
1685 // encoding, we raise the proper flags. Then we wait for the end
1686 // of the META group (0x0002) among which is "Transfer Syntax",
1687 // before switching the swap code to big endian. We have to postpone
1688 // the switching of the swap code since the META group is fully encoded
1689 // in little endian, and big endian coding only starts at the next
1690 // group. The corresponding code can be hard to analyse and adds
1691 // many additional unnecessary tests for regular tags.
1692 // * the second strategy consists in waiting for trouble, that shall
1693 // appear when we find the first group with big endian encoding. This
1694 // is easy to detect since the length of a "Group Length" tag (the
1695 // ones with zero as element number) has to be of 4 (0x0004). When we
1696 // encounter 1024 (0x0400) chances are the encoding changed and we
1697 // found a group with big endian encoding.
1698 // We shall use this second strategy. In order to make sure that we
1699 // can interpret the presence of an apparently big endian encoded
1700 // length of a "Group Length" without committing a big mistake, we
1701 // add an additional check: we look in the already parsed elements
1702 // for the presence of a "Transfer Syntax" whose value has to be "big
1703 // endian encoding". When this is the case, chances are we have got our
1704 // hands on a big endian encoded file: we switch the swap code to
1705 // big endian and proceed...
1706 if ( (element == 0x0000) && (length16 == 0x0400) )
1708 if ( ! IsExplicitVRBigEndianTransferSyntax() )
1710 dbg.Verbose(0, "gdcmDocument::FindLength", "not explicit VR");
1715 SwitchSwapToBigEndian();
1716 // Restore the unproperly loaded values i.e. the group, the element
1717 // and the dictionary entry depending on them.
1718 uint16_t correctGroup = SwapShort( entry->GetGroup() );
1719 uint16_t correctElem = SwapShort( entry->GetElement() );
1720 gdcmDictEntry* newTag = GetDictEntryByNumber( correctGroup,
1724 // This correct tag is not in the dictionary. Create a new one.
1725 newTag = NewVirtualDictEntry(correctGroup, correctElem);
1727 // FIXME this can create a memory leaks on the old entry that be
1728 // left unreferenced.
1729 entry->SetDictEntry( newTag );
1732 // Heuristic: well, some files are really ill-formed.
1733 if ( length16 == 0xffff)
1736 // Length16= 0xffff means that we deal with
1737 // 'Unknown Length' Sequence
1739 FixDocEntryFoundLength(entry, (uint32_t)length16);
1744 // Either implicit VR or a non DICOM conformal (see note below) explicit
1745 // VR that ommited the VR of (at least) this element. Farts happen.
1746 // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25
1747 // on Data elements "Implicit and Explicit VR Data Elements shall
1748 // not coexist in a Data Set and Data Sets nested within it".]
1749 // Length is on 4 bytes.
1751 FixDocEntryFoundLength(entry, ReadInt32());
1757 * \brief Find the Value Representation of the current Dicom Element.
1760 void gdcmDocument::FindDocEntryVR( gdcmDocEntry *entry )
1762 if ( Filetype != gdcmExplicitVR )
1769 long positionOnEntry = ftell(Fp);
1770 // Warning: we believe this is explicit VR (Value Representation) because
1771 // we used a heuristic that found "UL" in the first tag. Alas this
1772 // doesn't guarantee that all the tags will be in explicit VR. In some
1773 // cases (see e-film filtered files) one finds implicit VR tags mixed
1774 // within an explicit VR file. Hence we make sure the present tag
1775 // is in explicit VR and try to fix things if it happens not to be
1778 fread (&vr, (size_t)2,(size_t)1, Fp);
1781 if( !CheckDocEntryVR(entry, vr) )
1783 fseek(Fp, positionOnEntry, SEEK_SET);
1784 // When this element is known in the dictionary we shall use, e.g. for
1785 // the semantics (see the usage of IsAnInteger), the VR proposed by the
1786 // dictionary entry. Still we have to flag the element as implicit since
1787 // we know now our assumption on expliciteness is not furfilled.
1789 if ( entry->IsVRUnknown() )
1791 entry->SetVR("Implicit");
1793 entry->SetImplicitVR();
1798 * \brief Check the correspondance between the VR of the header entry
1799 * and the taken VR. If they are different, the header entry is
1800 * updated with the new VR.
1801 * @param Entry Header Entry to check
1802 * @param vr Dicom Value Representation
1803 * @return false if the VR is incorrect of if the VR isn't referenced
1804 * otherwise, it returns true
1806 bool gdcmDocument::CheckDocEntryVR(gdcmDocEntry *entry, gdcmVRKey vr)
1808 char msg[100]; // for sprintf
1809 bool realExplicit = true;
1811 // Assume we are reading a falsely explicit VR file i.e. we reached
1812 // a tag where we expect reading a VR but are in fact we read the
1813 // first to bytes of the length. Then we will interogate (through find)
1814 // the dicom_vr dictionary with oddities like "\004\0" which crashes
1815 // both GCC and VC++ implementations of the STL map. Hence when the
1816 // expected VR read happens to be non-ascii characters we consider
1817 // we hit falsely explicit VR tag.
1819 if ( (!isalpha(vr[0])) && (!isalpha(vr[1])) )
1821 realExplicit = false;
1824 // CLEANME searching the dicom_vr at each occurence is expensive.
1825 // PostPone this test in an optional integrity check at the end
1826 // of parsing or only in debug mode.
1827 if ( realExplicit && !gdcmGlobal::GetVR()->Count(vr) )
1829 realExplicit = false;
1832 if ( !realExplicit )
1834 // We thought this was explicit VR, but we end up with an
1835 // implicit VR tag. Let's backtrack.
1836 sprintf(msg,"Falsely explicit vr file (%04x,%04x)\n",
1837 entry->GetGroup(), entry->GetElement());
1838 dbg.Verbose(1, "gdcmDocument::FindVR: ",msg);
1840 if( entry->GetGroup() % 2 && entry->GetElement() == 0x0000)
1842 // Group length is UL !
1843 gdcmDictEntry* newEntry = NewVirtualDictEntry(
1844 entry->GetGroup(), entry->GetElement(),
1845 "UL", "FIXME", "Group Length");
1846 entry->SetDictEntry( newEntry );
1851 if ( entry->IsVRUnknown() )
1853 // When not a dictionary entry, we can safely overwrite the VR.
1854 if( entry->GetElement() == 0x0000 )
1856 // Group length is UL !
1864 else if ( entry->GetVR() != vr )
1866 // The VR present in the file and the dictionary disagree. We assume
1867 // the file writer knew best and use the VR of the file. Since it would
1868 // be unwise to overwrite the VR of a dictionary (since it would
1869 // compromise it's next user), we need to clone the actual DictEntry
1870 // and change the VR for the read one.
1871 gdcmDictEntry* newEntry = NewVirtualDictEntry(
1872 entry->GetGroup(), entry->GetElement(),
1873 vr, "FIXME", entry->GetName());
1874 entry->SetDictEntry(newEntry);
1881 * \brief Get the transformed value of the header entry. The VR value
1882 * is used to define the transformation to operate on the value
1883 * \warning NOT end user intended method !
1885 * @return Transformed entry value
1887 std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *entry)
1889 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1891 std::string val = ((gdcmValEntry *)entry)->GetValue();
1892 std::string vr = entry->GetVR();
1893 uint32_t length = entry->GetLength();
1894 std::ostringstream s;
1897 // When short integer(s) are expected, read and convert the following
1898 // n * 2 bytes properly i.e. as a multivaluated strings
1899 // (each single value is separated fromthe next one by '\'
1900 // as usual for standard multivaluated filels
1901 // Elements with Value Multiplicity > 1
1902 // contain a set of short integers (not a single one)
1904 if( vr == "US" || vr == "SS" )
1909 for (int i=0; i < nbInt; i++)
1915 newInt16 = ( val[2*i+0] & 0xFF ) + ( ( val[2*i+1] & 0xFF ) << 8);
1916 newInt16 = SwapShort( newInt16 );
1921 // When integer(s) are expected, read and convert the following
1922 // n * 4 bytes properly i.e. as a multivaluated strings
1923 // (each single value is separated fromthe next one by '\'
1924 // as usual for standard multivaluated filels
1925 // Elements with Value Multiplicity > 1
1926 // contain a set of integers (not a single one)
1927 else if( vr == "UL" || vr == "SL" )
1932 for (int i=0; i < nbInt; i++)
1938 newInt32 = ( val[4*i+0] & 0xFF )
1939 + (( val[4*i+1] & 0xFF ) << 8 )
1940 + (( val[4*i+2] & 0xFF ) << 16 )
1941 + (( val[4*i+3] & 0xFF ) << 24 );
1942 newInt32 = SwapLong( newInt32 );
1946 #ifdef GDCM_NO_ANSI_STRING_STREAM
1947 s << std::ends; // to avoid oddities on Solaris
1948 #endif //GDCM_NO_ANSI_STRING_STREAM
1952 return ((gdcmValEntry *)entry)->GetValue();
1956 * \brief Get the reverse transformed value of the header entry. The VR
1957 * value is used to define the reverse transformation to operate on
1959 * \warning NOT end user intended method !
1961 * @return Reverse transformed entry value
1963 std::string gdcmDocument::GetDocEntryUnvalue(gdcmDocEntry* entry)
1965 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1967 std::string vr = entry->GetVR();
1968 std::vector<std::string> tokens;
1969 std::ostringstream s;
1971 if ( vr == "US" || vr == "SS" )
1975 tokens.erase( tokens.begin(), tokens.end()); // clean any previous value
1976 Tokenize (((gdcmValEntry *)entry)->GetValue(), tokens, "\\");
1977 for (unsigned int i=0; i<tokens.size(); i++)
1979 newInt16 = atoi(tokens[i].c_str());
1980 s << ( newInt16 & 0xFF )
1981 << (( newInt16 >> 8 ) & 0xFF );
1985 if ( vr == "UL" || vr == "SL")
1989 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1990 Tokenize (((gdcmValEntry *)entry)->GetValue(), tokens, "\\");
1991 for (unsigned int i=0; i<tokens.size();i++)
1993 newInt32 = atoi(tokens[i].c_str());
1994 s << (char)( newInt32 & 0xFF )
1995 << (char)(( newInt32 >> 8 ) & 0xFF )
1996 << (char)(( newInt32 >> 16 ) & 0xFF )
1997 << (char)(( newInt32 >> 24 ) & 0xFF );
2002 #ifdef GDCM_NO_ANSI_STRING_STREAM
2003 s << std::ends; // to avoid oddities on Solaris
2004 #endif //GDCM_NO_ANSI_STRING_STREAM
2008 return ((gdcmValEntry *)entry)->GetValue();
2012 * \brief Skip a given Header Entry
2013 * \warning NOT end user intended method !
2016 void gdcmDocument::SkipDocEntry(gdcmDocEntry *entry)
2018 SkipBytes(entry->GetLength());
2022 * \brief Skips to the begining of the next Header Entry
2023 * \warning NOT end user intended method !
2026 void gdcmDocument::SkipToNextDocEntry(gdcmDocEntry *entry)
2028 fseek(Fp, (long)(entry->GetOffset()), SEEK_SET);
2029 fseek(Fp, (long)(entry->GetReadLength()), SEEK_CUR);
2033 * \brief When the length of an element value is obviously wrong (because
2034 * the parser went Jabberwocky) one can hope improving things by
2035 * applying some heuristics.
2037 void gdcmDocument::FixDocEntryFoundLength(gdcmDocEntry *entry,
2038 uint32_t foundLength)
2040 entry->SetReadLength( foundLength ); // will be updated only if a bug is found
2041 if ( foundLength == 0xffffffff)
2046 uint16_t gr = entry->GetGroup();
2047 uint16_t el = entry->GetElement();
2049 if ( foundLength % 2)
2051 std::ostringstream s;
2052 s << "Warning : Tag with uneven length "
2054 << " in x(" << std::hex << gr << "," << el <<")" << std::dec;
2055 dbg.Verbose(0, s.str().c_str());
2058 //////// Fix for some naughty General Electric images.
2059 // Allthough not recent many such GE corrupted images are still present
2060 // on Creatis hard disks. Hence this fix shall remain when such images
2061 // are no longer in user (we are talking a few years, here)...
2062 // Note: XMedCom probably uses such a trick since it is able to read
2063 // those pesky GE images ...
2064 if ( foundLength == 13)
2066 // Only happens for this length !
2067 if ( entry->GetGroup() != 0x0008
2068 || ( entry->GetElement() != 0x0070
2069 && entry->GetElement() != 0x0080 ) )
2072 entry->SetReadLength(10); /// \todo a bug is to be fixed !?
2076 //////// Fix for some brain-dead 'Leonardo' Siemens images.
2077 // Occurence of such images is quite low (unless one leaves close to a
2078 // 'Leonardo' source. Hence, one might consider commenting out the
2079 // following fix on efficiency reasons.
2080 else if ( entry->GetGroup() == 0x0009
2081 && ( entry->GetElement() == 0x1113
2082 || entry->GetElement() == 0x1114 ) )
2085 entry->SetReadLength(4); /// \todo a bug is to be fixed !?
2088 //////// Deal with sequences, but only on users request:
2089 else if ( entry->GetVR() == "SQ" && EnableSequences)
2091 foundLength = 0; // ReadLength is unchanged
2094 //////// We encountered a 'delimiter' element i.e. a tag of the form
2095 // "fffe|xxxx" which is just a marker. Delimiters length should not be
2096 // taken into account.
2097 else if( entry->GetGroup() == 0xfffe )
2099 // According to the norm, fffe|0000 shouldn't exist. BUT the Philips
2100 // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to
2101 // causes extra troubles...
2102 if( entry->GetElement() != 0x0000 )
2108 entry->SetUsableLength(foundLength);
2112 * \brief Apply some heuristics to predict whether the considered
2113 * element value contains/represents an integer or not.
2114 * @param Entry The element value on which to apply the predicate.
2115 * @return The result of the heuristical predicate.
2117 bool gdcmDocument::IsDocEntryAnInteger(gdcmDocEntry *entry)
2119 uint16_t element = entry->GetElement();
2120 uint16_t group = entry->GetGroup();
2121 std::string vr = entry->GetVR();
2122 uint32_t length = entry->GetLength();
2124 // When we have some semantics on the element we just read, and if we
2125 // a priori know we are dealing with an integer, then we shall be
2126 // able to swap it's element value properly.
2127 if ( element == 0 ) // This is the group length of the group
2135 // Allthough this should never happen, still some images have a
2136 // corrupted group length [e.g. have a glance at offset x(8336) of
2137 // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm].
2138 // Since for dicom compliant and well behaved headers, the present
2139 // test is useless (and might even look a bit paranoid), when we
2140 // encounter such an ill-formed image, we simply display a warning
2141 // message and proceed on parsing (while crossing fingers).
2142 std::ostringstream s;
2143 int filePosition = ftell(Fp);
2144 s << "Erroneous Group Length element length on : (" \
2145 << std::hex << group << " , " << element
2146 << ") -before- position x(" << filePosition << ")"
2147 << "lgt : " << length;
2148 dbg.Verbose(0, "gdcmDocument::IsDocEntryAnInteger", s.str().c_str() );
2152 if ( vr == "UL" || vr == "US" || vr == "SL" || vr == "SS" )
2161 * \brief Find the Length till the next sequence delimiter
2162 * \warning NOT end user intended method !
2166 uint32_t gdcmDocument::FindDocEntryLengthOB()
2168 // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
2169 long positionOnEntry = ftell(Fp);
2170 bool foundSequenceDelimiter = false;
2171 uint32_t totalLength = 0;
2173 while ( !foundSequenceDelimiter )
2175 uint16_t g = ReadInt16();
2176 uint16_t n = ReadInt16();
2182 // We have to decount the group and element we just read
2185 if ( g != 0xfffe || ( n != 0xe0dd && n != 0xe000 ) )
2187 dbg.Verbose(1, "gdcmDocument::FindLengthOB: neither an Item tag "
2188 "nor a Sequence delimiter tag.");
2189 fseek(Fp, positionOnEntry, SEEK_SET);
2196 foundSequenceDelimiter = true;
2199 uint32_t itemLength = ReadInt32();
2200 // We add 4 bytes since we just read the ItemLength with ReadInt32
2201 totalLength += itemLength + 4;
2202 SkipBytes(itemLength);
2204 if ( foundSequenceDelimiter )
2209 fseek(Fp, positionOnEntry, SEEK_SET);
2214 * \brief Reads a supposed to be 16 Bits integer
2215 * (swaps it depending on processor endianity)
2216 * @return read value
2218 uint16_t gdcmDocument::ReadInt16()
2221 size_t item_read = fread (&g, (size_t)2,(size_t)1, Fp);
2222 if ( item_read != 1 )
2226 dbg.Verbose(0, "gdcmDocument::ReadInt16", " File Error");
2237 * \brief Reads a supposed to be 32 Bits integer
2238 * (swaps it depending on processor endianity)
2239 * @return read value
2241 uint32_t gdcmDocument::ReadInt32()
2244 size_t item_read = fread (&g, (size_t)4,(size_t)1, Fp);
2245 if ( item_read != 1 )
2249 dbg.Verbose(0, "gdcmDocument::ReadInt32", " File Error");
2260 * \brief skips bytes inside the source file
2261 * \warning NOT end user intended method !
2264 void gdcmDocument::SkipBytes(uint32_t nBytes)
2266 //FIXME don't dump the returned value
2267 (void)fseek(Fp, (long)nBytes, SEEK_CUR);
2271 * \brief Loads all the needed Dictionaries
2272 * \warning NOT end user intended method !
2274 void gdcmDocument::Initialise()
2276 RefPubDict = gdcmGlobal::GetDicts()->GetDefaultPubDict();
2281 * \brief Discover what the swap code is (among little endian, big endian,
2282 * bad little endian, bad big endian).
2284 * @return false when we are absolutely sure
2285 * it's neither ACR-NEMA nor DICOM
2286 * true when we hope ours assuptions are OK
2288 bool gdcmDocument::CheckSwap()
2290 // The only guaranted way of finding the swap code is to find a
2291 // group tag since we know it's length has to be of four bytes i.e.
2292 // 0x00000004. Finding the swap code in then straigthforward. Trouble
2293 // occurs when we can't find such group...
2295 uint32_t x = 4; // x : for ntohs
2296 bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
2300 char deb[HEADER_LENGTH_TO_READ];
2302 // First, compare HostByteOrder and NetworkByteOrder in order to
2303 // determine if we shall need to swap bytes (i.e. the Endian type).
2304 if ( x == ntohs(x) )
2313 // The easiest case is the one of a DICOM header, since it possesses a
2314 // file preamble where it suffice to look for the string "DICM".
2315 int lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, Fp);
2316 (void)lgrLue; //FIXME not used
2318 char *entCur = deb + 128;
2319 if( memcmp(entCur, "DICM", (size_t)4) == 0 )
2321 dbg.Verbose(1, "gdcmDocument::CheckSwap:", "looks like DICOM Version3");
2323 // Next, determine the value representation (VR). Let's skip to the
2324 // first element (0002, 0000) and check there if we find "UL"
2325 // - or "OB" if the 1st one is (0002,0001) -,
2326 // in which case we (almost) know it is explicit VR.
2327 // WARNING: if it happens to be implicit VR then what we will read
2328 // is the length of the group. If this ascii representation of this
2329 // length happens to be "UL" then we shall believe it is explicit VR.
2330 // FIXME: in order to fix the above warning, we could read the next
2331 // element value (or a couple of elements values) in order to make
2332 // sure we are not commiting a big mistake.
2333 // We need to skip :
2334 // * the 128 bytes of File Preamble (often padded with zeroes),
2335 // * the 4 bytes of "DICM" string,
2336 // * the 4 bytes of the first tag (0002, 0000),or (0002, 0001)
2337 // i.e. a total of 136 bytes.
2341 // Sometimes (see : gdcmData/icone.dcm) group 0x0002 *is* Explicit VR,
2342 // but elem 0002,0010 (Transfert Syntax) tells us the file is
2343 // *Implicit* VR. -and it is !-
2345 if( memcmp(entCur, "UL", (size_t)2) == 0 ||
2346 memcmp(entCur, "OB", (size_t)2) == 0 ||
2347 memcmp(entCur, "UI", (size_t)2) == 0 ||
2348 memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later
2349 // when Write DCM *adds*
2351 // Use gdcmDocument::dicom_vr to test all the possibilities
2352 // instead of just checking for UL, OB and UI !? group 0000
2354 Filetype = gdcmExplicitVR;
2355 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2356 "explicit Value Representation");
2360 Filetype = gdcmImplicitVR;
2361 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2362 "not an explicit Value Representation");
2368 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2369 "HostByteOrder != NetworkByteOrder");
2374 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2375 "HostByteOrder = NetworkByteOrder");
2378 // Position the file position indicator at first tag (i.e.
2379 // after the file preamble and the "DICM" string).
2381 fseek (Fp, 132L, SEEK_SET);
2385 // Alas, this is not a DicomV3 file and whatever happens there is no file
2386 // preamble. We can reset the file position indicator to where the data
2387 // is (i.e. the beginning of the file).
2388 dbg.Verbose(1, "gdcmDocument::CheckSwap:", "not a DICOM Version3 file");
2391 // Our next best chance would be to be considering a 'clean' ACR/NEMA file.
2392 // By clean we mean that the length of the first tag is written down.
2393 // If this is the case and since the length of the first group HAS to be
2394 // four (bytes), then determining the proper swap code is straightforward.
2397 // We assume the array of char we are considering contains the binary
2398 // representation of a 32 bits integer. Hence the following dirty
2400 s32 = *((uint32_t *)(entCur));
2421 // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
2422 // It is time for despaired wild guesses.
2423 // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA,
2424 // i.e. the 'group length' element is not present :
2426 // check the supposed to be 'group number'
2427 // 0x0002 or 0x0004 or 0x0008
2428 // to determine ' SwapCode' value .
2429 // Only 0 or 4321 will be possible
2430 // (no oportunity to check for the formerly well known
2431 // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian'
2432 // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -4, 8-)
2433 // the file IS NOT ACR-NEMA nor DICOM V3
2434 // Find a trick to tell it the caller...
2436 s16 = *((uint16_t *)(deb));
2453 dbg.Verbose(0, "gdcmDocument::CheckSwap:",
2454 "ACR/NEMA unfound swap info (Really hopeless !)");
2455 Filetype = gdcmUnknown;
2458 // Then the only info we have is the net2host one.
2468 * \brief Restore the unproperly loaded values i.e. the group, the element
2469 * and the dictionary entry depending on them.
2471 void gdcmDocument::SwitchSwapToBigEndian()
2473 dbg.Verbose(1, "gdcmDocument::SwitchSwapToBigEndian",
2474 "Switching to BigEndian mode.");
2475 if ( SwapCode == 0 )
2479 else if ( SwapCode == 4321 )
2483 else if ( SwapCode == 3412 )
2487 else if ( SwapCode == 2143 )
2494 * \brief during parsing, Header Elements too long are not loaded in memory
2497 void gdcmDocument::SetMaxSizeLoadEntry(long newSize)
2503 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2505 MaxSizeLoadEntry = 0xffffffff;
2508 MaxSizeLoadEntry = newSize;
2513 * \brief Header Elements too long will not be printed
2514 * \todo See comments of \ref gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE
2517 void gdcmDocument::SetMaxSizePrintEntry(long newSize)
2519 //DOH !! This is exactly SetMaxSizeLoadEntry FIXME FIXME
2524 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2526 MaxSizePrintEntry = 0xffffffff;
2529 MaxSizePrintEntry = newSize;
2535 * \brief Read the next tag but WITHOUT loading it's value
2536 * (read the 'Group Number', the 'Element Number',
2537 * gets the Dict Entry
2538 * gets the VR, gets the length, gets the offset value)
2539 * @return On succes the newly created DocEntry, NULL on failure.
2541 gdcmDocEntry *gdcmDocument::ReadNextDocEntry()
2543 uint16_t g = ReadInt16();
2544 uint16_t n = ReadInt16();
2548 // We reached the EOF (or an error occured) therefore
2549 // header parsing has to be considered as finished.
2552 gdcmDocEntry *newEntry = NewDocEntryByNumber(g, n);
2554 FindDocEntryVR(newEntry);
2555 FindDocEntryLength(newEntry);
2563 newEntry->SetOffset(ftell(Fp));
2570 * \brief Generate a free gdcmTagKey i.e. a gdcmTagKey that is not present
2571 * in the TagHt dictionary.
2572 * @param group The generated tag must belong to this group.
2573 * @return The element of tag with given group which is fee.
2575 uint32_t gdcmDocument::GenerateFreeTagKeyInGroup(uint16_t group)
2577 for (uint32_t elem = 0; elem < UINT32_MAX; elem++)
2579 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, elem);
2580 if (TagHT.count(key) == 0)
2589 * \brief Assuming the internal file pointer \ref gdcmDocument::fp
2590 * is placed at the beginning of a tag check whether this
2591 * tag is (TestGroup, TestElement).
2592 * \warning On success the internal file pointer \ref gdcmDocument::fp
2593 * is modified to point after the tag.
2594 * On failure (i.e. when the tag wasn't the expected tag
2595 * (TestGroup, TestElement) the internal file pointer
2596 * \ref gdcmDocument::fp is restored to it's original position.
2597 * @param TestGroup The expected group of the tag.
2598 * @param TestElement The expected Element of the tag.
2599 * @return True on success, false otherwise.
2601 bool gdcmDocument::ReadTag(uint16_t testGroup, uint16_t testElement)
2603 long positionOnEntry = ftell(Fp);
2604 long currentPosition = ftell(Fp); // On debugging purposes
2606 //// Read the Item Tag group and element, and make
2607 // sure they are what we expected:
2608 uint16_t itemTagGroup = ReadInt16();
2609 uint16_t itemTagElement = ReadInt16();
2610 if ( itemTagGroup != testGroup || itemTagElement != testElement )
2612 std::ostringstream s;
2613 s << " We should have found tag (";
2614 s << std::hex << testGroup << "," << testElement << ")" << std::endl;
2615 s << " but instead we encountered tag (";
2616 s << std::hex << itemTagGroup << "," << itemTagElement << ")"
2618 s << " at address: " << (unsigned)currentPosition << std::endl;
2619 dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: wrong Item Tag found:");
2620 dbg.Verbose(0, s.str().c_str());
2621 fseek(Fp, positionOnEntry, SEEK_SET);
2629 * \brief Assuming the internal file pointer \ref gdcmDocument::fp
2630 * is placed at the beginning of a tag (TestGroup, TestElement),
2631 * read the length associated to the Tag.
2632 * \warning On success the internal file pointer \ref gdcmDocument::fp
2633 * is modified to point after the tag and it's length.
2634 * On failure (i.e. when the tag wasn't the expected tag
2635 * (TestGroup, TestElement) the internal file pointer
2636 * \ref gdcmDocument::fp is restored to it's original position.
2637 * @param TestGroup The expected group of the tag.
2638 * @param TestElement The expected Element of the tag.
2639 * @return On success returns the length associated to the tag. On failure
2642 uint32_t gdcmDocument::ReadTagLength(uint16_t testGroup, uint16_t testElement)
2644 long PositionOnEntry = ftell(Fp);
2645 (void)PositionOnEntry;
2647 if ( !ReadTag(testGroup, testElement) )
2652 //// Then read the associated Item Length
2653 long currentPosition = ftell(Fp);
2654 uint32_t itemLength = ReadInt32();
2656 std::ostringstream s;
2657 s << "Basic Item Length is: "
2658 << itemLength << std::endl;
2659 s << " at address: " << (unsigned)currentPosition << std::endl;
2660 dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: ", s.str().c_str());
2666 * \brief Parse pixel data from disk for multi-fragment Jpeg/Rle files
2667 * No other way so 'skip' the Data
2670 void gdcmDocument::Parse7FE0 ()
2672 gdcmDocEntry* element = GetDocEntryByNumber(0x0002, 0x0010);
2675 // Should warn user FIXME
2679 if ( IsImplicitVRLittleEndianTransferSyntax()
2680 || IsExplicitVRLittleEndianTransferSyntax()
2681 || IsExplicitVRBigEndianTransferSyntax() /// \todo 1.2.2 ??? A verifier !
2682 || IsDeflatedExplicitVRLittleEndianTransferSyntax() )
2687 // ---------------- for Parsing : Position on begining of Jpeg/RLE Pixels
2689 //// Read the Basic Offset Table Item Tag length...
2690 uint32_t itemLength = ReadTagLength(0xfffe, 0xe000);
2692 //// ... and then read length[s] itself[themselves]. We don't use
2693 // the values read (BTW what is the purpous of those lengths ?)
2694 if ( itemLength != 0 )
2696 // BTW, what is the purpous of those length anyhow !?
2697 char* basicOffsetTableItemValue = new char[itemLength + 1];
2698 fread(basicOffsetTableItemValue, itemLength, 1, Fp);
2700 for (unsigned int i=0; i < itemLength; i += 4 )
2702 uint32_t individualLength = str2num(&basicOffsetTableItemValue[i],uint32_t);
2703 std::ostringstream s;
2704 s << " Read one length: ";
2705 s << std::hex << individualLength << std::endl;
2706 dbg.Verbose(0, "gdcmDocument::Parse7FE0: ", s.str().c_str());
2710 if ( ! IsRLELossLessTransferSyntax() )
2714 //// We then skip (not reading them) all the fragments of images:
2715 while ( (itemLength = ReadTagLength(0xfffe, 0xe000)) )
2717 SkipBytes(itemLength);
2724 long rleSegmentLength[15], fragmentLength;
2726 // While we find some items:
2727 while ( (fragmentLength = ReadTagLength(0xfffe, 0xe000)) )
2729 // Parse fragments of the current Fragment (Frame)
2730 //------------------ scanning (not reading) fragment pixels
2731 uint32_t nbRleSegments = ReadInt32();
2733 //// Reading RLE Segments Offset Table
2734 uint32_t rleSegmentOffsetTable[15];
2735 for(int k=1; k<=15; k++)
2737 ftellRes = ftell(Fp);
2738 rleSegmentOffsetTable[k] = ReadInt32();
2741 // skipping (not reading) RLE Segments
2742 if ( nbRleSegments > 1)
2744 for(unsigned int k = 1; k <= nbRleSegments-1; k++)
2746 rleSegmentLength[k] = rleSegmentOffsetTable[k+1]
2747 - rleSegmentOffsetTable[k];
2748 ftellRes = ftell(Fp);
2749 SkipBytes(rleSegmentLength[k]);
2753 rleSegmentLength[nbRleSegments] = fragmentLength
2754 - rleSegmentOffsetTable[nbRleSegments];
2755 ftellRes = ftell(Fp);
2756 SkipBytes(rleSegmentLength[nbRleSegments]);
2759 // Make sure that at the end of the item we encounter a 'Sequence
2761 if ( ! ReadTag(0xfffe, 0xe0dd) )
2763 dbg.Verbose(0, "gdcmDocument::Parse7FE0: no sequence delimiter item");
2764 dbg.Verbose(0, " at end of RLE item sequence");
2772 * \brief Compares two documents, according to \ref gdcmDicomDir rules
2773 * \warning Does NOT work with ACR-NEMA files
2774 * \todo Find a trick to solve the pb (use RET fields ?)
2776 * @return true if 'smaller'
2778 bool gdcmDocument::operator<(gdcmDocument &document)
2781 std::string s1 = this->GetEntryByNumber(0x0010,0x0010);
2782 std::string s2 = document.GetEntryByNumber(0x0010,0x0010);
2794 s1 = this->GetEntryByNumber(0x0010,0x0020);
2795 s2 = document.GetEntryByNumber(0x0010,0x0020);
2806 // Study Instance UID
2807 s1 = this->GetEntryByNumber(0x0020,0x000d);
2808 s2 = document.GetEntryByNumber(0x0020,0x000d);
2819 // Serie Instance UID
2820 s1 = this->GetEntryByNumber(0x0020,0x000e);
2821 s2 = document.GetEntryByNumber(0x0020,0x000e);
2837 //-----------------------------------------------------------------------------