1 /*=========================================================================
4 Module: $RCSfile: gdcmDocument.cxx,v $
6 Date: $Date: 2004/07/19 05:59:12 $
7 Version: $Revision: 1.54 $
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 don'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 differente 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 )
329 //if( transfer.find( syntaxToCheck ) ) //should be faster
340 * \brief Determines if the Transfer Syntax of the present document
341 * corresponds to a Implicit Value Representation of
343 * \sa \ref gdcmDocument::IsGivenTransferSyntax.
344 * @return True when ImplicitVRLittleEndian found. False in all other cases.
346 bool gdcmDocument::IsImplicitVRLittleEndianTransferSyntax()
348 return IsGivenTransferSyntax(UI1_2_840_10008_1_2);
352 * \brief Determines if the Transfer Syntax was already encountered
353 * and if it corresponds to a ExplicitVRLittleEndian one.
354 * @return True when ExplicitVRLittleEndian found. False in all other cases.
356 bool gdcmDocument::IsExplicitVRLittleEndianTransferSyntax()
358 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1);
362 * \brief Determines if the Transfer Syntax was already encountered
363 * and if it corresponds to a DeflatedExplicitVRLittleEndian one.
364 * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases.
366 bool gdcmDocument::IsDeflatedExplicitVRLittleEndianTransferSyntax()
368 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1_99);
372 * \brief Determines if the Transfer Syntax was already encountered
373 * and if it corresponds to a Explicit VR Big Endian one.
374 * @return True when big endian found. False in all other cases.
376 bool gdcmDocument::IsExplicitVRBigEndianTransferSyntax()
378 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_2);
382 * \brief Determines if the Transfer Syntax was already encountered
383 * and if it corresponds to a JPEGBaseLineProcess1 one.
384 * @return True when JPEGBaseLineProcess1found. False in all other cases.
386 bool gdcmDocument::IsJPEGBaseLineProcess1TransferSyntax()
388 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_50);
392 * \brief Determines if the Transfer Syntax was already encountered
393 * and if it corresponds to a JPEGExtendedProcess2-4 one.
394 * @return True when JPEGExtendedProcess2-4 found. False in all other cases.
396 bool gdcmDocument::IsJPEGExtendedProcess2_4TransferSyntax()
398 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_51);
402 * \brief Determines if the Transfer Syntax was already encountered
403 * and if it corresponds to a JPEGExtendeProcess3-5 one.
404 * @return True when JPEGExtendedProcess3-5 found. False in all other cases.
406 bool gdcmDocument::IsJPEGExtendedProcess3_5TransferSyntax()
408 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_52);
412 * \brief Determines if the Transfer Syntax was already encountered
413 * and if it corresponds to a JPEGSpectralSelectionProcess6-8 one.
414 * @return True when JPEGSpectralSelectionProcess6-8 found. False in all
417 bool gdcmDocument::IsJPEGSpectralSelectionProcess6_8TransferSyntax()
419 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_53);
423 * \brief Determines if the Transfer Syntax was already encountered
424 * and if it corresponds to a RLE Lossless one.
425 * @return True when RLE Lossless found. False in all
428 bool gdcmDocument::IsRLELossLessTransferSyntax()
430 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_5);
434 * \brief Determines if Transfer Syntax was already encountered
435 * and if it corresponds to a JPEG Lossless one.
436 * @return True when RLE Lossless found. False in all
440 bool gdcmDocument::IsJPEGLossless()
442 return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_55)
443 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_57)
444 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_70) );
448 * \brief Determines if the Transfer Syntax was already encountered
449 * and if it corresponds to a JPEG2000 one
450 * @return True when JPEG2000 (Lossly or LossLess) found. False in all
453 bool gdcmDocument::IsJPEG2000()
455 return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_90)
456 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_91) );
460 * \brief Predicate for dicom version 3 file.
461 * @return True when the file is a dicom version 3.
463 bool gdcmDocument::IsDicomV3()
465 // Checking if Transfert Syntax exists is enough
466 // Anyway, it's to late check if the 'Preamble' was found ...
467 // And ... would it be a rich idea to check ?
468 // (some 'no Preamble' DICOM images exist !)
469 return GetDocEntryByNumber(0x0002, 0x0010) != NULL;
473 * \brief returns the File Type
474 * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
475 * @return the FileType code
477 FileType gdcmDocument::GetFileType()
483 * \brief opens the file
484 * @param exception_on_error
487 FILE *gdcmDocument::OpenFile(bool exception_on_error)
490 Fp = fopen(Filename.c_str(),"rb");
494 if(exception_on_error)
496 throw gdcmFileError("gdcmDocument::gdcmDocument(const char *, bool)");
500 dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file: ",
509 fread(&zero, (size_t)2, (size_t)1, Fp);
511 //ACR -- or DICOM with no Preamble --
512 if( zero == 0x0008 || zero == 0x0800 || zero == 0x0002 || zero == 0x0200 )
518 fseek(Fp, 126L, SEEK_CUR);
520 fread(dicm, (size_t)4, (size_t)1, Fp);
521 if( memcmp(dicm, "DICM", 4) == 0 )
527 dbg.Verbose(0, "gdcmDocument::OpenFile not DICOM/ACR", Filename.c_str());
531 dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file", Filename.c_str());
538 * \brief closes the file
539 * @return TRUE if the close was successfull
541 bool gdcmDocument::CloseFile()
543 int closed = fclose(Fp);
550 * \brief Writes in a file all the Header Entries (Dicom Elements)
551 * @param fp file pointer on an already open file
552 * @param filetype Type of the File to be written
553 * (ACR-NEMA, ExplicitVR, ImplicitVR)
554 * \return Always true.
556 void gdcmDocument::Write(FILE* fp,FileType filetype)
558 /// \todo move the following lines (and a lot of others, to be written)
559 /// to a future function CheckAndCorrectHeader
561 /// WARNING : Si on veut ecrire du DICOM V3 a partir d'un DcmHeader ACR-NEMA
562 /// no way (check : FileType est un champ de gdcmDocument ...)
563 /// a moins de se livrer a un tres complique ajout des champs manquants.
564 /// faire un CheckAndCorrectHeader (?)
566 if (filetype == gdcmImplicitVR)
568 std::string implicitVRTransfertSyntax = UI1_2_840_10008_1_2;
569 ReplaceOrCreateByNumber(implicitVRTransfertSyntax,0x0002, 0x0010);
571 /// \todo Refer to standards on page 21, chapter 6.2
572 /// "Value representation": values with a VR of UI shall be
573 /// padded with a single trailing null
574 /// in the following case we have to padd manually with a 0
576 SetEntryLengthByNumber(18, 0x0002, 0x0010);
579 if (filetype == gdcmExplicitVR)
581 std::string explicitVRTransfertSyntax = UI1_2_840_10008_1_2_1;
582 ReplaceOrCreateByNumber(explicitVRTransfertSyntax,0x0002, 0x0010);
584 /// \todo Refer to standards on page 21, chapter 6.2
585 /// "Value representation": values with a VR of UI shall be
586 /// padded with a single trailing null
587 /// Dans le cas suivant on doit pader manuellement avec un 0
589 SetEntryLengthByNumber(20, 0x0002, 0x0010);
593 * \todo rewrite later, if really usefull
594 * - 'Group Length' element is optional in DICOM
595 * - but un-updated odd groups lengthes can causes pb
598 * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) )
599 * UpdateGroupLength(false,filetype);
600 * if ( filetype == ACR)
601 * UpdateGroupLength(true,ACR);
604 gdcmElementSet::Write(fp, filetype); // This one is recursive
609 * \brief Modifies the value of a given Header Entry (Dicom Element)
610 * when it exists. Create it with the given value when unexistant.
611 * @param Value (string) Value to be set
612 * @param Group Group number of the Entry
613 * @param Elem Element number of the Entry
614 * \return pointer to the modified/created Header Entry (NULL when creation
618 gdcmValEntry * gdcmDocument::ReplaceOrCreateByNumber(
623 gdcmValEntry* valEntry = 0;
625 gdcmDocEntry* currentEntry = GetDocEntryByNumber( group, elem);
628 // The entry wasn't present and we simply create the required ValEntry:
629 currentEntry = NewDocEntryByNumber(group, elem);
632 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: call to"
633 " NewDocEntryByNumber failed.");
636 valEntry = new gdcmValEntry(currentEntry);
637 if ( !AddEntry(valEntry))
639 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: AddEntry"
640 " failed allthough this is a creation.");
645 valEntry = dynamic_cast< gdcmValEntry* >(currentEntry);
648 // We need to promote the gdcmDocEntry to a gdcmValEntry:
649 valEntry = new gdcmValEntry(currentEntry);
650 if (!RemoveEntry(currentEntry))
652 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: removal"
653 " of previous DocEntry failed.");
656 if ( !AddEntry(valEntry))
658 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: adding"
659 " promoted ValEntry failed.");
665 SetEntryByNumber(value, group, elem);
671 * \brief Modifies the value of a given Header Entry (Dicom Element)
672 * when it exists. Create it with the given value when unexistant.
673 * @param voidArea (binary) value to be set
674 * @param Group Group number of the Entry
675 * @param Elem Element number of the Entry
676 * \return pointer to the modified/created Header Entry (NULL when creation
679 gdcmBinEntry * gdcmDocument::ReplaceOrCreateByNumber(
686 gdcmDocEntry* a = GetDocEntryByNumber( group, elem);
689 a = NewBinEntryByNumber(group, elem);
695 b = new gdcmBinEntry(a);
697 b->SetVoidArea(voidArea);
699 SetEntryByNumber(voidArea, lgth, group, elem);
700 //b->SetVoidArea(voidArea); //what if b == 0 !!
706 * \brief Set a new value if the invoked element exists
707 * Seems to be useless !!!
708 * @param Value new element value
709 * @param Group group number of the Entry
710 * @param Elem element number of the Entry
713 bool gdcmDocument::ReplaceIfExistByNumber(const char* value, uint16_t group,
716 const std::string v = value;
717 SetEntryByNumber(v, group, elem);
722 //-----------------------------------------------------------------------------
726 * \brief Checks if a given Dicom Element exists within the H table
727 * @param group Group number of the searched Dicom Element
728 * @param element Element number of the searched Dicom Element
729 * @return number of occurences
731 int gdcmDocument::CheckIfEntryExistByNumber(uint16_t group, uint16_t element )
733 std::string key = gdcmDictEntry::TranslateToKey(group, element );
734 return TagHT.count(key);
738 * \brief Searches within Header Entries (Dicom Elements) parsed with
739 * the public and private dictionaries
740 * for the element value of a given tag.
741 * \warning Don't use any longer : use GetPubEntryByName
742 * @param tagName name of the searched element.
743 * @return Corresponding element value when it exists,
744 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
746 std::string gdcmDocument::GetEntryByName(TagName tagName)
748 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
754 return GetEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement());
758 * \brief Searches within Header Entries (Dicom Elements) parsed with
759 * the public and private dictionaries
760 * for the element value representation of a given tag.
762 * Obtaining the VR (Value Representation) might be needed by caller
763 * to convert the string typed content to caller's native type
764 * (think of C++ vs Python). The VR is actually of a higher level
765 * of semantics than just the native C++ type.
766 * @param tagName name of the searched element.
767 * @return Corresponding element value representation when it exists,
768 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
770 std::string gdcmDocument::GetEntryVRByName(TagName tagName)
772 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
773 if( dictEntry == NULL)
778 gdcmDocEntry* elem = GetDocEntryByNumber(dictEntry->GetGroup(),
779 dictEntry->GetElement());
780 return elem->GetVR();
785 * \brief Searches within Header Entries (Dicom Elements) parsed with
786 * the public and private dictionaries
787 * for the element value representation of a given tag.
788 * @param group Group number of the searched tag.
789 * @param element Element number of the searched tag.
790 * @return Corresponding element value representation when it exists,
791 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
793 std::string gdcmDocument::GetEntryByNumber(uint16_t group, uint16_t element)
795 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
796 /// \todo use map methods, instead of multimap JPR
797 if ( !TagHT.count(key))
802 return ((gdcmValEntry *)TagHT.find(key)->second)->GetValue();
806 * \brief Searches within Header Entries (Dicom Elements) parsed with
807 * the public and private dictionaries
808 * for the element value representation of a given tag..
810 * Obtaining the VR (Value Representation) might be needed by caller
811 * to convert the string typed content to caller's native type
812 * (think of C++ vs Python). The VR is actually of a higher level
813 * of semantics than just the native C++ type.
814 * @param group Group number of the searched tag.
815 * @param element Element number of the searched tag.
816 * @return Corresponding element value representation when it exists,
817 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
819 std::string gdcmDocument::GetEntryVRByNumber(uint16_t group, uint16_t element)
821 gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
826 return elem->GetVR();
830 * \brief Searches within Header Entries (Dicom Elements) parsed with
831 * the public and private dictionaries
832 * for the value length of a given tag..
833 * @param group Group number of the searched tag.
834 * @param element Element number of the searched tag.
835 * @return Corresponding element length; -2 if not found
837 int gdcmDocument::GetEntryLengthByNumber(uint16_t group, uint16_t element)
839 gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
842 return -2; //magic number
844 return elem->GetLength();
847 * \brief Sets the value (string) of the Header Entry (Dicom Element)
848 * @param content string value of the Dicom Element
849 * @param tagName name of the searched Dicom Element.
850 * @return true when found
852 bool gdcmDocument::SetEntryByName(std::string content,std::string tagName)
854 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
860 return SetEntryByNumber(content,dictEntry->GetGroup(),
861 dictEntry->GetElement());
865 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
866 * through it's (group, element) and modifies it's content with
868 * @param content new value (string) to substitute with
869 * @param group group number of the Dicom Element to modify
870 * @param element element number of the Dicom Element to modify
872 bool gdcmDocument::SetEntryByNumber(std::string content,
876 gdcmValEntry* valEntry = GetValEntryByNumber(group, element);
879 dbg.Verbose(0, "gdcmDocument::SetEntryByNumber: no corresponding",
880 " ValEntry (try promotion first).");
883 // Non even content must be padded with a space (020H).
884 if( content.length() % 2 )
888 valEntry->SetValue(content);
890 // Integers have a special treatement for their length:
891 gdcmVRKey vr = valEntry->GetVR();
892 if( vr == "US" || vr == "SS" )
894 valEntry->SetLength(2);
896 else if( vr == "UL" || vr == "SL" )
898 valEntry->SetLength(4);
902 valEntry->SetLength(content.length());
909 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
910 * through it's (group, element) and modifies it's content with
912 * @param content new value (void *) to substitute with
913 * @param lgth new value length
914 * @param group group number of the Dicom Element to modify
915 * @param element element number of the Dicom Element to modify
917 bool gdcmDocument::SetEntryByNumber(void *content,
922 (void)lgth; //not used
923 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
924 if ( !TagHT.count(key))
929 /* Hope Binaray field length is *never* wrong
930 if(lgth%2) // Non even length are padded with a space (020H).
933 //content = content + '\0'; // fing a trick to enlarge a binary field?
937 a = (gdcmBinEntry *)TagHT[key];
938 a->SetVoidArea(content);
939 //a->SetLength(lgth); // ???
944 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
945 * in the PubDocEntrySet of this instance
946 * through it's (group, element) and modifies it's length with
948 * \warning Use with extreme caution.
949 * @param l new length to substitute with
950 * @param group group number of the Entry to modify
951 * @param element element number of the Entry to modify
952 * @return true on success, false otherwise.
954 bool gdcmDocument::SetEntryLengthByNumber(uint32_t l,
958 /// \todo use map methods, instead of multimap JPR
959 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
960 if ( !TagHT.count(key) )
966 l++; // length must be even
968 ( ((TagHT.equal_range(key)).first)->second )->SetLength(l);
974 * \brief Gets (from Header) the offset of a 'non string' element value
975 * (LoadElementValues has already be executed)
976 * @param Group group number of the Entry
977 * @param Elem element number of the Entry
978 * @return File Offset of the Element Value
980 size_t gdcmDocument::GetEntryOffsetByNumber(uint16_t group, uint16_t elem)
982 gdcmDocEntry* entry = GetDocEntryByNumber(group, elem);
985 dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry present.");
988 return entry->GetOffset();
992 * \brief Gets (from Header) a 'non string' element value
993 * (LoadElementValues has already be executed)
994 * @param Group group number of the Entry
995 * @param Elem element number of the Entry
996 * @return Pointer to the 'non string' area
998 void * gdcmDocument::GetEntryVoidAreaByNumber(uint16_t group, uint16_t elem)
1000 gdcmDocEntry* entry = GetDocEntryByNumber(group, elem);
1003 dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry");
1006 return ((gdcmBinEntry *)entry)->GetVoidArea();
1010 * \brief Loads (from disk) the element content
1011 * when a string is not suitable
1012 * @param Group group number of the Entry
1013 * @param Elem element number of the Entry
1015 void *gdcmDocument::LoadEntryVoidArea(uint16_t group, uint16_t elem)
1017 gdcmDocEntry *docElement = GetDocEntryByNumber(group, elem);
1022 size_t o =(size_t)docElement->GetOffset();
1023 fseek(Fp, o, SEEK_SET);
1024 size_t l = docElement->GetLength();
1025 char* a = new char[l];
1028 dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a");
1031 size_t l2 = fread(a, 1, l , Fp);
1037 /// \todo Drop any already existing void area! JPR
1038 SetEntryVoidAreaByNumber(a, group, elem);
1043 * \brief Loads (from disk) the element content
1044 * when a string is not suitable
1045 * @param Element Entry whose voidArea is going to be loaded
1047 void *gdcmDocument::LoadEntryVoidArea(gdcmBinEntry *element)
1049 size_t o =(size_t)element->GetOffset();
1050 fseek(Fp, o, SEEK_SET);
1051 size_t l = element->GetLength();
1052 char* a = new char[l];
1055 dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a");
1058 element->SetVoidArea((void *)a);
1059 /// \todo check the result
1060 size_t l2 = fread(a, 1, l , Fp);
1071 * \brief Sets a 'non string' value to a given Dicom Element
1072 * @param area area containing the 'non string' value
1073 * @param group Group number of the searched Dicom Element
1074 * @param element Element number of the searched Dicom Element
1077 bool gdcmDocument::SetEntryVoidAreaByNumber(void * area,
1081 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
1082 if ( !TagHT.count(key))
1086 // This was for multimap ?
1087 (( gdcmBinEntry *)( ((TagHT.equal_range(key)).first)->second ))->SetVoidArea(area);
1093 * \brief Update the entries with the shadow dictionary.
1094 * Only non even entries are analyzed
1096 void gdcmDocument::UpdateShaEntries()
1098 //gdcmDictEntry *entry;
1101 /// \todo TODO : still any use to explore recursively the whole structure?
1103 for(ListTag::iterator it=listEntries.begin();
1104 it!=listEntries.end();
1107 // Odd group => from public dictionary
1108 if((*it)->GetGroup()%2==0)
1111 // Peer group => search the corresponding dict entry
1113 entry=RefShaDict->GetDictEntryByNumber((*it)->GetGroup(),(*it)->GetElement());
1117 if((*it)->IsImplicitVR())
1122 (*it)->SetValue(GetDocEntryUnvalue(*it)); // to go on compiling
1124 // Set the new entry and the new value
1125 (*it)->SetDictEntry(entry);
1126 CheckDocEntryVR(*it,vr);
1128 (*it)->SetValue(GetDocEntryValue(*it)); // to go on compiling
1133 // Remove precedent value transformation
1134 (*it)->SetDictEntry(NewVirtualDictEntry((*it)->GetGroup(),(*it)->GetElement(),vr));
1141 * \brief Searches within the Header Entries for a Dicom Element of
1143 * @param tagName name of the searched Dicom Element.
1144 * @return Corresponding Dicom Element when it exists, and NULL
1147 gdcmDocEntry* gdcmDocument::GetDocEntryByName(std::string const & tagName)
1149 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
1155 return GetDocEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement());
1159 * \brief retrieves a Dicom Element (the first one) using (group, element)
1160 * \warning (group, element) IS NOT an identifier inside the Dicom Header
1161 * if you think it's NOT UNIQUE, check the count number
1162 * and use iterators to retrieve ALL the Dicoms Elements within
1163 * a given couple (group, element)
1164 * @param group Group number of the searched Dicom Element
1165 * @param element Element number of the searched Dicom Element
1168 gdcmDocEntry* gdcmDocument::GetDocEntryByNumber(uint16_t group,
1171 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
1172 if ( !TagHT.count(key))
1176 return TagHT.find(key)->second;
1180 * \brief Same as \ref gdcmDocument::GetDocEntryByNumber except it only
1181 * returns a result when the corresponding entry is of type
1183 * @return When present, the corresponding ValEntry.
1185 gdcmValEntry* gdcmDocument::GetValEntryByNumber(uint16_t group,
1188 gdcmDocEntry* currentEntry = GetDocEntryByNumber(group, element);
1189 if ( !currentEntry )
1193 if ( gdcmValEntry* valEntry = dynamic_cast<gdcmValEntry*>(currentEntry) )
1197 dbg.Verbose(0, "gdcmDocument::GetValEntryByNumber: unfound ValEntry.");
1203 * \brief Loads the element while preserving the current
1204 * underlying file position indicator as opposed to
1205 * to LoadDocEntry that modifies it.
1206 * @param entry Header Entry whose value shall be loaded.
1209 void gdcmDocument::LoadDocEntrySafe(gdcmDocEntry * entry)
1211 long PositionOnEntry = ftell(Fp);
1212 LoadDocEntry(entry);
1213 fseek(Fp, PositionOnEntry, SEEK_SET);
1217 * \brief Swaps back the bytes of 4-byte long integer accordingly to
1219 * @return The properly swaped 32 bits integer.
1221 uint32_t gdcmDocument::SwapLong(uint32_t a)
1228 a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) |
1229 ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) );
1233 a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
1237 a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
1240 std::cout << "swapCode= " << SwapCode << std::endl;
1241 dbg.Error(" gdcmDocument::SwapLong : unset swap code");
1248 * \brief Unswaps back the bytes of 4-byte long integer accordingly to
1250 * @return The properly unswaped 32 bits integer.
1252 uint32_t gdcmDocument::UnswapLong(uint32_t a)
1258 * \brief Swaps the bytes so they agree with the processor order
1259 * @return The properly swaped 16 bits integer.
1261 uint16_t gdcmDocument::SwapShort(uint16_t a)
1263 if ( SwapCode == 4321 || SwapCode == 2143 )
1265 a =((( a << 8 ) & 0x0ff00 ) | (( a >> 8 ) & 0x00ff ) );
1271 * \brief Unswaps the bytes so they agree with the processor order
1272 * @return The properly unswaped 16 bits integer.
1274 uint16_t gdcmDocument::UnswapShort(uint16_t a)
1276 return SwapShort(a);
1279 //-----------------------------------------------------------------------------
1283 * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries)
1284 * @return length of the parsed set.
1287 long gdcmDocument::ParseDES(gdcmDocEntrySet *set,
1292 gdcmDocEntry *newDocEntry = 0;
1293 gdcmValEntry *newValEntry = 0;
1294 unsigned long l = 0;
1296 int depth = set->GetDepthLevel();
1299 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1303 newDocEntry = ReadNextDocEntry( );
1309 gdcmVRKey vr = newDocEntry->GetVR();
1313 if ( gdcmGlobal::GetVR()->IsVROfGdcmStringRepresentable(vr) )
1316 newValEntry = new gdcmValEntry(newDocEntry->GetDictEntry());
1317 newValEntry->Copy(newDocEntry);
1318 newValEntry->SetDepthLevel(depth);
1319 set->AddEntry(newValEntry);
1320 LoadDocEntry(newValEntry);
1321 if (newValEntry->isItemDelimitor())
1325 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1332 if ( ! gdcmGlobal::GetVR()->IsVROfGdcmBinaryRepresentable(vr) )
1334 ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR
1335 dbg.Verbose(0, "gdcmDocument::ParseDES: neither Valentry, "
1336 "nor BinEntry. Probably unknown VR.");
1339 ////// BinEntry or UNKOWN VR:
1340 gdcmBinEntry *bn = new gdcmBinEntry(newDocEntry->GetDictEntry());
1341 bn->Copy(newDocEntry);
1346 if (newDocEntry->GetGroup() == 0x7fe0 &&
1347 newDocEntry->GetElement() == 0x0010 )
1349 if (newDocEntry->GetReadLength()==0xffffffff)
1351 // Broken US.3405.1.dcm
1352 Parse7FE0(); // to skip the pixels
1353 // (multipart JPEG/RLE are trouble makers)
1357 SkipToNextDocEntry(newDocEntry);
1358 l = newDocEntry->GetFullLength();
1363 // to be sure we are at the beginning
1364 SkipToNextDocEntry(newDocEntry);
1365 l = newDocEntry->GetFullLength();
1371 l = newDocEntry->GetReadLength();
1372 if ( l != 0 ) // don't mess the delim_mode for zero-length sequence
1374 if ( l == 0xffffffff )
1383 // no other way to create it ...
1384 gdcmSeqEntry *sq = new gdcmSeqEntry(newDocEntry->GetDictEntry(),
1385 set->GetDepthLevel());
1386 sq->Copy(newDocEntry);
1387 sq->SetDelimitorMode(delim_mode);
1388 sq->SetDepthLevel(depth);
1391 { // Don't try to parse zero-length sequences
1392 long lgt = ParseSQ( sq,
1393 newDocEntry->GetOffset(),
1395 (void)lgt; //not used...
1398 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1406 return l; // Probably useless
1410 * \brief Parses a Sequence ( SeqEntry after SeqEntry)
1411 * @return parsed length for this level
1413 long gdcmDocument::ParseSQ(gdcmSeqEntry *set,
1414 long offset, long l_max, bool delim_mode)
1416 int SQItemNumber = 0;
1418 //int depth = set->GetDepthLevel();
1419 //(void)depth; //not used
1423 gdcmDocEntry *newDocEntry = ReadNextDocEntry();
1426 // FIXME Should warn user
1431 if ( newDocEntry->isSequenceDelimitor() )
1433 set->SetSequenceDelimitationItem( newDocEntry );
1437 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1442 gdcmSQItem *itemSQ = new gdcmSQItem(set->GetDepthLevel());
1443 itemSQ->AddEntry(newDocEntry);
1444 unsigned int l = newDocEntry->GetReadLength();
1446 if ( l == 0xffffffff )
1455 int lgr = ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod);
1456 (void)lgr; //FIXME not used
1458 set->AddEntry(itemSQ, SQItemNumber);
1460 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1466 int lgth = ftell(Fp) - offset;
1471 * \brief Loads the element content if its length doesn't exceed
1472 * the value specified with gdcmDocument::SetMaxSizeLoadEntry()
1473 * @param Entry Header Entry (Dicom Element) to be dealt with
1475 void gdcmDocument::LoadDocEntry(gdcmDocEntry* entry)
1478 uint16_t group = entry->GetGroup();
1479 std::string vr = entry->GetVR();
1480 uint32_t length = entry->GetLength();
1482 fseek(Fp, (long)entry->GetOffset(), SEEK_SET);
1484 // A SeQuence "contains" a set of Elements.
1485 // (fffe e000) tells us an Element is beginning
1486 // (fffe e00d) tells us an Element just ended
1487 // (fffe e0dd) tells us the current SeQuence just ended
1488 if( group == 0xfffe )
1490 // NO more value field for SQ !
1494 // When the length is zero things are easy:
1497 ((gdcmValEntry *)entry)->SetValue("");
1501 // The elements whose length is bigger than the specified upper bound
1502 // are not loaded. Instead we leave a short notice of the offset of
1503 // the element content and it's length.
1505 std::ostringstream s;
1506 if (length > MaxSizeLoadEntry)
1508 if (gdcmBinEntry* binEntryPtr = dynamic_cast< gdcmBinEntry* >(entry) )
1510 s << "gdcm::NotLoaded (BinEntry)";
1511 s << " Address:" << (long)entry->GetOffset();
1512 s << " Length:" << entry->GetLength();
1513 s << " x(" << std::hex << entry->GetLength() << ")";
1514 binEntryPtr->SetValue(s.str());
1516 // to be sure we are at the end of the value ...
1517 fseek(Fp, (long)entry->GetOffset()+(long)entry->GetLength(), SEEK_SET);
1519 return; //FIXME FIXME FIXME FIXME ????
1521 // Be carefull : a BinEntry IS_A ValEntry ...
1522 if (gdcmValEntry* valEntryPtr = dynamic_cast< gdcmValEntry* >(entry) )
1524 s << "gdcm::NotLoaded. (ValEntry)";
1525 s << " Address:" << (long)entry->GetOffset();
1526 s << " Length:" << entry->GetLength();
1527 s << " x(" << std::hex << entry->GetLength() << ")";
1528 valEntryPtr->SetValue(s.str());
1530 // to be sure we are at the end of the value ...
1531 fseek(Fp,(long)entry->GetOffset()+(long)entry->GetLength(),SEEK_SET);
1536 // When we find a BinEntry not very much can be done :
1537 if (gdcmBinEntry* binEntryPtr = dynamic_cast< gdcmBinEntry* >(entry) )
1540 LoadEntryVoidArea(binEntryPtr);
1541 s << "gdcm::Loaded (BinEntry)";
1542 binEntryPtr->SetValue(s.str());
1546 /// \todo Any compacter code suggested (?)
1547 if ( IsDocEntryAnInteger(entry) )
1550 std::ostringstream s;
1552 // When short integer(s) are expected, read and convert the following
1553 // n *two characters properly i.e. consider them as short integers as
1554 // opposed to strings.
1555 // Elements with Value Multiplicity > 1
1556 // contain a set of integers (not a single one)
1557 if (vr == "US" || vr == "SS")
1560 NewInt = ReadInt16();
1564 for (int i=1; i < nbInt; i++)
1567 NewInt = ReadInt16();
1572 // See above comment on multiple integers (mutatis mutandis).
1573 else if (vr == "UL" || vr == "SL")
1576 NewInt = ReadInt32();
1580 for (int i=1; i < nbInt; i++)
1583 NewInt = ReadInt32();
1588 #ifdef GDCM_NO_ANSI_STRING_STREAM
1589 s << std::ends; // to avoid oddities on Solaris
1590 #endif //GDCM_NO_ANSI_STRING_STREAM
1592 ((gdcmValEntry *)entry)->SetValue(s.str());
1596 // We need an additional byte for storing \0 that is not on disk
1597 //std::string newValue(length,0);
1598 //item_read = fread(&(newValue[0]), (size_t)length, (size_t)1, Fp);
1599 //rah !! I can't believe it could work, normally this is a const char* !!!
1600 char *str = new char[length+1];
1601 item_read = fread(str, (size_t)length, (size_t)1, Fp);
1603 std::string newValue = str;
1605 if ( gdcmValEntry* valEntry = dynamic_cast<gdcmValEntry* >(entry) )
1607 if ( item_read != 1 )
1609 dbg.Verbose(1, "gdcmDocument::LoadDocEntry",
1610 "unread element value");
1611 valEntry->SetValue("gdcm::UnRead");
1617 // Because of correspondance with the VR dic
1618 valEntry->SetValue(newValue);
1622 valEntry->SetValue(newValue);
1627 dbg.Error(true, "gdcmDocument::LoadDocEntry"
1628 "Should have a ValEntry, here !");
1634 * \brief Find the value Length of the passed Header Entry
1635 * @param Entry Header Entry whose length of the value shall be loaded.
1637 void gdcmDocument::FindDocEntryLength (gdcmDocEntry *entry)
1639 uint16_t element = entry->GetElement();
1640 std::string vr = entry->GetVR();
1644 if ( Filetype == gdcmExplicitVR && !entry->IsImplicitVR() )
1646 if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UN" )
1648 // The following reserved two bytes (see PS 3.5-2001, section
1649 // 7.1.2 Data element structure with explicit vr p27) must be
1650 // skipped before proceeding on reading the length on 4 bytes.
1651 fseek(Fp, 2L, SEEK_CUR);
1652 uint32_t length32 = ReadInt32();
1654 if ( vr == "OB" && length32 == 0xffffffff )
1656 uint32_t lengthOB = FindDocEntryLengthOB();
1659 // Computing the length failed (this happens with broken
1660 // files like gdcm-JPEG-LossLess3a.dcm). We still have a
1661 // chance to get the pixels by deciding the element goes
1662 // until the end of the file. Hence we artificially fix the
1663 // the length and proceed.
1664 long currentPosition = ftell(Fp);
1665 fseek(Fp,0L,SEEK_END);
1666 long lengthUntilEOF = ftell(Fp) - currentPosition;
1667 fseek(Fp, currentPosition, SEEK_SET);
1668 entry->SetLength(lengthUntilEOF);
1672 entry->SetLength(lengthOB);
1675 FixDocEntryFoundLength(entry, length32);
1679 // Length is encoded on 2 bytes.
1680 length16 = ReadInt16();
1682 // We can tell the current file is encoded in big endian (like
1683 // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag
1684 // and it's value is the one of the encoding of a big endian file.
1685 // In order to deal with such big endian encoded files, we have
1686 // (at least) two strategies:
1687 // * when we load the "Transfer Syntax" tag with value of big endian
1688 // encoding, we raise the proper flags. Then we wait for the end
1689 // of the META group (0x0002) among which is "Transfer Syntax",
1690 // before switching the swap code to big endian. We have to postpone
1691 // the switching of the swap code since the META group is fully encoded
1692 // in little endian, and big endian coding only starts at the next
1693 // group. The corresponding code can be hard to analyse and adds
1694 // many additional unnecessary tests for regular tags.
1695 // * the second strategy consists in waiting for trouble, that shall
1696 // appear when we find the first group with big endian encoding. This
1697 // is easy to detect since the length of a "Group Length" tag (the
1698 // ones with zero as element number) has to be of 4 (0x0004). When we
1699 // encounter 1024 (0x0400) chances are the encoding changed and we
1700 // found a group with big endian encoding.
1701 // We shall use this second strategy. In order to make sure that we
1702 // can interpret the presence of an apparently big endian encoded
1703 // length of a "Group Length" without committing a big mistake, we
1704 // add an additional check: we look in the already parsed elements
1705 // for the presence of a "Transfer Syntax" whose value has to be "big
1706 // endian encoding". When this is the case, chances are we have got our
1707 // hands on a big endian encoded file: we switch the swap code to
1708 // big endian and proceed...
1709 if ( (element == 0x0000) && (length16 == 0x0400) )
1711 if ( ! IsExplicitVRBigEndianTransferSyntax() )
1713 dbg.Verbose(0, "gdcmDocument::FindLength", "not explicit VR");
1718 SwitchSwapToBigEndian();
1719 // Restore the unproperly loaded values i.e. the group, the element
1720 // and the dictionary entry depending on them.
1721 uint16_t correctGroup = SwapShort( entry->GetGroup() );
1722 uint16_t correctElem = SwapShort( entry->GetElement() );
1723 gdcmDictEntry* newTag = GetDictEntryByNumber( correctGroup,
1727 // This correct tag is not in the dictionary. Create a new one.
1728 newTag = NewVirtualDictEntry(correctGroup, correctElem);
1730 // FIXME this can create a memory leaks on the old entry that be
1731 // left unreferenced.
1732 entry->SetDictEntry( newTag );
1735 // Heuristic: well, some files are really ill-formed.
1736 if ( length16 == 0xffff)
1739 // Length16= 0xffff means that we deal with
1740 // 'Unknown Length' Sequence
1742 FixDocEntryFoundLength(entry, (uint32_t)length16);
1747 // Either implicit VR or a non DICOM conformal (see note below) explicit
1748 // VR that ommited the VR of (at least) this element. Farts happen.
1749 // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25
1750 // on Data elements "Implicit and Explicit VR Data Elements shall
1751 // not coexist in a Data Set and Data Sets nested within it".]
1752 // Length is on 4 bytes.
1754 FixDocEntryFoundLength(entry, ReadInt32());
1760 * \brief Find the Value Representation of the current Dicom Element.
1763 void gdcmDocument::FindDocEntryVR( gdcmDocEntry *entry )
1765 if ( Filetype != gdcmExplicitVR )
1772 long positionOnEntry = ftell(Fp);
1773 // Warning: we believe this is explicit VR (Value Representation) because
1774 // we used a heuristic that found "UL" in the first tag. Alas this
1775 // doesn't guarantee that all the tags will be in explicit VR. In some
1776 // cases (see e-film filtered files) one finds implicit VR tags mixed
1777 // within an explicit VR file. Hence we make sure the present tag
1778 // is in explicit VR and try to fix things if it happens not to be
1781 fread (&vr, (size_t)2,(size_t)1, Fp);
1784 if( !CheckDocEntryVR(entry, vr) )
1786 fseek(Fp, positionOnEntry, SEEK_SET);
1787 // When this element is known in the dictionary we shall use, e.g. for
1788 // the semantics (see the usage of IsAnInteger), the VR proposed by the
1789 // dictionary entry. Still we have to flag the element as implicit since
1790 // we know now our assumption on expliciteness is not furfilled.
1792 if ( entry->IsVRUnknown() )
1794 entry->SetVR("Implicit");
1796 entry->SetImplicitVR();
1801 * \brief Check the correspondance between the VR of the header entry
1802 * and the taken VR. If they are different, the header entry is
1803 * updated with the new VR.
1804 * @param Entry Header Entry to check
1805 * @param vr Dicom Value Representation
1806 * @return false if the VR is incorrect of if the VR isn't referenced
1807 * otherwise, it returns true
1809 bool gdcmDocument::CheckDocEntryVR(gdcmDocEntry *entry, gdcmVRKey vr)
1811 char msg[100]; // for sprintf
1812 bool realExplicit = true;
1814 // Assume we are reading a falsely explicit VR file i.e. we reached
1815 // a tag where we expect reading a VR but are in fact we read the
1816 // first to bytes of the length. Then we will interogate (through find)
1817 // the dicom_vr dictionary with oddities like "\004\0" which crashes
1818 // both GCC and VC++ implementations of the STL map. Hence when the
1819 // expected VR read happens to be non-ascii characters we consider
1820 // we hit falsely explicit VR tag.
1822 if ( (!isalpha(vr[0])) && (!isalpha(vr[1])) )
1824 realExplicit = false;
1827 // CLEANME searching the dicom_vr at each occurence is expensive.
1828 // PostPone this test in an optional integrity check at the end
1829 // of parsing or only in debug mode.
1830 if ( realExplicit && !gdcmGlobal::GetVR()->Count(vr) )
1832 realExplicit = false;
1835 if ( !realExplicit )
1837 // We thought this was explicit VR, but we end up with an
1838 // implicit VR tag. Let's backtrack.
1839 sprintf(msg,"Falsely explicit vr file (%04x,%04x)\n",
1840 entry->GetGroup(), entry->GetElement());
1841 dbg.Verbose(1, "gdcmDocument::FindVR: ",msg);
1843 if( entry->GetGroup() % 2 && entry->GetElement() == 0x0000)
1845 // Group length is UL !
1846 gdcmDictEntry* newEntry = NewVirtualDictEntry(
1847 entry->GetGroup(), entry->GetElement(),
1848 "UL", "FIXME", "Group Length");
1849 entry->SetDictEntry( newEntry );
1854 if ( entry->IsVRUnknown() )
1856 // When not a dictionary entry, we can safely overwrite the VR.
1857 if( entry->GetElement() == 0x0000 )
1859 // Group length is UL !
1867 else if ( entry->GetVR() != vr )
1869 // The VR present in the file and the dictionary disagree. We assume
1870 // the file writer knew best and use the VR of the file. Since it would
1871 // be unwise to overwrite the VR of a dictionary (since it would
1872 // compromise it's next user), we need to clone the actual DictEntry
1873 // and change the VR for the read one.
1874 gdcmDictEntry* newEntry = NewVirtualDictEntry(
1875 entry->GetGroup(), entry->GetElement(),
1876 vr, "FIXME", entry->GetName());
1877 entry->SetDictEntry(newEntry);
1884 * \brief Get the transformed value of the header entry. The VR value
1885 * is used to define the transformation to operate on the value
1886 * \warning NOT end user intended method !
1888 * @return Transformed entry value
1890 std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *entry)
1892 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1894 std::string val = ((gdcmValEntry *)entry)->GetValue();
1895 std::string vr = entry->GetVR();
1896 uint32_t length = entry->GetLength();
1897 std::ostringstream s;
1900 // When short integer(s) are expected, read and convert the following
1901 // n * 2 bytes properly i.e. as a multivaluated strings
1902 // (each single value is separated fromthe next one by '\'
1903 // as usual for standard multivaluated filels
1904 // Elements with Value Multiplicity > 1
1905 // contain a set of short integers (not a single one)
1907 if( vr == "US" || vr == "SS" )
1912 for (int i=0; i < nbInt; i++)
1918 newInt16 = ( val[2*i+0] & 0xFF ) + ( ( val[2*i+1] & 0xFF ) << 8);
1919 newInt16 = SwapShort( newInt16 );
1924 // When integer(s) are expected, read and convert the following
1925 // n * 4 bytes properly i.e. as a multivaluated strings
1926 // (each single value is separated fromthe next one by '\'
1927 // as usual for standard multivaluated filels
1928 // Elements with Value Multiplicity > 1
1929 // contain a set of integers (not a single one)
1930 else if( vr == "UL" || vr == "SL" )
1935 for (int i=0; i < nbInt; i++)
1941 newInt32 = ( val[4*i+0] & 0xFF )
1942 + (( val[4*i+1] & 0xFF ) << 8 )
1943 + (( val[4*i+2] & 0xFF ) << 16 )
1944 + (( val[4*i+3] & 0xFF ) << 24 );
1945 newInt32 = SwapLong( newInt32 );
1949 #ifdef GDCM_NO_ANSI_STRING_STREAM
1950 s << std::ends; // to avoid oddities on Solaris
1951 #endif //GDCM_NO_ANSI_STRING_STREAM
1955 return ((gdcmValEntry *)entry)->GetValue();
1959 * \brief Get the reverse transformed value of the header entry. The VR
1960 * value is used to define the reverse transformation to operate on
1962 * \warning NOT end user intended method !
1964 * @return Reverse transformed entry value
1966 std::string gdcmDocument::GetDocEntryUnvalue(gdcmDocEntry* entry)
1968 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1970 std::string vr = entry->GetVR();
1971 std::vector<std::string> tokens;
1972 std::ostringstream s;
1974 if ( vr == "US" || vr == "SS" )
1978 tokens.erase( tokens.begin(), tokens.end()); // clean any previous value
1979 Tokenize (((gdcmValEntry *)entry)->GetValue(), tokens, "\\");
1980 for (unsigned int i=0; i<tokens.size(); i++)
1982 newInt16 = atoi(tokens[i].c_str());
1983 s << ( newInt16 & 0xFF )
1984 << (( newInt16 >> 8 ) & 0xFF );
1988 if ( vr == "UL" || vr == "SL")
1992 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1993 Tokenize (((gdcmValEntry *)entry)->GetValue(), tokens, "\\");
1994 for (unsigned int i=0; i<tokens.size();i++)
1996 newInt32 = atoi(tokens[i].c_str());
1997 s << (char)( newInt32 & 0xFF )
1998 << (char)(( newInt32 >> 8 ) & 0xFF )
1999 << (char)(( newInt32 >> 16 ) & 0xFF )
2000 << (char)(( newInt32 >> 24 ) & 0xFF );
2005 #ifdef GDCM_NO_ANSI_STRING_STREAM
2006 s << std::ends; // to avoid oddities on Solaris
2007 #endif //GDCM_NO_ANSI_STRING_STREAM
2011 return ((gdcmValEntry *)entry)->GetValue();
2015 * \brief Skip a given Header Entry
2016 * \warning NOT end user intended method !
2019 void gdcmDocument::SkipDocEntry(gdcmDocEntry *entry)
2021 SkipBytes(entry->GetLength());
2025 * \brief Skips to the begining of the next Header Entry
2026 * \warning NOT end user intended method !
2029 void gdcmDocument::SkipToNextDocEntry(gdcmDocEntry *entry)
2031 fseek(Fp, (long)(entry->GetOffset()), SEEK_SET);
2032 fseek(Fp, (long)(entry->GetReadLength()), SEEK_CUR);
2036 * \brief When the length of an element value is obviously wrong (because
2037 * the parser went Jabberwocky) one can hope improving things by
2038 * applying some heuristics.
2040 void gdcmDocument::FixDocEntryFoundLength(gdcmDocEntry *entry,
2041 uint32_t foundLength)
2043 entry->SetReadLength( foundLength ); // will be updated only if a bug is found
2044 if ( foundLength == 0xffffffff)
2049 uint16_t gr = entry->GetGroup();
2050 uint16_t el = entry->GetElement();
2052 if ( foundLength % 2)
2054 std::ostringstream s;
2055 s << "Warning : Tag with uneven length "
2057 << " in x(" << std::hex << gr << "," << el <<")" << std::dec;
2058 dbg.Verbose(0, s.str().c_str());
2061 //////// Fix for some naughty General Electric images.
2062 // Allthough not recent many such GE corrupted images are still present
2063 // on Creatis hard disks. Hence this fix shall remain when such images
2064 // are no longer in user (we are talking a few years, here)...
2065 // Note: XMedCom probably uses such a trick since it is able to read
2066 // those pesky GE images ...
2067 if ( foundLength == 13)
2069 // Only happens for this length !
2070 if ( entry->GetGroup() != 0x0008
2071 || ( entry->GetElement() != 0x0070
2072 && entry->GetElement() != 0x0080 ) )
2075 entry->SetReadLength(10); /// \todo a bug is to be fixed !?
2079 //////// Fix for some brain-dead 'Leonardo' Siemens images.
2080 // Occurence of such images is quite low (unless one leaves close to a
2081 // 'Leonardo' source. Hence, one might consider commenting out the
2082 // following fix on efficiency reasons.
2083 else if ( entry->GetGroup() == 0x0009
2084 && ( entry->GetElement() == 0x1113
2085 || entry->GetElement() == 0x1114 ) )
2088 entry->SetReadLength(4); /// \todo a bug is to be fixed !?
2091 //////// Deal with sequences, but only on users request:
2092 else if ( entry->GetVR() == "SQ" && EnableSequences)
2094 foundLength = 0; // ReadLength is unchanged
2097 //////// We encountered a 'delimiter' element i.e. a tag of the form
2098 // "fffe|xxxx" which is just a marker. Delimiters length should not be
2099 // taken into account.
2100 else if( entry->GetGroup() == 0xfffe )
2102 // According to the norm, fffe|0000 shouldn't exist. BUT the Philips
2103 // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to
2104 // causes extra troubles...
2105 if( entry->GetElement() != 0x0000 )
2111 entry->SetUsableLength(foundLength);
2115 * \brief Apply some heuristics to predict whether the considered
2116 * element value contains/represents an integer or not.
2117 * @param Entry The element value on which to apply the predicate.
2118 * @return The result of the heuristical predicate.
2120 bool gdcmDocument::IsDocEntryAnInteger(gdcmDocEntry *entry)
2122 uint16_t element = entry->GetElement();
2123 uint16_t group = entry->GetGroup();
2124 std::string vr = entry->GetVR();
2125 uint32_t length = entry->GetLength();
2127 // When we have some semantics on the element we just read, and if we
2128 // a priori know we are dealing with an integer, then we shall be
2129 // able to swap it's element value properly.
2130 if ( element == 0 ) // This is the group length of the group
2138 // Allthough this should never happen, still some images have a
2139 // corrupted group length [e.g. have a glance at offset x(8336) of
2140 // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm].
2141 // Since for dicom compliant and well behaved headers, the present
2142 // test is useless (and might even look a bit paranoid), when we
2143 // encounter such an ill-formed image, we simply display a warning
2144 // message and proceed on parsing (while crossing fingers).
2145 std::ostringstream s;
2146 int filePosition = ftell(Fp);
2147 s << "Erroneous Group Length element length on : (" \
2148 << std::hex << group << " , " << element
2149 << ") -before- position x(" << filePosition << ")"
2150 << "lgt : " << length;
2151 dbg.Verbose(0, "gdcmDocument::IsDocEntryAnInteger", s.str().c_str() );
2155 if ( vr == "UL" || vr == "US" || vr == "SL" || vr == "SS" )
2164 * \brief Find the Length till the next sequence delimiter
2165 * \warning NOT end user intended method !
2169 uint32_t gdcmDocument::FindDocEntryLengthOB()
2171 // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
2172 long positionOnEntry = ftell(Fp);
2173 bool foundSequenceDelimiter = false;
2174 uint32_t totalLength = 0;
2176 while ( !foundSequenceDelimiter )
2178 uint16_t g = ReadInt16();
2179 uint16_t n = ReadInt16();
2185 // We have to decount the group and element we just read
2188 if ( g != 0xfffe || ( n != 0xe0dd && n != 0xe000 ) )
2190 dbg.Verbose(1, "gdcmDocument::FindLengthOB: neither an Item tag "
2191 "nor a Sequence delimiter tag.");
2192 fseek(Fp, positionOnEntry, SEEK_SET);
2199 foundSequenceDelimiter = true;
2202 uint32_t itemLength = ReadInt32();
2203 // We add 4 bytes since we just read the ItemLength with ReadInt32
2204 totalLength += itemLength + 4;
2205 SkipBytes(itemLength);
2207 if ( foundSequenceDelimiter )
2212 fseek(Fp, positionOnEntry, SEEK_SET);
2217 * \brief Reads a supposed to be 16 Bits integer
2218 * (swaps it depending on processor endianity)
2219 * @return read value
2221 uint16_t gdcmDocument::ReadInt16()
2224 size_t item_read = fread (&g, (size_t)2,(size_t)1, Fp);
2225 if ( item_read != 1 )
2229 dbg.Verbose(0, "gdcmDocument::ReadInt16", " File Error");
2240 * \brief Reads a supposed to be 32 Bits integer
2241 * (swaps it depending on processor endianity)
2242 * @return read value
2244 uint32_t gdcmDocument::ReadInt32()
2247 size_t item_read = fread (&g, (size_t)4,(size_t)1, Fp);
2248 if ( item_read != 1 )
2252 dbg.Verbose(0, "gdcmDocument::ReadInt32", " File Error");
2263 * \brief skips bytes inside the source file
2264 * \warning NOT end user intended method !
2267 void gdcmDocument::SkipBytes(uint32_t nBytes)
2269 //FIXME don't dump the returned value
2270 (void)fseek(Fp, (long)nBytes, SEEK_CUR);
2274 * \brief Loads all the needed Dictionaries
2275 * \warning NOT end user intended method !
2277 void gdcmDocument::Initialise()
2279 RefPubDict = gdcmGlobal::GetDicts()->GetDefaultPubDict();
2284 * \brief Discover what the swap code is (among little endian, big endian,
2285 * bad little endian, bad big endian).
2287 * @return false when we are absolutely sure
2288 * it's neither ACR-NEMA nor DICOM
2289 * true when we hope ours assuptions are OK
2291 bool gdcmDocument::CheckSwap()
2293 // The only guaranted way of finding the swap code is to find a
2294 // group tag since we know it's length has to be of four bytes i.e.
2295 // 0x00000004. Finding the swap code in then straigthforward. Trouble
2296 // occurs when we can't find such group...
2298 uint32_t x = 4; // x : for ntohs
2299 bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
2303 char deb[HEADER_LENGTH_TO_READ];
2305 // First, compare HostByteOrder and NetworkByteOrder in order to
2306 // determine if we shall need to swap bytes (i.e. the Endian type).
2307 if ( x == ntohs(x) )
2316 // The easiest case is the one of a DICOM header, since it possesses a
2317 // file preamble where it suffice to look for the string "DICM".
2318 int lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, Fp);
2319 (void)lgrLue; //FIXME not used
2321 char *entCur = deb + 128;
2322 if( memcmp(entCur, "DICM", (size_t)4) == 0 )
2324 dbg.Verbose(1, "gdcmDocument::CheckSwap:", "looks like DICOM Version3");
2326 // Next, determine the value representation (VR). Let's skip to the
2327 // first element (0002, 0000) and check there if we find "UL"
2328 // - or "OB" if the 1st one is (0002,0001) -,
2329 // in which case we (almost) know it is explicit VR.
2330 // WARNING: if it happens to be implicit VR then what we will read
2331 // is the length of the group. If this ascii representation of this
2332 // length happens to be "UL" then we shall believe it is explicit VR.
2333 // FIXME: in order to fix the above warning, we could read the next
2334 // element value (or a couple of elements values) in order to make
2335 // sure we are not commiting a big mistake.
2336 // We need to skip :
2337 // * the 128 bytes of File Preamble (often padded with zeroes),
2338 // * the 4 bytes of "DICM" string,
2339 // * the 4 bytes of the first tag (0002, 0000),or (0002, 0001)
2340 // i.e. a total of 136 bytes.
2344 // Sometimes (see : gdcmData/icone.dcm) group 0x0002 *is* Explicit VR,
2345 // but elem 0002,0010 (Transfert Syntax) tells us the file is
2346 // *Implicit* VR. -and it is !-
2348 if( memcmp(entCur, "UL", (size_t)2) == 0 ||
2349 memcmp(entCur, "OB", (size_t)2) == 0 ||
2350 memcmp(entCur, "UI", (size_t)2) == 0 ||
2351 memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later
2352 // when Write DCM *adds*
2354 // Use gdcmDocument::dicom_vr to test all the possibilities
2355 // instead of just checking for UL, OB and UI !? group 0000
2357 Filetype = gdcmExplicitVR;
2358 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2359 "explicit Value Representation");
2363 Filetype = gdcmImplicitVR;
2364 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2365 "not an explicit Value Representation");
2371 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2372 "HostByteOrder != NetworkByteOrder");
2377 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2378 "HostByteOrder = NetworkByteOrder");
2381 // Position the file position indicator at first tag (i.e.
2382 // after the file preamble and the "DICM" string).
2384 fseek (Fp, 132L, SEEK_SET);
2388 // Alas, this is not a DicomV3 file and whatever happens there is no file
2389 // preamble. We can reset the file position indicator to where the data
2390 // is (i.e. the beginning of the file).
2391 dbg.Verbose(1, "gdcmDocument::CheckSwap:", "not a DICOM Version3 file");
2394 // Our next best chance would be to be considering a 'clean' ACR/NEMA file.
2395 // By clean we mean that the length of the first tag is written down.
2396 // If this is the case and since the length of the first group HAS to be
2397 // four (bytes), then determining the proper swap code is straightforward.
2400 // We assume the array of char we are considering contains the binary
2401 // representation of a 32 bits integer. Hence the following dirty
2403 s32 = *((uint32_t *)(entCur));
2424 // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
2425 // It is time for despaired wild guesses.
2426 // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA,
2427 // i.e. the 'group length' element is not present :
2429 // check the supposed to be 'group number'
2430 // 0x0002 or 0x0004 or 0x0008
2431 // to determine ' SwapCode' value .
2432 // Only 0 or 4321 will be possible
2433 // (no oportunity to check for the formerly well known
2434 // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian'
2435 // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -4, 8-)
2436 // the file IS NOT ACR-NEMA nor DICOM V3
2437 // Find a trick to tell it the caller...
2439 s16 = *((uint16_t *)(deb));
2456 dbg.Verbose(0, "gdcmDocument::CheckSwap:",
2457 "ACR/NEMA unfound swap info (Really hopeless !)");
2458 Filetype = gdcmUnknown;
2461 // Then the only info we have is the net2host one.
2471 * \brief Restore the unproperly loaded values i.e. the group, the element
2472 * and the dictionary entry depending on them.
2474 void gdcmDocument::SwitchSwapToBigEndian()
2476 dbg.Verbose(1, "gdcmDocument::SwitchSwapToBigEndian",
2477 "Switching to BigEndian mode.");
2478 if ( SwapCode == 0 )
2482 else if ( SwapCode == 4321 )
2486 else if ( SwapCode == 3412 )
2490 else if ( SwapCode == 2143 )
2497 * \brief during parsing, Header Elements too long are not loaded in memory
2500 void gdcmDocument::SetMaxSizeLoadEntry(long newSize)
2506 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2508 MaxSizeLoadEntry = 0xffffffff;
2511 MaxSizeLoadEntry = newSize;
2516 * \brief Header Elements too long will not be printed
2517 * \todo See comments of \ref gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE
2520 void gdcmDocument::SetMaxSizePrintEntry(long newSize)
2522 //DOH !! This is exactly SetMaxSizeLoadEntry FIXME FIXME
2527 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2529 MaxSizePrintEntry = 0xffffffff;
2532 MaxSizePrintEntry = newSize;
2538 * \brief Read the next tag but WITHOUT loading it's value
2539 * (read the 'Group Number', the 'Element Number',
2540 * gets the Dict Entry
2541 * gets the VR, gets the length, gets the offset value)
2542 * @return On succes the newly created DocEntry, NULL on failure.
2544 gdcmDocEntry *gdcmDocument::ReadNextDocEntry()
2546 uint16_t g = ReadInt16();
2547 uint16_t n = ReadInt16();
2551 // We reached the EOF (or an error occured) therefore
2552 // header parsing has to be considered as finished.
2555 gdcmDocEntry *newEntry = NewDocEntryByNumber(g, n);
2557 FindDocEntryVR(newEntry);
2558 FindDocEntryLength(newEntry);
2566 newEntry->SetOffset(ftell(Fp));
2573 * \brief Generate a free gdcmTagKey i.e. a gdcmTagKey that is not present
2574 * in the TagHt dictionary.
2575 * @param group The generated tag must belong to this group.
2576 * @return The element of tag with given group which is fee.
2578 uint32_t gdcmDocument::GenerateFreeTagKeyInGroup(uint16_t group)
2580 for (uint32_t elem = 0; elem < UINT32_MAX; elem++)
2582 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, elem);
2583 if (TagHT.count(key) == 0)
2592 * \brief Assuming the internal file pointer \ref gdcmDocument::fp
2593 * is placed at the beginning of a tag check whether this
2594 * tag is (TestGroup, TestElement).
2595 * \warning On success the internal file pointer \ref gdcmDocument::fp
2596 * is modified to point after the tag.
2597 * On failure (i.e. when the tag wasn't the expected tag
2598 * (TestGroup, TestElement) the internal file pointer
2599 * \ref gdcmDocument::fp is restored to it's original position.
2600 * @param TestGroup The expected group of the tag.
2601 * @param TestElement The expected Element of the tag.
2602 * @return True on success, false otherwise.
2604 bool gdcmDocument::ReadTag(uint16_t testGroup, uint16_t testElement)
2606 long positionOnEntry = ftell(Fp);
2607 long currentPosition = ftell(Fp); // On debugging purposes
2609 //// Read the Item Tag group and element, and make
2610 // sure they are what we expected:
2611 uint16_t itemTagGroup = ReadInt16();
2612 uint16_t itemTagElement = ReadInt16();
2613 if ( itemTagGroup != testGroup || itemTagElement != testElement )
2615 std::ostringstream s;
2616 s << " We should have found tag (";
2617 s << std::hex << testGroup << "," << testElement << ")" << std::endl;
2618 s << " but instead we encountered tag (";
2619 s << std::hex << itemTagGroup << "," << itemTagElement << ")"
2621 s << " at address: " << (unsigned)currentPosition << std::endl;
2622 dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: wrong Item Tag found:");
2623 dbg.Verbose(0, s.str().c_str());
2624 fseek(Fp, positionOnEntry, SEEK_SET);
2632 * \brief Assuming the internal file pointer \ref gdcmDocument::fp
2633 * is placed at the beginning of a tag (TestGroup, TestElement),
2634 * read the length associated to the Tag.
2635 * \warning On success the internal file pointer \ref gdcmDocument::fp
2636 * is modified to point after the tag and it's length.
2637 * On failure (i.e. when the tag wasn't the expected tag
2638 * (TestGroup, TestElement) the internal file pointer
2639 * \ref gdcmDocument::fp is restored to it's original position.
2640 * @param TestGroup The expected group of the tag.
2641 * @param TestElement The expected Element of the tag.
2642 * @return On success returns the length associated to the tag. On failure
2645 uint32_t gdcmDocument::ReadTagLength(uint16_t testGroup, uint16_t testElement)
2647 long PositionOnEntry = ftell(Fp);
2648 (void)PositionOnEntry;
2650 if ( !ReadTag(testGroup, testElement) )
2655 //// Then read the associated Item Length
2656 long currentPosition = ftell(Fp);
2657 uint32_t itemLength = ReadInt32();
2659 std::ostringstream s;
2660 s << "Basic Item Length is: "
2661 << itemLength << std::endl;
2662 s << " at address: " << (unsigned)currentPosition << std::endl;
2663 dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: ", s.str().c_str());
2669 * \brief Parse pixel data from disk for multi-fragment Jpeg/Rle files
2670 * No other way so 'skip' the Data
2673 void gdcmDocument::Parse7FE0 ()
2675 gdcmDocEntry* element = GetDocEntryByNumber(0x0002, 0x0010);
2678 // Should warn user FIXME
2682 if ( IsImplicitVRLittleEndianTransferSyntax()
2683 || IsExplicitVRLittleEndianTransferSyntax()
2684 || IsExplicitVRBigEndianTransferSyntax() /// \todo 1.2.2 ??? A verifier !
2685 || IsDeflatedExplicitVRLittleEndianTransferSyntax() )
2690 // ---------------- for Parsing : Position on begining of Jpeg/RLE Pixels
2692 //// Read the Basic Offset Table Item Tag length...
2693 uint32_t itemLength = ReadTagLength(0xfffe, 0xe000);
2695 //// ... and then read length[s] itself[themselves]. We don't use
2696 // the values read (BTW what is the purpous of those lengths ?)
2697 if ( itemLength != 0 )
2699 // BTW, what is the purpous of those length anyhow !?
2700 char* basicOffsetTableItemValue = new char[itemLength + 1];
2701 fread(basicOffsetTableItemValue, itemLength, 1, Fp);
2703 for (unsigned int i=0; i < itemLength; i += 4 )
2705 uint32_t individualLength = str2num(&basicOffsetTableItemValue[i],uint32_t);
2706 std::ostringstream s;
2707 s << " Read one length: ";
2708 s << std::hex << individualLength << std::endl;
2709 dbg.Verbose(0, "gdcmDocument::Parse7FE0: ", s.str().c_str());
2713 if ( ! IsRLELossLessTransferSyntax() )
2717 //// We then skip (not reading them) all the fragments of images:
2718 while ( (itemLength = ReadTagLength(0xfffe, 0xe000)) )
2720 SkipBytes(itemLength);
2727 long rleSegmentLength[15], fragmentLength;
2729 // While we find some items:
2730 while ( (fragmentLength = ReadTagLength(0xfffe, 0xe000)) )
2732 // Parse fragments of the current Fragment (Frame)
2733 //------------------ scanning (not reading) fragment pixels
2734 uint32_t nbRleSegments = ReadInt32();
2736 //// Reading RLE Segments Offset Table
2737 uint32_t rleSegmentOffsetTable[15];
2738 for(int k=1; k<=15; k++)
2740 ftellRes = ftell(Fp);
2741 rleSegmentOffsetTable[k] = ReadInt32();
2744 // skipping (not reading) RLE Segments
2745 if ( nbRleSegments > 1)
2747 for(unsigned int k = 1; k <= nbRleSegments-1; k++)
2749 rleSegmentLength[k] = rleSegmentOffsetTable[k+1]
2750 - rleSegmentOffsetTable[k];
2751 ftellRes = ftell(Fp);
2752 SkipBytes(rleSegmentLength[k]);
2756 rleSegmentLength[nbRleSegments] = fragmentLength
2757 - rleSegmentOffsetTable[nbRleSegments];
2758 ftellRes = ftell(Fp);
2759 SkipBytes(rleSegmentLength[nbRleSegments]);
2762 // Make sure that at the end of the item we encounter a 'Sequence
2764 if ( ! ReadTag(0xfffe, 0xe0dd) )
2766 dbg.Verbose(0, "gdcmDocument::Parse7FE0: no sequence delimiter item");
2767 dbg.Verbose(0, " at end of RLE item sequence");
2775 * \brief Compares two documents, according to \ref gdcmDicomDir rules
2776 * \warning Does NOT work with ACR-NEMA files
2777 * \todo Find a trick to solve the pb (use RET fields ?)
2779 * @return true if 'smaller'
2781 bool gdcmDocument::operator<(gdcmDocument &document)
2784 std::string s1 = this->GetEntryByNumber(0x0010,0x0010);
2785 std::string s2 = document.GetEntryByNumber(0x0010,0x0010);
2797 s1 = this->GetEntryByNumber(0x0010,0x0020);
2798 s2 = document.GetEntryByNumber(0x0010,0x0020);
2809 // Study Instance UID
2810 s1 = this->GetEntryByNumber(0x0020,0x000d);
2811 s2 = document.GetEntryByNumber(0x0020,0x000d);
2822 // Serie Instance UID
2823 s1 = this->GetEntryByNumber(0x0020,0x000e);
2824 s2 = document.GetEntryByNumber(0x0020,0x000e);
2840 //-----------------------------------------------------------------------------