1 /*=========================================================================
4 Module: $RCSfile: gdcmDocument.cxx,v $
6 Date: $Date: 2004/06/28 16:07:21 $
7 Version: $Revision: 1.41 $
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 = 0x7fffffff;// 4096;//
78 const unsigned int gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff;//64;
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 & inFilename,
98 bool exception_on_error,
99 bool enable_sequences,
101 : gdcmElementSet(-1) {
102 IgnoreShadow =ignore_shadow;
103 //enableSequences=enable_sequences;
104 (void)enable_sequences;
105 enableSequences=true; // JPR // TODO : remove params out of the constructor
106 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
107 Filename = inFilename;
110 if ( !OpenFile(exception_on_error))
113 dbg.Verbose(0, "gdcmDocument::gdcmDocument: starting parsing of file: ",
117 fseek(fp,0L,SEEK_END);
118 long lgt = ftell(fp);
122 long beg = ftell(fp);
127 long l=ParseDES( this, beg, lgt, false); // le Load sera fait a la volee
128 (void)l; //is l used anywhere ?
132 // Load 'non string' values
134 std::string PhotometricInterpretation = GetEntryByNumber(0x0028,0x0004);
135 if( PhotometricInterpretation == "PALETTE COLOR " ) {
136 LoadEntryVoidArea(0x0028,0x1200); // gray LUT
137 LoadEntryVoidArea(0x0028,0x1201); // R LUT
138 LoadEntryVoidArea(0x0028,0x1202); // G LUT
139 LoadEntryVoidArea(0x0028,0x1203); // B LUT
141 LoadEntryVoidArea(0x0028,0x1221); // Segmented Red Palette Color LUT Data
142 LoadEntryVoidArea(0x0028,0x1222); // Segmented Green Palette Color LUT Data
143 LoadEntryVoidArea(0x0028,0x1223); // Segmented Blue Palette Color LUT Data
145 //FIXME later : how to use it?
146 LoadEntryVoidArea(0x0028,0x3006); //LUT Data (CTX dependent)
150 // --------------------------------------------------------------
151 // Special Patch to allow gdcm to read ACR-LibIDO formated images
153 // if recognition code tells us we deal with a LibIDO image
154 // we switch lineNumber and columnNumber
157 RecCode = GetEntryByNumber(0x0008, 0x0010); // recognition code
158 if (RecCode == "ACRNEMA_LIBIDO_1.1" ||
159 RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares
160 // with "little-endian strings"
162 Filetype = gdcmACR_LIBIDO;
163 std::string rows = GetEntryByNumber(0x0028, 0x0010);
164 std::string columns = GetEntryByNumber(0x0028, 0x0011);
165 SetEntryByNumber(columns, 0x0028, 0x0010);
166 SetEntryByNumber(rows , 0x0028, 0x0011);
168 // ----------------- End of Special Patch ----------------
170 printLevel = 1; // 'Medium' print level by default
175 * @param exception_on_error
177 gdcmDocument::gdcmDocument(bool exception_on_error)
178 :gdcmElementSet(-1) {
179 (void)exception_on_error;
180 //enableSequences=0; // ?!? JPR
182 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
185 printLevel = 1; // 'Medium' print level by default
189 * \brief Canonical destructor.
191 gdcmDocument::~gdcmDocument (void) {
195 // Recursive clean up of sequences
196 for (TagDocEntryHT::iterator it = tagHT.begin(); it != tagHT.end(); ++it )
203 //-----------------------------------------------------------------------------
207 * \brief Prints The Dict Entries of THE public Dicom Dictionary
210 void gdcmDocument::PrintPubDict(std::ostream & os) {
211 RefPubDict->Print(os);
215 * \brief Prints The Dict Entries of THE shadow Dicom Dictionary
218 void gdcmDocument::PrintShaDict(std::ostream & os) {
219 RefShaDict->Print(os);
222 //-----------------------------------------------------------------------------
225 * \brief Get the public dictionary used
227 gdcmDict *gdcmDocument::GetPubDict(void) {
232 * \brief Get the shadow dictionary used
234 gdcmDict *gdcmDocument::GetShaDict(void) {
239 * \brief Set the shadow dictionary used
240 * \param dict dictionary to use in shadow
242 bool gdcmDocument::SetShaDict(gdcmDict *dict){
248 * \brief Set the shadow dictionary used
249 * \param dictName name of the dictionary to use in shadow
251 bool gdcmDocument::SetShaDict(DictKey dictName){
252 RefShaDict=gdcmGlobal::GetDicts()->GetDict(dictName);
257 * \brief This predicate, based on hopefully reasonable heuristics,
258 * decides whether or not the current gdcmDocument was properly parsed
259 * and contains the mandatory information for being considered as
260 * a well formed and usable Dicom/Acr File.
261 * @return true when gdcmDocument is the one of a reasonable Dicom/Acr file,
264 bool gdcmDocument::IsReadable(void) {
266 if(Filetype==gdcmUnknown) {
267 std::cout << " gdcmDocument::IsReadable: Filetype " << Filetype
268 << " " << "gdcmUnknown " << gdcmUnknown << std::endl; //JPR
269 dbg.Verbose(0, "gdcmDocument::IsReadable: wrong filetype");
274 dbg.Verbose(0, "gdcmDocument::IsReadable: no tags in internal"
284 * \brief Internal function that checks whether the Transfer Syntax given
285 * as argument is the one present in the current document.
286 * @param SyntaxToCheck The transfert syntax we need to check against.
287 * @return True when SyntaxToCheck corresponds to the Transfer Syntax of
288 * the current document. False either when the document contains
289 * no Transfer Syntax, or when the Tranfer Syntaxes don't match.
291 bool gdcmDocument::IsGivenTransferSyntax(const std::string & SyntaxToCheck)
293 gdcmDocEntry *Entry = GetDocEntryByNumber(0x0002, 0x0010);
297 // The entry might be present but not loaded (parsing and loading
298 // happen at differente stages): try loading and proceed with check...
299 LoadDocEntrySafe(Entry);
300 if (gdcmValEntry* ValEntry = dynamic_cast< gdcmValEntry* >(Entry) )
302 std::string Transfer = ValEntry->GetValue();
303 // The actual transfer (as read from disk) might be padded. We
304 // first need to remove the potential padding. We can make the
305 // weak assumption that padding was not executed with digits...
306 while ( ! isdigit(Transfer[Transfer.length()-1]) )
308 Transfer.erase(Transfer.length()-1, 1);
310 if ( Transfer == SyntaxToCheck )
317 * \brief Determines if the Transfer Syntax of the present document
318 * corresponds to a Implicit Value Representation of
320 * \sa \ref gdcmDocument::IsGivenTransferSyntax.
321 * @return True when ImplicitVRLittleEndian found. False in all other cases.
323 bool gdcmDocument::IsImplicitVRLittleEndianTransferSyntax(void)
325 return IsGivenTransferSyntax(UI1_2_840_10008_1_2);
329 * \brief Determines if the Transfer Syntax was already encountered
330 * and if it corresponds to a ExplicitVRLittleEndian one.
331 * @return True when ExplicitVRLittleEndian found. False in all other cases.
333 bool gdcmDocument::IsExplicitVRLittleEndianTransferSyntax(void)
335 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1);
339 * \brief Determines if the Transfer Syntax was already encountered
340 * and if it corresponds to a DeflatedExplicitVRLittleEndian one.
341 * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases.
343 bool gdcmDocument::IsDeflatedExplicitVRLittleEndianTransferSyntax(void)
345 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1_99);
349 * \brief Determines if the Transfer Syntax was already encountered
350 * and if it corresponds to a Explicit VR Big Endian one.
351 * @return True when big endian found. False in all other cases.
353 bool gdcmDocument::IsExplicitVRBigEndianTransferSyntax(void)
355 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_2);
359 * \brief Determines if the Transfer Syntax was already encountered
360 * and if it corresponds to a JPEGBaseLineProcess1 one.
361 * @return True when JPEGBaseLineProcess1found. False in all other cases.
363 bool gdcmDocument::IsJPEGBaseLineProcess1TransferSyntax(void)
365 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_50);
369 * \brief Determines if the Transfer Syntax was already encountered
370 * and if it corresponds to a JPEGExtendedProcess2-4 one.
371 * @return True when JPEGExtendedProcess2-4 found. False in all other cases.
373 bool gdcmDocument::IsJPEGExtendedProcess2_4TransferSyntax(void)
375 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_51);
379 * \brief Determines if the Transfer Syntax was already encountered
380 * and if it corresponds to a JPEGExtendeProcess3-5 one.
381 * @return True when JPEGExtendedProcess3-5 found. False in all other cases.
383 bool gdcmDocument::IsJPEGExtendedProcess3_5TransferSyntax(void)
385 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_52);
389 * \brief Determines if the Transfer Syntax was already encountered
390 * and if it corresponds to a JPEGSpectralSelectionProcess6-8 one.
391 * @return True when JPEGSpectralSelectionProcess6-8 found. False in all
394 bool gdcmDocument::IsJPEGSpectralSelectionProcess6_8TransferSyntax(void)
396 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_53);
400 * \brief Determines if the Transfer Syntax was already encountered
401 * and if it corresponds to a RLE Lossless one.
402 * @return True when RLE Lossless found. False in all
405 bool gdcmDocument::IsRLELossLessTransferSyntax(void)
407 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_5);
411 * \brief Determines if Transfer Syntax was already encountered
412 * and if it corresponds to a JPEG Lossless one.
413 * @return True when RLE Lossless found. False in all
417 bool gdcmDocument::IsJPEGLossless(void)
419 return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_55)
420 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_57)
421 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_70) );
425 * \brief Determines if the Transfer Syntax was already encountered
426 * and if it corresponds to a JPEG2000 one
427 * @return True when JPEG2000 (Lossly or LossLess) found. False in all
430 bool gdcmDocument::IsJPEG2000(void)
432 return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_90)
433 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_91) );
437 * \brief Predicate for dicom version 3 file.
438 * @return True when the file is a dicom version 3.
440 bool gdcmDocument::IsDicomV3(void) {
441 // Checking if Transfert Syntax exists is enough
442 // Anyway, it's to late check if the 'Preamble' was found ...
443 // And ... would it be a rich idea to check ?
444 // (some 'no Preamble' DICOM images exist !)
445 return (GetDocEntryByNumber(0x0002, 0x0010) != NULL);
449 * \brief returns the File Type
450 * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
451 * @return the FileType code
453 FileType gdcmDocument::GetFileType(void) {
458 * \brief opens the file
459 * @param exception_on_error
462 FILE *gdcmDocument::OpenFile(bool exception_on_error)
465 fp=fopen(Filename.c_str(),"rb");
469 if(exception_on_error)
470 throw gdcmFileError("gdcmDocument::gdcmDocument(const char *, bool)");
473 dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file: ",
482 fread(&zero, (size_t)2, (size_t)1, fp);
484 //ACR -- or DICOM with no Preamble --
485 if( zero == 0x0008 || zero == 0x0800 || zero == 0x0002 || zero == 0x0200)
489 fseek(fp, 126L, SEEK_CUR);
491 fread(dicm, (size_t)4, (size_t)1, fp);
492 if( memcmp(dicm, "DICM", 4) == 0 )
496 dbg.Verbose(0, "gdcmDocument::OpenFile not DICOM/ACR", Filename.c_str());
499 dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file", Filename.c_str());
505 * \brief closes the file
506 * @return TRUE if the close was successfull
508 bool gdcmDocument::CloseFile(void) {
509 int closed = fclose(fp);
517 * \brief Writes in a file all the Header Entries (Dicom Elements)
518 * @param fp file pointer on an already open file
519 * @param filetype Type of the File to be written
520 * (ACR-NEMA, ExplicitVR, ImplicitVR)
521 * \return Always true.
523 void gdcmDocument::Write(FILE* fp,FileType filetype) {
525 /// \todo move the following lines (and a lot of others, to be written)
526 /// to a future function CheckAndCorrectHeader
528 /// WARNING : Si on veut ecrire du DICOM V3 a partir d'un DcmHeader ACR-NEMA
529 /// no way (check : FileType est un champ de gdcmDocument ...)
530 /// a moins de se livrer a un tres complique ajout des champs manquants.
531 /// faire un CheckAndCorrectHeader (?)
533 if (filetype == gdcmImplicitVR)
535 std::string implicitVRTransfertSyntax = UI1_2_840_10008_1_2;
536 ReplaceOrCreateByNumber(implicitVRTransfertSyntax,0x0002, 0x0010);
538 /// \todo Refer to standards on page 21, chapter 6.2
539 /// "Value representation": values with a VR of UI shall be
540 /// padded with a single trailing null
541 /// in the following case we have to padd manually with a 0
543 SetEntryLengthByNumber(18, 0x0002, 0x0010);
546 if (filetype == gdcmExplicitVR)
548 std::string explicitVRTransfertSyntax = UI1_2_840_10008_1_2_1;
549 ReplaceOrCreateByNumber(explicitVRTransfertSyntax,0x0002, 0x0010);
551 /// \todo Refer to standards on page 21, chapter 6.2
552 /// "Value representation": values with a VR of UI shall be
553 /// padded with a single trailing null
554 /// Dans le cas suivant on doit pader manuellement avec un 0
556 SetEntryLengthByNumber(20, 0x0002, 0x0010);
560 * \todo rewrite later, if really usefull
561 * - 'Group Length' element is optional in DICOM
562 * - but un-updated odd groups lengthes can causes pb
565 * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) )
566 * UpdateGroupLength(false,filetype);
567 * if ( filetype == ACR)
568 * UpdateGroupLength(true,ACR);
571 gdcmElementSet::Write(fp,filetype); // This one is recursive
576 * \brief Modifies the value of a given Header Entry (Dicom Element)
577 * when it exists. Create it with the given value when unexistant.
578 * @param Value (string) Value to be set
579 * @param Group Group number of the Entry
580 * @param Elem Element number of the Entry
581 * \return pointer to the modified/created Header Entry (NULL when creation
585 gdcmValEntry * gdcmDocument::ReplaceOrCreateByNumber(
590 gdcmDocEntry* CurrentEntry;
591 gdcmValEntry* ValEntry;
593 CurrentEntry = GetDocEntryByNumber( Group, Elem);
596 // The entry wasn't present and we simply create the required ValEntry:
597 CurrentEntry = NewDocEntryByNumber(Group, Elem);
600 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: call to"
601 " NewDocEntryByNumber failed.");
604 ValEntry = new gdcmValEntry(CurrentEntry);
605 if ( !AddEntry(ValEntry))
607 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: AddEntry"
608 " failed allthough this is a creation.");
613 ValEntry = dynamic_cast< gdcmValEntry* >(CurrentEntry);
616 // We need to promote the gdcmDocEntry to a gdcmValEntry:
617 ValEntry = new gdcmValEntry(CurrentEntry);
618 if (!RemoveEntry(CurrentEntry))
620 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: removal"
621 " of previous DocEntry failed.");
624 if ( !AddEntry(ValEntry))
626 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: adding"
627 " promoted ValEntry failed.");
633 SetEntryByNumber(Value, Group, Elem);
639 * \brief Modifies the value of a given Header Entry (Dicom Element)
640 * when it exists. Create it with the given value when unexistant.
641 * @param voidArea (binary) value to be set
642 * @param Group Group number of the Entry
643 * @param Elem Element number of the Entry
644 * \return pointer to the modified/created Header Entry (NULL when creation
647 gdcmBinEntry * gdcmDocument::ReplaceOrCreateByNumber(
655 a = GetDocEntryByNumber( Group, Elem);
657 a =NewBinEntryByNumber(Group, Elem);
661 b = new gdcmBinEntry(a);
664 SetEntryByNumber(voidArea, lgth, Group, Elem);
665 b->SetVoidArea(voidArea);
673 * \brief Set a new value if the invoked element exists
674 * Seems to be useless !!!
675 * @param Value new element value
676 * @param Group group number of the Entry
677 * @param Elem element number of the Entry
680 bool gdcmDocument::ReplaceIfExistByNumber(char* Value, guint16 Group, guint16 Elem )
682 std::string v = Value;
683 SetEntryByNumber(v, Group, Elem);
687 //-----------------------------------------------------------------------------
691 * \brief Checks if a given Dicom Element exists within the H table
692 * @param group Group number of the searched Dicom Element
693 * @param element Element number of the searched Dicom Element
694 * @return number of occurences
696 int gdcmDocument::CheckIfEntryExistByNumber(guint16 group, guint16 element ) {
697 std::string key = gdcmDictEntry::TranslateToKey(group, element );
698 return tagHT.count(key);
702 * \brief Searches within Header Entries (Dicom Elements) parsed with
703 * the public and private dictionaries
704 * for the element value of a given tag.
705 * \warning Don't use any longer : use GetPubEntryByName
706 * @param tagName name of the searched element.
707 * @return Corresponding element value when it exists,
708 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
710 std::string gdcmDocument::GetEntryByName(TagName tagName) {
711 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
712 if( dictEntry == NULL)
715 return GetEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement());
719 * \brief Searches within Header Entries (Dicom Elements) parsed with
720 * the public and private dictionaries
721 * for the element value representation of a given tag.
723 * Obtaining the VR (Value Representation) might be needed by caller
724 * to convert the string typed content to caller's native type
725 * (think of C++ vs Python). The VR is actually of a higher level
726 * of semantics than just the native C++ type.
727 * @param tagName name of the searched element.
728 * @return Corresponding element value representation when it exists,
729 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
731 std::string gdcmDocument::GetEntryVRByName(TagName tagName) {
732 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
733 if( dictEntry == NULL)
736 gdcmDocEntry* elem = GetDocEntryByNumber(dictEntry->GetGroup(),
737 dictEntry->GetElement());
738 return elem->GetVR();
743 * \brief Searches within Header Entries (Dicom Elements) parsed with
744 * the public and private dictionaries
745 * for the element value representation of a given tag.
746 * @param group Group number of the searched tag.
747 * @param element Element number of the searched tag.
748 * @return Corresponding element value representation when it exists,
749 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
751 std::string gdcmDocument::GetEntryByNumber(guint16 group, guint16 element){
752 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
753 if ( ! tagHT.count(key))
755 return ((gdcmValEntry *)tagHT.find(key)->second)->GetValue();
759 * \brief Searches within Header Entries (Dicom Elements) parsed with
760 * the public and private dictionaries
761 * for the element value representation of a given tag..
763 * Obtaining the VR (Value Representation) might be needed by caller
764 * to convert the string typed content to caller's native type
765 * (think of C++ vs Python). The VR is actually of a higher level
766 * of semantics than just the native C++ type.
767 * @param group Group number of the searched tag.
768 * @param element Element number of the searched tag.
769 * @return Corresponding element value representation when it exists,
770 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
772 std::string gdcmDocument::GetEntryVRByNumber(guint16 group, guint16 element) {
773 gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
776 return elem->GetVR();
780 * \brief Searches within Header Entries (Dicom Elements) parsed with
781 * the public and private dictionaries
782 * for the value length of a given tag..
783 * @param group Group number of the searched tag.
784 * @param element Element number of the searched tag.
785 * @return Corresponding element length; -2 if not found
787 int gdcmDocument::GetEntryLengthByNumber(guint16 group, guint16 element) {
788 gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
791 return elem->GetLength();
794 * \brief Sets the value (string) of the Header Entry (Dicom Element)
795 * @param content string value of the Dicom Element
796 * @param tagName name of the searched Dicom Element.
797 * @return true when found
799 bool gdcmDocument::SetEntryByName(std::string content,std::string tagName) {
800 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
801 if( dictEntry == NULL)
804 return SetEntryByNumber(content,dictEntry->GetGroup(),
805 dictEntry->GetElement());
809 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
810 * through it's (group, element) and modifies it's content with
812 * @param content new value (string) to substitute with
813 * @param group group number of the Dicom Element to modify
814 * @param element element number of the Dicom Element to modify
816 bool gdcmDocument::SetEntryByNumber(std::string content,
820 gdcmValEntry* ValEntry = GetValEntryByNumber(group, element);
823 dbg.Verbose(0, "gdcmDocument::SetEntryByNumber: no corresponding",
824 " ValEntry (try promotion first).");
827 // Non even content must be padded with a space (020H).
828 if((content.length())%2) {
829 content = content + '\0';
831 ValEntry->SetValue(content);
833 // Integers have a special treatement for their length:
834 VRKey vr = ValEntry->GetVR();
835 if( (vr == "US") || (vr == "SS") )
836 ValEntry->SetLength(2);
837 else if( (vr == "UL") || (vr == "SL") )
838 ValEntry->SetLength(4);
840 ValEntry->SetLength(content.length());
846 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
847 * through it's (group, element) and modifies it's content with
849 * @param content new value (void *) to substitute with
850 * @param lgth new value length
851 * @param group group number of the Dicom Element to modify
852 * @param element element number of the Dicom Element to modify
854 bool gdcmDocument::SetEntryByNumber(void *content,
859 (void)lgth; //not used
860 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
861 if ( ! tagHT.count(key))
864 /* Hope Binray field length is never wrong
865 if(lgth%2) // Non even length are padded with a space (020H).
868 //content = content + '\0'; // fing a trick to enlarge a binary field?
872 a = (gdcmBinEntry *)tagHT[key];
873 a->SetVoidArea(content);
874 //a->SetLength(lgth); // ???
879 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
880 * in the PubDocEntrySet of this instance
881 * through it's (group, element) and modifies it's length with
883 * \warning Use with extreme caution.
884 * @param l new length to substitute with
885 * @param group group number of the Entry to modify
886 * @param element element number of the Entry to modify
887 * @return true on success, false otherwise.
889 bool gdcmDocument::SetEntryLengthByNumber(guint32 l,
893 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
894 if ( ! tagHT.count(key))
896 if (l%2) l++; // length must be even
897 ( ((tagHT.equal_range(key)).first)->second )->SetLength(l);
903 * \brief Gets (from Header) the offset of a 'non string' element value
904 * (LoadElementValues has already be executed)
905 * @param Group group number of the Entry
906 * @param Elem element number of the Entry
907 * @return File Offset of the Element Value
909 size_t gdcmDocument::GetEntryOffsetByNumber(guint16 Group, guint16 Elem)
911 gdcmDocEntry* Entry = GetDocEntryByNumber(Group, Elem);
914 dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry present.");
917 return Entry->GetOffset();
921 * \brief Gets (from Header) a 'non string' element value
922 * (LoadElementValues has already be executed)
923 * @param Group group number of the Entry
924 * @param Elem element number of the Entry
925 * @return Pointer to the 'non string' area
927 void * gdcmDocument::GetEntryVoidAreaByNumber(guint16 Group, guint16 Elem)
929 gdcmDocEntry* Entry = GetDocEntryByNumber(Group, Elem);
932 dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry");
935 return ((gdcmBinEntry *)Entry)->GetVoidArea();
939 * \brief Loads (from disk) the element content
940 * when a string is not suitable
941 * @param Group group number of the Entry
942 * @param Elem element number of the Entry
944 void *gdcmDocument::LoadEntryVoidArea(guint16 Group, guint16 Elem)
946 gdcmDocEntry * Element= GetDocEntryByNumber(Group, Elem);
949 size_t o =(size_t)Element->GetOffset();
950 fseek(fp, o, SEEK_SET);
951 size_t l = Element->GetLength();
952 char* a = new char[l];
954 dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a");
957 SetEntryVoidAreaByNumber(a, Group, Elem);
958 /// \todo check the result
959 size_t l2 = fread(a, 1, l ,fp);
968 * \brief Loads (from disk) the element content
969 * when a string is not suitable
970 * @param Element Entry whose voidArea is going to be loaded
972 void *gdcmDocument::LoadEntryVoidArea(gdcmBinEntry *Element)
974 size_t o =(size_t)Element->GetOffset();
975 fseek(fp, o, SEEK_SET);
976 size_t l = Element->GetLength();
977 char* a = new char[l];
979 dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a");
982 Element->SetVoidArea((void *)a);
983 /// \todo check the result
984 size_t l2 = fread(a, 1, l ,fp);
994 * \brief Sets a 'non string' value to a given Dicom Element
995 * @param area area containing the 'non string' value
996 * @param group Group number of the searched Dicom Element
997 * @param element Element number of the searched Dicom Element
1000 bool gdcmDocument::SetEntryVoidAreaByNumber(void * area,
1004 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
1005 if ( ! tagHT.count(key))
1007 // This was for multimap ?
1008 (( gdcmBinEntry *)( ((tagHT.equal_range(key)).first)->second ))->SetVoidArea(area);
1014 * \brief Update the entries with the shadow dictionary.
1015 * Only non even entries are analyzed
1017 void gdcmDocument::UpdateShaEntries(void) {
1018 //gdcmDictEntry *entry;
1021 /// \todo TODO : still any use to explore recursively the whole structure?
1023 for(ListTag::iterator it=listEntries.begin();
1024 it!=listEntries.end();
1027 // Odd group => from public dictionary
1028 if((*it)->GetGroup()%2==0)
1031 // Peer group => search the corresponding dict entry
1033 entry=RefShaDict->GetDictEntryByNumber((*it)->GetGroup(),(*it)->GetElement());
1037 if((*it)->IsImplicitVR())
1042 (*it)->SetValue(GetDocEntryUnvalue(*it)); // to go on compiling
1044 // Set the new entry and the new value
1045 (*it)->SetDictEntry(entry);
1046 CheckDocEntryVR(*it,vr);
1048 (*it)->SetValue(GetDocEntryValue(*it)); // to go on compiling
1053 // Remove precedent value transformation
1054 (*it)->SetDictEntry(NewVirtualDictEntry((*it)->GetGroup(),(*it)->GetElement(),vr));
1061 * \brief Searches within the Header Entries for a Dicom Element of
1063 * @param tagName name of the searched Dicom Element.
1064 * @return Corresponding Dicom Element when it exists, and NULL
1067 gdcmDocEntry *gdcmDocument::GetDocEntryByName(std::string tagName) {
1068 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
1069 if( dictEntry == NULL)
1072 return(GetDocEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement()));
1076 * \brief retrieves a Dicom Element (the first one) using (group, element)
1077 * \warning (group, element) IS NOT an identifier inside the Dicom Header
1078 * if you think it's NOT UNIQUE, check the count number
1079 * and use iterators to retrieve ALL the Dicoms Elements within
1080 * a given couple (group, element)
1081 * @param group Group number of the searched Dicom Element
1082 * @param element Element number of the searched Dicom Element
1085 gdcmDocEntry* gdcmDocument::GetDocEntryByNumber(guint16 group, guint16 element)
1087 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
1088 if ( ! tagHT.count(key))
1090 return tagHT.find(key)->second;
1094 * \brief Same as \ref gdcmDocument::GetDocEntryByNumber except it only
1095 * returns a result when the corresponding entry is of type
1097 * @return When present, the corresponding ValEntry.
1099 gdcmValEntry* gdcmDocument::GetValEntryByNumber(guint16 group, guint16 element)
1101 gdcmDocEntry* CurrentEntry = GetDocEntryByNumber(group, element);
1103 return (gdcmValEntry*)0;
1104 if ( gdcmValEntry* ValEntry = dynamic_cast<gdcmValEntry*>(CurrentEntry) )
1108 dbg.Verbose(0, "gdcmDocument::GetValEntryByNumber: unfound ValEntry.");
1109 return (gdcmValEntry*)0;
1113 * \brief Loads the element while preserving the current
1114 * underlying file position indicator as opposed to
1115 * to LoadDocEntry that modifies it.
1116 * @param entry Header Entry whose value shall be loaded.
1119 void gdcmDocument::LoadDocEntrySafe(gdcmDocEntry * entry) {
1120 long PositionOnEntry = ftell(fp);
1121 LoadDocEntry(entry);
1122 fseek(fp, PositionOnEntry, SEEK_SET);
1126 * \brief Swaps back the bytes of 4-byte long integer accordingly to
1128 * @return The properly swaped 32 bits integer.
1130 guint32 gdcmDocument::SwapLong(guint32 a) {
1135 a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) |
1136 ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) );
1140 a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
1144 a=( ((a<<8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
1147 std::cout << "swapCode= " << sw << std::endl;
1148 dbg.Error(" gdcmDocument::SwapLong : unset swap code");
1155 * \brief Unswaps back the bytes of 4-byte long integer accordingly to
1157 * @return The properly unswaped 32 bits integer.
1159 guint32 gdcmDocument::UnswapLong(guint32 a) {
1164 * \brief Swaps the bytes so they agree with the processor order
1165 * @return The properly swaped 16 bits integer.
1167 guint16 gdcmDocument::SwapShort(guint16 a) {
1168 if ( (sw==4321) || (sw==2143) )
1169 a =(((a<<8) & 0x0ff00) | ((a>>8)&0x00ff));
1174 * \brief Unswaps the bytes so they agree with the processor order
1175 * @return The properly unswaped 16 bits integer.
1177 guint16 gdcmDocument::UnswapShort(guint16 a) {
1178 return SwapShort(a);
1181 //-----------------------------------------------------------------------------
1185 * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries)
1186 * @return length of the parsed set.
1189 long gdcmDocument::ParseDES(gdcmDocEntrySet *set, long offset, long l_max, bool delim_mode) {
1191 gdcmDocEntry *NewDocEntry = (gdcmDocEntry *)0;
1192 gdcmValEntry *NewValEntry = (gdcmValEntry *)0;
1196 unsigned long l = 0;
1199 depth = set->GetDepthLevel();
1202 if ( !delim_mode && ftell(fp)-offset >= l_max) {
1205 NewDocEntry = ReadNextDocEntry( );
1208 // NewDocEntry->Print(); cout << endl; //JPR
1209 vr = NewDocEntry->GetVR();
1213 if ( gdcmGlobal::GetVR()->IsVROfGdcmStringRepresentable(vr) )
1216 NewValEntry = new gdcmValEntry(NewDocEntry->GetDictEntry());
1217 NewValEntry->Copy(NewDocEntry);
1218 NewValEntry->SetDepthLevel(depth);
1219 set->AddEntry(NewValEntry);
1220 LoadDocEntry(NewValEntry);
1221 if (NewValEntry->isItemDelimitor())
1223 if ( !delim_mode && ftell(fp)-offset >= l_max)
1230 if ( ! gdcmGlobal::GetVR()->IsVROfGdcmBinaryRepresentable(vr) )
1232 ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR
1233 dbg.Verbose(0, "gdcmDocument::ParseDES: neither Valentry, "
1234 "nor BinEntry. Probably unknown VR.");
1237 ////// BinEntry or UNKOWN VR:
1238 bn = new gdcmBinEntry(NewDocEntry->GetDictEntry());
1239 bn->Copy(NewDocEntry);
1244 if (NewDocEntry->GetGroup() == 0x7fe0 &&
1245 NewDocEntry->GetElement() == 0x0010 )
1247 if (NewDocEntry->GetReadLength()==0xffffffff)
1249 // Broken US.3405.1.dcm
1250 Parse7FE0(); // to skip the pixels
1251 // (multipart JPEG/RLE are trouble makers)
1255 SkipToNextDocEntry(NewDocEntry);
1256 l = NewDocEntry->GetFullLength();
1261 // to be sure we are at the beginning
1262 SkipToNextDocEntry(NewDocEntry);
1263 l = NewDocEntry->GetFullLength();
1269 l=NewDocEntry->GetReadLength();
1270 if (l != 0) // don't mess the delim_mode for zero-length sequence
1271 if (l == 0xffffffff)
1275 // no other way to create it ...
1276 sq = new gdcmSeqEntry(NewDocEntry->GetDictEntry(),
1277 set->GetDepthLevel());
1278 sq->Copy(NewDocEntry);
1279 sq->SetDelimitorMode(delim_mode);
1280 sq->SetDepthLevel(depth);
1283 { // Don't try to parse zero-length sequences
1284 long lgt = ParseSQ( sq,
1285 NewDocEntry->GetOffset(),
1287 (void)lgt; //not used...
1290 if ( !delim_mode && ftell(fp)-offset >= l_max)
1297 return l; // Probably useless
1301 * \brief Parses a Sequence ( SeqEntry after SeqEntry)
1302 * @return parsed length for this level
1304 long gdcmDocument::ParseSQ(gdcmSeqEntry *set,
1305 long offset, long l_max, bool delim_mode)
1307 int SQItemNumber = 0;
1309 gdcmDocEntry *NewDocEntry = (gdcmDocEntry *)0;
1314 int depth = set->GetDepthLevel();
1315 (void)depth; //not used
1319 NewDocEntry = ReadNextDocEntry();
1323 if (NewDocEntry->isSequenceDelimitor()) {
1324 set->SetSequenceDelimitationItem(NewDocEntry);
1328 if (!delim_mode && (ftell(fp)-offset) >= l_max) {
1332 itemSQ = new gdcmSQItem(set->GetDepthLevel());
1333 itemSQ->AddEntry(NewDocEntry);
1334 l= NewDocEntry->GetReadLength();
1336 if (l == 0xffffffff)
1341 lgr=ParseDES(itemSQ, NewDocEntry->GetOffset(), l, dlm_mod);
1343 set->AddEntry(itemSQ,SQItemNumber);
1345 if (!delim_mode && (ftell(fp)-offset) >= l_max) {
1349 lgth = ftell(fp) - offset;
1354 * \brief Loads the element content if its length doesn't exceed
1355 * the value specified with gdcmDocument::SetMaxSizeLoadEntry()
1356 * @param Entry Header Entry (Dicom Element) to be dealt with
1358 void gdcmDocument::LoadDocEntry(gdcmDocEntry *Entry)
1361 guint16 group = Entry->GetGroup();
1362 std::string vr= Entry->GetVR();
1363 guint32 length = Entry->GetLength();
1365 fseek(fp, (long)Entry->GetOffset(), SEEK_SET);
1367 // A SeQuence "contains" a set of Elements.
1368 // (fffe e000) tells us an Element is beginning
1369 // (fffe e00d) tells us an Element just ended
1370 // (fffe e0dd) tells us the current SeQuence just ended
1371 if( group == 0xfffe ) {
1372 // NO more value field for SQ !
1376 // When the length is zero things are easy:
1377 if ( length == 0 ) {
1378 ((gdcmValEntry *)Entry)->SetValue("");
1382 // The elements whose length is bigger than the specified upper bound
1383 // are not loaded. Instead we leave a short notice of the offset of
1384 // the element content and it's length.
1386 std::ostringstream s;
1387 if (length > MaxSizeLoadEntry) {
1388 if (gdcmBinEntry* BinEntryPtr = dynamic_cast< gdcmBinEntry* >(Entry) )
1390 s << "gdcm::NotLoaded (BinEntry)";
1391 s << " Address:" << (long)Entry->GetOffset();
1392 s << " Length:" << Entry->GetLength();
1393 s << " x(" << std::hex << Entry->GetLength() << ")";
1394 BinEntryPtr->SetValue(s.str());
1396 // to be sure we are at the end of the value ...
1397 fseek(fp,(long)Entry->GetOffset()+(long)Entry->GetLength(),SEEK_SET);
1399 // Be carefull : a BinEntry IS_A ValEntry ...
1400 if (gdcmValEntry* ValEntryPtr = dynamic_cast< gdcmValEntry* >(Entry) )
1402 s << "gdcm::NotLoaded. (ValEntry)";
1403 s << " Address:" << (long)Entry->GetOffset();
1404 s << " Length:" << Entry->GetLength();
1405 s << " x(" << std::hex << Entry->GetLength() << ")";
1406 ValEntryPtr->SetValue(s.str());
1408 // to be sure we are at the end of the value ...
1409 fseek(fp,(long)Entry->GetOffset()+(long)Entry->GetLength(),SEEK_SET);
1413 // When we find a BinEntry not very much can be done :
1414 if (gdcmBinEntry* BinEntryPtr = dynamic_cast< gdcmBinEntry* >(Entry) ) {
1416 LoadEntryVoidArea(BinEntryPtr);
1417 s << "gdcm::Loaded (BinEntry)";
1418 BinEntryPtr->SetValue(s.str());
1423 // Any compacter code suggested (?)
1424 if ( IsDocEntryAnInteger(Entry) ) {
1426 std::ostringstream s;
1428 // When short integer(s) are expected, read and convert the following
1429 // n *two characters properly i.e. as short integers as opposed to strings.
1430 // Elements with Value Multiplicity > 1
1431 // contain a set of integers (not a single one)
1432 if (vr == "US" || vr == "SS") {
1434 NewInt = ReadInt16();
1437 for (int i=1; i < nbInt; i++) {
1439 NewInt = ReadInt16();
1444 // When integer(s) are expected, read and convert the following
1445 // n * four characters properly i.e. as integers as opposed to strings.
1446 // Elements with Value Multiplicity > 1
1447 // contain a set of integers (not a single one)
1448 else if (vr == "UL" || vr == "SL") {
1450 NewInt = ReadInt32();
1453 for (int i=1; i < nbInt; i++) {
1455 NewInt = ReadInt32();
1460 #ifdef GDCM_NO_ANSI_STRING_STREAM
1461 s << std::ends; // to avoid oddities on Solaris
1462 #endif //GDCM_NO_ANSI_STRING_STREAM
1464 ((gdcmValEntry *)Entry)->SetValue(s.str());
1468 // We need an additional byte for storing \0 that is not on disk
1469 std::string NewValue(length,0);
1470 item_read = fread(&(NewValue[0]), (size_t)length, (size_t)1, fp);
1471 if (gdcmValEntry* ValEntry = dynamic_cast< gdcmValEntry* >(Entry) ) {
1472 if ( item_read != 1 ) {
1473 dbg.Verbose(1, "gdcmDocument::LoadElementValue","unread element value");
1474 ValEntry->SetValue("gdcm::UnRead");
1478 if( (vr == "UI") ) // Because of correspondance with the VR dic
1479 ValEntry->SetValue(NewValue.c_str());
1481 ValEntry->SetValue(NewValue);
1484 std::cout << "Should have a ValEntry, here !" << std::endl;
1491 * \brief Find the value Length of the passed Header Entry
1492 * @param Entry Header Entry whose length of the value shall be loaded.
1494 void gdcmDocument::FindDocEntryLength (gdcmDocEntry *Entry) {
1495 guint16 element = Entry->GetElement();
1496 //guint16 group = Entry->GetGroup(); //FIXME
1497 std::string vr = Entry->GetVR();
1501 if ( (Filetype == gdcmExplicitVR) && (! Entry->IsImplicitVR()) )
1503 if ( (vr=="OB") || (vr=="OW") || (vr=="SQ") || (vr=="UN") )
1505 // The following reserved two bytes (see PS 3.5-2001, section
1506 // 7.1.2 Data element structure with explicit vr p27) must be
1507 // skipped before proceeding on reading the length on 4 bytes.
1508 fseek(fp, 2L, SEEK_CUR);
1509 guint32 length32 = ReadInt32();
1511 if ( (vr == "OB") && (length32 == 0xffffffff) )
1513 Entry->SetLength(FindDocEntryLengthOB());
1516 FixDocEntryFoundLength(Entry, length32);
1520 // Length is encoded on 2 bytes.
1521 length16 = ReadInt16();
1523 // We can tell the current file is encoded in big endian (like
1524 // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag
1525 // and it's value is the one of the encoding of a big endian file.
1526 // In order to deal with such big endian encoded files, we have
1527 // (at least) two strategies:
1528 // * when we load the "Transfer Syntax" tag with value of big endian
1529 // encoding, we raise the proper flags. Then we wait for the end
1530 // of the META group (0x0002) among which is "Transfer Syntax",
1531 // before switching the swap code to big endian. We have to postpone
1532 // the switching of the swap code since the META group is fully encoded
1533 // in little endian, and big endian coding only starts at the next
1534 // group. The corresponding code can be hard to analyse and adds
1535 // many additional unnecessary tests for regular tags.
1536 // * the second strategy consists in waiting for trouble, that shall
1537 // appear when we find the first group with big endian encoding. This
1538 // is easy to detect since the length of a "Group Length" tag (the
1539 // ones with zero as element number) has to be of 4 (0x0004). When we
1540 // encounter 1024 (0x0400) chances are the encoding changed and we
1541 // found a group with big endian encoding.
1542 // We shall use this second strategy. In order to make sure that we
1543 // can interpret the presence of an apparently big endian encoded
1544 // length of a "Group Length" without committing a big mistake, we
1545 // add an additional check: we look in the already parsed elements
1546 // for the presence of a "Transfer Syntax" whose value has to be "big
1547 // endian encoding". When this is the case, chances are we have got our
1548 // hands on a big endian encoded file: we switch the swap code to
1549 // big endian and proceed...
1550 if ( (element == 0x0000) && (length16 == 0x0400) )
1552 if ( ! IsExplicitVRBigEndianTransferSyntax() )
1554 dbg.Verbose(0, "gdcmDocument::FindLength", "not explicit VR");
1559 SwitchSwapToBigEndian();
1560 // Restore the unproperly loaded values i.e. the group, the element
1561 // and the dictionary entry depending on them.
1562 guint16 CorrectGroup = SwapShort(Entry->GetGroup());
1563 guint16 CorrectElem = SwapShort(Entry->GetElement());
1564 gdcmDictEntry * NewTag = GetDictEntryByNumber(CorrectGroup,
1568 // This correct tag is not in the dictionary. Create a new one.
1569 NewTag = NewVirtualDictEntry(CorrectGroup, CorrectElem);
1571 // FIXME this can create a memory leaks on the old entry that be
1572 // left unreferenced.
1573 Entry->SetDictEntry(NewTag);
1576 // Heuristic: well, some files are really ill-formed.
1577 if ( length16 == 0xffff)
1580 // Length16= 0xffff means that we deal with
1581 // 'Unknown Length' Sequence
1583 FixDocEntryFoundLength(Entry, (guint32)length16);
1588 // Either implicit VR or a non DICOM conformal (see note below) explicit
1589 // VR that ommited the VR of (at least) this element. Farts happen.
1590 // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25
1591 // on Data elements "Implicit and Explicit VR Data Elements shall
1592 // not coexist in a Data Set and Data Sets nested within it".]
1593 // Length is on 4 bytes.
1595 FixDocEntryFoundLength(Entry, ReadInt32());
1601 * \brief Find the Value Representation of the current Dicom Element.
1604 void gdcmDocument::FindDocEntryVR( gdcmDocEntry *Entry)
1606 if (Filetype != gdcmExplicitVR)
1611 long PositionOnEntry = ftell(fp);
1612 // Warning: we believe this is explicit VR (Value Representation) because
1613 // we used a heuristic that found "UL" in the first tag. Alas this
1614 // doesn't guarantee that all the tags will be in explicit VR. In some
1615 // cases (see e-film filtered files) one finds implicit VR tags mixed
1616 // within an explicit VR file. Hence we make sure the present tag
1617 // is in explicit VR and try to fix things if it happens not to be
1620 (void)fread (&VR, (size_t)2,(size_t)1, fp);
1622 if(!CheckDocEntryVR(Entry,VR))
1624 fseek(fp, PositionOnEntry, SEEK_SET);
1625 // When this element is known in the dictionary we shall use, e.g. for
1626 // the semantics (see the usage of IsAnInteger), the VR proposed by the
1627 // dictionary entry. Still we have to flag the element as implicit since
1628 // we know now our assumption on expliciteness is not furfilled.
1630 if ( Entry->IsVRUnknown() )
1631 Entry->SetVR("Implicit");
1632 Entry->SetImplicitVR();
1637 * \brief Check the correspondance between the VR of the header entry
1638 * and the taken VR. If they are different, the header entry is
1639 * updated with the new VR.
1640 * @param Entry Header Entry to check
1641 * @param vr Dicom Value Representation
1642 * @return false if the VR is incorrect of if the VR isn't referenced
1643 * otherwise, it returns true
1645 bool gdcmDocument::CheckDocEntryVR(gdcmDocEntry *Entry, VRKey vr)
1647 char msg[100]; // for sprintf
1648 bool RealExplicit = true;
1650 // Assume we are reading a falsely explicit VR file i.e. we reached
1651 // a tag where we expect reading a VR but are in fact we read the
1652 // first to bytes of the length. Then we will interogate (through find)
1653 // the dicom_vr dictionary with oddities like "\004\0" which crashes
1654 // both GCC and VC++ implementations of the STL map. Hence when the
1655 // expected VR read happens to be non-ascii characters we consider
1656 // we hit falsely explicit VR tag.
1658 if ( (!isalpha(vr[0])) && (!isalpha(vr[1])) )
1659 RealExplicit = false;
1661 // CLEANME searching the dicom_vr at each occurence is expensive.
1662 // PostPone this test in an optional integrity check at the end
1663 // of parsing or only in debug mode.
1664 if ( RealExplicit && !gdcmGlobal::GetVR()->Count(vr) )
1665 RealExplicit= false;
1667 if ( !RealExplicit )
1669 // We thought this was explicit VR, but we end up with an
1670 // implicit VR tag. Let's backtrack.
1671 sprintf(msg,"Falsely explicit vr file (%04x,%04x)\n",
1672 Entry->GetGroup(),Entry->GetElement());
1673 dbg.Verbose(1, "gdcmDocument::FindVR: ",msg);
1674 if (Entry->GetGroup()%2 && Entry->GetElement() == 0x0000) { // Group length is UL !
1675 gdcmDictEntry* NewEntry = NewVirtualDictEntry(
1676 Entry->GetGroup(),Entry->GetElement(),
1677 "UL","FIXME","Group Length");
1678 Entry->SetDictEntry(NewEntry);
1683 if ( Entry->IsVRUnknown() )
1685 // When not a dictionary entry, we can safely overwrite the VR.
1686 if (Entry->GetElement() == 0x0000) { // Group length is UL !
1692 else if ( Entry->GetVR() != vr )
1694 // The VR present in the file and the dictionary disagree. We assume
1695 // the file writer knew best and use the VR of the file. Since it would
1696 // be unwise to overwrite the VR of a dictionary (since it would
1697 // compromise it's next user), we need to clone the actual DictEntry
1698 // and change the VR for the read one.
1699 gdcmDictEntry* NewEntry = NewVirtualDictEntry(
1700 Entry->GetGroup(),Entry->GetElement(),
1701 vr,"FIXME",Entry->GetName());
1702 Entry->SetDictEntry(NewEntry);
1708 * \brief Get the transformed value of the header entry. The VR value
1709 * is used to define the transformation to operate on the value
1710 * \warning NOT end user intended method !
1712 * @return Transformed entry value
1714 std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *Entry)
1716 if ( (IsDocEntryAnInteger(Entry)) && (Entry->IsImplicitVR()) )
1718 std::string val=((gdcmValEntry *)Entry)->GetValue();
1719 std::string vr=Entry->GetVR();
1720 guint32 length = Entry->GetLength();
1721 std::ostringstream s;
1724 // When short integer(s) are expected, read and convert the following
1725 // n * 2 bytes properly i.e. as a multivaluated strings
1726 // (each single value is separated fromthe next one by '\'
1727 // as usual for standard multivaluated filels
1728 // Elements with Value Multiplicity > 1
1729 // contain a set of short integers (not a single one)
1731 if (vr == "US" || vr == "SS")
1736 for (int i=0; i < nbInt; i++)
1740 NewInt16 = (val[2*i+0]&0xFF)+((val[2*i+1]&0xFF)<<8);
1741 NewInt16 = SwapShort(NewInt16);
1746 // When integer(s) are expected, read and convert the following
1747 // n * 4 bytes properly i.e. as a multivaluated strings
1748 // (each single value is separated fromthe next one by '\'
1749 // as usual for standard multivaluated filels
1750 // Elements with Value Multiplicity > 1
1751 // contain a set of integers (not a single one)
1752 else if (vr == "UL" || vr == "SL")
1757 for (int i=0; i < nbInt; i++)
1761 NewInt32= (val[4*i+0]&0xFF)+((val[4*i+1]&0xFF)<<8)+
1762 ((val[4*i+2]&0xFF)<<16)+((val[4*i+3]&0xFF)<<24);
1763 NewInt32=SwapLong(NewInt32);
1767 #ifdef GDCM_NO_ANSI_STRING_STREAM
1768 s << std::ends; // to avoid oddities on Solaris
1769 #endif //GDCM_NO_ANSI_STRING_STREAM
1773 return ((gdcmValEntry *)Entry)->GetValue();
1777 * \brief Get the reverse transformed value of the header entry. The VR
1778 * value is used to define the reverse transformation to operate on
1780 * \warning NOT end user intended method !
1782 * @return Reverse transformed entry value
1784 std::string gdcmDocument::GetDocEntryUnvalue(gdcmDocEntry *Entry)
1786 if ( (IsDocEntryAnInteger(Entry)) && (Entry->IsImplicitVR()) )
1788 std::string vr=Entry->GetVR();
1789 std::ostringstream s;
1790 std::vector<std::string> tokens;
1792 if (vr == "US" || vr == "SS")
1796 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1797 Tokenize (((gdcmValEntry *)Entry)->GetValue(), tokens, "\\");
1798 for (unsigned int i=0; i<tokens.size();i++)
1800 NewInt16 = atoi(tokens[i].c_str());
1801 s<<(NewInt16&0xFF)<<((NewInt16>>8)&0xFF);
1805 if (vr == "UL" || vr == "SL")
1809 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1810 Tokenize (((gdcmValEntry *)Entry)->GetValue(), tokens, "\\");
1811 for (unsigned int i=0; i<tokens.size();i++)
1813 NewInt32 = atoi(tokens[i].c_str());
1814 s<<(char)(NewInt32&0xFF)<<(char)((NewInt32>>8)&0xFF)
1815 <<(char)((NewInt32>>16)&0xFF)<<(char)((NewInt32>>24)&0xFF);
1820 #ifdef GDCM_NO_ANSI_STRING_STREAM
1821 s << std::ends; // to avoid oddities on Solaris
1822 #endif //GDCM_NO_ANSI_STRING_STREAM
1826 return ((gdcmValEntry *)Entry)->GetValue();
1830 * \brief Skip a given Header Entry
1831 * \warning NOT end user intended method !
1834 void gdcmDocument::SkipDocEntry(gdcmDocEntry *entry)
1836 SkipBytes(entry->GetLength());
1840 * \brief Skips to the begining of the next Header Entry
1841 * \warning NOT end user intended method !
1844 void gdcmDocument::SkipToNextDocEntry(gdcmDocEntry *entry)
1846 (void)fseek(fp, (long)(entry->GetOffset()), SEEK_SET);
1847 (void)fseek(fp, (long)(entry->GetReadLength()), SEEK_CUR);
1851 * \brief When the length of an element value is obviously wrong (because
1852 * the parser went Jabberwocky) one can hope improving things by
1853 * applying some heuristics.
1855 void gdcmDocument::FixDocEntryFoundLength(gdcmDocEntry *Entry,
1856 guint32 FoundLength)
1858 Entry->SetReadLength(FoundLength); // will be updated only if a bug is found
1859 if ( FoundLength == 0xffffffff) {
1863 guint16 gr =Entry->GetGroup();
1864 guint16 el =Entry->GetElement();
1866 if (FoundLength%2) {
1867 std::ostringstream s;
1868 s << "Warning : Tag with uneven length "
1870 << " in x(" << std::hex << gr << "," << el <<")" << std::dec;
1871 dbg.Verbose(0, s.str().c_str());
1874 //////// Fix for some naughty General Electric images.
1875 // Allthough not recent many such GE corrupted images are still present
1876 // on Creatis hard disks. Hence this fix shall remain when such images
1877 // are no longer in user (we are talking a few years, here)...
1878 // Note: XMedCom probably uses such a trick since it is able to read
1879 // those pesky GE images ...
1880 if (FoundLength == 13) { // Only happens for this length !
1881 if ( (Entry->GetGroup() != 0x0008)
1882 || ( (Entry->GetElement() != 0x0070)
1883 && (Entry->GetElement() != 0x0080) ) )
1886 Entry->SetReadLength(10); /// \todo a bug is to be fixed !?
1890 //////// Fix for some brain-dead 'Leonardo' Siemens images.
1891 // Occurence of such images is quite low (unless one leaves close to a
1892 // 'Leonardo' source. Hence, one might consider commenting out the
1893 // following fix on efficiency reasons.
1895 if ( (Entry->GetGroup() == 0x0009)
1896 && ( (Entry->GetElement() == 0x1113)
1897 || (Entry->GetElement() == 0x1114) ) )
1900 Entry->SetReadLength(4); /// \todo a bug is to be fixed !?
1903 //////// Deal with sequences, but only on users request:
1905 if ( ( Entry->GetVR() == "SQ") && enableSequences)
1907 FoundLength = 0; // ReadLength is unchanged
1910 //////// We encountered a 'delimiter' element i.e. a tag of the form
1911 // "fffe|xxxx" which is just a marker. Delimiters length should not be
1912 // taken into account.
1914 if(Entry->GetGroup() == 0xfffe)
1916 // According to the norm, fffe|0000 shouldn't exist. BUT the Philips
1917 // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to
1918 // causes extra troubles...
1919 if( Entry->GetElement() != 0x0000 )
1925 Entry->SetUsableLength(FoundLength);
1929 * \brief Apply some heuristics to predict whether the considered
1930 * element value contains/represents an integer or not.
1931 * @param Entry The element value on which to apply the predicate.
1932 * @return The result of the heuristical predicate.
1934 bool gdcmDocument::IsDocEntryAnInteger(gdcmDocEntry *Entry) {
1935 guint16 element = Entry->GetElement();
1936 guint16 group = Entry->GetGroup();
1937 std::string vr = Entry->GetVR();
1938 guint32 length = Entry->GetLength();
1940 // When we have some semantics on the element we just read, and if we
1941 // a priori know we are dealing with an integer, then we shall be
1942 // able to swap it's element value properly.
1943 if ( element == 0 ) // This is the group length of the group
1949 // Allthough this should never happen, still some images have a
1950 // corrupted group length [e.g. have a glance at offset x(8336) of
1951 // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm].
1952 // Since for dicom compliant and well behaved headers, the present
1953 // test is useless (and might even look a bit paranoid), when we
1954 // encounter such an ill-formed image, we simply display a warning
1955 // message and proceed on parsing (while crossing fingers).
1956 std::ostringstream s;
1957 int filePosition = ftell(fp);
1958 s << "Erroneous Group Length element length on : (" \
1959 << std::hex << group << " , " << element
1960 << ") -before- position x(" << filePosition << ")"
1961 << "lgt : " << length;
1962 dbg.Verbose(0, "gdcmDocument::IsDocEntryAnInteger", s.str().c_str() );
1966 if ( (vr == "UL") || (vr == "US") || (vr == "SL") || (vr == "SS") )
1973 * \brief Find the Length till the next sequence delimiter
1974 * \warning NOT end user intended method !
1978 guint32 gdcmDocument::FindDocEntryLengthOB(void) {
1979 // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
1982 long PositionOnEntry = ftell(fp);
1983 bool FoundSequenceDelimiter = false;
1984 guint32 TotalLength = 0;
1987 while ( ! FoundSequenceDelimiter)
1993 TotalLength += 4; // We even have to decount the group and element
1995 if ( g != 0xfffe && g!=0xb00c ) //for bogus header
1997 char msg[100]; // for sprintf. Sorry
1998 sprintf(msg,"wrong group (%04x) for an item sequence (%04x,%04x)\n",g, g,n);
1999 dbg.Verbose(1, "gdcmDocument::FindLengthOB: ",msg);
2003 if ( n == 0xe0dd || ( g==0xb00c && n==0x0eb6 ) ) // for bogus header
2004 FoundSequenceDelimiter = true;
2005 else if ( n != 0xe000 )
2007 char msg[100]; // for sprintf. Sorry
2008 sprintf(msg,"wrong element (%04x) for an item sequence (%04x,%04x)\n",
2010 dbg.Verbose(1, "gdcmDocument::FindLengthOB: ",msg);
2014 ItemLength = ReadInt32();
2015 TotalLength += ItemLength + 4; // We add 4 bytes since we just read
2016 // the ItemLength with ReadInt32
2017 SkipBytes(ItemLength);
2019 fseek(fp, PositionOnEntry, SEEK_SET);
2024 * \brief Reads a supposed to be 16 Bits integer
2025 * (swaps it depending on processor endianity)
2026 * @return read value
2028 guint16 gdcmDocument::ReadInt16() {
2031 item_read = fread (&g, (size_t)2,(size_t)1, fp);
2032 if ( item_read != 1 ) {
2034 dbg.Verbose(0, "gdcmDocument::ReadInt16", " File Error");
2044 * \brief Reads a supposed to be 32 Bits integer
2045 * (swaps it depending on processor endianity)
2046 * @return read value
2048 guint32 gdcmDocument::ReadInt32() {
2051 item_read = fread (&g, (size_t)4,(size_t)1, fp);
2052 if ( item_read != 1 ) {
2054 dbg.Verbose(0, "gdcmDocument::ReadInt32", " File Error");
2064 * \brief skips bytes inside the source file
2065 * \warning NOT end user intended method !
2068 void gdcmDocument::SkipBytes(guint32 NBytes) {
2069 //FIXME don't dump the returned value
2070 (void)fseek(fp, (long)NBytes, SEEK_CUR);
2074 * \brief Loads all the needed Dictionaries
2075 * \warning NOT end user intended method !
2077 void gdcmDocument::Initialise(void)
2079 RefPubDict = gdcmGlobal::GetDicts()->GetDefaultPubDict();
2084 * \brief Discover what the swap code is (among little endian, big endian,
2085 * bad little endian, bad big endian).
2087 * @return false when we are absolutely sure
2088 * it's neither ACR-NEMA nor DICOM
2089 * true when we hope ours assuptions are OK
2091 bool gdcmDocument::CheckSwap() {
2093 // The only guaranted way of finding the swap code is to find a
2094 // group tag since we know it's length has to be of four bytes i.e.
2095 // 0x00000004. Finding the swap code in then straigthforward. Trouble
2096 // occurs when we can't find such group...
2098 guint32 x=4; // x : for ntohs
2099 bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
2105 char deb[HEADER_LENGTH_TO_READ];
2107 // First, compare HostByteOrder and NetworkByteOrder in order to
2108 // determine if we shall need to swap bytes (i.e. the Endian type).
2114 // The easiest case is the one of a DICOM header, since it possesses a
2115 // file preamble where it suffice to look for the string "DICM".
2116 lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, fp);
2119 if(memcmp(entCur, "DICM", (size_t)4) == 0) {
2120 dbg.Verbose(1, "gdcmDocument::CheckSwap:", "looks like DICOM Version3");
2122 // Next, determine the value representation (VR). Let's skip to the
2123 // first element (0002, 0000) and check there if we find "UL"
2124 // - or "OB" if the 1st one is (0002,0001) -,
2125 // in which case we (almost) know it is explicit VR.
2126 // WARNING: if it happens to be implicit VR then what we will read
2127 // is the length of the group. If this ascii representation of this
2128 // length happens to be "UL" then we shall believe it is explicit VR.
2129 // FIXME: in order to fix the above warning, we could read the next
2130 // element value (or a couple of elements values) in order to make
2131 // sure we are not commiting a big mistake.
2132 // We need to skip :
2133 // * the 128 bytes of File Preamble (often padded with zeroes),
2134 // * the 4 bytes of "DICM" string,
2135 // * the 4 bytes of the first tag (0002, 0000),or (0002, 0001)
2136 // i.e. a total of 136 bytes.
2140 // Sometimes (see : gdcmData/icone.dcm) group 0x0002 *is* Explicit VR,
2141 // but elem 0002,0010 (Transfert Syntax) tells us the file is
2142 // *Implicit* VR. -and it is !-
2144 if( (memcmp(entCur, "UL", (size_t)2) == 0) ||
2145 (memcmp(entCur, "OB", (size_t)2) == 0) ||
2146 (memcmp(entCur, "UI", (size_t)2) == 0) ||
2147 (memcmp(entCur, "CS", (size_t)2) == 0) ) // CS, to remove later
2148 // when Write DCM *adds*
2150 // Use gdcmDocument::dicom_vr to test all the possibilities
2151 // instead of just checking for UL, OB and UI !? group 0000
2153 Filetype = gdcmExplicitVR;
2154 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2155 "explicit Value Representation");
2159 Filetype = gdcmImplicitVR;
2160 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2161 "not an explicit Value Representation");
2167 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2168 "HostByteOrder != NetworkByteOrder");
2173 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2174 "HostByteOrder = NetworkByteOrder");
2177 // Position the file position indicator at first tag (i.e.
2178 // after the file preamble and the "DICM" string).
2180 fseek (fp, 132L, SEEK_SET);
2184 // Alas, this is not a DicomV3 file and whatever happens there is no file
2185 // preamble. We can reset the file position indicator to where the data
2186 // is (i.e. the beginning of the file).
2187 dbg.Verbose(1, "gdcmDocument::CheckSwap:", "not a DICOM Version3 file");
2190 // Our next best chance would be to be considering a 'clean' ACR/NEMA file.
2191 // By clean we mean that the length of the first tag is written down.
2192 // If this is the case and since the length of the first group HAS to be
2193 // four (bytes), then determining the proper swap code is straightforward.
2196 // We assume the array of char we are considering contains the binary
2197 // representation of a 32 bits integer. Hence the following dirty
2199 s32 = *((guint32 *)(entCur));
2220 // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
2221 // It is time for despaired wild guesses.
2222 // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA,
2223 // i.e. the 'group length' element is not present :
2225 // check the supposed to be 'group number'
2226 // 0x0002 or 0x0004 or 0x0008
2227 // to determine ' sw' value .
2228 // Only 0 or 4321 will be possible
2229 // (no oportunity to check for the formerly well known
2230 // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian'
2231 // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -4, 8-)
2232 // the file IS NOT ACR-NEMA nor DICOM V3
2233 // Find a trick to tell it the caller...
2235 s16 = *((guint16 *)(deb));
2251 dbg.Verbose(0, "gdcmDocument::CheckSwap:",
2252 "ACR/NEMA unfound swap info (Really hopeless !)");
2253 Filetype = gdcmUnknown;
2257 // Then the only info we have is the net2host one.
2267 * \brief Restore the unproperly loaded values i.e. the group, the element
2268 * and the dictionary entry depending on them.
2270 void gdcmDocument::SwitchSwapToBigEndian(void)
2272 dbg.Verbose(1, "gdcmDocument::SwitchSwapToBigEndian",
2273 "Switching to BigEndian mode.");
2294 * \brief during parsing, Header Elements too long are not loaded in memory
2297 void gdcmDocument::SetMaxSizeLoadEntry(long NewSize)
2301 if ((guint32)NewSize >= (guint32)0xffffffff)
2303 MaxSizeLoadEntry = 0xffffffff;
2306 MaxSizeLoadEntry = NewSize;
2311 * \brief Header Elements too long will not be printed
2312 * \todo See comments of \ref gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE
2315 void gdcmDocument::SetMaxSizePrintEntry(long NewSize)
2319 if ((guint32)NewSize >= (guint32)0xffffffff)
2321 MaxSizePrintEntry = 0xffffffff;
2324 MaxSizePrintEntry = NewSize;
2330 * \brief Read the next tag but WITHOUT loading it's value
2331 * (read the 'Group Number', the 'Element Number',
2332 * gets the Dict Entry
2333 * gets the VR, gets the length, gets the offset value)
2334 * @return On succes the newly created DocEntry, NULL on failure.
2336 gdcmDocEntry *gdcmDocument::ReadNextDocEntry(void) {
2338 gdcmDocEntry *NewEntry;
2343 // We reached the EOF (or an error occured) therefore
2344 // header parsing has to be considered as finished.
2345 return (gdcmDocEntry *)0;
2347 NewEntry = NewDocEntryByNumber(g, n);
2348 FindDocEntryVR(NewEntry);
2349 FindDocEntryLength(NewEntry);
2356 NewEntry->SetOffset(ftell(fp));
2362 * \brief Generate a free TagKey i.e. a TagKey that is not present
2363 * in the TagHt dictionary.
2364 * @param group The generated tag must belong to this group.
2365 * @return The element of tag with given group which is fee.
2367 guint32 gdcmDocument::GenerateFreeTagKeyInGroup(guint16 group)
2369 for (guint32 elem = 0; elem < UINT32_MAX; elem++)
2371 TagKey key = gdcmDictEntry::TranslateToKey(group, elem);
2372 if (tagHT.count(key) == 0)
2380 * \brief Assuming the internal file pointer \ref gdcmDocument::fp
2381 * is placed at the beginning of a tag (TestGroup, TestElement),
2382 * read the length associated to the Tag.
2383 * \warning On success the internal file pointer \ref gdcmDocument::fp
2384 * is modified to point after the tag and it's length.
2385 * On failure (i.e. when the tag wasn't the expected tag
2386 * (TestGroup, TestElement) the internal file pointer
2387 * \ref gdcmDocument::fp is restored to it's original position.
2388 * @param TestGroup The expected group of the tag.
2389 * @param TestElement The expected Element of the tag.
2390 * @return On success returns the length associated to the tag. On failure
2393 guint32 gdcmDocument::ReadTagLength(guint16 TestGroup, guint16 TestElement)
2395 guint16 ItemTagGroup;
2396 guint16 ItemTagElement;
2397 long PositionOnEntry = ftell(fp);
2398 long CurrentPosition = ftell(fp); // On debugging purposes
2400 //// Read the Item Tag group and element, and make
2401 // sure they are respectively 0xfffe and 0xe000:
2402 ItemTagGroup = ReadInt16();
2403 ItemTagElement = ReadInt16();
2404 if ( (ItemTagGroup != TestGroup) || (ItemTagElement != TestElement ) )
2406 std::ostringstream s;
2407 s << " We should have found tag (";
2408 s << std::hex << TestGroup << "," << TestElement << ")" << std::endl;
2409 s << " but instead we encountered tag (";
2410 s << std::hex << ItemTagGroup << "," << ItemTagElement << ")"
2412 s << " at address: " << (unsigned)CurrentPosition << std::endl;
2413 dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: wrong Item Tag found:");
2414 dbg.Verbose(0, s.str().c_str());
2415 fseek(fp, PositionOnEntry, SEEK_SET);
2419 //// Then read the associated Item Length
2420 CurrentPosition=ftell(fp);
2422 ItemLength = ReadInt32();
2424 std::ostringstream s;
2425 s << "Basic Item Length is: "
2426 << ItemLength << std::endl;
2427 s << " at address: " << (unsigned)CurrentPosition << std::endl;
2428 dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: ", s.str().c_str());
2434 * \brief Read the length of an exptected Item tag i.e. (0xfffe, 0xe000).
2435 * \sa \ref gdcmDocument::ReadTagLength
2436 * \warning See warning of \ref gdcmDocument::ReadTagLength
2437 * @return On success returns the length associated to the item tag.
2438 * On failure returns 0.
2440 guint32 gdcmDocument::ReadItemTagLength(void)
2442 return ReadTagLength(0xfffe, 0xe000);
2446 * \brief Read the length of an expected Sequence Delimiter tag i.e.
2448 * \sa \ref gdcmDocument::ReadTagLength
2449 * \warning See warning of \ref gdcmDocument::ReadTagLength
2450 * @return On success returns the length associated to the Sequence
2451 * Delimiter tag. On failure returns 0.
2453 guint32 gdcmDocument::ReadSequenceDelimiterTagLength(void)
2455 return ReadTagLength(0xfffe, 0xe0dd);
2460 * \brief Parse pixel data from disk for multi-fragment Jpeg/Rle files
2461 * No other way so 'skip' the Data
2465 void gdcmDocument::Parse7FE0 (void)
2467 gdcmDocEntry* Element = GetDocEntryByNumber(0x0002, 0x0010);
2471 if ( IsImplicitVRLittleEndianTransferSyntax()
2472 || IsExplicitVRLittleEndianTransferSyntax()
2473 || IsExplicitVRBigEndianTransferSyntax() /// \todo 1.2.2 ??? A verifier !
2474 || IsDeflatedExplicitVRLittleEndianTransferSyntax() )
2477 // ---------------- for Parsing : Position on begining of Jpeg/RLE Pixels
2479 //// Read the Basic Offset Table Item Tag length...
2480 guint32 ItemLength = ReadItemTagLength();
2482 //// ... and then read length[s] itself[themselves]. We don't use
2483 // the values read (BTW what is the purpous of those lengths ?)
2484 if (ItemLength != 0) {
2485 // BTW, what is the purpous of those length anyhow !?
2486 char * BasicOffsetTableItemValue = new char[ItemLength + 1];
2487 fread(BasicOffsetTableItemValue, ItemLength, 1, fp);
2488 for (unsigned int i=0; i < ItemLength; i += 4){
2489 guint32 IndividualLength;
2490 IndividualLength = str2num(&BasicOffsetTableItemValue[i],guint32);
2491 std::ostringstream s;
2492 s << " Read one length: ";
2493 s << std::hex << IndividualLength << std::endl;
2494 dbg.Verbose(0, "gdcmDocument::Parse7FE0: ", s.str().c_str());
2498 if ( ! IsRLELossLessTransferSyntax() )
2502 //// We then skip (not reading them) all the fragments of images:
2503 while ( (ItemLength = ReadItemTagLength()) )
2505 SkipBytes(ItemLength);
2513 long RleSegmentLength[15], fragmentLength;
2515 // while 'Sequence Delimiter Item' (fffe,e0dd) not found
2516 while ( (fragmentLength = ReadSequenceDelimiterTagLength()) )
2518 // Parse fragments of the current Fragment (Frame)
2519 //------------------ scanning (not reading) fragment pixels
2520 guint32 nbRleSegments = ReadInt32();
2521 printf(" Nb of RLE Segments : %d\n",nbRleSegments);
2523 //// Reading RLE Segments Offset Table
2524 guint32 RleSegmentOffsetTable[15];
2525 for(int k=1; k<=15; k++) {
2527 RleSegmentOffsetTable[k] = ReadInt32();
2528 printf(" at : %x Offset Segment %d : %d (%x)\n",
2529 (unsigned)ftellRes,k,RleSegmentOffsetTable[k],
2530 RleSegmentOffsetTable[k]);
2533 // skipping (not reading) RLE Segments
2534 if (nbRleSegments>1) {
2535 for(unsigned int k=1; k<=nbRleSegments-1; k++) {
2536 RleSegmentLength[k]= RleSegmentOffsetTable[k+1]
2537 - RleSegmentOffsetTable[k];
2539 printf (" Segment %d : Length = %d x(%x) Start at %x\n",
2540 k,(unsigned)RleSegmentLength[k],
2541 (unsigned)RleSegmentLength[k], (unsigned)ftellRes);
2542 SkipBytes(RleSegmentLength[k]);
2546 RleSegmentLength[nbRleSegments]= fragmentLength
2547 - RleSegmentOffsetTable[nbRleSegments];
2549 printf (" Segment %d : Length = %d x(%x) Start at %x\n",
2550 nbRleSegments,(unsigned)RleSegmentLength[nbRleSegments],
2551 (unsigned)RleSegmentLength[nbRleSegments],(unsigned)ftellRes);
2552 SkipBytes(RleSegmentLength[nbRleSegments]);
2560 * \brief Compares two documents, according to \ref gdcmDicomDir rules
2561 * \warning Does NOT work with ACR-NEMA files
2562 * \todo Find a trick to solve the pb (use RET fields ?)
2564 * @return true if 'smaller'
2566 bool gdcmDocument::operator<(gdcmDocument &document)
2571 s1=this->GetEntryByNumber(0x0010,0x0010);
2572 s2=document.GetEntryByNumber(0x0010,0x0010);
2580 s1=this->GetEntryByNumber(0x0010,0x0020);
2581 s2=document.GetEntryByNumber(0x0010,0x0020);
2588 // Study Instance UID
2589 s1=this->GetEntryByNumber(0x0020,0x000d);
2590 s2=document.GetEntryByNumber(0x0020,0x000d);
2597 // Serie Instance UID
2598 s1=this->GetEntryByNumber(0x0020,0x000e);
2599 s2=document.GetEntryByNumber(0x0020,0x000e);
2611 //-----------------------------------------------------------------------------