2 //-----------------------------------------------------------------------------
3 #include "gdcmParser.h"
11 #include <netinet/in.h>
14 #ifdef GDCM_NO_ANSI_STRING_STREAM
16 # define ostringstream ostrstream
22 #define UI1_2_840_10008_1_2 "1.2.840.10008.1.2"
23 #define UI1_2_840_10008_1_2_1 "1.2.840.10008.1.2.1"
24 #define UI1_2_840_10008_1_2_2 "1.2.840.10008.1.2.2"
25 #define UI1_2_840_10008_1_2_1_99 "1.2.840.10008.1.2.1.99"
29 // ---> Warning : This fourth field is NOT part
30 // of the 'official' Dicom Dictionnary
31 // and should NOT be used.
32 // (Not defined for all the groups
33 // may be removed in a future release)
36 // META Meta Information
48 // NMI Nuclear Medicine
50 // BFS Basic Film Session
52 // BIB Basic Image Box
67 //-----------------------------------------------------------------------------
68 // Refer to gdcmParser::CheckSwap()
69 const unsigned int gdcmParser::HEADER_LENGTH_TO_READ = 256;
71 // Refer to gdcmParser::SetMaxSizeLoadEntry()
72 const unsigned int gdcmParser::MAX_SIZE_LOAD_ELEMENT_VALUE = 4096;
74 // Refer to gdcmParser::SetMaxSizePrintEntry()
75 // TODO : Right now, better see "define, in gdcmHederEntry.cxx
76 const unsigned int gdcmParser::MAX_SIZE_PRINT_ELEMENT_VALUE = 64;
78 //-----------------------------------------------------------------------------
79 // Constructor / Destructor
84 * @param exception_on_error
85 * @param enable_sequences = true to allow the header
86 * to be parsed *inside* the SeQuences,
87 * when they have an actual length
88 * @param ignore_shadow to allow skipping the shadow elements,
89 * to save memory space.
90 * \warning The TRUE value for this param has to be used
91 * with a FALSE value for the 'enable_sequence' param.
92 * ('public elements' may be embedded in 'shadow Sequences')
94 gdcmParser::gdcmParser(const char *InFilename,
95 bool exception_on_error,
96 bool enable_sequences,
98 enableSequences=enable_sequences;
99 ignoreShadow =ignore_shadow;
101 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
102 filename = InFilename;
105 if ( !OpenFile(exception_on_error))
112 wasUpdated = 0; // will be set to 1 if user adds an entry
113 printLevel = 1; // 'Medium' print level by default
117 * \ingroup gdcmParser
119 * @param exception_on_error
121 gdcmParser::gdcmParser(bool exception_on_error) {
124 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
127 wasUpdated = 0; // will be set to 1 if user adds an entry
128 printLevel = 1; // 'Medium' print level by default
132 * \ingroup gdcmParser
133 * \brief Canonical destructor.
135 gdcmParser::~gdcmParser (void) {
140 //-----------------------------------------------------------------------------
143 * \ingroup gdcmParser
144 * \brief Prints the Header Entries (Dicom Elements)
145 * both from the H Table and the chained list
148 void gdcmParser::PrintEntry(std::ostream & os) {
149 std::ostringstream s;
151 for (ListTag::iterator i = listEntries.begin();
152 i != listEntries.end();
155 (*i)->SetPrintLevel(printLevel);
162 * \ingroup gdcmParser
163 * \brief Prints The Dict Entries of THE public Dicom Dictionnry
166 void gdcmParser::PrintPubDict(std::ostream & os) {
167 RefPubDict->Print(os);
171 * \ingroup gdcmParser
172 * \brief Prints The Dict Entries of THE shadow Dicom Dictionnry
175 void gdcmParser::PrintShaDict(std::ostream & os) {
176 RefShaDict->Print(os);
179 //-----------------------------------------------------------------------------
182 * \ingroup gdcmParser
183 * \brief Get the public dictionary used
185 gdcmDict *gdcmParser::GetPubDict(void) {
190 * \ingroup gdcmParser
191 * \brief Get the shadow dictionary used
193 gdcmDict *gdcmParser::GetShaDict(void) {
198 * \ingroup gdcmParser
199 * \brief Set the shadow dictionary used
200 * \param dict dictionary to use in shadow
202 bool gdcmParser::SetShaDict(gdcmDict *dict){
208 * \ingroup gdcmParser
209 * \brief Set the shadow dictionary used
210 * \param dictName name of the dictionary to use in shadow
212 bool gdcmParser::SetShaDict(DictKey dictName){
213 RefShaDict=gdcmGlobal::GetDicts()->GetDict(dictName);
218 * \ingroup gdcmParser
219 * \brief This predicate, based on hopefully reasonable heuristics,
220 * decides whether or not the current gdcmParser was properly parsed
221 * and contains the mandatory information for being considered as
222 * a well formed and usable Dicom/Acr File.
223 * @return true when gdcmParser is the one of a reasonable Dicom/Acr file,
226 bool gdcmParser::IsReadable(void) {
227 if(filetype==Unknown) {
230 if(listEntries.size()<=0) {
238 * \ingroup gdcmParser
239 * \brief Determines if the Transfer Syntax was already encountered
240 * and if it corresponds to a ImplicitVRLittleEndian one.
242 * @return True when ImplicitVRLittleEndian found. False in all other cases.
244 bool gdcmParser::IsImplicitVRLittleEndianTransferSyntax(void) {
245 gdcmHeaderEntry *Element = GetHeaderEntryByNumber(0x0002, 0x0010);
248 LoadHeaderEntrySafe(Element);
250 std::string Transfer = Element->GetValue();
251 if ( Transfer == UI1_2_840_10008_1_2 )
257 * \ingroup gdcmParser
258 * \brief Determines if the Transfer Syntax was already encountered
259 * and if it corresponds to a ExplicitVRLittleEndian one.
261 * @return True when ExplicitVRLittleEndian found. False in all other cases.
263 bool gdcmParser::IsExplicitVRLittleEndianTransferSyntax(void) {
264 gdcmHeaderEntry* Element = GetHeaderEntryByNumber(0x0002, 0x0010);
267 LoadHeaderEntrySafe(Element);
269 std::string Transfer = Element->GetValue();
270 if ( Transfer == UI1_2_840_10008_1_2_1 )
276 * \ingroup gdcmParser
277 * \brief Determines if the Transfer Syntax was already encountered
278 * and if it corresponds to a DeflatedExplicitVRLittleEndian one.
280 * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases.
282 bool gdcmParser::IsDeflatedExplicitVRLittleEndianTransferSyntax(void) {
283 gdcmHeaderEntry* Element = GetHeaderEntryByNumber(0x0002, 0x0010);
286 LoadHeaderEntrySafe(Element);
288 std::string Transfer = Element->GetValue();
289 if ( Transfer == UI1_2_840_10008_1_2_1_99 )
295 * \ingroup gdcmParser
296 * \brief Determines if the Transfer Syntax was already encountered
297 * and if it corresponds to a Explicit VR Big Endian one.
299 * @return True when big endian found. False in all other cases.
301 bool gdcmParser::IsExplicitVRBigEndianTransferSyntax(void) {
302 gdcmHeaderEntry* Element = GetHeaderEntryByNumber(0x0002, 0x0010);
305 LoadHeaderEntrySafe(Element);
307 std::string Transfer = Element->GetValue();
308 if ( Transfer == UI1_2_840_10008_1_2_2 ) //1.2.2 ??? A verifier !
314 * \ingroup gdcmParser
315 * \brief returns the File Type
316 * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
319 FileType gdcmParser::GetFileType(void) {
324 * \ingroup gdcmParser
325 * \brief opens the file
326 * @param exception_on_error
329 FILE *gdcmParser::OpenFile(bool exception_on_error)
332 fp=fopen(filename.c_str(),"rb");
333 if(exception_on_error)
336 throw gdcmFileError("gdcmParser::gdcmParser(const char *, bool)");
342 fread(&zero, (size_t)2, (size_t)1, fp);
344 //ACR -- or DICOM with no Preamble --
345 if( zero == 0x0008 || zero == 0x0800 || zero == 0x0002 || zero == 0x0200)
349 fseek(fp, 126L, SEEK_CUR);
351 fread(dicm, (size_t)4, (size_t)1, fp);
352 if( memcmp(dicm, "DICM", 4) == 0 )
356 dbg.Verbose(0, "gdcmParser::gdcmParser not DICOM/ACR", filename.c_str());
359 dbg.Verbose(0, "gdcmParser::gdcmParser cannot open file", filename.c_str());
365 * \ingroup gdcmParser
366 * \brief closes the file
367 * @return TRUE if the close was successfull
369 bool gdcmParser::CloseFile(void) {
370 int closed = fclose(fp);
378 * \ingroup gdcmParser
380 * @param fp file pointer on an already open file
381 * @param type type of the File to be written
382 * (ACR-NEMA, ExplicitVR, ImplicitVR)
383 * @return always "True" ?!
385 bool gdcmParser::Write(FILE *fp, FileType type) {
387 // TODO The stuff was rewritten using the chained list instead
389 // so we could remove the GroupHT from the gdcmParser
393 // TODO : move the following lines (and a lot of others, to be written)
394 // to a future function CheckAndCorrectHeader
397 // Comment pourrait-on savoir si le DcmHeader vient d'un fichier DicomV3 ou non
398 // (FileType est un champ de gdcmParser ...)
399 // WARNING : Si on veut ecrire du DICOM V3 a partir d'un DcmHeader ACR-NEMA
401 // a moins de se livrer a un tres complique ajout des champs manquants.
402 // faire un CheckAndCorrectHeader (?)
405 if (type == ImplicitVR)
407 std::string implicitVRTransfertSyntax = UI1_2_840_10008_1_2;
408 ReplaceOrCreateByNumber(implicitVRTransfertSyntax,0x0002, 0x0010);
410 //FIXME Refer to standards on page 21, chapter 6.2 "Value representation":
411 // values with a VR of UI shall be padded with a single trailing null
412 // Dans le cas suivant on doit pader manuellement avec un 0
414 SetEntryLengthByNumber(18, 0x0002, 0x0010);
417 if (type == ExplicitVR)
419 std::string explicitVRTransfertSyntax = UI1_2_840_10008_1_2_1;
420 ReplaceOrCreateByNumber(explicitVRTransfertSyntax,0x0002, 0x0010);
422 //FIXME Refer to standards on page 21, chapter 6.2 "Value representation":
423 // values with a VR of UI shall be padded with a single trailing null
424 // Dans le cas suivant on doit pader manuellement avec un 0
426 SetEntryLengthByNumber(20, 0x0002, 0x0010);
429 /* TODO : rewrite later
431 if ( (type == ImplicitVR) || (type == ExplicitVR) )
432 UpdateGroupLength(false,type);
434 UpdateGroupLength(true,ACR);
437 WriteEntries(type, fp);
442 * \ingroup gdcmParser
443 * \brief Modifies the value of a given Header Entry (Dicom Element)
444 * if it exists; Creates it with the given value if it doesn't
445 * @param Value passed as a std::string
450 bool gdcmParser::ReplaceOrCreateByNumber(std::string Value,
453 if (CheckIfEntryExistByNumber(Group, Elem) == 0) {
454 gdcmHeaderEntry *a =NewHeaderEntryByNumber(Group, Elem);
459 SetEntryByNumber(Value, Group, Elem);
464 * \ingroup gdcmParser
465 * \brief Modifies the value of a given Header Entry (Dicom Element)
466 * if it exists; Creates it with the given value if it doesn't
467 * @param Value passed as a char*
473 bool gdcmParser::ReplaceOrCreateByNumber(char* Value, guint16 Group, guint16 Elem ) {
474 gdcmHeaderEntry* nvHeaderEntry=NewHeaderEntryByNumber(Group, Elem);
479 AddHeaderEntry(nvHeaderEntry);
481 std::string v = Value;
482 SetEntryByNumber(v, Group, Elem);
487 * \ingroup gdcmParser
488 * \brief Set a new value if the invoked element exists
489 * Seems to be useless !!!
495 bool gdcmParser::ReplaceIfExistByNumber(char* Value, guint16 Group, guint16 Elem )
497 std::string v = Value;
498 SetEntryByNumber(v, Group, Elem);
502 //-----------------------------------------------------------------------------
505 * \ingroup gdcmParser
506 * \brief Checks if a given Dicom Element exists
507 * \ within the H table
508 * @param group Group number of the searched Dicom Element
509 * @param element Element number of the searched Dicom Element
510 * @return number of occurences
512 int gdcmParser::CheckIfEntryExistByNumber(guint16 group, guint16 element ) {
513 std::string key = gdcmDictEntry::TranslateToKey(group, element );
514 return (tagHT.count(key));
518 * \ingroup gdcmParser
519 * \brief Searches within Header Entries (Dicom Elements) parsed with
520 * the public and private dictionaries
521 * for the element value of a given tag.
522 * \warning Don't use any longer : use GetPubEntryByName
523 * @param tagName name of the searched element.
524 * @return Corresponding element value when it exists,
525 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
527 std::string gdcmParser::GetEntryByName(std::string tagName) {
528 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
529 if( dictEntry == NULL)
532 return(GetEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement()));
536 * \ingroup gdcmParser
537 * \brief Searches within Header Entries (Dicom Elements) parsed with
538 * the public and private dictionaries
539 * for the element value representation of a given tag.
541 * Obtaining the VR (Value Representation) might be needed by caller
542 * to convert the string typed content to caller's native type
543 * (think of C++ vs Python). The VR is actually of a higher level
544 * of semantics than just the native C++ type.
545 * @param tagName name of the searched element.
546 * @return Corresponding element value representation when it exists,
547 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
549 std::string gdcmParser::GetEntryVRByName(std::string tagName) {
550 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
551 if( dictEntry == NULL)
554 gdcmHeaderEntry* elem = GetHeaderEntryByNumber(dictEntry->GetGroup(),
555 dictEntry->GetElement());
556 return elem->GetVR();
560 * \ingroup gdcmParser
561 * \brief Searches within Header Entries (Dicom Elements) parsed with
562 * the public and private dictionaries
563 * for the element value representation of a given tag.
564 * @param group Group of the searched tag.
565 * @param element Element of the searched tag.
566 * @return Corresponding element value representation when it exists,
567 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
569 std::string gdcmParser::GetEntryByNumber(guint16 group, guint16 element){
570 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
571 if ( ! tagHT.count(key))
573 return tagHT.find(key)->second->GetValue();
577 * \ingroup gdcmParser
578 * \brief Searches within Header Entries (Dicom Elements) parsed with
579 * the public and private dictionaries
580 * for the element value representation of a given tag..
582 * Obtaining the VR (Value Representation) might be needed by caller
583 * to convert the string typed content to caller's native type
584 * (think of C++ vs Python). The VR is actually of a higher level
585 * of semantics than just the native C++ type.
586 * @param group Group of the searched tag.
587 * @param element Element of the searched tag.
588 * @return Corresponding element value representation when it exists,
589 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
591 std::string gdcmParser::GetEntryVRByNumber(guint16 group, guint16 element) {
592 gdcmHeaderEntry* elem = GetHeaderEntryByNumber(group, element);
595 return elem->GetVR();
599 * \ingroup gdcmParser
600 * \brief Sets the value (string) of the Header Entry (Dicom Element)
601 * @param content string value of the Dicom Element
602 * @param tagName name of the searched Dicom Element.
603 * @return true when found
605 bool gdcmParser::SetEntryByName(std::string content,std::string tagName) {
606 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
607 if( dictEntry == NULL)
610 return(SetEntryByNumber(content,dictEntry->GetGroup(),
611 dictEntry->GetElement()));
615 * \ingroup gdcmParser
616 * \brief Accesses an existing gdcmHeaderEntry (i.e. a Dicom Element)
617 * through it's (group, element) and modifies it's content with
619 * \warning Don't use any longer : use SetPubEntryByNumber
620 * @param content new value to substitute with
621 * @param group group of the Dicom Element to modify
622 * @param element element of the Dicom Element to modify
624 bool gdcmParser::SetEntryByNumber(std::string content,
628 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
629 if ( ! tagHT.count(key))
631 int l = content.length();
632 if(l%2) // Odd length are padded with a space (020H).
635 content = content + '\0';
640 TagHeaderEntryHT::iterator p2;
641 // DO NOT remove the following lines : they explain the stuff
642 //p= tagHT.equal_range(key); // get a pair of iterators first-last synonym
643 //p2=p.first; // iterator on the first synonym
644 //a=p2->second; // H Table target column (2-nd col)
647 a = ((tagHT.equal_range(key)).first)->second;
649 a-> SetValue(content);
651 std::string vr = a->GetVR();
654 if( (vr == "US") || (vr == "SS") )
656 else if( (vr == "UL") || (vr == "SL") )
666 * \ingroup gdcmParser
667 * \brief Accesses an existing gdcmHeaderEntry (i.e. a Dicom Element)
668 * in the PubHeaderEntrySet of this instance
669 * through it's (group, element) and modifies it's length with
671 * \warning Use with extreme caution.
672 * @param length new length to substitute with
673 * @param group group of the entry to modify
674 * @param element element of the Entry to modify
675 * @return 1 on success, 0 otherwise.
678 bool gdcmParser::SetEntryLengthByNumber(guint32 length,
682 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
683 if ( ! tagHT.count(key))
685 if (length%2) length++; // length must be even
686 ( ((tagHT.equal_range(key)).first)->second )->SetLength(length);
692 * \ingroup gdcmParser
693 * \brief Gets (from Header) the offset of a 'non string' element value
694 * \ (LoadElementValues has already be executed)
697 * @return File Offset of the Element Value
699 size_t gdcmParser::GetEntryOffsetByNumber(guint16 Group, guint16 Elem)
701 gdcmHeaderEntry* Entry = GetHeaderEntryByNumber(Group, Elem);
704 dbg.Verbose(1, "gdcmParser::GetHeaderEntryByNumber",
705 "failed to Locate gdcmHeaderEntry");
708 return Entry->GetOffset();
712 * \ingroup gdcmParser
713 * \brief Gets (from Header) a 'non string' element value
714 * \ (LoadElementValues has already be executed)
717 * @return Pointer to the 'non string' area
719 void * gdcmParser::GetEntryVoidAreaByNumber(guint16 Group, guint16 Elem)
721 gdcmHeaderEntry* Entry = GetHeaderEntryByNumber(Group, Elem);
724 dbg.Verbose(1, "gdcmParser::GetHeaderEntryByNumber",
725 "failed to Locate gdcmHeaderEntry");
728 return Entry->GetVoidArea();
732 * \ingroup gdcmParser
733 * \brief Loads (from disk) the element content
734 * when a string is not suitable
738 void *gdcmParser::LoadEntryVoidArea(guint16 Group, guint16 Elem)
740 gdcmHeaderEntry * Element= GetHeaderEntryByNumber(Group, Elem);
743 size_t o =(size_t)Element->GetOffset();
744 fseek(fp, o, SEEK_SET);
745 int l=Element->GetLength();
746 void * a = malloc(l);
750 SetEntryVoidAreaByNumber(a, Group, Elem);
751 // TODO check the result
752 size_t l2 = fread(a, 1, l ,fp);
762 * \ingroup gdcmParser
763 * \brief Sets a 'non string' value to a given Dicom Element
765 * @param group Group number of the searched Dicom Element
766 * @param element Element number of the searched Dicom Element
769 bool gdcmParser::SetEntryVoidAreaByNumber(void * area,
773 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
774 if ( ! tagHT.count(key))
776 ( ((tagHT.equal_range(key)).first)->second )->SetVoidArea(area);
781 * \ingroup gdcmParser
782 * \brief Update the entries with the shadow dictionary. Only odd entries are
785 void gdcmParser::UpdateShaEntries(void) {
786 gdcmDictEntry *entry;
789 for(ListTag::iterator it=listEntries.begin();
790 it!=listEntries.end();
793 // Odd group => from public dictionary
794 if((*it)->GetGroup()%2==0)
797 // Peer group => search the corresponding dict entry
799 entry=RefShaDict->GetDictEntryByNumber((*it)->GetGroup(),(*it)->GetElement());
803 if((*it)->IsImplicitVR())
808 (*it)->SetValue(GetHeaderEntryUnvalue(*it));
810 // Set the new entry and the new value
811 (*it)->SetDictEntry(entry);
812 CheckHeaderEntryVR(*it,vr);
814 (*it)->SetValue(GetHeaderEntryValue(*it));
818 // Remove precedent value transformation
819 (*it)->SetDictEntry(NewVirtualDictEntry((*it)->GetGroup(),(*it)->GetElement(),vr));
825 * \ingroup gdcmParser
826 * \brief Searches within the Header Entries for a Dicom Element of
828 * @param tagName name of the searched Dicom Element.
829 * @return Corresponding Dicom Element when it exists, and NULL
832 gdcmHeaderEntry *gdcmParser::GetHeaderEntryByName(std::string tagName) {
833 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
834 if( dictEntry == NULL)
837 return(GetHeaderEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement()));
841 * \ingroup gdcmParser
842 * \brief retrieves a Dicom Element (the first one) using (group, element)
843 * \ warning (group, element) IS NOT an identifier inside the Dicom Header
844 * if you think it's NOT UNIQUE, check the count number
845 * and use iterators to retrieve ALL the Dicoms Elements within
846 * a given couple (group, element)
847 * @param group Group number of the searched Dicom Element
848 * @param element Element number of the searched Dicom Element
851 gdcmHeaderEntry* gdcmParser::GetHeaderEntryByNumber(guint16 group, guint16 element)
853 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
854 if ( ! tagHT.count(key))
856 return tagHT.find(key)->second;
860 * \ingroup gdcmParser
861 * \brief retrieves the Dicom Elements (all of them) using (group, element)
862 * @param group Group number of the searched Dicom Element.
863 * @param element Element number of the searched Dicom Element.
864 * @return a range (i.e.pair<,>) containing all elements whose key is group|element)
867 IterHT gdcmParser::GetHeaderEntrySameNumber(guint16 group, guint16 element){
868 TagKey key = gdcmDictEntry::TranslateToKey(group, element);
869 return (tagHT.equal_range(key));
873 * \ingroup gdcmParser
874 * \brief Loads the element while preserving the current
875 * underlying file position indicator as opposed to
876 * to LoadHeaderEntry that modifies it.
877 * @param entry Header Entry whose value shall be loaded.
880 void gdcmParser::LoadHeaderEntrySafe(gdcmHeaderEntry * entry) {
881 long PositionOnEntry = ftell(fp);
882 LoadHeaderEntry(entry);
883 fseek(fp, PositionOnEntry, SEEK_SET);
887 * \ingroup gdcmParser
888 * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader
889 * \warning : to be re-written using the chained list instead of the H table.
890 * \warning : DO NOT use (doesn't work any longer because of the multimap)
891 * \todo : to be re-written using the chained list instead of the H table
892 * @param SkipSequence TRUE if we don't want to write Sequences (ACR-NEMA Files)
893 * @param type Type of the File (ExplicitVR,ImplicitVR, ACR, ...)
895 void gdcmParser::UpdateGroupLength(bool SkipSequence, FileType type) {
899 gdcmHeaderEntry *elem;
901 std::string str_trash;
904 GroupHT groupHt; // to hold the length of each group
907 // typedef std::map<GroupKey, int> GroupHT;
909 gdcmHeaderEntry *elemZ;
911 // for each Tag in the DCM Header
913 for (TagHeaderEntryHT::iterator tag2 = tagHT.begin();
918 gr = elem->GetGroup();
919 el = elem->GetElement();
922 sprintf(trash, "%04x", gr);
923 key = trash; // generate 'group tag'
925 // if the caller decided not to take SEQUENCEs into account
926 // e.g : he wants to write an ACR-NEMA File
928 if (SkipSequence && vr == "SQ")
931 // Still unsolved problem :
932 // we cannot find the 'Sequence Delimitation Item'
933 // since it's at the end of the Hash Table
936 // pas SEQUENCE en ACR-NEMA
938 // --> la descente a l'interieur' des SQ
939 // devrait etre faite avec une liste chainee, pas avec une HTable...
941 if ( groupHt.count(key) == 0) // we just read the first elem of a given group
943 if (el == 0x0000) // the first elem is 0x0000
945 groupHt[key] = 0; // initialize group length
949 groupHt[key] = 2 + 2 + 4 + elem->GetLength(); // non 0x0000 first group elem
952 else // any elem but the first
954 if (type == ExplicitVR)
956 if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") )
958 groupHt[key] += 4; // explicit VR AND OB, OW, SQ : 4 more bytes
961 groupHt[key] += 2 + 2 + 4 + elem->GetLength();
965 unsigned short int gr_bid;
967 for (GroupHT::iterator g = groupHt.begin(); // for each group we found
971 // FIXME: g++ -Wall -Wstrict-prototypes reports on following line:
972 // warning: unsigned int format, different type arg
973 sscanf(g->first.c_str(),"%x",&gr_bid);
974 tk = g->first + "|0000"; // generate the element full tag
976 if ( tagHT.count(tk) == 0) // if element 0x0000 not found
978 gdcmDictEntry * tagZ = new gdcmDictEntry(gr_bid, 0x0000, "UL");
979 elemZ = new gdcmHeaderEntry(tagZ);
981 AddHeaderEntry(elemZ); // create it
985 elemZ=GetHeaderEntryByNumber(gr_bid, 0x0000);
987 sprintf(trash ,"%d",g->second);
989 elemZ->SetValue(str_trash);
994 * \ingroup gdcmParser
995 * \brief writes on disc according to the requested format
996 * \ (ACR-NEMA, ExplicitVR, ImplicitVR) the image
997 * \ warning does NOT add the missing elements in the header :
998 * \ it's up to the user doing it !
999 * \ (function CheckHeaderCoherence to be written)
1000 * \ warning DON'T try, right now, to write a DICOM image
1001 * \ from an ACR Header (meta elements will be missing!)
1002 * @param type type of the File to be written
1003 * (ACR-NEMA, ExplicitVR, ImplicitVR)
1004 * @param _fp already open file pointer
1006 void gdcmParser::WriteEntries(FileType type, FILE * _fp)
1015 std::vector<std::string> tokens;
1017 // TODO : function CheckHeaderCoherence to be written
1019 // uses now listEntries to iterate, not TagHt!
1021 // pb : gdcmParser.Add does NOT update listEntries
1022 // TODO : find a trick (in STL?) to do it, at low cost !
1026 // TODO (?) tester les echecs en ecriture (apres chaque fwrite)
1029 for (ListTag::iterator tag2=listEntries.begin();
1030 tag2 != listEntries.end();
1033 gr = (*tag2)->GetGroup();
1034 el = (*tag2)->GetElement();
1035 lgr = (*tag2)->GetReadLength();
1036 val = (*tag2)->GetValue().c_str();
1037 vr = (*tag2)->GetVR();
1040 if (gr < 0x0008) continue; // ignore pure DICOM V3 groups
1041 if (gr %2) continue; // ignore shadow groups
1042 if (vr == "SQ" ) continue; // ignore Sequences
1043 // TODO : find a trick to *skip* the SeQuences !
1044 // Not only ignore the SQ element
1045 if (gr == 0xfffe ) continue; // ignore delimiters
1048 fwrite ( &gr,(size_t)2 ,(size_t)1 ,_fp); //group
1049 fwrite ( &el,(size_t)2 ,(size_t)1 ,_fp); //element
1051 // === Deal with the length
1052 // --------------------
1054 // if ( (type == ExplicitVR) && (gr <= 0x0002) ) // ?!? < 2
1055 if ( (type == ExplicitVR) || (type == DICOMDIR) )
1058 guint16 z=0, shortLgr;
1059 if (vr == "Unknown") { // Unknown was 'written'
1061 fwrite ( &shortLgr,(size_t)2 ,(size_t)1 ,_fp);
1062 fwrite ( &z, (size_t)2 ,(size_t)1 ,_fp);
1064 if (gr != 0xfffe) { // NO value for 'delimiters'
1065 if (vr == "Unknown") // Unknown was 'written'
1066 fwrite(&z,(size_t)2 ,(size_t)1 ,_fp);
1068 fwrite (vr.c_str(),(size_t)2 ,(size_t)1 ,_fp);
1071 if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") || gr == 0xfffe) // JPR
1074 fwrite ( &z, (size_t)2 ,(size_t)1 ,_fp);
1075 fwrite ( &lgr,(size_t)4 ,(size_t)1 ,_fp);
1080 fwrite ( &shortLgr,(size_t)2 ,(size_t)1 ,_fp);
1086 fwrite ( &lgr,(size_t)4 ,(size_t)1 ,_fp);
1089 // === Deal with the value
1090 // -------------------
1091 if (vr == "SQ") continue; // no "value" to write for the SEQuences
1092 if (gr == 0xfffe)continue;
1094 if (vr == "US" || vr == "SS")
1096 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1097 Tokenize ((*tag2)->GetValue(), tokens, "\\");
1098 for (unsigned int i=0; i<tokens.size();i++)
1100 val_uint16 = atoi(tokens[i].c_str());
1102 fwrite ( ptr,(size_t)2 ,(size_t)1 ,_fp);
1107 if (vr == "UL" || vr == "SL")
1109 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1110 Tokenize ((*tag2)->GetValue(), tokens, "\\");
1111 for (unsigned int i=0; i<tokens.size();i++)
1113 val_uint32 = atoi(tokens[i].c_str());
1115 fwrite ( ptr,(size_t)4 ,(size_t)1 ,_fp);
1121 // Pixels are never loaded in the element !
1122 // we stop writting when Pixel are processed
1123 // FIX : we loose trailing elements (RAB, right now)
1125 if ((gr == GrPixel) && (el == NumPixel) ) {
1127 if (compte == countGrPixel) // we passed *all* the GrPixel,NumPixel
1130 fwrite ( val,(size_t)lgr ,(size_t)1 ,_fp); // Elem value
1135 * \ingroup gdcmParser
1136 * \brief Swaps back the bytes of 4-byte long integer accordingly to
1138 * @return The properly swaped 32 bits integer.
1140 guint32 gdcmParser::SwapLong(guint32 a) {
1145 a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) |
1146 ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) );
1150 a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
1154 a=( ((a<<8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
1157 dbg.Error(" gdcmParser::SwapLong : unset swap code");
1164 * \ingroup gdcmParser
1165 * \brief Unswaps back the bytes of 4-byte long integer accordingly to
1167 * @return The properly unswaped 32 bits integer.
1169 guint32 gdcmParser::UnswapLong(guint32 a) {
1170 return (SwapLong(a));
1174 * \ingroup gdcmParser
1175 * \brief Swaps the bytes so they agree with the processor order
1176 * @return The properly swaped 16 bits integer.
1178 guint16 gdcmParser::SwapShort(guint16 a) {
1179 if ( (sw==4321) || (sw==2143) )
1180 a =(((a<<8) & 0x0ff00) | ((a>>8)&0x00ff));
1185 * \ingroup gdcmParser
1186 * \brief Unswaps the bytes so they agree with the processor order
1187 * @return The properly unswaped 16 bits integer.
1189 guint16 gdcmParser::UnswapShort(guint16 a) {
1190 return (SwapShort(a));
1193 //-----------------------------------------------------------------------------
1196 * \ingroup gdcmParser
1197 * \brief Parses the header of the file but WITHOUT loading element values.
1198 * @return false if file is not ACR-NEMA / DICOM
1200 bool gdcmParser::ParseHeader(bool exception_on_error) throw(gdcmFormatError) {
1206 gdcmHeaderEntry *newHeaderEntry = (gdcmHeaderEntry *)0;
1207 while ( (newHeaderEntry = ReadNextHeaderEntry()) ) {
1208 SkipHeaderEntry(newHeaderEntry);
1209 if ( (ignoreShadow==0) || (newHeaderEntry->GetGroup()%2) == 0) {
1210 AddHeaderEntry(newHeaderEntry);
1217 * \ingroup gdcmParser
1218 * \brief Loads the element values of all the Header Entries pointed in the
1219 * public Chained List.
1221 void gdcmParser::LoadHeaderEntries(void) {
1223 for (ListTag::iterator i = GetListEntry().begin();
1224 i != GetListEntry().end();
1227 LoadHeaderEntry(*i);
1232 // Load 'non string' values
1233 std::string PhotometricInterpretation = GetEntryByNumber(0x0028,0x0004);
1234 if( PhotometricInterpretation == "PALETTE COLOR " ) {
1235 LoadEntryVoidArea(0x0028,0x1200); // gray LUT
1236 LoadEntryVoidArea(0x0028,0x1201); // R LUT
1237 LoadEntryVoidArea(0x0028,0x1202); // G LUT
1238 LoadEntryVoidArea(0x0028,0x1203); // B LUT
1240 LoadEntryVoidArea(0x0028,0x1221); // Segmented Red Palette Color LUT Data
1241 LoadEntryVoidArea(0x0028,0x1222); // Segmented Green Palette Color LUT Data
1242 LoadEntryVoidArea(0x0028,0x1223); // Segmented Blue Palette Color LUT Data
1245 // --------------------------------------------------------------
1246 // Special Patch to allow gdcm to read ACR-LibIDO formated images
1248 // if recognition code tells us we deal with a LibIDO image
1249 // we switch lineNumber and columnNumber
1251 std::string RecCode;
1252 RecCode = GetEntryByNumber(0x0008, 0x0010); // recognition code
1253 if (RecCode == "ACRNEMA_LIBIDO_1.1" ||
1254 RecCode == "CANRME_AILIBOD1_1." )
1256 filetype = ACR_LIBIDO;
1257 std::string rows = GetEntryByNumber(0x0028, 0x0010);
1258 std::string columns = GetEntryByNumber(0x0028, 0x0011);
1259 SetEntryByNumber(columns, 0x0028, 0x0010);
1260 SetEntryByNumber(rows , 0x0028, 0x0011);
1262 // ----------------- End of Special Patch ----------------
1266 * \ingroup gdcmParser
1267 * \brief Loads the element content if its length doesn't exceed
1268 * the value specified with gdcmParser::SetMaxSizeLoadEntry()
1269 * @param Entry Header Entry (Dicom Element) to be dealt with
1271 void gdcmParser::LoadHeaderEntry(gdcmHeaderEntry *Entry) {
1273 guint16 group = Entry->GetGroup();
1274 std::string vr= Entry->GetVR();
1275 guint32 length = Entry->GetLength();
1276 bool SkipLoad = false;
1278 fseek(fp, (long)Entry->GetOffset(), SEEK_SET);
1280 // the test was commented out to 'go inside' the SeQuences
1281 // we don't any longer skip them !
1283 // if( vr == "SQ" ) // (DO NOT remove this comment)
1286 // A SeQuence "contains" a set of Elements.
1287 // (fffe e000) tells us an Element is beginning
1288 // (fffe e00d) tells us an Element just ended
1289 // (fffe e0dd) tells us the current SeQuence just ended
1290 if( group == 0xfffe )
1294 Entry->SetLength(0);
1295 Entry->SetValue("gdcm::Skipped");
1299 // When the length is zero things are easy:
1300 if ( length == 0 ) {
1301 Entry->SetValue("");
1305 // The elements whose length is bigger than the specified upper bound
1306 // are not loaded. Instead we leave a short notice of the offset of
1307 // the element content and it's length.
1308 if (length > MaxSizeLoadEntry) {
1309 std::ostringstream s;
1310 s << "gdcm::NotLoaded.";
1311 s << " Address:" << (long)Entry->GetOffset();
1312 s << " Length:" << Entry->GetLength();
1313 s << " x(" << std::hex << Entry->GetLength() << ")";
1314 Entry->SetValue(s.str());
1318 // When integer(s) are expected, read and convert the following
1319 // n *(two or four bytes)
1320 // properly i.e. as integers as opposed to strings.
1321 // Elements with Value Multiplicity > 1
1322 // contain a set of integers (not a single one)
1324 // Any compacter code suggested (?)
1325 if ( IsHeaderEntryAnInteger(Entry) ) {
1327 std::ostringstream s;
1329 if (vr == "US" || vr == "SS") {
1331 NewInt = ReadInt16();
1334 for (int i=1; i < nbInt; i++) {
1336 NewInt = ReadInt16();
1341 else if (vr == "UL" || vr == "SL") {
1343 NewInt = ReadInt32();
1346 for (int i=1; i < nbInt; i++) {
1348 NewInt = ReadInt32();
1353 #ifdef GDCM_NO_ANSI_STRING_STREAM
1354 s << std::ends; // to avoid oddities on Solaris
1355 #endif //GDCM_NO_ANSI_STRING_STREAM
1357 Entry->SetValue(s.str());
1361 // We need an additional byte for storing \0 that is not on disk
1362 std::string NewValue(length,0);
1363 item_read = fread(&(NewValue[0]), (size_t)length, (size_t)1, fp);
1364 if ( item_read != 1 ) {
1365 dbg.Verbose(1, "gdcmParser::LoadElementValue","unread element value");
1366 Entry->SetValue("gdcm::UnRead");
1370 if( (vr == "UI") ) // Because of correspondance with the VR dic
1371 Entry->SetValue(NewValue.c_str()); // ??? JPR ???
1373 Entry->SetValue(NewValue);
1377 * \ingroup gdcmParser
1378 * \brief add a new Dicom Element pointer to
1379 * the H Table and to the chained List
1380 * \warning push_bash in listEntries ONLY during ParseHeader
1381 * \TODO something to allow further Elements addition,
1382 * \ when position to be taken care of
1383 * @param newHeaderEntry
1385 void gdcmParser::AddHeaderEntry(gdcmHeaderEntry *newHeaderEntry) {
1386 tagHT.insert( PairHT( newHeaderEntry->GetKey(),newHeaderEntry) );
1387 listEntries.push_back(newHeaderEntry);
1392 * \ingroup gdcmParser
1394 * @param Entry Header Entry whose length of the value shall be loaded.
1398 void gdcmParser::FindHeaderEntryLength (gdcmHeaderEntry *Entry) {
1399 guint16 element = Entry->GetElement();
1400 guint16 group = Entry->GetGroup();
1401 std::string vr = Entry->GetVR();
1403 if( (element == NumPixel) && (group == GrPixel) )
1406 dbg.Verbose(2, "gdcmParser::FindLength: ",
1407 "we reached (GrPixel,NumPixel)");
1410 if ( (filetype == ExplicitVR) && (! Entry->IsImplicitVR()) )
1412 if ( (vr=="OB") || (vr=="OW") || (vr=="SQ") || (vr=="UN") )
1414 // The following reserved two bytes (see PS 3.5-2001, section
1415 // 7.1.2 Data element structure with explicit vr p27) must be
1416 // skipped before proceeding on reading the length on 4 bytes.
1417 fseek(fp, 2L, SEEK_CUR);
1418 guint32 length32 = ReadInt32();
1420 if ( (vr == "OB") && (length32 == 0xffffffff) )
1422 Entry->SetLength(FindHeaderEntryLengthOB());
1425 FixHeaderEntryFoundLength(Entry, length32);
1429 // Length is encoded on 2 bytes.
1430 length16 = ReadInt16();
1432 // We can tell the current file is encoded in big endian (like
1433 // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag
1434 // and it's value is the one of the encoding of a big endian file.
1435 // In order to deal with such big endian encoded files, we have
1436 // (at least) two strategies:
1437 // * when we load the "Transfer Syntax" tag with value of big endian
1438 // encoding, we raise the proper flags. Then we wait for the end
1439 // of the META group (0x0002) among which is "Transfer Syntax",
1440 // before switching the swap code to big endian. We have to postpone
1441 // the switching of the swap code since the META group is fully encoded
1442 // in little endian, and big endian coding only starts at the next
1443 // group. The corresponding code can be hard to analyse and adds
1444 // many additional unnecessary tests for regular tags.
1445 // * the second strategy consists in waiting for trouble, that shall
1446 // appear when we find the first group with big endian encoding. This
1447 // is easy to detect since the length of a "Group Length" tag (the
1448 // ones with zero as element number) has to be of 4 (0x0004). When we
1449 // encounter 1024 (0x0400) chances are the encoding changed and we
1450 // found a group with big endian encoding.
1451 // We shall use this second strategy. In order to make sure that we
1452 // can interpret the presence of an apparently big endian encoded
1453 // length of a "Group Length" without committing a big mistake, we
1454 // add an additional check: we look in the already parsed elements
1455 // for the presence of a "Transfer Syntax" whose value has to be "big
1456 // endian encoding". When this is the case, chances are we have got our
1457 // hands on a big endian encoded file: we switch the swap code to
1458 // big endian and proceed...
1459 if ( (element == 0x0000) && (length16 == 0x0400) )
1461 if ( ! IsExplicitVRBigEndianTransferSyntax() )
1463 dbg.Verbose(0, "gdcmParser::FindLength", "not explicit VR");
1468 SwitchSwapToBigEndian();
1469 // Restore the unproperly loaded values i.e. the group, the element
1470 // and the dictionary entry depending on them.
1471 guint16 CorrectGroup = SwapShort(Entry->GetGroup());
1472 guint16 CorrectElem = SwapShort(Entry->GetElement());
1473 gdcmDictEntry * NewTag = GetDictEntryByNumber(CorrectGroup,
1477 // This correct tag is not in the dictionary. Create a new one.
1478 NewTag = NewVirtualDictEntry(CorrectGroup, CorrectElem);
1480 // FIXME this can create a memory leaks on the old entry that be
1481 // left unreferenced.
1482 Entry->SetDictEntry(NewTag);
1485 // Heuristic: well some files are really ill-formed.
1486 if ( length16 == 0xffff)
1489 //dbg.Verbose(0, "gdcmParser::FindLength",
1490 // "Erroneous element length fixed.");
1491 // Actually, length= 0xffff means that we deal with
1492 // Unknown Sequence Length
1495 FixHeaderEntryFoundLength(Entry, (guint32)length16);
1500 // Either implicit VR or a non DICOM conformal (see not below) explicit
1501 // VR that ommited the VR of (at least) this element. Farts happen.
1502 // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25
1503 // on Data elements "Implicit and Explicit VR Data Elements shall
1504 // not coexist in a Data Set and Data Sets nested within it".]
1505 // Length is on 4 bytes.
1506 FixHeaderEntryFoundLength(Entry, ReadInt32());
1512 * \ingroup gdcmParser
1513 * \brief Find the Value Representation of the current Dicom Element.
1516 void gdcmParser::FindHeaderEntryVR( gdcmHeaderEntry *Entry)
1518 if (filetype != ExplicitVR)
1524 long PositionOnEntry = ftell(fp);
1525 // Warning: we believe this is explicit VR (Value Representation) because
1526 // we used a heuristic that found "UL" in the first tag. Alas this
1527 // doesn't guarantee that all the tags will be in explicit VR. In some
1528 // cases (see e-film filtered files) one finds implicit VR tags mixed
1529 // within an explicit VR file. Hence we make sure the present tag
1530 // is in explicit VR and try to fix things if it happens not to be
1533 lgrLue=fread (&VR, (size_t)2,(size_t)1, fp);
1535 if(!CheckHeaderEntryVR(Entry,VR))
1537 fseek(fp, PositionOnEntry, SEEK_SET);
1538 // When this element is known in the dictionary we shall use, e.g. for
1539 // the semantics (see the usage of IsAnInteger), the VR proposed by the
1540 // dictionary entry. Still we have to flag the element as implicit since
1541 // we know now our assumption on expliciteness is not furfilled.
1543 if ( Entry->IsVRUnknown() )
1544 Entry->SetVR("Implicit");
1545 Entry->SetImplicitVR();
1550 * \ingroup gdcmParser
1551 * \brief Check the correspondance between the VR of the header entry
1552 * and the taken VR. If they are different, the header entry is
1553 * updated with the new VR.
1556 * @return false if the VR is incorrect of if the VR isn't referenced
1557 * otherwise, it returns true
1559 bool gdcmParser::CheckHeaderEntryVR(gdcmHeaderEntry *Entry, VRKey vr)
1561 char msg[100]; // for sprintf
1562 bool RealExplicit = true;
1564 // Assume we are reading a falsely explicit VR file i.e. we reached
1565 // a tag where we expect reading a VR but are in fact we read the
1566 // first to bytes of the length. Then we will interogate (through find)
1567 // the dicom_vr dictionary with oddities like "\004\0" which crashes
1568 // both GCC and VC++ implementations of the STL map. Hence when the
1569 // expected VR read happens to be non-ascii characters we consider
1570 // we hit falsely explicit VR tag.
1572 if ( (!isalpha(vr[0])) && (!isalpha(vr[1])) )
1573 RealExplicit = false;
1575 // CLEANME searching the dicom_vr at each occurence is expensive.
1576 // PostPone this test in an optional integrity check at the end
1577 // of parsing or only in debug mode.
1578 if ( RealExplicit && !gdcmGlobal::GetVR()->Count(vr) )
1579 RealExplicit= false;
1581 if ( !RealExplicit )
1583 // We thought this was explicit VR, but we end up with an
1584 // implicit VR tag. Let's backtrack.
1585 sprintf(msg,"Falsely explicit vr file (%04x,%04x)\n",
1586 Entry->GetGroup(),Entry->GetElement());
1587 dbg.Verbose(1, "gdcmParser::FindVR: ",msg);
1592 if ( Entry->IsVRUnknown() )
1594 // When not a dictionary entry, we can safely overwrite the VR.
1597 else if ( Entry->GetVR() != vr )
1599 // The VR present in the file and the dictionary disagree. We assume
1600 // the file writer knew best and use the VR of the file. Since it would
1601 // be unwise to overwrite the VR of a dictionary (since it would
1602 // compromise it's next user), we need to clone the actual DictEntry
1603 // and change the VR for the read one.
1604 gdcmDictEntry* NewEntry = NewVirtualDictEntry(
1605 Entry->GetGroup(),Entry->GetElement(),
1606 vr,"FIXME",Entry->GetName());
1607 Entry->SetDictEntry(NewEntry);
1613 * \ingroup gdcmParser
1614 * \brief Get the transformed value of the header entry. The VR value
1615 * is used to define the transformation to operate on the value
1616 * \warning NOT end user intended method !
1618 * @return Transformed entry value
1620 std::string gdcmParser::GetHeaderEntryValue(gdcmHeaderEntry *Entry)
1622 if ( (IsHeaderEntryAnInteger(Entry)) && (Entry->IsImplicitVR()) )
1624 std::string val=Entry->GetValue();
1625 std::string vr=Entry->GetVR();
1626 guint32 length = Entry->GetLength();
1627 std::ostringstream s;
1630 if (vr == "US" || vr == "SS")
1635 for (int i=0; i < nbInt; i++)
1639 NewInt16 = (val[2*i+0]&0xFF)+((val[2*i+1]&0xFF)<<8);
1640 NewInt16 = SwapShort(NewInt16);
1645 else if (vr == "UL" || vr == "SL")
1650 for (int i=0; i < nbInt; i++)
1654 NewInt32= (val[4*i+0]&0xFF)+((val[4*i+1]&0xFF)<<8)+
1655 ((val[4*i+2]&0xFF)<<16)+((val[4*i+3]&0xFF)<<24);
1656 NewInt32=SwapLong(NewInt32);
1661 #ifdef GDCM_NO_ANSI_STRING_STREAM
1662 s << std::ends; // to avoid oddities on Solaris
1663 #endif //GDCM_NO_ANSI_STRING_STREAM
1667 return(Entry->GetValue());
1671 * \ingroup gdcmParser
1672 * \brief Get the reverse transformed value of the header entry. The VR
1673 * value is used to define the reverse transformation to operate on
1675 * \warning NOT end user intended method !
1677 * @return Reverse transformed entry value
1679 std::string gdcmParser::GetHeaderEntryUnvalue(gdcmHeaderEntry *Entry)
1681 if ( (IsHeaderEntryAnInteger(Entry)) && (Entry->IsImplicitVR()) )
1683 std::string vr=Entry->GetVR();
1684 std::ostringstream s;
1685 std::vector<std::string> tokens;
1687 if (vr == "US" || vr == "SS")
1691 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1692 Tokenize (Entry->GetValue(), tokens, "\\");
1693 for (unsigned int i=0; i<tokens.size();i++)
1695 NewInt16 = atoi(tokens[i].c_str());
1696 s<<(NewInt16&0xFF)<<((NewInt16>>8)&0xFF);
1700 if (vr == "UL" || vr == "SL")
1704 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1705 Tokenize (Entry->GetValue(), tokens, "\\");
1706 for (unsigned int i=0; i<tokens.size();i++)
1708 NewInt32 = atoi(tokens[i].c_str());
1709 s<<(char)(NewInt32&0xFF)<<(char)((NewInt32>>8)&0xFF)
1710 <<(char)((NewInt32>>16)&0xFF)<<(char)((NewInt32>>24)&0xFF);
1715 #ifdef GDCM_NO_ANSI_STRING_STREAM
1716 s << std::ends; // to avoid oddities on Solaris
1717 #endif //GDCM_NO_ANSI_STRING_STREAM
1721 return(Entry->GetValue());
1725 * \ingroup gdcmParser
1726 * \brief Skip a given Header Entry
1727 * \warning NOT end user intended method !
1730 void gdcmParser::SkipHeaderEntry(gdcmHeaderEntry *entry)
1732 SkipBytes(entry->GetLength());
1736 * \ingroup gdcmParser
1737 * \brief When the length of an element value is obviously wrong (because
1738 * the parser went Jabberwocky) one can hope improving things by
1739 * applying this heuristic.
1741 void gdcmParser::FixHeaderEntryFoundLength(gdcmHeaderEntry *Entry, guint32 FoundLength)
1743 Entry->SetReadLength(FoundLength); // will be updated only if a bug is found
1745 if ( FoundLength == 0xffffffff) {
1749 guint16 gr =Entry->GetGroup();
1750 guint16 el =Entry->GetElement();
1752 if (FoundLength%2) {
1753 std::cout << "Warning : Tag with uneven length " << FoundLength
1754 << "in x(" << std::hex << gr << "," << el <<")" << std::endl;
1757 // Sorry for the patch!
1758 // XMedCom did the trick to read some nasty GE images ...
1759 if (FoundLength == 13) {
1760 // The following 'if' will be removed when there is no more
1761 // images on Creatis HDs with a 13 length for Manufacturer...
1762 if ( (Entry->GetGroup() != 0x0008) ||
1763 ( (Entry->GetElement() != 0x0070) && (Entry->GetElement() != 0x0080) ) ){
1764 // end of remove area
1766 Entry->SetReadLength(10); // a bug is to be fixed
1770 // to fix some garbage 'Leonardo' Siemens images
1771 // May be commented out to avoid overhead
1772 else if ( (Entry->GetGroup() == 0x0009) &&
1773 ( (Entry->GetElement() == 0x1113) || (Entry->GetElement() == 0x1114) ) ){
1775 Entry->SetReadLength(4); // a bug is to be fixed
1779 // to try to 'go inside' SeQuences (with length), and not to skip them
1780 else if ( Entry->GetVR() == "SQ")
1782 if (enableSequences) // only if the user does want to !
1783 FoundLength =0; // ReadLength is unchanged
1786 // a SeQuence Element is beginning
1787 // Let's forget it's length
1788 // (we want to 'go inside')
1790 // Pb : *normaly* fffe|e000 is just a marker, its length *should be* zero
1791 // in gdcm-MR-PHILIPS-16-Multi-Seq.dcm we find lengthes as big as 28800
1792 // if we set the length to zero IsHeaderEntryAnInteger() breaks...
1793 // if we don't, we lost 28800 characters from the Header :-(
1795 else if(Entry->GetGroup() == 0xfffe)
1797 // sometimes, length seems to be wrong
1798 FoundLength =0; // some more clever checking to be done !
1800 // only gdcm-MR-PHILIPS-16-Multi-Seq.dcm
1801 // causes troubles :-(
1804 Entry->SetUsableLength(FoundLength);
1808 * \ingroup gdcmParser
1809 * \brief Apply some heuristics to predict wether the considered
1810 * element value contains/represents an integer or not.
1811 * @param Entry The element value on which to apply the predicate.
1812 * @return The result of the heuristical predicate.
1814 bool gdcmParser::IsHeaderEntryAnInteger(gdcmHeaderEntry *Entry) {
1815 guint16 element = Entry->GetElement();
1816 guint16 group = Entry->GetGroup();
1817 std::string vr = Entry->GetVR();
1818 guint32 length = Entry->GetLength();
1819 // When we have some semantics on the element we just read, and if we
1820 // a priori know we are dealing with an integer, then we shall be
1821 // able to swap it's element value properly.
1822 if ( element == 0 ) // This is the group length of the group
1828 std::ostringstream s;
1829 int filePosition = ftell(fp);
1830 s << "Erroneous Group Length element length on : (" \
1831 << std::hex << group << " , " << element
1832 << ") -before- position x(" << filePosition << ")"
1833 << "lgt : " << length;
1834 // These 2 lines commented out : a *very dirty* patch
1835 // to go on PrintHeader'ing gdcm-MR-PHILIPS-16-Multi-Seq.dcm.
1836 // have a glance at offset x(8336) ...
1837 // For *regular* headers, the test is useless..
1838 // lets's print a warning message and go on,
1839 // instead of giving up with an error message
1841 //std::cout << s.str().c_str() << std::endl;
1843 // dbg.Error("gdcmParser::IsHeaderEntryAnInteger",
1844 // s.str().c_str());
1847 if ( (vr == "UL") || (vr == "US") || (vr == "SL") || (vr == "SS") )
1854 * \ingroup gdcmParser
1859 guint32 gdcmParser::FindHeaderEntryLengthOB(void) {
1860 // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
1863 long PositionOnEntry = ftell(fp);
1864 bool FoundSequenceDelimiter = false;
1865 guint32 TotalLength = 0;
1868 while ( ! FoundSequenceDelimiter)
1874 TotalLength += 4; // We even have to decount the group and element
1876 if ( g != 0xfffe && g!=0xb00c ) /*for bogus header */
1878 char msg[100]; // for sprintf. Sorry
1879 sprintf(msg,"wrong group (%04x) for an item sequence (%04x,%04x)\n",g, g,n);
1880 dbg.Verbose(1, "gdcmParser::FindLengthOB: ",msg);
1884 if ( n == 0xe0dd || ( g==0xb00c && n==0x0eb6 ) ) /* for bogus header */
1885 FoundSequenceDelimiter = true;
1886 else if ( n != 0xe000 )
1888 char msg[100]; // for sprintf. Sorry
1889 sprintf(msg,"wrong element (%04x) for an item sequence (%04x,%04x)\n",
1891 dbg.Verbose(1, "gdcmParser::FindLengthOB: ",msg);
1895 ItemLength = ReadInt32();
1896 TotalLength += ItemLength + 4; // We add 4 bytes since we just read
1897 // the ItemLength with ReadInt32
1898 SkipBytes(ItemLength);
1900 fseek(fp, PositionOnEntry, SEEK_SET);
1905 * \ingroup gdcmParser
1906 * \brief Reads a supposed to be 16 Bits integer
1907 * \ (swaps it depending on processor endianity)
1909 * @return read value
1911 guint16 gdcmParser::ReadInt16(void) {
1914 item_read = fread (&g, (size_t)2,(size_t)1, fp);
1915 if ( item_read != 1 ) {
1917 dbg.Verbose(0, "gdcmParser::ReadInt16", " File Error");
1927 * \ingroup gdcmParser
1928 * \brief Reads a supposed to be 32 Bits integer
1929 * \ (swaps it depending on processor endianity)
1931 * @return read value
1933 guint32 gdcmParser::ReadInt32(void) {
1936 item_read = fread (&g, (size_t)4,(size_t)1, fp);
1937 if ( item_read != 1 ) {
1939 dbg.Verbose(0, "gdcmParser::ReadInt32", " File Error");
1949 * \ingroup gdcmParser
1954 void gdcmParser::SkipBytes(guint32 NBytes) {
1955 //FIXME don't dump the returned value
1956 (void)fseek(fp, (long)NBytes, SEEK_CUR);
1960 * \ingroup gdcmParser
1963 void gdcmParser::Initialise(void)
1965 RefPubDict = gdcmGlobal::GetDicts()->GetDefaultPubDict();
1966 RefShaDict = (gdcmDict*)0;
1970 * \ingroup gdcmParser
1971 * \brief Discover what the swap code is (among little endian, big endian,
1972 * bad little endian, bad big endian).
1974 * @return false when we are absolutely sure
1975 * it's neither ACR-NEMA nor DICOM
1976 * true when we hope ours assuptions are OK
1978 bool gdcmParser::CheckSwap() {
1980 // The only guaranted way of finding the swap code is to find a
1981 // group tag since we know it's length has to be of four bytes i.e.
1982 // 0x00000004. Finding the swap code in then straigthforward. Trouble
1983 // occurs when we can't find such group...
1985 guint32 x=4; // x : for ntohs
1986 bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
1992 char deb[HEADER_LENGTH_TO_READ];
1994 // First, compare HostByteOrder and NetworkByteOrder in order to
1995 // determine if we shall need to swap bytes (i.e. the Endian type).
2001 // The easiest case is the one of a DICOM header, since it possesses a
2002 // file preamble where it suffice to look for the string "DICM".
2003 lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, fp);
2006 if(memcmp(entCur, "DICM", (size_t)4) == 0) {
2007 dbg.Verbose(1, "gdcmParser::CheckSwap:", "looks like DICOM Version3");
2009 // Next, determine the value representation (VR). Let's skip to the
2010 // first element (0002, 0000) and check there if we find "UL"
2011 // - or "OB" if the 1st one is (0002,0001) -,
2012 // in which case we (almost) know it is explicit VR.
2013 // WARNING: if it happens to be implicit VR then what we will read
2014 // is the length of the group. If this ascii representation of this
2015 // length happens to be "UL" then we shall believe it is explicit VR.
2016 // FIXME: in order to fix the above warning, we could read the next
2017 // element value (or a couple of elements values) in order to make
2018 // sure we are not commiting a big mistake.
2019 // We need to skip :
2020 // * the 128 bytes of File Preamble (often padded with zeroes),
2021 // * the 4 bytes of "DICM" string,
2022 // * the 4 bytes of the first tag (0002, 0000),or (0002, 0001)
2023 // i.e. a total of 136 bytes.
2026 // Use gdcmParser::dicom_vr to test all the possibilities
2027 // instead of just checking for UL, OB and UI !?
2030 // Sometimes (see : gdcmData/icone.dcm) group 0x0002 *is* Explicit VR,
2031 // but elem 0002,0010 (Transfert Syntax) tells us the file is *Implicit* VR.
2034 // The following test is *absolutely useless*, since everything *goes right*
2035 // with a *100 % wrong* assumption !!!
2037 if( (memcmp(entCur, "UL", (size_t)2) == 0) ||
2038 (memcmp(entCur, "OB", (size_t)2) == 0) ||
2039 (memcmp(entCur, "UI", (size_t)2) == 0) ||
2040 (memcmp(entCur, "CS", (size_t)2) == 0) ) // CS, to remove later
2041 // when Write DCM *adds*
2044 filetype = ExplicitVR;
2045 dbg.Verbose(1, "gdcmParser::CheckSwap:",
2046 "explicit Value Representation");
2050 filetype = ImplicitVR;
2051 dbg.Verbose(1, "gdcmParser::CheckSwap:",
2052 "not an explicit Value Representation");
2058 dbg.Verbose(1, "gdcmParser::CheckSwap:",
2059 "HostByteOrder != NetworkByteOrder");
2064 dbg.Verbose(1, "gdcmParser::CheckSwap:",
2065 "HostByteOrder = NetworkByteOrder");
2068 // Position the file position indicator at first tag (i.e.
2069 // after the file preamble and the "DICM" string).
2071 fseek (fp, 132L, SEEK_SET);
2075 // Alas, this is not a DicomV3 file and whatever happens there is no file
2076 // preamble. We can reset the file position indicator to where the data
2077 // is (i.e. the beginning of the file).
2078 dbg.Verbose(1, "gdcmParser::CheckSwap:", "not a DICOM Version3 file");
2081 // Our next best chance would be to be considering a 'clean' ACR/NEMA file.
2082 // By clean we mean that the length of the first tag is written down.
2083 // If this is the case and since the length of the first group HAS to be
2084 // four (bytes), then determining the proper swap code is straightforward.
2087 // We assume the array of char we are considering contains the binary
2088 // representation of a 32 bits integer. Hence the following dirty
2090 s32 = *((guint32 *)(entCur));
2111 // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
2112 // It is time for despaired wild guesses.
2113 // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA,
2114 // i.e. the 'group length' element is not present :
2116 // check the supposed to be 'group number'
2117 // 0x0002 or 0x0004 or 0x0008
2118 // to determine ' sw' value .
2119 // Only 0 or 4321 will be possible
2120 // (no oportunity to check for the formerly well known
2121 // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian'
2122 // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -4, 8-)
2123 // the file IS NOT ACR-NEMA nor DICOM V3
2124 // Find a trick to tell it the caller...
2126 s16 = *((guint16 *)(deb));
2142 dbg.Verbose(0, "gdcmParser::CheckSwap:",
2143 "ACR/NEMA unfound swap info (Really hopeless !)");
2148 // Then the only info we have is the net2host one.
2158 * \ingroup gdcmParser
2161 void gdcmParser::SwitchSwapToBigEndian(void)
2163 dbg.Verbose(1, "gdcmParser::SwitchSwapToBigEndian",
2164 "Switching to BigEndian mode.");
2185 * \ingroup gdcmParser
2189 void gdcmParser::SetMaxSizeLoadEntry(long NewSize)
2193 if ((guint32)NewSize >= (guint32)0xffffffff)
2195 MaxSizeLoadEntry = 0xffffffff;
2198 MaxSizeLoadEntry = NewSize;
2203 * \ingroup gdcmParser
2205 * \warning TODO : not yet usable
2206 * (see MAX_SIZE_PRINT_ELEMENT_VALUE
2207 * in gdcmHeaderEntry gdcmLoadEntry)
2211 void gdcmParser::SetMaxSizePrintEntry(long NewSize)
2215 if ((guint32)NewSize >= (guint32)0xffffffff)
2217 MaxSizePrintEntry = 0xffffffff;
2220 MaxSizePrintEntry = NewSize;
2224 * \ingroup gdcmParser
2225 * \brief Searches both the public and the shadow dictionary (when they
2226 * exist) for the presence of the DictEntry with given name.
2227 * The public dictionary has precedence on the shadow one.
2228 * @param Name name of the searched DictEntry
2229 * @return Corresponding DictEntry when it exists, NULL otherwise.
2231 gdcmDictEntry *gdcmParser::GetDictEntryByName(std::string Name)
2233 gdcmDictEntry *found = (gdcmDictEntry *)0;
2234 if (!RefPubDict && !RefShaDict)
2236 dbg.Verbose(0, "gdcmParser::GetDictEntry",
2237 "we SHOULD have a default dictionary");
2241 found = RefPubDict->GetDictEntryByName(Name);
2247 found = RefShaDict->GetDictEntryByName(Name);
2255 * \ingroup gdcmParser
2256 * \brief Searches both the public and the shadow dictionary (when they
2257 * exist) for the presence of the DictEntry with given
2258 * group and element. The public dictionary has precedence on the
2260 * @param group group of the searched DictEntry
2261 * @param element element of the searched DictEntry
2262 * @return Corresponding DictEntry when it exists, NULL otherwise.
2264 gdcmDictEntry *gdcmParser::GetDictEntryByNumber(guint16 group,guint16 element)
2266 gdcmDictEntry *found = (gdcmDictEntry *)0;
2267 if (!RefPubDict && !RefShaDict)
2269 dbg.Verbose(0, "gdcmParser::GetDictEntry",
2270 "we SHOULD have a default dictionary");
2274 found = RefPubDict->GetDictEntryByNumber(group, element);
2280 found = RefShaDict->GetDictEntryByNumber(group, element);
2288 * \ingroup gdcmParser
2289 * \brief Read the next tag but WITHOUT loading it's value
2290 * @return On succes the newly created HeaderEntry, NULL on failure.
2292 gdcmHeaderEntry *gdcmParser::ReadNextHeaderEntry(void) {
2294 gdcmHeaderEntry *NewEntry;
2300 // We reached the EOF (or an error occured) therefore
2301 // header parsing has to be considered as finished.
2302 return (gdcmHeaderEntry *)0;
2304 /* Pb : how to propagate the element length (used in SkipHeaderEntry)
2305 // direct call to SkipBytes ?
2307 if (ignoreShadow == 1 && g%2 ==1) //JPR
2308 // if user wants to skip shadow groups
2309 // and current element *is* a shadow element
2310 // we don't create anything
2311 return (gdcmHeaderEntry *)1; // to tell caller it's NOT finished
2313 NewEntry = NewHeaderEntryByNumber(g, n);
2314 FindHeaderEntryVR(NewEntry);
2315 FindHeaderEntryLength(NewEntry);
2321 NewEntry->SetOffset(ftell(fp));
2326 * \ingroup gdcmParser
2327 * \brief Build a new Element Value from all the low level arguments.
2328 * Check for existence of dictionary entry, and build
2329 * a default one when absent.
2330 * @param Name Name of the underlying DictEntry
2332 gdcmHeaderEntry *gdcmParser::NewHeaderEntryByName(std::string Name)
2334 gdcmDictEntry *NewTag = GetDictEntryByName(Name);
2336 NewTag = NewVirtualDictEntry(0xffff, 0xffff, "LO", "Unknown", Name);
2338 gdcmHeaderEntry* NewEntry = new gdcmHeaderEntry(NewTag);
2341 dbg.Verbose(1, "gdcmParser::ObtainHeaderEntryByName",
2342 "failed to allocate gdcmHeaderEntry");
2343 return (gdcmHeaderEntry *)0;
2349 * \ingroup gdcmParser
2350 * \brief Request a new virtual dict entry to the dict set
2351 * @param group group of the underlying DictEntry
2352 * @param elem element of the underlying DictEntry
2353 * @param vr VR of the underlying DictEntry
2354 * @param fourth owner group
2355 * @param name english name
2357 gdcmDictEntry *gdcmParser::NewVirtualDictEntry(guint16 group, guint16 element,
2362 return gdcmGlobal::GetDicts()->NewVirtualDictEntry(group,element,vr,fourth,name);
2366 * \ingroup gdcmParser
2367 * \brief Build a new Element Value from all the low level arguments.
2368 * Check for existence of dictionary entry, and build
2369 * a default one when absent.
2370 * @param Group group of the underlying DictEntry
2371 * @param Elem element of the underlying DictEntry
2373 gdcmHeaderEntry *gdcmParser::NewHeaderEntryByNumber(guint16 Group, guint16 Elem)
2375 // Find out if the tag we encountered is in the dictionaries:
2376 gdcmDictEntry *DictEntry = GetDictEntryByNumber(Group, Elem);
2378 DictEntry = NewVirtualDictEntry(Group, Elem);
2380 gdcmHeaderEntry *NewEntry = new gdcmHeaderEntry(DictEntry);
2383 dbg.Verbose(1, "gdcmParser::NewHeaderEntryByNumber",
2384 "failed to allocate gdcmHeaderEntry");
2390 // Never used; commented out, waiting for removal.
2392 * \ingroup gdcmParser
2393 * \brief Small utility function that creates a new manually crafted
2394 * (as opposed as read from the file) gdcmHeaderEntry with user
2395 * specified name and adds it to the public tag hash table.
2396 * \note A fake TagKey is generated so the PubDict can keep it's coherence.
2397 * @param NewTagName The name to be given to this new tag.
2398 * @param VR The Value Representation to be given to this new tag.
2399 * @return The newly hand crafted Element Value.
2401 //gdcmHeaderEntry *gdcmParser::NewManualHeaderEntryToPubDict(std::string NewTagName,
2404 // gdcmHeaderEntry *NewEntry = NULL;
2405 // guint32 StuffGroup = 0xffff; // Group to be stuffed with additional info
2406 // guint32 FreeElem = 0;
2407 // gdcmDictEntry *DictEntry = NULL;
2409 // FreeElem = GenerateFreeTagKeyInGroup(StuffGroup);
2410 // if (FreeElem == UINT32_MAX)
2412 // dbg.Verbose(1, "gdcmHeader::NewManualHeaderEntryToPubDict",
2413 // "Group 0xffff in Public Dict is full");
2417 // DictEntry = NewVirtualDictEntry(StuffGroup, FreeElem,
2418 // VR, "GDCM", NewTagName);
2419 // NewEntry = new gdcmHeaderEntry(DictEntry);
2420 // AddHeaderEntry(NewEntry);
2425 * \ingroup gdcmParser
2426 * \brief Generate a free TagKey i.e. a TagKey that is not present
2427 * in the TagHt dictionary.
2428 * @param group The generated tag must belong to this group.
2429 * @return The element of tag with given group which is fee.
2431 guint32 gdcmParser::GenerateFreeTagKeyInGroup(guint16 group)
2433 for (guint32 elem = 0; elem < UINT32_MAX; elem++)
2435 TagKey key = gdcmDictEntry::TranslateToKey(group, elem);
2436 if (tagHT.count(key) == 0)
2442 //-----------------------------------------------------------------------------