1 /*=========================================================================
4 Module: $RCSfile: gdcmDocument.cxx,v $
6 Date: $Date: 2005/01/06 14:49:16 $
7 Version: $Revision: 1.156 $
9 Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de
10 l'Image). All rights reserved. See Doc/License.txt or
11 http://www.creatis.insa-lyon.fr/Public/Gdcm/License.html for details.
13 This software is distributed WITHOUT ANY WARRANTY; without even
14 the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 PURPOSE. See the above copyright notices for more information.
17 =========================================================================*/
19 #include "gdcmDocument.h"
20 #include "gdcmValEntry.h"
21 #include "gdcmBinEntry.h"
22 #include "gdcmSeqEntry.h"
23 #include "gdcmGlobal.h"
25 #include "gdcmDebug.h"
27 #include "gdcmException.h"
28 #include "gdcmDictSet.h"
29 #include "gdcmRLEFramesInfo.h"
30 #include "gdcmJPEGFragmentsInfo.h"
31 #include "gdcmDocEntrySet.h"
32 #include "gdcmSQItem.h"
38 #if defined(_MSC_VER) || defined(__BORLANDC__)
41 #include <netinet/in.h>
46 //-----------------------------------------------------------------------------
47 static const char *TransferSyntaxStrings[] = {
48 // Implicit VR Little Endian
50 // Implicit VR Little Endian DLX G.E?
52 // Explicit VR Little Endian
53 "1.2.840.10008.1.2.1",
54 // Deflated Explicit VR Little Endian
55 "1.2.840.10008.1.2.1.99",
56 // Explicit VR Big Endian
57 "1.2.840.10008.1.2.2",
58 // JPEG Baseline (Process 1)
59 "1.2.840.10008.1.2.4.50",
60 // JPEG Extended (Process 2 & 4)
61 "1.2.840.10008.1.2.4.51",
62 // JPEG Extended (Process 3 & 5)
63 "1.2.840.10008.1.2.4.52",
64 // JPEG Spectral Selection, Non-Hierarchical (Process 6 & 8)
65 "1.2.840.10008.1.2.4.53",
66 // JPEG Full Progression, Non-Hierarchical (Process 10 & 12)
67 "1.2.840.10008.1.2.4.55",
68 // JPEG Lossless, Non-Hierarchical (Process 14)
69 "1.2.840.10008.1.2.4.57",
70 // JPEG Lossless, Hierarchical, First-Order Prediction (Process 14, [Selection Value 1])
71 "1.2.840.10008.1.2.4.70",
73 "1.2.840.10008.1.2.4.90",
75 "1.2.840.10008.1.2.4.91",
77 "1.2.840.10008.1.2.5",
79 "Unknown Transfer Syntax"
82 //-----------------------------------------------------------------------------
83 // Refer to Document::CheckSwap()
84 //const unsigned int Document::HEADER_LENGTH_TO_READ = 256;
86 // Refer to Document::SetMaxSizeLoadEntry()
87 const unsigned int Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096
88 const unsigned int Document::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff;
90 //-----------------------------------------------------------------------------
91 // Constructor / Destructor
95 * @param filename file to be opened for parsing
97 Document::Document( std::string const & filename ) : ElementSet(-1)
99 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
109 dbg.Verbose(0, "Document::Document: starting parsing of file: ",
111 Fp->seekg( 0, std::ios::beg);
113 Fp->seekg(0, std::ios::end);
114 long lgt = Fp->tellg();
116 Fp->seekg( 0, std::ios::beg);
118 long beg = Fp->tellg();
121 ParseDES( this, beg, lgt, false); // le Load sera fait a la volee
123 Fp->seekg( 0, std::ios::beg);
125 // Load 'non string' values
127 std::string PhotometricInterpretation = GetEntryByNumber(0x0028,0x0004);
128 if( PhotometricInterpretation == "PALETTE COLOR " )
130 LoadEntryBinArea(0x0028,0x1200); // gray LUT
131 /// FIXME FIXME FIXME
132 /// The tags refered by the three following lines used to be CORRECTLY
133 /// defined as having an US Value Representation in the public
134 /// dictionnary. BUT the semantics implied by the three following
135 /// lines state that the corresponding tag contents are in fact
136 /// the ones of a BinEntry.
137 /// In order to fix things "Quick and Dirty" the dictionnary was
138 /// altered on PURPOUS but now contains a WRONG value.
139 /// In order to fix things and restore the dictionary to its
140 /// correct value, one needs to decided of the semantics by deciding
141 /// wether the following tags are either:
142 /// - multivaluated US, and hence loaded as ValEntry, but afterwards
143 /// also used as BinEntry, which requires the proper conversion,
144 /// - OW, and hence loaded as BinEntry, but afterwards also used
145 /// as ValEntry, which requires the proper conversion.
146 LoadEntryBinArea(0x0028,0x1201); // R LUT
147 LoadEntryBinArea(0x0028,0x1202); // G LUT
148 LoadEntryBinArea(0x0028,0x1203); // B LUT
150 // Segmented Red Palette Color LUT Data
151 LoadEntryBinArea(0x0028,0x1221);
152 // Segmented Green Palette Color LUT Data
153 LoadEntryBinArea(0x0028,0x1222);
154 // Segmented Blue Palette Color LUT Data
155 LoadEntryBinArea(0x0028,0x1223);
157 //FIXME later : how to use it?
158 LoadEntryBinArea(0x0028,0x3006); //LUT Data (CTX dependent)
162 // --------------------------------------------------------------
163 // Specific code to allow gdcm to read ACR-LibIDO formated images
164 // Note: ACR-LibIDO is an extension of the ACR standard that was
165 // used at CREATIS. For the time being (say a couple years)
166 // we keep this kludge to allow a smooth move to gdcm for
167 // CREATIS developpers (sorry folks).
169 // if recognition code tells us we deal with a LibIDO image
170 // we switch lineNumber and columnNumber
173 RecCode = GetEntryByNumber(0x0008, 0x0010); // recognition code
174 if (RecCode == "ACRNEMA_LIBIDO_1.1" ||
175 RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares
176 // with "little-endian strings"
178 Filetype = ACR_LIBIDO;
179 std::string rows = GetEntryByNumber(0x0028, 0x0010);
180 std::string columns = GetEntryByNumber(0x0028, 0x0011);
181 SetEntryByNumber(columns, 0x0028, 0x0010);
182 SetEntryByNumber(rows , 0x0028, 0x0011);
184 // ----------------- End of ACR-LibIDO kludge ------------------
188 * \brief This default constructor doesn't parse the file. You should
189 * then invoke \ref Document::SetFileName and then the parsing.
191 Document::Document() : ElementSet(-1)
195 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
198 Filetype = ExplicitVR;
202 * \brief Canonical destructor.
204 Document::~Document ()
213 //-----------------------------------------------------------------------------
217 * \brief Prints The Dict Entries of THE public Dicom Dictionary
220 void Document::PrintPubDict(std::ostream & os)
222 RefPubDict->SetPrintLevel(PrintLevel);
223 RefPubDict->Print(os);
227 * \brief Prints The Dict Entries of THE shadow Dicom Dictionary
230 void Document::PrintShaDict(std::ostream & os)
232 RefShaDict->SetPrintLevel(PrintLevel);
233 RefShaDict->Print(os);
236 //-----------------------------------------------------------------------------
239 * \brief Get the public dictionary used
241 Dict* Document::GetPubDict()
247 * \brief Get the shadow dictionary used
249 Dict* Document::GetShaDict()
255 * \brief Set the shadow dictionary used
256 * \param dict dictionary to use in shadow
258 bool Document::SetShaDict(Dict *dict)
265 * \brief Set the shadow dictionary used
266 * \param dictName name of the dictionary to use in shadow
268 bool Document::SetShaDict(DictKey const & dictName)
270 RefShaDict = Global::GetDicts()->GetDict(dictName);
275 * \brief This predicate, based on hopefully reasonable heuristics,
276 * decides whether or not the current Document was properly parsed
277 * and contains the mandatory information for being considered as
278 * a well formed and usable Dicom/Acr File.
279 * @return true when Document is the one of a reasonable Dicom/Acr file,
282 bool Document::IsReadable()
284 if( Filetype == Unknown)
286 dbg.Verbose(0, "Document::IsReadable: wrong filetype");
292 dbg.Verbose(0, "Document::IsReadable: no tags in internal"
301 * \brief Accessor to the Transfer Syntax (when present) of the
302 * current document (it internally handles reading the
303 * value from disk when only parsing occured).
304 * @return The encountered Transfer Syntax of the current document.
306 TransferSyntaxType Document::GetTransferSyntax()
308 DocEntry *entry = GetDocEntryByNumber(0x0002, 0x0010);
314 // The entry might be present but not loaded (parsing and loading
315 // happen at different stages): try loading and proceed with check...
316 LoadDocEntrySafe(entry);
317 if (ValEntry* valEntry = dynamic_cast< ValEntry* >(entry) )
319 std::string transfer = valEntry->GetValue();
320 // The actual transfer (as read from disk) might be padded. We
321 // first need to remove the potential padding. We can make the
322 // weak assumption that padding was not executed with digits...
323 if ( transfer.length() == 0 )
325 // for brain damaged headers
328 while ( !isdigit((unsigned char)transfer[transfer.length()-1]) )
330 transfer.erase(transfer.length()-1, 1);
332 for (int i = 0; TransferSyntaxStrings[i] != NULL; i++)
334 if ( TransferSyntaxStrings[i] == transfer )
336 return TransferSyntaxType(i);
343 bool Document::IsJPEGLossless()
345 TransferSyntaxType r = GetTransferSyntax();
346 return r == JPEGFullProgressionProcess10_12
347 || r == JPEGLosslessProcess14
348 || r == JPEGLosslessProcess14_1;
352 * \brief Determines if the Transfer Syntax was already encountered
353 * and if it corresponds to a JPEG2000 one
354 * @return True when JPEG2000 (Lossly or LossLess) found. False in all
357 bool Document::IsJPEG2000()
359 TransferSyntaxType r = GetTransferSyntax();
360 return r == JPEG2000Lossless || r == JPEG2000;
364 * \brief Determines if the Transfer Syntax corresponds to any form
365 * of Jpeg encoded Pixel data.
366 * @return True when any form of JPEG found. False otherwise.
368 bool Document::IsJPEG()
370 TransferSyntaxType r = GetTransferSyntax();
371 return r == JPEGBaselineProcess1
372 || r == JPEGExtendedProcess2_4
373 || r == JPEGExtendedProcess3_5
374 || r == JPEGSpectralSelectionProcess6_8
380 * \brief Determines if the Transfer Syntax corresponds to encapsulated
381 * of encoded Pixel Data (as opposed to native).
382 * @return True when encapsulated. False when native.
384 bool Document::IsEncapsulate()
386 TransferSyntaxType r = GetTransferSyntax();
387 return IsJPEG() || r == RLELossless;
391 * \brief Predicate for dicom version 3 file.
392 * @return True when the file is a dicom version 3.
394 bool Document::IsDicomV3()
396 // Checking if Transfert Syntax exists is enough
397 // Anyway, it's to late check if the 'Preamble' was found ...
398 // And ... would it be a rich idea to check ?
399 // (some 'no Preamble' DICOM images exist !)
400 return GetDocEntryByNumber(0x0002, 0x0010) != NULL;
404 * \brief returns the File Type
405 * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
406 * @return the FileType code
408 FileType Document::GetFileType()
414 * \brief Tries to open the file \ref Document::Filename and
415 * checks the preamble when existing.
416 * @return The FILE pointer on success.
418 std::ifstream* Document::OpenFile()
420 if (Filename.length() == 0)
428 "Document::OpenFile is already opened when opening: ",
432 Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary);
436 "Document::OpenFile cannot open file: ",
444 Fp->read((char*)&zero, (size_t)2);
451 //ACR -- or DICOM with no Preamble; may start with a Shadow Group --
453 zero == 0x0001 || zero == 0x0100 || zero == 0x0002 || zero == 0x0200 ||
454 zero == 0x0003 || zero == 0x0300 || zero == 0x0004 || zero == 0x0400 ||
455 zero == 0x0005 || zero == 0x0500 || zero == 0x0006 || zero == 0x0600 ||
456 zero == 0x0007 || zero == 0x0700 || zero == 0x0008 || zero == 0x0800 )
462 Fp->seekg(126L, std::ios::cur);
464 Fp->read(dicm, (size_t)4);
470 if( memcmp(dicm, "DICM", 4) == 0 )
477 "Document::OpenFile not DICOM/ACR (missing preamble)",
484 * \brief closes the file
485 * @return TRUE if the close was successfull
487 bool Document::CloseFile()
496 return true; //FIXME how do we detect a non-close ifstream ?
500 * \brief Writes in a file all the Header Entries (Dicom Elements)
501 * @param fp file pointer on an already open file
502 * @param filetype Type of the File to be written
503 * (ACR-NEMA, ExplicitVR, ImplicitVR)
504 * \return Always true.
506 void Document::WriteContent(std::ofstream* fp, FileType filetype)
508 /// \todo move the following lines (and a lot of others, to be written)
509 /// to a future function CheckAndCorrectHeader
510 /// (necessary if user wants to write a DICOM V3 file
511 /// starting from an ACR-NEMA (V2) Header
513 if ( filetype == ImplicitVR || filetype == ExplicitVR )
515 // writing Dicom File Preamble
516 char filePreamble[128];
517 memset(filePreamble, 0, 128);
518 fp->write(filePreamble, 128);
519 fp->write("DICM", 4);
523 * \todo rewrite later, if really usefull
524 * - 'Group Length' element is optional in DICOM
525 * - but un-updated odd groups lengthes can causes pb
528 * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) )
529 * UpdateGroupLength(false,filetype);
530 * if ( filetype == ACR)
531 * UpdateGroupLength(true,ACR);
534 ElementSet::WriteContent(fp, filetype); // This one is recursive
538 * \brief Modifies the value of a given Doc Entry (Dicom Element)
539 * when it exists. Create it with the given value when unexistant.
540 * @param value (string) Value to be set
541 * @param group Group number of the Entry
542 * @param elem Element number of the Entry
543 * @param vr V(alue) R(epresentation) of the Entry -if private Entry-
544 * \return pointer to the modified/created Header Entry (NULL when creation
547 ValEntry* Document::ReplaceOrCreateByNumber(
548 std::string const & value,
553 ValEntry* valEntry = 0;
554 DocEntry* currentEntry = GetDocEntryByNumber( group, elem);
558 // check if (group,element) DictEntry exists
559 // if it doesn't, create an entry in DictSet::VirtualEntry
562 // Find out if the tag we received is in the dictionaries:
563 Dict *pubDict = Global::GetDicts()->GetDefaultPubDict();
564 DictEntry* dictEntry = pubDict->GetDictEntryByNumber(group, elem);
567 currentEntry = NewDocEntryByNumber(group, elem, vr);
571 currentEntry = NewDocEntryByNumber(group, elem);
576 dbg.Verbose(0, "Document::ReplaceOrCreateByNumber: call to"
577 " NewDocEntryByNumber failed.");
581 valEntry = new ValEntry(currentEntry);
584 if ( !AddEntry(valEntry))
587 dbg.Verbose(0, "Document::ReplaceOrCreateByNumber: AddEntry"
588 " failed allthough this is a creation.");
594 valEntry = dynamic_cast< ValEntry* >(currentEntry);
595 if ( !valEntry ) // Euuuuh? It wasn't a ValEntry
596 // then we change it to a ValEntry ?
597 // Shouldn't it be considered as an error ?
599 // We need to promote the DocEntry to a ValEntry:
600 valEntry = new ValEntry(currentEntry);
601 if (!RemoveEntry(currentEntry))
604 dbg.Verbose(0, "Document::ReplaceOrCreateByNumber: removal"
605 " of previous DocEntry failed.");
608 if ( !AddEntry(valEntry))
611 dbg.Verbose(0, "Document::ReplaceOrCreateByNumber: adding"
612 " promoted ValEntry failed.");
618 SetEntryByNumber(value, group, elem);
624 * \brief Modifies the value of a given Header Entry (Dicom Element)
625 * when it exists. Create it with the given value when unexistant.
626 * A copy of the binArea is made to be kept in the Document.
627 * @param binArea (binary) value to be set
628 * @param Group Group number of the Entry
629 * @param Elem Element number of the Entry
630 * @param vr V(alue) R(epresentation) of the Entry -if private Entry-
631 * \return pointer to the modified/created Header Entry (NULL when creation
634 BinEntry* Document::ReplaceOrCreateByNumber(
641 BinEntry* binEntry = 0;
642 DocEntry* currentEntry = GetDocEntryByNumber( group, elem);
646 // check if (group,element) DictEntry exists
647 // if it doesn't, create an entry in DictSet::VirtualEntry
650 // Find out if the tag we received is in the dictionaries:
651 Dict *pubDict = Global::GetDicts()->GetDefaultPubDict();
652 DictEntry *dictEntry = pubDict->GetDictEntryByNumber(group, elem);
656 currentEntry = NewDocEntryByNumber(group, elem, vr);
660 currentEntry = NewDocEntryByNumber(group, elem);
664 dbg.Verbose(0, "Document::ReplaceOrCreateByNumber: call to"
665 " NewDocEntryByNumber failed.");
668 binEntry = new BinEntry(currentEntry);
669 if ( !AddEntry(binEntry))
671 dbg.Verbose(0, "Document::ReplaceOrCreateByNumber: AddEntry"
672 " failed allthough this is a creation.");
678 binEntry = dynamic_cast< BinEntry* >(currentEntry);
679 if ( !binEntry ) // Euuuuh? It wasn't a BinEntry
680 // then we change it to a BinEntry ?
681 // Shouldn't it be considered as an error ?
683 // We need to promote the DocEntry to a BinEntry:
684 binEntry = new BinEntry(currentEntry);
685 if (!RemoveEntry(currentEntry))
687 dbg.Verbose(0, "Document::ReplaceOrCreateByNumber: removal"
688 " of previous DocEntry failed.");
691 if ( !AddEntry(binEntry))
693 dbg.Verbose(0, "Document::ReplaceOrCreateByNumber: adding"
694 " promoted BinEntry failed.");
701 if (lgth>0 && binArea)
703 tmpArea = new uint8_t[lgth];
704 memcpy(tmpArea,binArea,lgth);
710 if (!SetEntryByNumber(tmpArea, lgth, group, elem))
723 * \brief Modifies the value of a given Header Entry (Dicom Element)
724 * when it exists. Create it when unexistant.
725 * @param Group Group number of the Entry
726 * @param Elem Element number of the Entry
727 * \return pointer to the modified/created SeqEntry (NULL when creation
730 SeqEntry* Document::ReplaceOrCreateByNumber( uint16_t group, uint16_t elem)
733 DocEntry* a = GetDocEntryByNumber( group, elem);
736 a = NewSeqEntryByNumber(group, elem);
742 b = new SeqEntry(a, 1); // FIXME : 1 (Depth)
749 * \brief Set a new value if the invoked element exists
750 * Seems to be useless !!!
751 * @param value new element value
752 * @param group group number of the Entry
753 * @param elem element number of the Entry
756 bool Document::ReplaceIfExistByNumber(std::string const & value,
757 uint16_t group, uint16_t elem )
759 SetEntryByNumber(value, group, elem);
764 std::string Document::GetTransferSyntaxValue(TransferSyntaxType type)
766 return TransferSyntaxStrings[type];
769 //-----------------------------------------------------------------------------
773 * \brief Checks if a given Dicom Element exists within the H table
774 * @param group Group number of the searched Dicom Element
775 * @param element Element number of the searched Dicom Element
776 * @return true is found
778 bool Document::CheckIfEntryExistByNumber(uint16_t group, uint16_t element )
780 const std::string &key = DictEntry::TranslateToKey(group, element );
781 return TagHT.count(key) != 0;
785 * \brief Searches within Header Entries (Dicom Elements) parsed with
786 * the public and private dictionaries
787 * for the element value of a given tag.
788 * \warning Don't use any longer : use GetPubEntryByName
789 * @param tagName name of the searched element.
790 * @return Corresponding element value when it exists,
791 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
793 std::string Document::GetEntryByName(TagName const & tagName)
795 DictEntry* dictEntry = RefPubDict->GetDictEntryByName(tagName);
801 return GetEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement());
805 * \brief Searches within Header Entries (Dicom Elements) parsed with
806 * the public and private dictionaries
807 * for the element value representation of a given tag.
809 * Obtaining the VR (Value Representation) might be needed by caller
810 * to convert the string typed content to caller's native type
811 * (think of C++ vs Python). The VR is actually of a higher level
812 * of semantics than just the native C++ type.
813 * @param tagName name of the searched element.
814 * @return Corresponding element value representation when it exists,
815 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
817 std::string Document::GetEntryVRByName(TagName const& tagName)
819 DictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
820 if( dictEntry == NULL)
825 DocEntry* elem = GetDocEntryByNumber(dictEntry->GetGroup(),
826 dictEntry->GetElement());
827 return elem->GetVR();
831 * \brief Searches within Header Entries (Dicom Elements) parsed with
832 * the public and private dictionaries
833 * for the element value representation of a given tag.
834 * @param group Group number of the searched tag.
835 * @param element Element number of the searched tag.
836 * @return Corresponding element value representation when it exists,
837 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
839 std::string Document::GetEntryByNumber(uint16_t group, uint16_t element)
841 TagKey key = DictEntry::TranslateToKey(group, element);
842 if ( !TagHT.count(key))
847 return ((ValEntry *)TagHT.find(key)->second)->GetValue();
851 * \brief Searches within Header Entries (Dicom Elements) parsed with
852 * the public and private dictionaries
853 * for the element value representation of a given tag..
855 * Obtaining the VR (Value Representation) might be needed by caller
856 * to convert the string typed content to caller's native type
857 * (think of C++ vs Python). The VR is actually of a higher level
858 * of semantics than just the native C++ type.
859 * @param group Group number of the searched tag.
860 * @param element Element number of the searched tag.
861 * @return Corresponding element value representation when it exists,
862 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
864 std::string Document::GetEntryVRByNumber(uint16_t group, uint16_t element)
866 DocEntry* elem = GetDocEntryByNumber(group, element);
871 return elem->GetVR();
875 * \brief Searches within Header Entries (Dicom Elements) parsed with
876 * the public and private dictionaries
877 * for the value length of a given tag..
878 * @param group Group number of the searched tag.
879 * @param element Element number of the searched tag.
880 * @return Corresponding element length; -2 if not found
882 int Document::GetEntryLengthByNumber(uint16_t group, uint16_t element)
884 DocEntry* elem = GetDocEntryByNumber(group, element);
887 return -2; //magic number
889 return elem->GetLength();
892 * \brief Sets the value (string) of the Header Entry (Dicom Element)
893 * @param content string value of the Dicom Element
894 * @param tagName name of the searched Dicom Element.
895 * @return true when found
897 bool Document::SetEntryByName( std::string const & content,
898 TagName const & tagName)
900 DictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
906 return SetEntryByNumber(content,dictEntry->GetGroup(),
907 dictEntry->GetElement());
911 * \brief Accesses an existing DocEntry (i.e. a Dicom Element)
912 * through it's (group, element) and modifies it's content with
914 * @param content new value (string) to substitute with
915 * @param group group number of the Dicom Element to modify
916 * @param element element number of the Dicom Element to modify
918 bool Document::SetEntryByNumber(std::string const& content,
919 uint16_t group, uint16_t element)
921 ValEntry* valEntry = GetValEntryByNumber(group, element);
924 dbg.Verbose(0, "Document::SetEntryByNumber: no corresponding",
925 " ValEntry (try promotion first).");
928 // Non even content must be padded with a space (020H)...
929 valEntry->SetValue(content);
934 * \brief Accesses an existing DocEntry (i.e. a Dicom Element)
935 * through it's (group, element) and modifies it's content with
937 * @param content new value (void* -> uint8_t*) to substitute with
938 * @param lgth new value length
939 * @param group group number of the Dicom Element to modify
940 * @param element element number of the Dicom Element to modify
942 bool Document::SetEntryByNumber(uint8_t*content, int lgth,
943 uint16_t group, uint16_t element)
945 (void)lgth; //not used
946 TagKey key = DictEntry::TranslateToKey(group, element);
947 if ( !TagHT.count(key))
952 /* Hope Binary field length is *never* wrong
953 if(lgth%2) // Non even length are padded with a space (020H).
956 //content = content + '\0'; // fing a trick to enlarge a binary field?
959 BinEntry* entry = (BinEntry *)TagHT[key];
960 entry->SetBinArea(content);
961 entry->SetLength(lgth);
962 entry->SetValue(GDCM_BINLOADED);
968 * \brief Gets (from Header) a 'non string' element value
969 * (LoadElementValues has already be executed)
970 * @param group group number of the Entry
971 * @param elem element number of the Entry
972 * @return Pointer to the 'non string' area
974 void* Document::GetEntryBinAreaByNumber(uint16_t group, uint16_t elem)
976 DocEntry* entry = GetDocEntryByNumber(group, elem);
979 dbg.Verbose(1, "Document::GetDocEntryByNumber: no entry");
982 if ( BinEntry* binEntry = dynamic_cast<BinEntry*>(entry) )
984 return binEntry->GetBinArea();
991 * \brief Loads (from disk) the element content
992 * when a string is not suitable
993 * @param group group number of the Entry
994 * @param elem element number of the Entry
996 void Document::LoadEntryBinArea(uint16_t group, uint16_t elem)
998 // Search the corresponding DocEntry
999 DocEntry *docElement = GetDocEntryByNumber(group, elem);
1003 BinEntry *binElement = dynamic_cast<BinEntry *>(docElement);
1007 LoadEntryBinArea(binElement);
1011 * \brief Loads (from disk) the element content
1012 * when a string is not suitable
1013 * @param element Entry whose binArea is going to be loaded
1015 void Document::LoadEntryBinArea(BinEntry* element)
1017 if(element->GetBinArea())
1020 bool openFile = !Fp;
1024 size_t o =(size_t)element->GetOffset();
1025 Fp->seekg(o, std::ios::beg);
1027 size_t l = element->GetLength();
1028 uint8_t* a = new uint8_t[l];
1031 dbg.Verbose(0, "Document::LoadEntryBinArea cannot allocate a");
1035 /// \todo check the result
1036 Fp->read((char*)a, l);
1037 if( Fp->fail() || Fp->eof()) //Fp->gcount() == 1
1043 element->SetBinArea(a);
1050 * \brief Sets a 'non string' value to a given Dicom Element
1051 * @param area area containing the 'non string' value
1052 * @param group Group number of the searched Dicom Element
1053 * @param element Element number of the searched Dicom Element
1056 /*bool Document::SetEntryBinAreaByNumber(uint8_t* area,
1057 uint16_t group, uint16_t element)
1059 DocEntry* currentEntry = GetDocEntryByNumber(group, element);
1060 if ( !currentEntry )
1065 if ( BinEntry* binEntry = dynamic_cast<BinEntry*>(currentEntry) )
1067 binEntry->SetBinArea( area );
1075 * \brief Searches within the Header Entries for a Dicom Element of
1077 * @param tagName name of the searched Dicom Element.
1078 * @return Corresponding Dicom Element when it exists, and NULL
1081 DocEntry* Document::GetDocEntryByName(TagName const & tagName)
1083 DictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
1089 return GetDocEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement());
1093 * \brief retrieves a Dicom Element (the first one) using (group, element)
1094 * \warning (group, element) IS NOT an identifier inside the Dicom Header
1095 * if you think it's NOT UNIQUE, check the count number
1096 * and use iterators to retrieve ALL the Dicoms Elements within
1097 * a given couple (group, element)
1098 * @param group Group number of the searched Dicom Element
1099 * @param element Element number of the searched Dicom Element
1102 DocEntry* Document::GetDocEntryByNumber(uint16_t group, uint16_t element)
1104 TagKey key = DictEntry::TranslateToKey(group, element);
1105 if ( !TagHT.count(key))
1109 return TagHT.find(key)->second;
1113 * \brief Same as \ref Document::GetDocEntryByNumber except it only
1114 * returns a result when the corresponding entry is of type
1116 * @return When present, the corresponding ValEntry.
1118 ValEntry* Document::GetValEntryByNumber(uint16_t group, uint16_t element)
1120 DocEntry* currentEntry = GetDocEntryByNumber(group, element);
1121 if ( !currentEntry )
1125 if ( ValEntry* valEntry = dynamic_cast<ValEntry*>(currentEntry) )
1129 dbg.Verbose(0, "Document::GetValEntryByNumber: unfound ValEntry.");
1135 * \brief Loads the element while preserving the current
1136 * underlying file position indicator as opposed to
1137 * to LoadDocEntry that modifies it.
1138 * @param entry Header Entry whose value shall be loaded.
1141 void Document::LoadDocEntrySafe(DocEntry * entry)
1145 long PositionOnEntry = Fp->tellg();
1146 LoadDocEntry(entry);
1147 Fp->seekg(PositionOnEntry, std::ios::beg);
1152 * \brief Swaps back the bytes of 4-byte long integer accordingly to
1154 * @return The properly swaped 32 bits integer.
1156 uint32_t Document::SwapLong(uint32_t a)
1163 a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) |
1164 ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) );
1168 a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
1172 a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
1175 //std::cout << "swapCode= " << SwapCode << std::endl;
1176 dbg.Error(" Document::SwapLong : unset swap code");
1183 * \brief Unswaps back the bytes of 4-byte long integer accordingly to
1185 * @return The properly unswaped 32 bits integer.
1187 uint32_t Document::UnswapLong(uint32_t a)
1193 * \brief Swaps the bytes so they agree with the processor order
1194 * @return The properly swaped 16 bits integer.
1196 uint16_t Document::SwapShort(uint16_t a)
1198 if ( SwapCode == 4321 || SwapCode == 2143 )
1200 a = ((( a << 8 ) & 0x0ff00 ) | (( a >> 8 ) & 0x00ff ) );
1206 * \brief Unswaps the bytes so they agree with the processor order
1207 * @return The properly unswaped 16 bits integer.
1209 uint16_t Document::UnswapShort(uint16_t a)
1211 return SwapShort(a);
1214 //-----------------------------------------------------------------------------
1218 * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries)
1219 * @return length of the parsed set.
1221 void Document::ParseDES(DocEntrySet *set, long offset,
1222 long l_max, bool delim_mode)
1224 DocEntry *newDocEntry = 0;
1228 if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max)
1232 newDocEntry = ReadNextDocEntry( );
1238 VRKey vr = newDocEntry->GetVR();
1242 if ( Global::GetVR()->IsVROfGdcmStringRepresentable(vr) )
1244 /////////////////////// ValEntry
1245 ValEntry* newValEntry =
1246 new ValEntry( newDocEntry->GetDictEntry() ); //LEAK
1247 newValEntry->Copy( newDocEntry );
1249 // When "set" is a Document, then we are at the top of the
1250 // hierarchy and the Key is simply of the form ( group, elem )...
1251 if (Document* dummy = dynamic_cast< Document* > ( set ) )
1254 newValEntry->SetKey( newValEntry->GetKey() );
1256 // ...but when "set" is a SQItem, we are inserting this new
1257 // valEntry in a sequence item. Hence the key has the
1258 // generalized form (refer to \ref BaseTagKey):
1259 if (SQItem* parentSQItem = dynamic_cast< SQItem* > ( set ) )
1261 newValEntry->SetKey( parentSQItem->GetBaseTagKey()
1262 + newValEntry->GetKey() );
1265 LoadDocEntry( newValEntry );
1266 bool delimitor=newValEntry->IsItemDelimitor();
1267 if( !set->AddEntry( newValEntry ) )
1269 // If here expect big troubles
1270 delete newValEntry; //otherwise mem leak
1278 if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max)
1286 if ( ! Global::GetVR()->IsVROfGdcmBinaryRepresentable(vr) )
1288 ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR
1289 dbg.Verbose(0, "Document::ParseDES: neither Valentry, "
1290 "nor BinEntry. Probably unknown VR.");
1293 //////////////////// BinEntry or UNKOWN VR:
1294 BinEntry* newBinEntry = new BinEntry( newDocEntry ); //LEAK
1296 // When "this" is a Document the Key is simply of the
1297 // form ( group, elem )...
1298 if (Document* dummy = dynamic_cast< Document* > ( set ) )
1301 newBinEntry->SetKey( newBinEntry->GetKey() );
1303 // but when "this" is a SQItem, we are inserting this new
1304 // valEntry in a sequence item, and the kay has the
1305 // generalized form (refer to \ref BaseTagKey):
1306 if (SQItem* parentSQItem = dynamic_cast< SQItem* > ( set ) )
1308 newBinEntry->SetKey( parentSQItem->GetBaseTagKey()
1309 + newBinEntry->GetKey() );
1312 LoadDocEntry( newBinEntry );
1313 if( !set->AddEntry( newBinEntry ) )
1315 //Expect big troubles if here
1320 if ( ( newDocEntry->GetGroup() == 0x7fe0 )
1321 && ( newDocEntry->GetElement() == 0x0010 ) )
1323 TransferSyntaxType ts = GetTransferSyntax();
1324 if ( ts == RLELossless )
1326 long PositionOnEntry = Fp->tellg();
1327 Fp->seekg( newDocEntry->GetOffset(), std::ios::beg );
1329 Fp->seekg( PositionOnEntry, std::ios::beg );
1331 else if ( IsJPEG() )
1333 long PositionOnEntry = Fp->tellg();
1334 Fp->seekg( newDocEntry->GetOffset(), std::ios::beg );
1335 ComputeJPEGFragmentInfo();
1336 Fp->seekg( PositionOnEntry, std::ios::beg );
1340 // Just to make sure we are at the beginning of next entry.
1341 SkipToNextDocEntry(newDocEntry);
1342 //delete newDocEntry;
1347 unsigned long l = newDocEntry->GetReadLength();
1348 if ( l != 0 ) // don't mess the delim_mode for zero-length sequence
1350 if ( l == 0xffffffff )
1359 // no other way to create it ...
1360 SeqEntry* newSeqEntry =
1361 new SeqEntry( newDocEntry->GetDictEntry() );
1362 newSeqEntry->Copy( newDocEntry );
1363 newSeqEntry->SetDelimitorMode( delim_mode );
1365 // At the top of the hierarchy, stands a Document. When "set"
1366 // is a Document, then we are building the first depth level.
1367 // Hence the SeqEntry we are building simply has a depth
1369 if (Document* dummy = dynamic_cast< Document* > ( set ) )
1372 newSeqEntry->SetDepthLevel( 1 );
1373 newSeqEntry->SetKey( newSeqEntry->GetKey() );
1375 // But when "set" is allready a SQItem, we are building a nested
1376 // sequence, and hence the depth level of the new SeqEntry
1377 // we are building, is one level deeper:
1378 if (SQItem* parentSQItem = dynamic_cast< SQItem* > ( set ) )
1380 newSeqEntry->SetDepthLevel( parentSQItem->GetDepthLevel() + 1 );
1381 newSeqEntry->SetKey( parentSQItem->GetBaseTagKey()
1382 + newSeqEntry->GetKey() );
1386 { // Don't try to parse zero-length sequences
1387 ParseSQ( newSeqEntry,
1388 newDocEntry->GetOffset(),
1391 set->AddEntry( newSeqEntry );
1392 if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max)
1403 * \brief Parses a Sequence ( SeqEntry after SeqEntry)
1404 * @return parsed length for this level
1406 void Document::ParseSQ( SeqEntry* seqEntry,
1407 long offset, long l_max, bool delim_mode)
1409 int SQItemNumber = 0;
1414 DocEntry* newDocEntry = ReadNextDocEntry();
1417 // FIXME Should warn user
1422 if ( newDocEntry->IsSequenceDelimitor() )
1424 seqEntry->SetSequenceDelimitationItem( newDocEntry );
1428 if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max)
1434 SQItem *itemSQ = new SQItem( seqEntry->GetDepthLevel() );
1435 std::ostringstream newBase;
1436 newBase << seqEntry->GetKey()
1440 itemSQ->SetBaseTagKey( newBase.str() );
1441 unsigned int l = newDocEntry->GetReadLength();
1443 if ( l == 0xffffffff )
1452 ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod);
1455 seqEntry->AddEntry( itemSQ, SQItemNumber );
1457 if ( !delim_mode && ((long)(Fp->tellg())-offset ) >= l_max )
1465 * \brief Loads the element content if its length doesn't exceed
1466 * the value specified with Document::SetMaxSizeLoadEntry()
1467 * @param entry Header Entry (Dicom Element) to be dealt with
1469 void Document::LoadDocEntry(DocEntry* entry)
1471 uint16_t group = entry->GetGroup();
1472 std::string vr = entry->GetVR();
1473 uint32_t length = entry->GetLength();
1475 Fp->seekg((long)entry->GetOffset(), std::ios::beg);
1477 // A SeQuence "contains" a set of Elements.
1478 // (fffe e000) tells us an Element is beginning
1479 // (fffe e00d) tells us an Element just ended
1480 // (fffe e0dd) tells us the current SeQuence just ended
1481 if( group == 0xfffe )
1483 // NO more value field for SQ !
1487 // When the length is zero things are easy:
1490 ((ValEntry *)entry)->SetValue("");
1494 // The elements whose length is bigger than the specified upper bound
1495 // are not loaded. Instead we leave a short notice of the offset of
1496 // the element content and it's length.
1498 std::ostringstream s;
1499 if (length > MaxSizeLoadEntry)
1501 if (BinEntry* binEntryPtr = dynamic_cast< BinEntry* >(entry) )
1503 //s << "gdcm::NotLoaded (BinEntry)";
1504 s << GDCM_NOTLOADED;
1505 s << " Address:" << (long)entry->GetOffset();
1506 s << " Length:" << entry->GetLength();
1507 s << " x(" << std::hex << entry->GetLength() << ")";
1508 binEntryPtr->SetValue(s.str());
1510 // Be carefull : a BinEntry IS_A ValEntry ...
1511 else if (ValEntry* valEntryPtr = dynamic_cast< ValEntry* >(entry) )
1513 // s << "gdcm::NotLoaded. (ValEntry)";
1514 s << GDCM_NOTLOADED;
1515 s << " Address:" << (long)entry->GetOffset();
1516 s << " Length:" << entry->GetLength();
1517 s << " x(" << std::hex << entry->GetLength() << ")";
1518 valEntryPtr->SetValue(s.str());
1523 std::cout<< "MaxSizeLoadEntry exceeded, neither a BinEntry "
1524 << "nor a ValEntry ?! Should never print that !" << std::endl;
1527 // to be sure we are at the end of the value ...
1528 Fp->seekg((long)entry->GetOffset()+(long)entry->GetLength(),
1533 // When we find a BinEntry not very much can be done :
1534 if (BinEntry* binEntryPtr = dynamic_cast< BinEntry* >(entry) )
1536 s << GDCM_BINLOADED;
1537 binEntryPtr->SetValue(s.str());
1538 LoadEntryBinArea(binEntryPtr); // last one, not to erase length !
1542 /// \todo Any compacter code suggested (?)
1543 if ( IsDocEntryAnInteger(entry) )
1547 // When short integer(s) are expected, read and convert the following
1548 // n *two characters properly i.e. consider them as short integers as
1549 // opposed to strings.
1550 // Elements with Value Multiplicity > 1
1551 // contain a set of integers (not a single one)
1552 if (vr == "US" || vr == "SS")
1555 NewInt = ReadInt16();
1559 for (int i=1; i < nbInt; i++)
1562 NewInt = ReadInt16();
1567 // See above comment on multiple integers (mutatis mutandis).
1568 else if (vr == "UL" || vr == "SL")
1571 NewInt = ReadInt32();
1575 for (int i=1; i < nbInt; i++)
1578 NewInt = ReadInt32();
1583 #ifdef GDCM_NO_ANSI_STRING_STREAM
1584 s << std::ends; // to avoid oddities on Solaris
1585 #endif //GDCM_NO_ANSI_STRING_STREAM
1587 ((ValEntry *)entry)->SetValue(s.str());
1591 // FIXME: We need an additional byte for storing \0 that is not on disk
1592 char *str = new char[length+1];
1593 Fp->read(str, (size_t)length);
1594 str[length] = '\0'; //this is only useful when length is odd
1595 // Special DicomString call to properly handle \0 and even length
1596 std::string newValue;
1599 newValue = Util::DicomString(str, length+1);
1600 //dbg.Verbose(0, "Warning: bad length: ", length );
1601 dbg.Verbose(0, "For string :", newValue.c_str());
1602 // Since we change the length of string update it length
1603 entry->SetReadLength(length+1);
1607 newValue = Util::DicomString(str, length);
1611 if ( ValEntry* valEntry = dynamic_cast<ValEntry* >(entry) )
1613 if ( Fp->fail() || Fp->eof())//Fp->gcount() == 1
1615 dbg.Verbose(1, "Document::LoadDocEntry",
1616 "unread element value");
1617 valEntry->SetValue(GDCM_UNREAD);
1623 // Because of correspondance with the VR dic
1624 valEntry->SetValue(newValue);
1628 valEntry->SetValue(newValue);
1633 dbg.Error(true, "Document::LoadDocEntry"
1634 "Should have a ValEntry, here !");
1640 * \brief Find the value Length of the passed Header Entry
1641 * @param entry Header Entry whose length of the value shall be loaded.
1643 void Document::FindDocEntryLength( DocEntry *entry )
1644 throw ( FormatError )
1646 uint16_t element = entry->GetElement();
1647 std::string vr = entry->GetVR();
1650 if ( Filetype == ExplicitVR && !entry->IsImplicitVR() )
1652 if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UN" )
1654 // The following reserved two bytes (see PS 3.5-2003, section
1655 // "7.1.2 Data element structure with explicit vr", p 27) must be
1656 // skipped before proceeding on reading the length on 4 bytes.
1657 Fp->seekg( 2L, std::ios::cur);
1658 uint32_t length32 = ReadInt32();
1660 if ( (vr == "OB" || vr == "OW") && length32 == 0xffffffff )
1665 /// \todo rename that to FindDocEntryLengthOBOrOW since
1666 /// the above test is on both OB and OW...
1667 lengthOB = FindDocEntryLengthOB();
1669 catch ( FormatUnexpected )
1671 // Computing the length failed (this happens with broken
1672 // files like gdcm-JPEG-LossLess3a.dcm). We still have a
1673 // chance to get the pixels by deciding the element goes
1674 // until the end of the file. Hence we artificially fix the
1675 // the length and proceed.
1676 long currentPosition = Fp->tellg();
1677 Fp->seekg(0L,std::ios::end);
1678 long lengthUntilEOF = (long)(Fp->tellg())-currentPosition;
1679 Fp->seekg(currentPosition, std::ios::beg);
1680 entry->SetLength(lengthUntilEOF);
1683 entry->SetLength(lengthOB);
1686 FixDocEntryFoundLength(entry, length32);
1690 // Length is encoded on 2 bytes.
1691 length16 = ReadInt16();
1693 // We can tell the current file is encoded in big endian (like
1694 // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag
1695 // and it's value is the one of the encoding of a big endian file.
1696 // In order to deal with such big endian encoded files, we have
1697 // (at least) two strategies:
1698 // * when we load the "Transfer Syntax" tag with value of big endian
1699 // encoding, we raise the proper flags. Then we wait for the end
1700 // of the META group (0x0002) among which is "Transfer Syntax",
1701 // before switching the swap code to big endian. We have to postpone
1702 // the switching of the swap code since the META group is fully encoded
1703 // in little endian, and big endian coding only starts at the next
1704 // group. The corresponding code can be hard to analyse and adds
1705 // many additional unnecessary tests for regular tags.
1706 // * the second strategy consists in waiting for trouble, that shall
1707 // appear when we find the first group with big endian encoding. This
1708 // is easy to detect since the length of a "Group Length" tag (the
1709 // ones with zero as element number) has to be of 4 (0x0004). When we
1710 // encounter 1024 (0x0400) chances are the encoding changed and we
1711 // found a group with big endian encoding.
1712 // We shall use this second strategy. In order to make sure that we
1713 // can interpret the presence of an apparently big endian encoded
1714 // length of a "Group Length" without committing a big mistake, we
1715 // add an additional check: we look in the already parsed elements
1716 // for the presence of a "Transfer Syntax" whose value has to be "big
1717 // endian encoding". When this is the case, chances are we have got our
1718 // hands on a big endian encoded file: we switch the swap code to
1719 // big endian and proceed...
1720 if ( element == 0x0000 && length16 == 0x0400 )
1722 TransferSyntaxType ts = GetTransferSyntax();
1723 if ( ts != ExplicitVRBigEndian )
1725 throw FormatError( "Document::FindDocEntryLength()",
1726 " not explicit VR." );
1730 SwitchSwapToBigEndian();
1731 // Restore the unproperly loaded values i.e. the group, the element
1732 // and the dictionary entry depending on them.
1733 uint16_t correctGroup = SwapShort( entry->GetGroup() );
1734 uint16_t correctElem = SwapShort( entry->GetElement() );
1735 DictEntry* newTag = GetDictEntryByNumber( correctGroup,
1739 // This correct tag is not in the dictionary. Create a new one.
1740 newTag = NewVirtualDictEntry(correctGroup, correctElem);
1742 // FIXME this can create a memory leaks on the old entry that be
1743 // left unreferenced.
1744 entry->SetDictEntry( newTag );
1747 // Heuristic: well, some files are really ill-formed.
1748 if ( length16 == 0xffff)
1750 // 0xffff means that we deal with 'Unknown Length' Sequence
1753 FixDocEntryFoundLength( entry, (uint32_t)length16 );
1758 // Either implicit VR or a non DICOM conformal (see note below) explicit
1759 // VR that ommited the VR of (at least) this element. Farts happen.
1760 // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25
1761 // on Data elements "Implicit and Explicit VR Data Elements shall
1762 // not coexist in a Data Set and Data Sets nested within it".]
1763 // Length is on 4 bytes.
1765 FixDocEntryFoundLength( entry, ReadInt32() );
1771 * \brief Find the Value Representation of the current Dicom Element.
1774 std::string Document::FindDocEntryVR()
1776 if ( Filetype != ExplicitVR )
1777 return(GDCM_UNKNOWN);
1779 long positionOnEntry = Fp->tellg();
1780 // Warning: we believe this is explicit VR (Value Representation) because
1781 // we used a heuristic that found "UL" in the first tag. Alas this
1782 // doesn't guarantee that all the tags will be in explicit VR. In some
1783 // cases (see e-film filtered files) one finds implicit VR tags mixed
1784 // within an explicit VR file. Hence we make sure the present tag
1785 // is in explicit VR and try to fix things if it happens not to be
1789 Fp->read (vr, (size_t)2);
1792 if( !CheckDocEntryVR(vr) )
1794 Fp->seekg(positionOnEntry, std::ios::beg);
1795 return(GDCM_UNKNOWN);
1801 * \brief Check the correspondance between the VR of the header entry
1802 * and the taken VR. If they are different, the header entry is
1803 * updated with the new VR.
1804 * @param entry Header Entry to check
1805 * @param vr Dicom Value Representation
1806 * @return false if the VR is incorrect of if the VR isn't referenced
1807 * otherwise, it returns true
1809 bool Document::CheckDocEntryVR(VRKey vr)
1811 // CLEANME searching the dicom_vr at each occurence is expensive.
1812 // PostPone this test in an optional integrity check at the end
1813 // of parsing or only in debug mode.
1814 if ( !Global::GetVR()->IsValidVR(vr) )
1821 * \brief Get the transformed value of the header entry. The VR value
1822 * is used to define the transformation to operate on the value
1823 * \warning NOT end user intended method !
1824 * @param entry entry to tranform
1825 * @return Transformed entry value
1827 std::string Document::GetDocEntryValue(DocEntry *entry)
1829 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1831 std::string val = ((ValEntry *)entry)->GetValue();
1832 std::string vr = entry->GetVR();
1833 uint32_t length = entry->GetLength();
1834 std::ostringstream s;
1837 // When short integer(s) are expected, read and convert the following
1838 // n * 2 bytes properly i.e. as a multivaluated strings
1839 // (each single value is separated fromthe next one by '\'
1840 // as usual for standard multivaluated filels
1841 // Elements with Value Multiplicity > 1
1842 // contain a set of short integers (not a single one)
1844 if( vr == "US" || vr == "SS" )
1849 for (int i=0; i < nbInt; i++)
1855 newInt16 = ( val[2*i+0] & 0xFF ) + ( ( val[2*i+1] & 0xFF ) << 8);
1856 newInt16 = SwapShort( newInt16 );
1861 // When integer(s) are expected, read and convert the following
1862 // n * 4 bytes properly i.e. as a multivaluated strings
1863 // (each single value is separated fromthe next one by '\'
1864 // as usual for standard multivaluated filels
1865 // Elements with Value Multiplicity > 1
1866 // contain a set of integers (not a single one)
1867 else if( vr == "UL" || vr == "SL" )
1872 for (int i=0; i < nbInt; i++)
1878 newInt32 = ( val[4*i+0] & 0xFF )
1879 + (( val[4*i+1] & 0xFF ) << 8 )
1880 + (( val[4*i+2] & 0xFF ) << 16 )
1881 + (( val[4*i+3] & 0xFF ) << 24 );
1882 newInt32 = SwapLong( newInt32 );
1886 #ifdef GDCM_NO_ANSI_STRING_STREAM
1887 s << std::ends; // to avoid oddities on Solaris
1888 #endif //GDCM_NO_ANSI_STRING_STREAM
1892 return ((ValEntry *)entry)->GetValue();
1896 * \brief Get the reverse transformed value of the header entry. The VR
1897 * value is used to define the reverse transformation to operate on
1899 * \warning NOT end user intended method !
1900 * @param entry Entry to reverse transform
1901 * @return Reverse transformed entry value
1903 std::string Document::GetDocEntryUnvalue(DocEntry* entry)
1905 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1907 std::string vr = entry->GetVR();
1908 std::vector<std::string> tokens;
1909 std::ostringstream s;
1911 if ( vr == "US" || vr == "SS" )
1915 tokens.erase( tokens.begin(), tokens.end()); // clean any previous value
1916 Util::Tokenize (((ValEntry *)entry)->GetValue(), tokens, "\\");
1917 for (unsigned int i=0; i<tokens.size(); i++)
1919 newInt16 = atoi(tokens[i].c_str());
1920 s << ( newInt16 & 0xFF )
1921 << (( newInt16 >> 8 ) & 0xFF );
1925 if ( vr == "UL" || vr == "SL")
1929 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1930 Util::Tokenize (((ValEntry *)entry)->GetValue(), tokens, "\\");
1931 for (unsigned int i=0; i<tokens.size();i++)
1933 newInt32 = atoi(tokens[i].c_str());
1934 s << (char)( newInt32 & 0xFF )
1935 << (char)(( newInt32 >> 8 ) & 0xFF )
1936 << (char)(( newInt32 >> 16 ) & 0xFF )
1937 << (char)(( newInt32 >> 24 ) & 0xFF );
1942 #ifdef GDCM_NO_ANSI_STRING_STREAM
1943 s << std::ends; // to avoid oddities on Solaris
1944 #endif //GDCM_NO_ANSI_STRING_STREAM
1948 return ((ValEntry *)entry)->GetValue();
1952 * \brief Skip a given Header Entry
1953 * \warning NOT end user intended method !
1954 * @param entry entry to skip
1956 void Document::SkipDocEntry(DocEntry *entry)
1958 SkipBytes(entry->GetLength());
1962 * \brief Skips to the begining of the next Header Entry
1963 * \warning NOT end user intended method !
1964 * @param entry entry to skip
1966 void Document::SkipToNextDocEntry(DocEntry *entry)
1968 Fp->seekg((long)(entry->GetOffset()), std::ios::beg);
1969 Fp->seekg( (long)(entry->GetReadLength()), std::ios::cur);
1973 * \brief When the length of an element value is obviously wrong (because
1974 * the parser went Jabberwocky) one can hope improving things by
1975 * applying some heuristics.
1976 * @param entry entry to check
1977 * @param foundLength fist assumption about length
1979 void Document::FixDocEntryFoundLength(DocEntry *entry,
1980 uint32_t foundLength)
1982 entry->SetReadLength( foundLength ); // will be updated only if a bug is found
1983 if ( foundLength == 0xffffffff)
1988 uint16_t gr = entry->GetGroup();
1989 uint16_t el = entry->GetElement();
1991 if ( foundLength % 2)
1993 std::ostringstream s;
1994 s << "Warning : Tag with uneven length "
1996 << " in x(" << std::hex << gr << "," << el <<")" << std::dec;
1997 dbg.Verbose(0, s.str().c_str());
2000 //////// Fix for some naughty General Electric images.
2001 // Allthough not recent many such GE corrupted images are still present
2002 // on Creatis hard disks. Hence this fix shall remain when such images
2003 // are no longer in user (we are talking a few years, here)...
2004 // Note: XMedCom probably uses such a trick since it is able to read
2005 // those pesky GE images ...
2006 if ( foundLength == 13)
2008 // Only happens for this length !
2009 if ( entry->GetGroup() != 0x0008
2010 || ( entry->GetElement() != 0x0070
2011 && entry->GetElement() != 0x0080 ) )
2014 entry->SetReadLength(10); /// \todo a bug is to be fixed !?
2018 //////// Fix for some brain-dead 'Leonardo' Siemens images.
2019 // Occurence of such images is quite low (unless one leaves close to a
2020 // 'Leonardo' source. Hence, one might consider commenting out the
2021 // following fix on efficiency reasons.
2022 else if ( entry->GetGroup() == 0x0009
2023 && ( entry->GetElement() == 0x1113
2024 || entry->GetElement() == 0x1114 ) )
2027 entry->SetReadLength(4); /// \todo a bug is to be fixed !?
2030 else if ( entry->GetVR() == "SQ" )
2032 foundLength = 0; // ReadLength is unchanged
2035 //////// We encountered a 'delimiter' element i.e. a tag of the form
2036 // "fffe|xxxx" which is just a marker. Delimiters length should not be
2037 // taken into account.
2038 else if( entry->GetGroup() == 0xfffe )
2040 // According to the norm, fffe|0000 shouldn't exist. BUT the Philips
2041 // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to
2042 // causes extra troubles...
2043 if( entry->GetElement() != 0x0000 )
2049 entry->SetUsableLength(foundLength);
2053 * \brief Apply some heuristics to predict whether the considered
2054 * element value contains/represents an integer or not.
2055 * @param entry The element value on which to apply the predicate.
2056 * @return The result of the heuristical predicate.
2058 bool Document::IsDocEntryAnInteger(DocEntry *entry)
2060 uint16_t element = entry->GetElement();
2061 uint16_t group = entry->GetGroup();
2062 const std::string & vr = entry->GetVR();
2063 uint32_t length = entry->GetLength();
2065 // When we have some semantics on the element we just read, and if we
2066 // a priori know we are dealing with an integer, then we shall be
2067 // able to swap it's element value properly.
2068 if ( element == 0 ) // This is the group length of the group
2076 // Allthough this should never happen, still some images have a
2077 // corrupted group length [e.g. have a glance at offset x(8336) of
2078 // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm].
2079 // Since for dicom compliant and well behaved headers, the present
2080 // test is useless (and might even look a bit paranoid), when we
2081 // encounter such an ill-formed image, we simply display a warning
2082 // message and proceed on parsing (while crossing fingers).
2083 std::ostringstream s;
2084 long filePosition = Fp->tellg();
2085 s << "Erroneous Group Length element length on : (" \
2086 << std::hex << group << " , " << element
2087 << ") -before- position x(" << filePosition << ")"
2088 << "lgt : " << length;
2089 dbg.Verbose(0, "Document::IsDocEntryAnInteger", s.str().c_str() );
2093 if ( vr == "UL" || vr == "US" || vr == "SL" || vr == "SS" )
2101 * \brief Find the Length till the next sequence delimiter
2102 * \warning NOT end user intended method !
2106 uint32_t Document::FindDocEntryLengthOB()
2107 throw( FormatUnexpected )
2109 // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
2110 long positionOnEntry = Fp->tellg();
2111 bool foundSequenceDelimiter = false;
2112 uint32_t totalLength = 0;
2114 while ( !foundSequenceDelimiter )
2120 group = ReadInt16();
2123 catch ( FormatError )
2125 throw FormatError("Document::FindDocEntryLengthOB()",
2126 " group or element not present.");
2129 // We have to decount the group and element we just read
2132 if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) )
2134 dbg.Verbose(1, "Document::FindDocEntryLengthOB: neither an Item "
2135 "tag nor a Sequence delimiter tag.");
2136 Fp->seekg(positionOnEntry, std::ios::beg);
2137 throw FormatUnexpected("Document::FindDocEntryLengthOB()",
2138 "Neither an Item tag nor a Sequence "
2142 if ( elem == 0xe0dd )
2144 foundSequenceDelimiter = true;
2147 uint32_t itemLength = ReadInt32();
2148 // We add 4 bytes since we just read the ItemLength with ReadInt32
2149 totalLength += itemLength + 4;
2150 SkipBytes(itemLength);
2152 if ( foundSequenceDelimiter )
2157 Fp->seekg( positionOnEntry, std::ios::beg);
2162 * \brief Reads a supposed to be 16 Bits integer
2163 * (swaps it depending on processor endianity)
2164 * @return read value
2166 uint16_t Document::ReadInt16()
2167 throw( FormatError )
2170 Fp->read ((char*)&g, (size_t)2);
2173 throw FormatError( "Document::ReadInt16()", " file error." );
2177 throw FormatError( "Document::ReadInt16()", "EOF." );
2184 * \brief Reads a supposed to be 32 Bits integer
2185 * (swaps it depending on processor endianity)
2186 * @return read value
2188 uint32_t Document::ReadInt32()
2189 throw( FormatError )
2192 Fp->read ((char*)&g, (size_t)4);
2195 throw FormatError( "Document::ReadInt32()", " file error." );
2199 throw FormatError( "Document::ReadInt32()", "EOF." );
2206 * \brief skips bytes inside the source file
2207 * \warning NOT end user intended method !
2210 void Document::SkipBytes(uint32_t nBytes)
2212 //FIXME don't dump the returned value
2213 Fp->seekg((long)nBytes, std::ios::cur);
2217 * \brief Loads all the needed Dictionaries
2218 * \warning NOT end user intended method !
2220 void Document::Initialise()
2222 RefPubDict = Global::GetDicts()->GetDefaultPubDict();
2224 RLEInfo = new RLEFramesInfo;
2225 JPEGInfo = new JPEGFragmentsInfo;
2230 * \brief Discover what the swap code is (among little endian, big endian,
2231 * bad little endian, bad big endian).
2233 * @return false when we are absolutely sure
2234 * it's neither ACR-NEMA nor DICOM
2235 * true when we hope ours assuptions are OK
2237 bool Document::CheckSwap()
2239 // The only guaranted way of finding the swap code is to find a
2240 // group tag since we know it's length has to be of four bytes i.e.
2241 // 0x00000004. Finding the swap code in then straigthforward. Trouble
2242 // occurs when we can't find such group...
2244 uint32_t x = 4; // x : for ntohs
2245 bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
2251 // First, compare HostByteOrder and NetworkByteOrder in order to
2252 // determine if we shall need to swap bytes (i.e. the Endian type).
2253 if ( x == ntohs(x) )
2262 // The easiest case is the one of a DICOM header, since it possesses a
2263 // file preamble where it suffice to look for the string "DICM".
2266 char *entCur = deb + 128;
2267 if( memcmp(entCur, "DICM", (size_t)4) == 0 )
2269 dbg.Verbose(1, "Document::CheckSwap:", "looks like DICOM Version3");
2271 // Next, determine the value representation (VR). Let's skip to the
2272 // first element (0002, 0000) and check there if we find "UL"
2273 // - or "OB" if the 1st one is (0002,0001) -,
2274 // in which case we (almost) know it is explicit VR.
2275 // WARNING: if it happens to be implicit VR then what we will read
2276 // is the length of the group. If this ascii representation of this
2277 // length happens to be "UL" then we shall believe it is explicit VR.
2278 // FIXME: in order to fix the above warning, we could read the next
2279 // element value (or a couple of elements values) in order to make
2280 // sure we are not commiting a big mistake.
2281 // We need to skip :
2282 // * the 128 bytes of File Preamble (often padded with zeroes),
2283 // * the 4 bytes of "DICM" string,
2284 // * the 4 bytes of the first tag (0002, 0000),or (0002, 0001)
2285 // i.e. a total of 136 bytes.
2289 // Sometimes (see : gdcmData/icone.dcm) group 0x0002 *is* Explicit VR,
2290 // but elem 0002,0010 (Transfert Syntax) tells us the file is
2291 // *Implicit* VR. -and it is !-
2293 if( memcmp(entCur, "UL", (size_t)2) == 0 ||
2294 memcmp(entCur, "OB", (size_t)2) == 0 ||
2295 memcmp(entCur, "UI", (size_t)2) == 0 ||
2296 memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later
2297 // when Write DCM *adds*
2299 // Use Document::dicom_vr to test all the possibilities
2300 // instead of just checking for UL, OB and UI !? group 0000
2302 Filetype = ExplicitVR;
2303 dbg.Verbose(1, "Document::CheckSwap:",
2304 "explicit Value Representation");
2308 Filetype = ImplicitVR;
2309 dbg.Verbose(1, "Document::CheckSwap:",
2310 "not an explicit Value Representation");
2316 dbg.Verbose(1, "Document::CheckSwap:",
2317 "HostByteOrder != NetworkByteOrder");
2322 dbg.Verbose(1, "Document::CheckSwap:",
2323 "HostByteOrder = NetworkByteOrder");
2326 // Position the file position indicator at first tag (i.e.
2327 // after the file preamble and the "DICM" string).
2328 Fp->seekg(0, std::ios::beg);
2329 Fp->seekg ( 132L, std::ios::beg);
2333 // Alas, this is not a DicomV3 file and whatever happens there is no file
2334 // preamble. We can reset the file position indicator to where the data
2335 // is (i.e. the beginning of the file).
2336 dbg.Verbose(1, "Document::CheckSwap:", "not a DICOM Version3 file");
2337 Fp->seekg(0, std::ios::beg);
2339 // Our next best chance would be to be considering a 'clean' ACR/NEMA file.
2340 // By clean we mean that the length of the first tag is written down.
2341 // If this is the case and since the length of the first group HAS to be
2342 // four (bytes), then determining the proper swap code is straightforward.
2345 // We assume the array of char we are considering contains the binary
2346 // representation of a 32 bits integer. Hence the following dirty
2348 s32 = *((uint32_t *)(entCur));
2369 // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
2370 // It is time for despaired wild guesses.
2371 // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA,
2372 // i.e. the 'group length' element is not present :
2374 // check the supposed-to-be 'group number'
2375 // in ( 0x0001 .. 0x0008 )
2376 // to determine ' SwapCode' value .
2377 // Only 0 or 4321 will be possible
2378 // (no oportunity to check for the formerly well known
2379 // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian'
2380 // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -3, 4, ..., 8-)
2381 // the file IS NOT ACR-NEMA nor DICOM V3
2382 // Find a trick to tell it the caller...
2384 s16 = *((uint16_t *)(deb));
2411 dbg.Verbose(0, "Document::CheckSwap:",
2412 "ACR/NEMA unfound swap info (Really hopeless !)");
2416 // Then the only info we have is the net2host one.
2428 * \brief Restore the unproperly loaded values i.e. the group, the element
2429 * and the dictionary entry depending on them.
2431 void Document::SwitchSwapToBigEndian()
2433 dbg.Verbose(1, "Document::SwitchSwapToBigEndian",
2434 "Switching to BigEndian mode.");
2435 if ( SwapCode == 0 )
2439 else if ( SwapCode == 4321 )
2443 else if ( SwapCode == 3412 )
2447 else if ( SwapCode == 2143 )
2454 * \brief during parsing, Header Elements too long are not loaded in memory
2457 void Document::SetMaxSizeLoadEntry(long newSize)
2463 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2465 MaxSizeLoadEntry = 0xffffffff;
2468 MaxSizeLoadEntry = newSize;
2473 * \brief Header Elements too long will not be printed
2474 * \todo See comments of \ref Document::MAX_SIZE_PRINT_ELEMENT_VALUE
2477 void Document::SetMaxSizePrintEntry(long newSize)
2479 //DOH !! This is exactly SetMaxSizeLoadEntry FIXME FIXME
2484 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2486 MaxSizePrintEntry = 0xffffffff;
2489 MaxSizePrintEntry = newSize;
2495 * \brief Handle broken private tag from Philips NTSCAN
2496 * where the endianess is being switch to BigEndian for no
2500 void Document::HandleBrokenEndian(uint16_t group, uint16_t elem)
2502 // Endian reversion. Some files contain groups of tags with reversed endianess.
2503 static int reversedEndian = 0;
2504 // try to fix endian switching in the middle of headers
2505 if ((group == 0xfeff) && (elem == 0x00e0))
2507 // start endian swap mark for group found
2509 SwitchSwapToBigEndian();
2514 else if ((group == 0xfffe) && (elem == 0xe00d) && reversedEndian)
2516 // end of reversed endian group
2518 SwitchSwapToBigEndian();
2523 * \brief Read the next tag but WITHOUT loading it's value
2524 * (read the 'Group Number', the 'Element Number',
2525 * gets the Dict Entry
2526 * gets the VR, gets the length, gets the offset value)
2527 * @return On succes the newly created DocEntry, NULL on failure.
2529 DocEntry* Document::ReadNextDocEntry()
2536 group = ReadInt16();
2539 catch ( FormatError e )
2541 // We reached the EOF (or an error occured) therefore
2542 // header parsing has to be considered as finished.
2547 HandleBrokenEndian(group, elem);
2548 std::string vr=FindDocEntryVR();
2550 DocEntry *newEntry = NewDocEntryByNumber(group, elem, vr);
2551 if( vr == GDCM_UNKNOWN )
2553 if( Filetype == ExplicitVR )
2555 // We thought this was explicit VR, but we end up with an
2556 // implicit VR tag. Let's backtrack.
2558 msg = Util::Format("Falsely explicit vr file (%04x,%04x)\n",
2559 newEntry->GetGroup(), newEntry->GetElement());
2560 dbg.Verbose(1, "Document::FindVR: ", msg.c_str());
2562 newEntry->SetImplicitVR();
2567 FindDocEntryLength(newEntry);
2569 catch ( FormatError e )
2577 newEntry->SetOffset(Fp->tellg());
2584 * \brief Generate a free TagKey i.e. a TagKey that is not present
2585 * in the TagHt dictionary.
2586 * @param group The generated tag must belong to this group.
2587 * @return The element of tag with given group which is fee.
2589 uint32_t Document::GenerateFreeTagKeyInGroup(uint16_t group)
2591 for (uint32_t elem = 0; elem < UINT32_MAX; elem++)
2593 TagKey key = DictEntry::TranslateToKey(group, elem);
2594 if (TagHT.count(key) == 0)
2603 * \brief Assuming the internal file pointer \ref Document::Fp
2604 * is placed at the beginning of a tag check whether this
2605 * tag is (TestGroup, TestElement).
2606 * \warning On success the internal file pointer \ref Document::Fp
2607 * is modified to point after the tag.
2608 * On failure (i.e. when the tag wasn't the expected tag
2609 * (TestGroup, TestElement) the internal file pointer
2610 * \ref Document::Fp is restored to it's original position.
2611 * @param testGroup The expected group of the tag.
2612 * @param testElement The expected Element of the tag.
2613 * @return True on success, false otherwise.
2615 bool Document::ReadTag(uint16_t testGroup, uint16_t testElement)
2617 long positionOnEntry = Fp->tellg();
2618 long currentPosition = Fp->tellg(); // On debugging purposes
2620 //// Read the Item Tag group and element, and make
2621 // sure they are what we expected:
2622 uint16_t itemTagGroup;
2623 uint16_t itemTagElement;
2626 itemTagGroup = ReadInt16();
2627 itemTagElement = ReadInt16();
2629 catch ( FormatError e )
2631 //std::cerr << e << std::endl;
2634 if ( itemTagGroup != testGroup || itemTagElement != testElement )
2636 std::ostringstream s;
2637 s << " We should have found tag (";
2638 s << std::hex << testGroup << "," << testElement << ")" << std::endl;
2639 s << " but instead we encountered tag (";
2640 s << std::hex << itemTagGroup << "," << itemTagElement << ")"
2642 s << " at address: " << (unsigned)currentPosition << std::endl;
2643 dbg.Verbose(0, "Document::ReadItemTagLength: wrong Item Tag found:");
2644 dbg.Verbose(0, s.str().c_str());
2645 Fp->seekg(positionOnEntry, std::ios::beg);
2653 * \brief Assuming the internal file pointer \ref Document::Fp
2654 * is placed at the beginning of a tag (TestGroup, TestElement),
2655 * read the length associated to the Tag.
2656 * \warning On success the internal file pointer \ref Document::Fp
2657 * is modified to point after the tag and it's length.
2658 * On failure (i.e. when the tag wasn't the expected tag
2659 * (TestGroup, TestElement) the internal file pointer
2660 * \ref Document::Fp is restored to it's original position.
2661 * @param testGroup The expected group of the tag.
2662 * @param testElement The expected Element of the tag.
2663 * @return On success returns the length associated to the tag. On failure
2666 uint32_t Document::ReadTagLength(uint16_t testGroup, uint16_t testElement)
2668 long positionOnEntry = Fp->tellg();
2669 (void)positionOnEntry;
2671 if ( !ReadTag(testGroup, testElement) )
2676 //// Then read the associated Item Length
2677 long currentPosition = Fp->tellg();
2678 uint32_t itemLength = ReadInt32();
2680 std::ostringstream s;
2681 s << "Basic Item Length is: "
2682 << itemLength << std::endl;
2683 s << " at address: " << (unsigned)currentPosition << std::endl;
2684 dbg.Verbose(0, "Document::ReadItemTagLength: ", s.str().c_str());
2690 * \brief When parsing the Pixel Data of an encapsulated file, read
2691 * the basic offset table (when present, and BTW dump it).
2693 void Document::ReadAndSkipEncapsulatedBasicOffsetTable()
2695 //// Read the Basic Offset Table Item Tag length...
2696 uint32_t itemLength = ReadTagLength(0xfffe, 0xe000);
2698 // When present, read the basic offset table itself.
2699 // Notes: - since the presence of this basic offset table is optional
2700 // we can't rely on it for the implementation, and we will simply
2701 // trash it's content (when present).
2702 // - still, when present, we could add some further checks on the
2703 // lengths, but we won't bother with such fuses for the time being.
2704 if ( itemLength != 0 )
2706 char* basicOffsetTableItemValue = new char[itemLength + 1];
2707 Fp->read(basicOffsetTableItemValue, itemLength);
2710 for (unsigned int i=0; i < itemLength; i += 4 )
2712 uint32_t individualLength = str2num( &basicOffsetTableItemValue[i],
2714 std::ostringstream s;
2715 s << " Read one length: ";
2716 s << std::hex << individualLength << std::endl;
2718 "Document::ReadAndSkipEncapsulatedBasicOffsetTable: ",
2723 delete[] basicOffsetTableItemValue;
2728 * \brief Parse pixel data from disk of [multi-]fragment RLE encoding.
2729 * Compute the RLE extra information and store it in \ref RLEInfo
2730 * for later pixel retrieval usage.
2732 void Document::ComputeRLEInfo()
2734 TransferSyntaxType ts = GetTransferSyntax();
2735 if ( ts != RLELossless )
2740 // Encoded pixel data: for the time being we are only concerned with
2741 // Jpeg or RLE Pixel data encodings.
2742 // As stated in PS 3.5-2003, section 8.2 p44:
2743 // "If sent in Encapsulated Format (i.e. other than the Native Format) the
2744 // value representation OB is used".
2745 // Hence we expect an OB value representation. Concerning OB VR,
2746 // the section PS 3.5-2003, section A.4.c p 58-59, states:
2747 // "For the Value Representations OB and OW, the encoding shall meet the
2748 // following specifications depending on the Data element tag:"
2750 // - the first item in the sequence of items before the encoded pixel
2751 // data stream shall be basic offset table item. The basic offset table
2752 // item value, however, is not required to be present"
2754 ReadAndSkipEncapsulatedBasicOffsetTable();
2756 // Encapsulated RLE Compressed Images (see PS 3.5-2003, Annex G)
2757 // Loop on the individual frame[s] and store the information
2758 // on the RLE fragments in a RLEFramesInfo.
2759 // Note: - when only a single frame is present, this is a
2761 // - when more than one frame are present, then we are in
2762 // the case of a multi-frame image.
2764 while ( (frameLength = ReadTagLength(0xfffe, 0xe000)) )
2766 // Parse the RLE Header and store the corresponding RLE Segment
2767 // Offset Table information on fragments of this current Frame.
2768 // Note that the fragment pixels themselves are not loaded
2769 // (but just skipped).
2770 long frameOffset = Fp->tellg();
2772 uint32_t nbRleSegments = ReadInt32();
2773 if ( nbRleSegments > 16 )
2775 // There should be at most 15 segments (refer to RLEFrame class)
2776 dbg.Verbose(0, "Document::ComputeRLEInfo: too many segments.");
2779 uint32_t rleSegmentOffsetTable[16];
2780 for( int k = 1; k <= 15; k++ )
2782 rleSegmentOffsetTable[k] = ReadInt32();
2785 // Deduce from both the RLE Header and the frameLength the
2786 // fragment length, and again store this info in a
2788 long rleSegmentLength[15];
2789 // skipping (not reading) RLE Segments
2790 if ( nbRleSegments > 1)
2792 for(unsigned int k = 1; k <= nbRleSegments-1; k++)
2794 rleSegmentLength[k] = rleSegmentOffsetTable[k+1]
2795 - rleSegmentOffsetTable[k];
2796 SkipBytes(rleSegmentLength[k]);
2800 rleSegmentLength[nbRleSegments] = frameLength
2801 - rleSegmentOffsetTable[nbRleSegments];
2802 SkipBytes(rleSegmentLength[nbRleSegments]);
2804 // Store the collected info
2805 RLEFrame* newFrameInfo = new RLEFrame;
2806 newFrameInfo->NumberFragments = nbRleSegments;
2807 for( unsigned int uk = 1; uk <= nbRleSegments; uk++ )
2809 newFrameInfo->Offset[uk] = frameOffset + rleSegmentOffsetTable[uk];
2810 newFrameInfo->Length[uk] = rleSegmentLength[uk];
2812 RLEInfo->Frames.push_back( newFrameInfo );
2815 // Make sure that at the end of the item we encounter a 'Sequence
2817 if ( !ReadTag(0xfffe, 0xe0dd) )
2819 dbg.Verbose(0, "Document::ComputeRLEInfo: no sequence delimiter ");
2820 dbg.Verbose(0, " item at end of RLE item sequence");
2825 * \brief Parse pixel data from disk of [multi-]fragment Jpeg encoding.
2826 * Compute the jpeg extra information (fragment[s] offset[s] and
2827 * length) and store it[them] in \ref JPEGInfo for later pixel
2830 void Document::ComputeJPEGFragmentInfo()
2832 // If you need to, look for comments of ComputeRLEInfo().
2838 ReadAndSkipEncapsulatedBasicOffsetTable();
2840 // Loop on the fragments[s] and store the parsed information in a
2842 long fragmentLength;
2843 while ( (fragmentLength = ReadTagLength(0xfffe, 0xe000)) )
2845 long fragmentOffset = Fp->tellg();
2847 // Store the collected info
2848 JPEGFragment* newFragment = new JPEGFragment;
2849 newFragment->Offset = fragmentOffset;
2850 newFragment->Length = fragmentLength;
2851 JPEGInfo->Fragments.push_back( newFragment );
2853 SkipBytes( fragmentLength );
2856 // Make sure that at the end of the item we encounter a 'Sequence
2858 if ( !ReadTag(0xfffe, 0xe0dd) )
2860 dbg.Verbose(0, "Document::ComputeRLEInfo: no sequence delimiter ");
2861 dbg.Verbose(0, " item at end of JPEG item sequence");
2866 * \brief Walk recursively the given \ref DocEntrySet, and feed
2867 * the given hash table (\ref TagDocEntryHT) with all the
2868 * \ref DocEntry (Dicom entries) encountered.
2869 * This method does the job for \ref BuildFlatHashTable.
2870 * @param builtHT Where to collect all the \ref DocEntry encountered
2871 * when recursively walking the given set.
2872 * @param set The structure to be traversed (recursively).
2874 void Document::BuildFlatHashTableRecurse( TagDocEntryHT& builtHT,
2877 if (ElementSet* elementSet = dynamic_cast< ElementSet* > ( set ) )
2879 TagDocEntryHT const & currentHT = elementSet->GetTagHT();
2880 for( TagDocEntryHT::const_iterator i = currentHT.begin();
2881 i != currentHT.end();
2884 DocEntry* entry = i->second;
2885 if ( SeqEntry* seqEntry = dynamic_cast<SeqEntry*>(entry) )
2887 const ListSQItem& items = seqEntry->GetSQItems();
2888 for( ListSQItem::const_iterator item = items.begin();
2889 item != items.end();
2892 BuildFlatHashTableRecurse( builtHT, *item );
2896 builtHT[entry->GetKey()] = entry;
2901 if (SQItem* SQItemSet = dynamic_cast< SQItem* > ( set ) )
2903 const ListDocEntry& currentList = SQItemSet->GetDocEntries();
2904 for (ListDocEntry::const_iterator i = currentList.begin();
2905 i != currentList.end();
2908 DocEntry* entry = *i;
2909 if ( SeqEntry* seqEntry = dynamic_cast<SeqEntry*>(entry) )
2911 const ListSQItem& items = seqEntry->GetSQItems();
2912 for( ListSQItem::const_iterator item = items.begin();
2913 item != items.end();
2916 BuildFlatHashTableRecurse( builtHT, *item );
2920 builtHT[entry->GetKey()] = entry;
2927 * \brief Build a \ref TagDocEntryHT (i.e. a std::map<>) from the current
2930 * The structure used by a Document (through \ref ElementSet),
2931 * in order to hold the parsed entries of a Dicom header, is a recursive
2932 * one. This is due to the fact that the sequences (when present)
2933 * can be nested. Additionaly, the sequence items (represented in
2934 * gdcm as \ref SQItem) add an extra complexity to the data
2935 * structure. Hence, a gdcm user whishing to visit all the entries of
2936 * a Dicom header will need to dig in the gdcm internals (which
2937 * implies exposing all the internal data structures to the API).
2938 * In order to avoid this burden to the user, \ref BuildFlatHashTable
2939 * recursively builds a temporary hash table, which holds all the
2940 * Dicom entries in a flat structure (a \ref TagDocEntryHT i.e. a
2942 * \warning Of course there is NO integrity constrain between the
2943 * returned \ref TagDocEntryHT and the \ref ElementSet used
2944 * to build it. Hence if the underlying \ref ElementSet is
2945 * altered, then it is the caller responsability to invoke
2946 * \ref BuildFlatHashTable again...
2947 * @return The flat std::map<> we juste build.
2949 TagDocEntryHT* Document::BuildFlatHashTable()
2951 TagDocEntryHT* FlatHT = new TagDocEntryHT;
2952 BuildFlatHashTableRecurse( *FlatHT, this );
2959 * \brief Compares two documents, according to \ref DicomDir rules
2960 * \warning Does NOT work with ACR-NEMA files
2961 * \todo Find a trick to solve the pb (use RET fields ?)
2963 * @return true if 'smaller'
2965 bool Document::operator<(Document &document)
2968 std::string s1 = GetEntryByNumber(0x0010,0x0010);
2969 std::string s2 = document.GetEntryByNumber(0x0010,0x0010);
2981 s1 = GetEntryByNumber(0x0010,0x0020);
2982 s2 = document.GetEntryByNumber(0x0010,0x0020);
2993 // Study Instance UID
2994 s1 = GetEntryByNumber(0x0020,0x000d);
2995 s2 = document.GetEntryByNumber(0x0020,0x000d);
3006 // Serie Instance UID
3007 s1 = GetEntryByNumber(0x0020,0x000e);
3008 s2 = document.GetEntryByNumber(0x0020,0x000e);
3025 * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader
3026 * @param filetype Type of the File to be written
3028 int Document::ComputeGroup0002Length( FileType filetype )
3033 int groupLength = 0;
3034 bool found0002 = false;
3036 // for each zero-level Tag in the DCM Header
3040 entry = GetNextEntry();
3043 gr = entry->GetGroup();
3049 el = entry->GetElement();
3050 vr = entry->GetVR();
3052 if (filetype == ExplicitVR)
3054 if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") )
3056 groupLength += 4; // explicit VR AND OB, OW, SQ : 4 more bytes
3059 groupLength += 2 + 2 + 4 + entry->GetLength();
3061 else if (found0002 )
3064 entry = GetNextEntry();
3069 } // end namespace gdcm
3071 //-----------------------------------------------------------------------------