1 // $Header: /cvs/public/gdcm/src/Attic/gdcmHeader.cxx,v 1.64 2003/05/07 12:49:10 frog Exp $
9 #include <netinet/in.h>
11 #include <cctype> // for isalpha
14 #include "gdcmHeader.h"
16 // Refer to gdcmHeader::CheckSwap()
17 #define HEADER_LENGTH_TO_READ 256
18 // Refer to gdcmHeader::SetMaxSizeLoadElementValue()
19 #define _MaxSizeLoadElementValue_ 1024
21 void gdcmHeader::Initialise(void) {
22 dicom_vr = gdcmGlobal::GetVR();
23 Dicts = gdcmGlobal::GetDicts();
24 RefPubDict = Dicts->GetDefaultPubDict();
25 RefShaDict = (gdcmDict*)0;
28 gdcmHeader::gdcmHeader(const char *InFilename, bool exception_on_error) {
29 SetMaxSizeLoadElementValue(_MaxSizeLoadElementValue_);
30 filename = InFilename;
32 if ( !OpenFile(exception_on_error))
39 bool gdcmHeader::OpenFile(bool exception_on_error)
40 throw(gdcmFileError) {
41 fp=fopen(filename.c_str(),"rb");
42 if(exception_on_error) {
44 throw gdcmFileError("gdcmHeader::gdcmHeader(const char *, bool)");
48 dbg.Verbose(0, "gdcmHeader::gdcmHeader cannot open file", filename.c_str());
52 bool gdcmHeader::CloseFile(void) {
53 int closed = fclose(fp);
60 gdcmHeader::~gdcmHeader (void) {
61 dicom_vr = (gdcmVR*)0;
62 Dicts = (gdcmDictSet*)0;
63 RefPubDict = (gdcmDict*)0;
64 RefShaDict = (gdcmDict*)0;
70 // META Meta Information
82 // NMI Nuclear Medicine
84 // BFS Basic Film Session
86 // BIB Basic Image Box
101 * \ingroup gdcmHeader
102 * \brief Discover what the swap code is (among little endian, big endian,
103 * bad little endian, bad big endian).
106 void gdcmHeader::CheckSwap()
108 // The only guaranted way of finding the swap code is to find a
109 // group tag since we know it's length has to be of four bytes i.e.
110 // 0x00000004. Finding the swap code in then straigthforward. Trouble
111 // occurs when we can't find such group...
113 guint32 x=4; // x : pour ntohs
114 bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
118 char deb[HEADER_LENGTH_TO_READ];
120 // First, compare HostByteOrder and NetworkByteOrder in order to
121 // determine if we shall need to swap bytes (i.e. the Endian type).
127 // The easiest case is the one of a DICOM header, since it possesses a
128 // file preamble where it suffice to look for the string "DICM".
129 lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, fp);
132 if(memcmp(entCur, "DICM", (size_t)4) == 0) {
133 dbg.Verbose(1, "gdcmHeader::CheckSwap:", "looks like DICOM Version3");
134 // Next, determine the value representation (VR). Let's skip to the
135 // first element (0002, 0000) and check there if we find "UL"
136 // - or "OB" if the 1st one is (0002,0001) -,
137 // in which case we (almost) know it is explicit VR.
138 // WARNING: if it happens to be implicit VR then what we will read
139 // is the length of the group. If this ascii representation of this
140 // length happens to be "UL" then we shall believe it is explicit VR.
141 // FIXME: in order to fix the above warning, we could read the next
142 // element value (or a couple of elements values) in order to make
143 // sure we are not commiting a big mistake.
145 // * the 128 bytes of File Preamble (often padded with zeroes),
146 // * the 4 bytes of "DICM" string,
147 // * the 4 bytes of the first tag (0002, 0000),or (0002, 0001)
148 // i.e. a total of 136 bytes.
151 // Use gdcmHeader::dicom_vr to test all the possibilities
152 // instead of just checking for UL, OB and UI !?
153 if( (memcmp(entCur, "UL", (size_t)2) == 0) ||
154 (memcmp(entCur, "OB", (size_t)2) == 0) ||
155 (memcmp(entCur, "UI", (size_t)2) == 0) )
157 filetype = ExplicitVR;
158 dbg.Verbose(1, "gdcmHeader::CheckSwap:",
159 "explicit Value Representation");
161 filetype = ImplicitVR;
162 dbg.Verbose(1, "gdcmHeader::CheckSwap:",
163 "not an explicit Value Representation");
168 dbg.Verbose(1, "gdcmHeader::CheckSwap:",
169 "HostByteOrder != NetworkByteOrder");
172 dbg.Verbose(1, "gdcmHeader::CheckSwap:",
173 "HostByteOrder = NetworkByteOrder");
176 // Position the file position indicator at first tag (i.e.
177 // after the file preamble and the "DICM" string).
179 fseek (fp, 132L, SEEK_SET);
183 // Alas, this is not a DicomV3 file and whatever happens there is no file
184 // preamble. We can reset the file position indicator to where the data
185 // is (i.e. the beginning of the file).
186 dbg.Verbose(1, "gdcmHeader::CheckSwap:", "not a DICOM Version3 file");
189 // Our next best chance would be to be considering a 'clean' ACR/NEMA file.
190 // By clean we mean that the length of the first tag is written down.
191 // If this is the case and since the length of the first group HAS to be
192 // four (bytes), then determining the proper swap code is straightforward.
195 // We assume the array of char we are considering contains the binary
196 // representation of a 32 bits integer. Hence the following dirty
198 s = *((guint32 *)(entCur));
218 dbg.Verbose(0, "gdcmHeader::CheckSwap:",
219 "ACR/NEMA unfound swap info (time to raise bets)");
222 // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
223 // It is time for despaired wild guesses. So, let's assume this file
224 // happens to be 'dirty' ACR/NEMA, i.e. the length of the group is
225 // not present. Then the only info we have is the net2host one.
234 void gdcmHeader::SwitchSwapToBigEndian(void) {
235 dbg.Verbose(1, "gdcmHeader::SwitchSwapToBigEndian",
236 "Switching to BigEndian mode.");
254 * \ingroup gdcmHeader
255 * \brief Find the value representation of the current tag.
257 void gdcmHeader::FindVR( gdcmElValue *ElVal) {
258 if (filetype != ExplicitVR)
264 long PositionOnEntry = ftell(fp);
265 // Warning: we believe this is explicit VR (Value Representation) because
266 // we used a heuristic that found "UL" in the first tag. Alas this
267 // doesn't guarantee that all the tags will be in explicit VR. In some
268 // cases (see e-film filtered files) one finds implicit VR tags mixed
269 // within an explicit VR file. Hence we make sure the present tag
270 // is in explicit VR and try to fix things if it happens not to be
272 bool RealExplicit = true;
274 lgrLue=fread (&VR, (size_t)2,(size_t)1, fp);
278 // Assume we are reading a falsely explicit VR file i.e. we reached
279 // a tag where we expect reading a VR but are in fact we read the
280 // first to bytes of the length. Then we will interogate (through find)
281 // the dicom_vr dictionary with oddities like "\004\0" which crashes
282 // both GCC and VC++ implementations of the STL map. Hence when the
283 // expected VR read happens to be non-ascii characters we consider
284 // we hit falsely explicit VR tag.
286 if ( (!isalpha(VR[0])) && (!isalpha(VR[1])) )
287 RealExplicit = false;
289 // CLEANME searching the dicom_vr at each occurence is expensive.
290 // PostPone this test in an optional integrity check at the end
291 // of parsing or only in debug mode.
292 if ( RealExplicit && !dicom_vr->Count(vr) )
295 if ( RealExplicit ) {
296 if ( ElVal->IsVrUnknown() ) {
297 // When not a dictionary entry, we can safely overwrite the vr.
301 if ( ElVal->GetVR() == vr ) {
302 // The vr we just read and the dictionary agree. Nothing to do.
305 // The vr present in the file and the dictionary disagree. We assume
306 // the file writer knew best and use the vr of the file. Since it would
307 // be unwise to overwrite the vr of a dictionary (since it would
308 // compromise it's next user), we need to clone the actual DictEntry
309 // and change the vr for the read one.
310 gdcmDictEntry* NewTag = new gdcmDictEntry(ElVal->GetGroup(),
315 ElVal->SetDictEntry(NewTag);
319 // We thought this was explicit VR, but we end up with an
320 // implicit VR tag. Let's backtrack.
321 dbg.Verbose(1, "gdcmHeader::FindVR:", "Falsely explicit vr file");
322 fseek(fp, PositionOnEntry, SEEK_SET);
323 // When this element is known in the dictionary we shall use, e.g. for
324 // the semantics (see the usage of IsAnInteger), the vr proposed by the
325 // dictionary entry. Still we have to flag the element as implicit since
326 // we know now our assumption on expliciteness is not furfilled.
328 if ( ElVal->IsVrUnknown() )
329 ElVal->SetVR("Implicit");
330 ElVal->SetImplicitVr();
334 * \ingroup gdcmHeader
335 * \brief Determines if the Transfer Syntax was allready encountered
336 * and if it corresponds to a ImplicitVRLittleEndian one.
338 * @return True when ImplicitVRLittleEndian found. False in all other cases.
340 bool gdcmHeader::IsImplicitVRLittleEndianTransferSyntax(void) {
341 gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010);
344 LoadElementValueSafe(Element);
345 string Transfer = Element->GetValue();
346 if ( Transfer == "1.2.840.10008.1.2" )
352 * \ingroup gdcmHeader
353 * \brief Determines if the Transfer Syntax was allready encountered
354 * and if it corresponds to a ExplicitVRLittleEndian one.
356 * @return True when ExplicitVRLittleEndian found. False in all other cases.
358 bool gdcmHeader::IsExplicitVRLittleEndianTransferSyntax(void) {
359 gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010);
362 LoadElementValueSafe(Element);
363 string Transfer = Element->GetValue();
364 if ( Transfer == "1.2.840.10008.1.2.1" )
370 * \ingroup gdcmHeader
371 * \brief Determines if the Transfer Syntax was allready encountered
372 * and if it corresponds to a DeflatedExplicitVRLittleEndian one.
374 * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases.
376 bool gdcmHeader::IsDeflatedExplicitVRLittleEndianTransferSyntax(void) {
377 gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010);
380 LoadElementValueSafe(Element);
381 string Transfer = Element->GetValue();
382 if ( Transfer == "1.2.840.10008.1.2.1.99" )
388 * \ingroup gdcmHeader
389 * \brief Determines if the Transfer Syntax was allready encountered
390 * and if it corresponds to a Explicit VR Big Endian one.
392 * @return True when big endian found. False in all other cases.
394 bool gdcmHeader::IsExplicitVRBigEndianTransferSyntax(void) {
395 gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010);
398 LoadElementValueSafe(Element);
399 string Transfer = Element->GetValue();
400 if ( Transfer == "1.2.840.10008.1.2.2" ) //1.2.2 ??? A verifier !
406 * \ingroup gdcmHeader
407 * \brief Determines if the Transfer Syntax was allready encountered
408 * and if it corresponds to a JPEGBaseLineProcess1 one.
410 * @return True when JPEGBaseLineProcess1found. False in all other cases.
412 bool gdcmHeader::IsJPEGBaseLineProcess1TransferSyntax(void) {
413 gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010);
416 LoadElementValueSafe(Element);
417 string Transfer = Element->GetValue();
418 if ( Transfer == "1.2.840.10008.1.2.4.50" )
423 // faire qq chose d'intelligent a la place de ça
425 bool gdcmHeader::IsJPEGLossless(void) {
426 gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010);
429 LoadElementValueSafe(Element);
430 const char * Transfert = Element->GetValue().c_str();
431 printf("TransfertSyntx %s\n",Transfert);
432 if ( memcmp(Transfert+strlen(Transfert)-2 ,"70",2)==0) return true;
433 if ( memcmp(Transfert+strlen(Transfert)-2 ,"55",2)==0) return true;
439 * \ingroup gdcmHeader
440 * \brief Determines if the Transfer Syntax was allready encountered
441 * and if it corresponds to a JPEGExtendedProcess2-4 one.
443 * @return True when JPEGExtendedProcess2-4 found. False in all other cases.
445 bool gdcmHeader::IsJPEGExtendedProcess2_4TransferSyntax(void) {
446 gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010);
449 LoadElementValueSafe(Element);
450 string Transfer = Element->GetValue();
451 if ( Transfer == "1.2.840.10008.1.2.4.51" )
457 * \ingroup gdcmHeader
458 * \brief Determines if the Transfer Syntax was allready encountered
459 * and if it corresponds to a JPEGExtendeProcess3-5 one.
461 * @return True when JPEGExtendedProcess3-5 found. False in all other cases.
463 bool gdcmHeader::IsJPEGExtendedProcess3_5TransferSyntax(void) {
464 gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010);
467 LoadElementValueSafe(Element);
468 string Transfer = Element->GetValue();
469 if ( Transfer == "1.2.840.10008.1.2.4.52" )
475 * \ingroup gdcmHeader
476 * \brief Determines if the Transfer Syntax was allready encountered
477 * and if it corresponds to a JPEGSpectralSelectionProcess6-8 one.
479 * @return True when JPEGSpectralSelectionProcess6-8 found. False in all
482 bool gdcmHeader::IsJPEGSpectralSelectionProcess6_8TransferSyntax(void) {
483 gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010);
486 LoadElementValueSafe(Element);
487 string Transfer = Element->GetValue();
488 if ( Transfer == "1.2.840.10008.1.2.4.53" )
493 * \ingroup gdcmHeader
494 * \brief Predicate for dicom version 3 file.
495 * @return True when the file is a dicom version 3.
497 bool gdcmHeader::IsDicomV3(void) {
498 if ( (filetype == ExplicitVR)
499 || (filetype == ImplicitVR) )
505 * \ingroup gdcmHeader
506 * \brief When the length of an element value is obviously wrong (because
507 * the parser went Jabberwocky) one can hope improving things by
508 * applying this heuristic.
510 void gdcmHeader::FixFoundLength(gdcmElValue * ElVal, guint32 FoundLength) {
511 if ( FoundLength == 0xffffffff)
513 ElVal->SetLength(FoundLength);
516 guint32 gdcmHeader::FindLengthOB(void) {
517 // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
520 long PositionOnEntry = ftell(fp);
521 bool FoundSequenceDelimiter = false;
522 guint32 TotalLength = 0;
525 while ( ! FoundSequenceDelimiter) {
530 TotalLength += 4; // We even have to decount the group and element
532 dbg.Verbose(1, "gdcmHeader::FindLengthOB: ",
533 "wrong group for an item sequence.");
538 FoundSequenceDelimiter = true;
539 else if ( n != 0xe000) {
540 dbg.Verbose(1, "gdcmHeader::FindLengthOB: ",
541 "wrong element for an item sequence.");
545 ItemLength = ReadInt32();
546 TotalLength += ItemLength + 4; // We add 4 bytes since we just read
547 // the ItemLength with ReadInt32
548 SkipBytes(ItemLength);
550 fseek(fp, PositionOnEntry, SEEK_SET);
554 void gdcmHeader::FindLength(gdcmElValue * ElVal) {
555 guint16 element = ElVal->GetElement();
556 string vr = ElVal->GetVR();
559 if ( (filetype == ExplicitVR) && ! ElVal->IsImplicitVr() ) {
561 if ( (vr=="OB") || (vr=="OW") || (vr=="SQ") || (vr=="UN") ) {
562 // The following reserved two bytes (see PS 3.5-2001, section
563 // 7.1.2 Data element structure with explicit vr p27) must be
564 // skipped before proceeding on reading the length on 4 bytes.
565 fseek(fp, 2L, SEEK_CUR);
566 guint32 length32 = ReadInt32();
567 if ( (vr == "OB") && (length32 == 0xffffffff) ) {
568 ElVal->SetLength(FindLengthOB());
571 FixFoundLength(ElVal, length32);
575 // Length is encoded on 2 bytes.
576 length16 = ReadInt16();
578 // We can tell the current file is encoded in big endian (like
579 // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag
580 // and it's value is the one of the encoding of a big endian file.
581 // In order to deal with such big endian encoded files, we have
582 // (at least) two strategies:
583 // * when we load the "Transfer Syntax" tag with value of big endian
584 // encoding, we raise the proper flags. Then we wait for the end
585 // of the META group (0x0002) among which is "Transfer Syntax",
586 // before switching the swap code to big endian. We have to postpone
587 // the switching of the swap code since the META group is fully encoded
588 // in little endian, and big endian coding only starts at the next
589 // group. The corresponding code can be hard to analyse and adds
590 // many additional unnecessary tests for regular tags.
591 // * the second strategy consists in waiting for trouble, that shall
592 // appear when we find the first group with big endian encoding. This
593 // is easy to detect since the length of a "Group Length" tag (the
594 // ones with zero as element number) has to be of 4 (0x0004). When we
595 // encouter 1024 (0x0400) chances are the encoding changed and we
596 // found a group with big endian encoding.
597 // We shall use this second strategy. In order to make sure that we
598 // can interpret the presence of an apparently big endian encoded
599 // length of a "Group Length" without committing a big mistake, we
600 // add an additional check: we look in the allready parsed elements
601 // for the presence of a "Transfer Syntax" whose value has to be "big
602 // endian encoding". When this is the case, chances are we have got our
603 // hands on a big endian encoded file: we switch the swap code to
604 // big endian and proceed...
605 if ( (element == 0x000) && (length16 == 0x0400) ) {
606 if ( ! IsExplicitVRBigEndianTransferSyntax() ) {
607 dbg.Verbose(0, "gdcmHeader::FindLength", "not explicit VR");
612 SwitchSwapToBigEndian();
613 // Restore the unproperly loaded values i.e. the group, the element
614 // and the dictionary entry depending on them.
615 guint16 CorrectGroup = SwapShort(ElVal->GetGroup());
616 guint16 CorrectElem = SwapShort(ElVal->GetElement());
617 gdcmDictEntry * NewTag = GetDictEntryByNumber(CorrectGroup,
620 // This correct tag is not in the dictionary. Create a new one.
621 NewTag = new gdcmDictEntry(CorrectGroup, CorrectElem);
623 // FIXME this can create a memory leaks on the old entry that be
624 // left unreferenced.
625 ElVal->SetDictEntry(NewTag);
628 // Heuristic: well some files are really ill-formed.
629 if ( length16 == 0xffff) {
631 dbg.Verbose(0, "gdcmHeader::FindLength",
632 "Erroneous element length fixed.");
634 FixFoundLength(ElVal, (guint32)length16);
638 // Either implicit VR or a non DICOM conformal (see not below) explicit
639 // VR that ommited the VR of (at least) this element. Farts happen.
640 // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25
641 // on Data elements "Implicit and Explicit VR Data Elements shall
642 // not coexist in a Data Set and Data Sets nested within it".]
643 // Length is on 4 bytes.
644 FixFoundLength(ElVal, ReadInt32());
648 * \ingroup gdcmHeader
649 * \brief Swaps back the bytes of 4-byte long integer accordingly to
652 * @return The suggested integer.
654 guint32 gdcmHeader::SwapLong(guint32 a) {
659 a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) |
660 ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) );
664 a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
668 a=( ((a<<8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
671 dbg.Error(" gdcmHeader::SwapLong : unset swap code");
678 * \ingroup gdcmHeader
679 * \brief Swaps the bytes so they agree with the processor order
680 * @return The properly swaped 16 bits integer.
682 guint16 gdcmHeader::SwapShort(guint16 a) {
683 if ( (sw==4321) || (sw==2143) )
684 a =(((a<<8) & 0x0ff00) | ((a>>8)&0x00ff));
688 void gdcmHeader::SkipBytes(guint32 NBytes) {
689 //FIXME don't dump the returned value
690 (void)fseek(fp, (long)NBytes, SEEK_CUR);
693 void gdcmHeader::SkipElementValue(gdcmElValue * ElVal) {
694 SkipBytes(ElVal->GetLength());
697 void gdcmHeader::SetMaxSizeLoadElementValue(long NewSize) {
700 if ((guint32)NewSize >= (guint32)0xffffffff) {
701 MaxSizeLoadElementValue = 0xffffffff;
704 MaxSizeLoadElementValue = NewSize;
708 * \ingroup gdcmHeader
709 * \brief Loads the element content if it's length is not bigger
710 * than the value specified with
711 * gdcmHeader::SetMaxSizeLoadElementValue()
713 void gdcmHeader::LoadElementValue(gdcmElValue * ElVal) {
715 guint16 group = ElVal->GetGroup();
716 guint16 elem = ElVal->GetElement();
717 string vr = ElVal->GetVR();
718 guint32 length = ElVal->GetLength();
719 bool SkipLoad = false;
721 fseek(fp, (long)ElVal->GetOffset(), SEEK_SET);
723 // FIXME Sequences not treated yet !
725 // Ne faudrait-il pas au contraire trouver immediatement
726 // une maniere 'propre' de traiter les sequences (vr = SQ)
727 // car commencer par les ignorer risque de conduire a qq chose
728 // qui pourrait ne pas etre generalisable
729 // Well, I'm expecting your code !!!
734 // Heuristic : a sequence "contains" a set of tags (called items). It looks
735 // like the last tag of a sequence (the one that terminates the sequence)
736 // has a group of 0xfffe (with a dummy length).
737 if( group == 0xfffe )
742 ElVal->SetValue("gdcm::Skipped");
746 // When the length is zero things are easy:
752 // The elements whose length is bigger than the specified upper bound
753 // are not loaded. Instead we leave a short notice of the offset of
754 // the element content and it's length.
755 if (length > MaxSizeLoadElementValue) {
757 s << "gdcm::NotLoaded.";
758 s << " Address:" << (long)ElVal->GetOffset();
759 s << " Length:" << ElVal->GetLength();
760 ElVal->SetValue(s.str());
764 // When an integer is expected, read and convert the following two or
765 // four bytes properly i.e. as an integer as opposed to a string.
767 // pour les elements de Value Multiplicity > 1
768 // on aura en fait une serie d'entiers
770 // on devrait pouvoir faire + compact (?)
772 if ( IsAnInteger(ElVal) ) {
776 if (vr == "US" || vr == "SS") {
778 NewInt = ReadInt16();
781 for (int i=1; i < nbInt; i++) {
783 NewInt = ReadInt16();
788 } else if (vr == "UL" || vr == "SL") {
790 NewInt = ReadInt32();
793 for (int i=1; i < nbInt; i++) {
795 NewInt = ReadInt32();
800 ElVal->SetValue(s.str());
804 // We need an additional byte for storing \0 that is not on disk
805 char* NewValue = (char*)malloc(length+1);
807 dbg.Verbose(1, "LoadElementValue: Failed to allocate NewValue");
812 item_read = fread(NewValue, (size_t)length, (size_t)1, fp);
813 if ( item_read != 1 ) {
815 dbg.Verbose(1, "gdcmHeader::LoadElementValue","unread element value");
816 ElVal->SetValue("gdcm::UnRead");
819 ElVal->SetValue(NewValue);
824 * \ingroup gdcmHeader
825 * \brief Loads the element while preserving the current
826 * underlying file position indicator as opposed to
827 * to LoadElementValue that modifies it.
828 * @param ElVal Element whose value shall be loaded.
831 void gdcmHeader::LoadElementValueSafe(gdcmElValue * ElVal) {
832 long PositionOnEntry = ftell(fp);
833 LoadElementValue(ElVal);
834 fseek(fp, PositionOnEntry, SEEK_SET);
838 guint16 gdcmHeader::ReadInt16(void) {
841 item_read = fread (&g, (size_t)2,(size_t)1, fp);
843 if ( item_read != 1 ) {
844 dbg.Verbose(1, "gdcmHeader::ReadInt16", " File read error");
852 guint32 gdcmHeader::ReadInt32(void) {
855 item_read = fread (&g, (size_t)4,(size_t)1, fp);
857 if ( item_read != 1 ) {
858 dbg.Verbose(1, "gdcmHeader::ReadInt32", " File read error");
867 gdcmElValue* gdcmHeader::GetElValueByNumber(guint16 Group, guint16 Elem) {
869 gdcmElValue* elValue = PubElValSet.GetElementByNumber(Group, Elem);
871 dbg.Verbose(1, "gdcmHeader::GetElValueByNumber",
872 "failed to Locate gdcmElValue");
873 return (gdcmElValue*)0;
879 * \ingroup gdcmHeader
880 * \brief Build a new Element Value from all the low level arguments.
881 * Check for existence of dictionary entry, and build
882 * a default one when absent.
883 * @param Group group of the underlying DictEntry
884 * @param Elem element of the underlying DictEntry
886 gdcmElValue* gdcmHeader::NewElValueByNumber(guint16 Group, guint16 Elem) {
887 // Find out if the tag we encountered is in the dictionaries:
888 gdcmDictEntry * NewTag = GetDictEntryByNumber(Group, Elem);
890 NewTag = new gdcmDictEntry(Group, Elem);
892 gdcmElValue* NewElVal = new gdcmElValue(NewTag);
894 dbg.Verbose(1, "gdcmHeader::NewElValueByNumber",
895 "failed to allocate gdcmElValue");
896 return (gdcmElValue*)0;
902 * \ingroup gdcmHeader
906 int gdcmHeader::ReplaceOrCreateByNumber(string Value, guint16 Group, guint16 Elem ) {
908 gdcmElValue* nvElValue=NewElValueByNumber(Group, Elem);
909 PubElValSet.Add(nvElValue);
910 PubElValSet.SetElValueByNumber(Value, Group, Elem);
916 * \ingroup gdcmHeader
917 * \brief Build a new Element Value from all the low level arguments.
918 * Check for existence of dictionary entry, and build
919 * a default one when absent.
920 * @param Name Name of the underlying DictEntry
922 gdcmElValue* gdcmHeader::NewElValueByName(string Name) {
924 gdcmDictEntry * NewTag = GetDictEntryByName(Name);
926 NewTag = new gdcmDictEntry(0xffff, 0xffff, "LO", "Unknown", Name);
928 gdcmElValue* NewElVal = new gdcmElValue(NewTag);
930 dbg.Verbose(1, "gdcmHeader::ObtainElValueByName",
931 "failed to allocate gdcmElValue");
932 return (gdcmElValue*)0;
938 * \ingroup gdcmHeader
939 * \brief Read the next tag but WITHOUT loading it's value
940 * @return On succes the newly created ElValue, NULL on failure.
942 gdcmElValue * gdcmHeader::ReadNextElement(void) {
945 gdcmElValue * NewElVal;
950 // We reached the EOF (or an error occured) and header parsing
951 // has to be considered as finished.
952 return (gdcmElValue *)0;
954 NewElVal = NewElValueByNumber(g, n);
956 FindLength(NewElVal);
959 return (gdcmElValue *)0;
960 NewElVal->SetOffset(ftell(fp));
965 * \ingroup gdcmHeader
966 * \brief Apply some heuristics to predict wether the considered
967 * element value contains/represents an integer or not.
968 * @param ElVal The element value on which to apply the predicate.
969 * @return The result of the heuristical predicate.
971 bool gdcmHeader::IsAnInteger(gdcmElValue * ElVal) {
972 guint16 group = ElVal->GetGroup();
973 guint16 element = ElVal->GetElement();
974 string vr = ElVal->GetVR();
975 guint32 length = ElVal->GetLength();
977 // When we have some semantics on the element we just read, and if we
978 // a priori know we are dealing with an integer, then we shall be
979 // able to swap it's element value properly.
980 if ( element == 0 ) { // This is the group length of the group
984 printf("Erroneous Group Length element length (%04x , %04x) : %d\n",
985 group, element,length);
987 dbg.Error("gdcmHeader::IsAnInteger",
988 "Erroneous Group Length element length.");
992 if ( (vr == "UL") || (vr == "US") || (vr == "SL") || (vr == "SS") )
999 * \ingroup gdcmHeader
1000 * \brief Recover the offset (from the beginning of the file) of the pixels.
1002 size_t gdcmHeader::GetPixelOffset(void) {
1003 // If this file complies with the norm we should encounter the
1004 // "Image Location" tag (0x0028, 0x0200). This tag contains the
1005 // the group that contains the pixel data (hence the "Pixel Data"
1006 // is found by indirection through the "Image Location").
1007 // Inside the group pointed by "Image Location" the searched element
1008 // is conventionally the element 0x0010 (when the norm is respected).
1009 // When the "Image Location" is absent we default to group 0x7fe0.
1012 string ImageLocation = GetPubElValByName("Image Location");
1013 if ( ImageLocation == "gdcm::Unfound" ) {
1016 grPixel = (guint16) atoi( ImageLocation.c_str() );
1018 if (grPixel != 0x7fe0)
1019 // This is a kludge for old dirty Philips imager.
1023 gdcmElValue* PixelElement = PubElValSet.GetElementByNumber(grPixel,
1026 return PixelElement->GetOffset();
1032 * \ingroup gdcmHeader
1033 * \brief Searches both the public and the shadow dictionary (when they
1034 * exist) for the presence of the DictEntry with given
1035 * group and element. The public dictionary has precedence on the
1037 * @param group group of the searched DictEntry
1038 * @param element element of the searched DictEntry
1039 * @return Corresponding DictEntry when it exists, NULL otherwise.
1041 gdcmDictEntry * gdcmHeader::GetDictEntryByNumber(guint16 group,
1043 gdcmDictEntry * found = (gdcmDictEntry*)0;
1044 if (!RefPubDict && !RefShaDict) {
1045 dbg.Verbose(0, "gdcmHeader::GetDictEntry",
1046 "we SHOULD have a default dictionary");
1049 found = RefPubDict->GetTagByNumber(group, element);
1054 found = RefShaDict->GetTagByNumber(group, element);
1062 * \ingroup gdcmHeader
1063 * \brief Searches both the public and the shadow dictionary (when they
1064 * exist) for the presence of the DictEntry with given name.
1065 * The public dictionary has precedence on the shadow one.
1066 * @param Name name of the searched DictEntry
1067 * @return Corresponding DictEntry when it exists, NULL otherwise.
1069 gdcmDictEntry * gdcmHeader::GetDictEntryByName(string Name) {
1070 gdcmDictEntry * found = (gdcmDictEntry*)0;
1071 if (!RefPubDict && !RefShaDict) {
1072 dbg.Verbose(0, "gdcmHeader::GetDictEntry",
1073 "we SHOULD have a default dictionary");
1076 found = RefPubDict->GetTagByName(Name);
1081 found = RefShaDict->GetTagByName(Name);
1089 * \ingroup gdcmHeader
1090 * \brief Searches within the public dictionary for element value of
1092 * @param group Group of the researched tag.
1093 * @param element Element of the researched tag.
1094 * @return Corresponding element value when it exists, and the string
1095 * "gdcm::Unfound" otherwise.
1097 string gdcmHeader::GetPubElValByNumber(guint16 group, guint16 element) {
1098 return PubElValSet.GetElValueByNumber(group, element);
1102 * \ingroup gdcmHeader
1103 * \brief Searches within the public dictionary for element value
1104 * representation of a given tag.
1106 * Obtaining the VR (Value Representation) might be needed by caller
1107 * to convert the string typed content to caller's native type
1108 * (think of C++ vs Python). The VR is actually of a higher level
1109 * of semantics than just the native C++ type.
1110 * @param group Group of the researched tag.
1111 * @param element Element of the researched tag.
1112 * @return Corresponding element value representation when it exists,
1113 * and the string "gdcm::Unfound" otherwise.
1115 string gdcmHeader::GetPubElValRepByNumber(guint16 group, guint16 element) {
1116 gdcmElValue* elem = PubElValSet.GetElementByNumber(group, element);
1118 return "gdcm::Unfound";
1119 return elem->GetVR();
1123 * \ingroup gdcmHeader
1124 * \brief Searches within the public dictionary for element value of
1126 * @param TagName name of the researched element.
1127 * @return Corresponding element value when it exists, and the string
1128 * "gdcm::Unfound" otherwise.
1130 string gdcmHeader::GetPubElValByName(string TagName) {
1131 return PubElValSet.GetElValueByName(TagName);
1135 * \ingroup gdcmHeader
1136 * \brief Searches within the elements parsed with the public dictionary for
1137 * the element value representation of a given tag.
1139 * Obtaining the VR (Value Representation) might be needed by caller
1140 * to convert the string typed content to caller's native type
1141 * (think of C++ vs Python). The VR is actually of a higher level
1142 * of semantics than just the native C++ type.
1143 * @param TagName name of the researched element.
1144 * @return Corresponding element value representation when it exists,
1145 * and the string "gdcm::Unfound" otherwise.
1147 string gdcmHeader::GetPubElValRepByName(string TagName) {
1148 gdcmElValue* elem = PubElValSet.GetElementByName(TagName);
1150 return "gdcm::Unfound";
1151 return elem->GetVR();
1155 * \ingroup gdcmHeader
1156 * \brief Searches within elements parsed with the SHADOW dictionary
1157 * for the element value of a given tag.
1158 * @param group Group of the researched tag.
1159 * @param element Element of the researched tag.
1160 * @return Corresponding element value representation when it exists,
1161 * and the string "gdcm::Unfound" otherwise.
1163 string gdcmHeader::GetShaElValByNumber(guint16 group, guint16 element) {
1164 return ShaElValSet.GetElValueByNumber(group, element);
1168 * \ingroup gdcmHeader
1169 * \brief Searches within the elements parsed with the SHADOW dictionary
1170 * for the element value representation of a given tag.
1172 * Obtaining the VR (Value Representation) might be needed by caller
1173 * to convert the string typed content to caller's native type
1174 * (think of C++ vs Python). The VR is actually of a higher level
1175 * of semantics than just the native C++ type.
1176 * @param group Group of the researched tag.
1177 * @param element Element of the researched tag.
1178 * @return Corresponding element value representation when it exists,
1179 * and the string "gdcm::Unfound" otherwise.
1181 string gdcmHeader::GetShaElValRepByNumber(guint16 group, guint16 element) {
1182 gdcmElValue* elem = ShaElValSet.GetElementByNumber(group, element);
1184 return "gdcm::Unfound";
1185 return elem->GetVR();
1189 * \ingroup gdcmHeader
1190 * \brief Searches within the elements parsed with the shadow dictionary
1191 * for an element value of given tag.
1192 * @param TagName name of the researched element.
1193 * @return Corresponding element value when it exists, and the string
1194 * "gdcm::Unfound" otherwise.
1196 string gdcmHeader::GetShaElValByName(string TagName) {
1197 return ShaElValSet.GetElValueByName(TagName);
1201 * \ingroup gdcmHeader
1202 * \brief Searches within the elements parsed with the shadow dictionary for
1203 * the element value representation of a given tag.
1205 * Obtaining the VR (Value Representation) might be needed by caller
1206 * to convert the string typed content to caller's native type
1207 * (think of C++ vs Python). The VR is actually of a higher level
1208 * of semantics than just the native C++ type.
1209 * @param TagName name of the researched element.
1210 * @return Corresponding element value representation when it exists,
1211 * and the string "gdcm::Unfound" otherwise.
1213 string gdcmHeader::GetShaElValRepByName(string TagName) {
1214 gdcmElValue* elem = ShaElValSet.GetElementByName(TagName);
1216 return "gdcm::Unfound";
1217 return elem->GetVR();
1221 * \ingroup gdcmHeader
1222 * \brief Searches within elements parsed with the public dictionary
1223 * and then within the elements parsed with the shadow dictionary
1224 * for the element value of a given tag.
1225 * @param group Group of the researched tag.
1226 * @param element Element of the researched tag.
1227 * @return Corresponding element value representation when it exists,
1228 * and the string "gdcm::Unfound" otherwise.
1230 string gdcmHeader::GetElValByNumber(guint16 group, guint16 element) {
1231 string pub = GetPubElValByNumber(group, element);
1234 return GetShaElValByNumber(group, element);
1238 * \ingroup gdcmHeader
1239 * \brief Searches within elements parsed with the public dictionary
1240 * and then within the elements parsed with the shadow dictionary
1241 * for the element value representation of a given tag.
1243 * Obtaining the VR (Value Representation) might be needed by caller
1244 * to convert the string typed content to caller's native type
1245 * (think of C++ vs Python). The VR is actually of a higher level
1246 * of semantics than just the native C++ type.
1247 * @param group Group of the researched tag.
1248 * @param element Element of the researched tag.
1249 * @return Corresponding element value representation when it exists,
1250 * and the string "gdcm::Unfound" otherwise.
1252 string gdcmHeader::GetElValRepByNumber(guint16 group, guint16 element) {
1253 string pub = GetPubElValRepByNumber(group, element);
1256 return GetShaElValRepByNumber(group, element);
1260 * \ingroup gdcmHeader
1261 * \brief Searches within elements parsed with the public dictionary
1262 * and then within the elements parsed with the shadow dictionary
1263 * for the element value of a given tag.
1264 * @param TagName name of the researched element.
1265 * @return Corresponding element value when it exists,
1266 * and the string "gdcm::Unfound" otherwise.
1268 string gdcmHeader::GetElValByName(string TagName) {
1269 string pub = GetPubElValByName(TagName);
1272 return GetShaElValByName(TagName);
1276 * \ingroup gdcmHeader
1277 * \brief Searches within elements parsed with the public dictionary
1278 * and then within the elements parsed with the shadow dictionary
1279 * for the element value representation of a given tag.
1281 * Obtaining the VR (Value Representation) might be needed by caller
1282 * to convert the string typed content to caller's native type
1283 * (think of C++ vs Python). The VR is actually of a higher level
1284 * of semantics than just the native C++ type.
1285 * @param TagName name of the researched element.
1286 * @return Corresponding element value representation when it exists,
1287 * and the string "gdcm::Unfound" otherwise.
1289 string gdcmHeader::GetElValRepByName(string TagName) {
1290 string pub = GetPubElValRepByName(TagName);
1293 return GetShaElValRepByName(TagName);
1297 * \ingroup gdcmHeader
1298 * \brief Accesses an existing gdcmElValue in the PubElValSet of this instance
1299 * through it's (group, element) and modifies it's content with
1301 * @param content new value to substitute with
1302 * @param group group of the ElVal to modify
1303 * @param element element of the ElVal to modify
1305 int gdcmHeader::SetPubElValByNumber(string content, guint16 group,
1308 //TODO : homogeneiser les noms : SetPubElValByNumber qui appelle PubElValSet.SetElValueByNumber
1309 // pourquoi pas SetPubElValueByNumber ??
1312 return ( PubElValSet.SetElValueByNumber (content, group, element) );
1316 * \ingroup gdcmHeader
1317 * \brief Accesses an existing gdcmElValue in the PubElValSet of this instance
1318 * through tag name and modifies it's content with the given value.
1319 * @param content new value to substitute with
1320 * @param TagName name of the tag to be modified
1322 int gdcmHeader::SetPubElValByName(string content, string TagName) {
1323 return ( PubElValSet.SetElValueByName (content, TagName) );
1327 * \ingroup gdcmHeader
1328 * \brief Accesses an existing gdcmElValue in the PubElValSet of this instance
1329 * through it's (group, element) and modifies it's length with
1331 * \warning Use with extreme caution.
1332 * @param length new length to substitute with
1333 * @param group group of the ElVal to modify
1334 * @param element element of the ElVal to modify
1335 * @return 1 on success, 0 otherwise.
1338 int gdcmHeader::SetPubElValLengthByNumber(guint32 length, guint16 group,
1340 return ( PubElValSet.SetElValueLengthByNumber (length, group, element) );
1344 * \ingroup gdcmHeader
1345 * \brief Accesses an existing gdcmElValue in the ShaElValSet of this instance
1346 * through it's (group, element) and modifies it's content with
1348 * @param content new value to substitute with
1349 * @param group group of the ElVal to modify
1350 * @param element element of the ElVal to modify
1351 * @return 1 on success, 0 otherwise.
1353 int gdcmHeader::SetShaElValByNumber(string content,
1354 guint16 group, guint16 element) {
1355 return ( ShaElValSet.SetElValueByNumber (content, group, element) );
1359 * \ingroup gdcmHeader
1360 * \brief Accesses an existing gdcmElValue in the ShaElValSet of this instance
1361 * through tag name and modifies it's content with the given value.
1362 * @param content new value to substitute with
1363 * @param TagName name of the tag to be modified
1365 int gdcmHeader::SetShaElValByName(string content, string TagName) {
1366 return ( ShaElValSet.SetElValueByName (content, TagName) );
1370 * \ingroup gdcmHeader
1371 * \brief Parses the header of the file but WITHOUT loading element values.
1373 void gdcmHeader::ParseHeader(bool exception_on_error) throw(gdcmFormatError) {
1374 gdcmElValue * newElValue = (gdcmElValue *)0;
1378 while ( (newElValue = ReadNextElement()) ) {
1379 SkipElementValue(newElValue);
1380 PubElValSet.Add(newElValue);
1385 * \ingroup gdcmHeader
1386 * \brief Retrieve the number of columns of image.
1387 * @return The encountered size when found, 0 by default.
1389 int gdcmHeader::GetXSize(void) {
1390 // We cannot check for "Columns" because the "Columns" tag is present
1391 // both in IMG (0028,0011) and OLY (6000,0011) sections of the dictionary.
1392 string StrSize = GetPubElValByNumber(0x0028,0x0011);
1393 if (StrSize == "gdcm::Unfound")
1395 return atoi(StrSize.c_str());
1399 * \ingroup gdcmHeader
1400 * \brief Retrieve the number of lines of image.
1401 * \warning The defaulted value is 1 as opposed to gdcmHeader::GetXSize()
1402 * @return The encountered size when found, 1 by default.
1404 int gdcmHeader::GetYSize(void) {
1405 // We cannot check for "Rows" because the "Rows" tag is present
1406 // both in IMG (0028,0010) and OLY (6000,0010) sections of the dictionary.
1407 string StrSize = GetPubElValByNumber(0x0028,0x0010);
1408 if (StrSize != "gdcm::Unfound")
1409 return atoi(StrSize.c_str());
1413 // The Rows (0028,0010) entry is optional for ACR/NEMA. It might
1414 // hence be a signal (1d image). So we default to 1:
1419 * \ingroup gdcmHeader
1420 * \brief Retrieve the number of planes of volume or the number
1421 * of frames of a multiframe.
1422 * \warning When present we consider the "Number of Frames" as the third
1423 * dimension. When absent we consider the third dimension as
1424 * being the "Planes" tag content.
1425 * @return The encountered size when found, 1 by default.
1427 int gdcmHeader::GetZSize(void) {
1428 // Both in DicomV3 and ACR/Nema the consider the "Number of Frames"
1429 // as the third dimension.
1430 string StrSize = GetPubElValByNumber(0x0028,0x0008);
1431 if (StrSize != "gdcm::Unfound")
1432 return atoi(StrSize.c_str());
1434 // We then consider the "Planes" entry as the third dimension [we
1435 // cannot retrieve by name since "Planes tag is present both in
1436 // IMG (0028,0012) and OLY (6000,0012) sections of the dictionary].
1437 StrSize = GetPubElValByNumber(0x0028,0x0012);
1438 if (StrSize != "gdcm::Unfound")
1439 return atoi(StrSize.c_str());
1444 * \ingroup gdcmHeader
1445 * \brief Build the Pixel Type of the image.
1446 * Possible values are:
1447 * - U8 unsigned 8 bit,
1448 * - S8 signed 8 bit,
1449 * - U16 unsigned 16 bit,
1450 * - S16 signed 16 bit,
1451 * - U32 unsigned 32 bit,
1452 * - S32 signed 32 bit,
1453 * \warning 12 bit images appear as 16 bit.
1456 string gdcmHeader::GetPixelType(void) {
1458 BitsAlloc = GetElValByName("Bits Allocated");
1459 if (BitsAlloc == "gdcm::Unfound") {
1460 dbg.Verbose(0, "gdcmHeader::GetPixelType: unfound Bits Allocated");
1461 BitsAlloc = string("16");
1463 if (BitsAlloc == "12")
1464 BitsAlloc = string("16");
1467 Signed = GetElValByName("Pixel Representation");
1468 if (Signed == "gdcm::Unfound") {
1469 dbg.Verbose(0, "gdcmHeader::GetPixelType: unfound Pixel Representation");
1470 BitsAlloc = string("0");
1473 Signed = string("U");
1475 Signed = string("S");
1477 return( BitsAlloc + Signed);
1481 * \ingroup gdcmHeader
1482 * \brief This predicate, based on hopefully reasonnable heuristics,
1483 * decides whether or not the current gdcmHeader was properly parsed
1484 * and contains the mandatory information for being considered as
1485 * a well formed and usable image.
1486 * @return true when gdcmHeader is the one of a reasonable Dicom file,
1489 bool gdcmHeader::IsReadable(void) {
1490 if ( GetElValByName("Image Dimensions") != "gdcm::Unfound"
1491 && atoi(GetElValByName("Image Dimensions").c_str()) > 4 ) {
1494 if ( GetElValByName("Bits Allocated") == "gdcm::Unfound" )
1496 if ( GetElValByName("Bits Stored") == "gdcm::Unfound" )
1498 if ( GetElValByName("High Bit") == "gdcm::Unfound" )
1500 if ( GetElValByName("Pixel Representation") == "gdcm::Unfound" )
1506 * \ingroup gdcmHeader
1507 * \brief Small utility function that creates a new manually crafted
1508 * (as opposed as read from the file) gdcmElValue with user
1509 * specified name and adds it to the public tag hash table.
1510 * \note A fake TagKey is generated so the PubDict can keep it's coherence.
1511 * @param NewTagName The name to be given to this new tag.
1512 * @param VR The Value Representation to be given to this new tag.
1513 * @ return The newly hand crafted Element Value.
1515 gdcmElValue* gdcmHeader::NewManualElValToPubDict(string NewTagName, string VR) {
1516 gdcmElValue* NewElVal = (gdcmElValue*)0;
1517 guint32 StuffGroup = 0xffff; // Group to be stuffed with additional info
1518 guint32 FreeElem = 0;
1519 gdcmDictEntry* NewEntry = (gdcmDictEntry*)0;
1521 FreeElem = PubElValSet.GenerateFreeTagKeyInGroup(StuffGroup);
1522 if (FreeElem == UINT32_MAX) {
1523 dbg.Verbose(1, "gdcmHeader::NewManualElValToPubDict",
1524 "Group 0xffff in Public Dict is full");
1525 return (gdcmElValue*)0;
1527 NewEntry = new gdcmDictEntry(StuffGroup, FreeElem,
1528 VR, "GDCM", NewTagName);
1529 NewElVal = new gdcmElValue(NewEntry);
1530 PubElValSet.Add(NewElVal);
1536 * \ingroup gdcmHeader
1537 * \brief Loads the element values of all the elements present in the
1538 * public tag based hash table.
1540 void gdcmHeader::LoadElements(void) {
1542 TagElValueHT ht = PubElValSet.GetTagHt();
1543 for (TagElValueHT::iterator tag = ht.begin(); tag != ht.end(); ++tag) {
1544 LoadElementValue(tag->second);
1548 void gdcmHeader::PrintPubElVal(ostream & os) {
1549 PubElValSet.Print(os);
1552 void gdcmHeader::PrintPubDict(ostream & os) {
1553 RefPubDict->Print(os);
1556 int gdcmHeader::Write(FILE * fp, FileType type) {
1557 return PubElValSet.Write(fp, type);