X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=src%2FgdcmHeader.cxx;h=d9e26448923776f6773671057bd477c546380af6;hb=f5d6e67978dff569e38c19df24f14aea254ecd7b;hp=b886afe215bc2203ee6cd864b901a1e67519d354;hpb=8e657a0766489604ade10a101d0d8995bcb79280;p=gdcm.git diff --git a/src/gdcmHeader.cxx b/src/gdcmHeader.cxx index b886afe2..d9e26448 100644 --- a/src/gdcmHeader.cxx +++ b/src/gdcmHeader.cxx @@ -1,572 +1,1267 @@ -#include "gdcm.h" +// $Header: /cvs/public/gdcm/src/Attic/gdcmHeader.cxx,v 1.104 2003/10/30 17:04:21 jpr Exp $ + +#include "gdcmHeader.h" + #include +#include // For nthos: #ifdef _MSC_VER #include #else #include #endif -#include // for isalpha -#include +#include // for isalpha #include #include "gdcmUtil.h" +#include "gdcmTS.h" -#define HEADER_LENGHT_TO_READ 256 // on ne lit plus que le debut +// Refer to gdcmHeader::CheckSwap() +#define HEADER_LENGTH_TO_READ 256 +// Refer to gdcmHeader::SetMaxSizeLoadElementValue() +//#define _MaxSizeLoadElementValue_ 1024 +#define _MaxSizeLoadElementValue_ 4096 +/** + * \ingroup gdcmHeader + * \brief + */ +void gdcmHeader::Initialise(void) { + dicom_vr = gdcmGlobal::GetVR(); + dicom_ts = gdcmGlobal::GetTS(); + Dicts = gdcmGlobal::GetDicts(); + RefPubDict = Dicts->GetDefaultPubDict(); + RefShaDict = (gdcmDict*)0; +} -namespace Error { - struct FileReadError { - FileReadError(FILE* fp, const char* Mesg) { - if (feof(fp)) - dbg.Verbose(1, "EOF encountered :", Mesg); - if (ferror(fp)) - dbg.Verbose(1, "Error on reading :", Mesg); - } - }; +/** + * \ingroup gdcmHeader + * \brief + * @param InFilename + * @param exception_on_error + */ +gdcmHeader::gdcmHeader(const char *InFilename, bool exception_on_error) { + SetMaxSizeLoadElementValue(_MaxSizeLoadElementValue_); + filename = InFilename; + Initialise(); + if ( !OpenFile(exception_on_error)) + return; + ParseHeader(); + LoadElements(); + CloseFile(); +} + +/** + * \ingroup gdcmHeader + * \brief + * @param exception_on_error + */ +gdcmHeader::gdcmHeader(bool exception_on_error) { + SetMaxSizeLoadElementValue(_MaxSizeLoadElementValue_); + Initialise(); } -//FIXME: this looks dirty to me... -#define str2num(str, typeNum) *((typeNum *)(str)) +/** + * \ingroup gdcmHeader + * \brief + * @param exception_on_error + * @return + */ + bool gdcmHeader::OpenFile(bool exception_on_error) + throw(gdcmFileError) { + fp=fopen(filename.c_str(),"rb"); + if(exception_on_error) { + if(!fp) + throw gdcmFileError("gdcmHeader::gdcmHeader(const char *, bool)"); + } -VRHT * gdcmHeader::dicom_vr = (VRHT*)0; -gdcmDictSet* gdcmHeader::Dicts = new gdcmDictSet(); + if ( fp ) { + guint16 zero; + fread(&zero, (size_t)2, (size_t)1, fp); -void gdcmHeader::Initialise(void) { - if (!gdcmHeader::dicom_vr) - InitVRDict(); - RefPubDict = gdcmHeader::Dicts->GetDefaultPublicDict(); - RefShaDict = (gdcmDict*)0; + //ACR -- or DICOM with no Preamble + if( zero == 0x0008 || zero == 0x0800 || zero == 0x0002 || zero == 0x0200) + return true; + //DICOM + fseek(fp, 126L, SEEK_CUR); + char dicm[4]; + fread(dicm, (size_t)4, (size_t)1, fp); + if( memcmp(dicm, "DICM", 4) == 0 ) + return true; + fclose(fp); + dbg.Verbose(0, "gdcmHeader::gdcmHeader not DICOM/ACR", filename.c_str()); + } + else { + dbg.Verbose(0, "gdcmHeader::gdcmHeader cannot open file", filename.c_str()); + } + return false; } -gdcmHeader::gdcmHeader (const char* InFilename) { - filename = InFilename; - Initialise(); - fp=fopen(InFilename,"rw"); - dbg.Error(!fp, "gdcmHeader::gdcmHeader cannot open file", InFilename); - ParseHeader(); +/** + * \ingroup gdcmHeader + * \brief + * @return TRUE if the close was successfull + */ +bool gdcmHeader::CloseFile(void) { + int closed = fclose(fp); + fp = (FILE *)0; + if (! closed) + return false; + return true; } +/** + * \ingroup gdcmHeader + * \brief Canonical destructor. + */ gdcmHeader::~gdcmHeader (void) { - fclose(fp); - return; -} - -void gdcmHeader::InitVRDict (void) { - if (dicom_vr) { - dbg.Verbose(0, "gdcmHeader::InitVRDict:", "VR dictionary allready set"); - return; - } - VRHT *vr = new VRHT; - (*vr)["AE"] = "Application Entity"; // 16 car max - (*vr)["AS"] = "Age String"; // 4 car fixe - (*vr)["AT"] = "Attribute Tag"; // 2 unsigned short int - (*vr)["CS"] = "Code String"; // 16 car max - (*vr)["DA"] = "Date"; // 8 car fixe - (*vr)["DS"] = "Decimal String"; // Decimal codé Binaire 16 max - (*vr)["DT"] = "Date Time"; // 26 car max - (*vr)["FL"] = "Floating Point Single"; // 4 octets IEEE 754:1985 - (*vr)["FD"] = "Floating Point Double"; // 8 octets IEEE 754:1985 - (*vr)["IS"] = "Integer String"; // en format externe 12 max - (*vr)["LO"] = "Long String"; // 64 octets max - (*vr)["LT"] = "Long Text"; // 10240 max - (*vr)["OB"] = "Other Byte String"; - (*vr)["OW"] = "Other Word String"; - (*vr)["PN"] = "Person Name"; - (*vr)["SH"] = "Short String"; // 16 car max - (*vr)["SL"] = "Signed Long"; - (*vr)["SQ"] = "Sequence of Items"; // Not Applicable - (*vr)["SS"] = "Signed Short"; // 2 octets - (*vr)["ST"] = "Short Text"; // 1024 car max - (*vr)["TM"] = "Time"; // 16 car max - (*vr)["UI"] = "Unique Identifier"; // 64 car max - (*vr)["UN"] = "Unknown"; - (*vr)["UT"] = "Unlimited Text"; // 2 puissance 32 -1 car max - (*vr)["UL"] = "Unsigned Long "; // 4 octets fixe - (*vr)["US"] = "Unsigned Short "; // 2 octets fixe - dicom_vr = vr; -} - -/** - * \ingroup gdcmHeader - * \brief La seule maniere sure que l'on aie pour determiner - * si on est en LITTLE_ENDIAN, BIG-ENDIAN, - * BAD-LITTLE-ENDIAN, BAD-BIG-ENDIAN - * est de trouver l'element qui donne la longueur d'un 'GROUP' - * (on sait que la longueur de cet element vaut 0x00000004) - * et de regarder comment cette longueur est codee en memoire - * - * Le probleme vient de ce que parfois, il n'y en a pas ... - * - * On fait alors le pari qu'on a a faire a du LITTLE_ENDIAN propre. - * (Ce qui est la norme -pas respectee- depuis ACR-NEMA) - * Si ce n'est pas le cas, on ne peut rien faire. + dicom_vr = (gdcmVR*)0; + Dicts = (gdcmDictSet*)0; + RefPubDict = (gdcmDict*)0; + RefShaDict = (gdcmDict*)0; + return; +} + +// Fourth semantics: +// +// ---> Warning : This fourth field is NOT part +// of the 'official' Dicom Dictionnary +// and should NOT be used. +// (Not defined for all the groups +// may be removed in a future release) +// +// CMD Command +// META Meta Information +// DIR Directory +// ID +// PAT Patient +// ACQ Acquisition +// REL Related +// IMG Image +// SDY Study +// VIS Visit +// WAV Waveform +// PRC +// DEV Device +// NMI Nuclear Medicine +// MED +// BFS Basic Film Session +// BFB Basic Film Box +// BIB Basic Image Box +// BAB +// IOB +// PJ +// PRINTER +// RT Radio Therapy +// DVH +// SSET +// RES Results +// CRV Curve +// OLY Overlays +// PXL Pixels +// DL Delimiters +// + +/** + * \ingroup gdcmHeader + * \brief Discover what the swap code is (among little endian, big endian, + * bad little endian, bad big endian). * - * (il faudrait avoir des fonctions auxquelles - * on passe le code Swap en parametre, pour faire des essais 'manuels') */ void gdcmHeader::CheckSwap() { - guint32 s; - guint32 x=4; // x : pour ntohs - bool net2host; // true when HostByteOrder is the same as NetworkByteOrder - - int lgrLue; - char * entCur; - char deb[HEADER_LENGHT_TO_READ]; - - // First, compare HostByteOrder and NetworkByteOrder in order to - // determine if we shall need to swap bytes (i.e. the Endian type). - if (x==ntohs(x)) - net2host = true; - else - net2host = false; - - // The easiest case is the one of a DICOM header, since it possesses a - // file preamble where it suffice to look for the sting "DICM". - lgrLue = fread(deb, 1, HEADER_LENGHT_TO_READ, fp); - - entCur = deb + 128; - if(memcmp(entCur, "DICM", (size_t)4) == 0) { - filetype = TrueDicom; - dbg.Verbose(1, "gdcmHeader::CheckSwap:", "looks like DICOM Version3"); - } else { - filetype = Unknown; - dbg.Verbose(1, "gdcmHeader::CheckSwap:", "not a DICOM Version3 file"); - } - - if(filetype == TrueDicom) { - // Next, determine the value representation (VR). Let's skip to the - // first element (0002, 0000) and check there if we find "UL", in - // which case we (almost) know it is explicit VR. - // WARNING: if it happens to be implicit VR then what we will read - // is the length of the group. If this ascii representation of this - // length happens to be "UL" then we shall believe it is explicit VR. - // FIXME: in order to fix the above warning, we could read the next - // element value (or a couple of elements values) in order to make - // sure we are not commiting a big mistake. - // We need to skip : - // * the 128 bytes of File Preamble (often padded with zeroes), - // * the 4 bytes of "DICM" string, - // * the 4 bytes of the first tag (0002, 0000), - // i.e. a total of 136 bytes. - entCur = deb + 136; - if(memcmp(entCur, "UL", (size_t)2) == 0) { - filetype = ExplicitVR; - dbg.Verbose(1, "gdcmHeader::CheckSwap:", - "explicit Value Representation"); - } else { - filetype = ImplicitVR; - dbg.Verbose(1, "gdcmHeader::CheckSwap:", - "not an explicit Value Representation"); - } - - if (net2host) { - sw = 4321; - dbg.Verbose(1, "gdcmHeader::CheckSwap:", - "HostByteOrder != NetworkByteOrder"); - } else { - sw = 0; - dbg.Verbose(1, "gdcmHeader::CheckSwap:", - "HostByteOrder = NetworkByteOrder"); - } - - // Position the file position indicator at first tag (i.e. - // after the file preamble and the "DICM" string). - rewind(fp); - fseek (fp, 132L, SEEK_SET); - return; - } // End of TrueDicom - - // Alas, this is not a DicomV3 file and whatever happens there is no file - // preamble. We can reset the file position indicator to where the data - // is (i.e. the beginning of the file). - rewind(fp); - - // Our next best chance would be to be considering a 'clean' ACR/NEMA file. - // By clean we mean that the length of the first tag is written down. - // If this is the case and since the length of the first group HAS to be - // four (bytes), then determining the proper swap code is straightforward. - - entCur = deb + 4; - s = str2num(entCur, int); - - switch (s) { - case 0x00040000 : - sw=3412; - filetype = ACR; - return; - case 0x04000000 : - sw=4321; - filetype = ACR; - return; - case 0x00000400 : - sw=2143; - filetype = ACR; - return; - case 0x00000004 : - sw=0; - filetype = ACR; - return; - default : - dbg.Verbose(0, "gdcmHeader::CheckSwap:", - "ACE/NEMA unfound swap info (time to raise bets)"); - } - - // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file. - // It is time for despaired wild guesses. So, let's assume this file - // happens to be 'dirty' ACR/NEMA, i.e. the length of the group it - // not present. Then the only info we have is the net2host one. - //FIXME Si c'est du RAW, ca degagera + tard - - if (! net2host ) - sw = 0; - else - sw = 4321; - return; + // The only guaranted way of finding the swap code is to find a + // group tag since we know it's length has to be of four bytes i.e. + // 0x00000004. Finding the swap code in then straigthforward. Trouble + // occurs when we can't find such group... + guint32 s; + guint32 x=4; // x : for ntohs + bool net2host; // true when HostByteOrder is the same as NetworkByteOrder + + int lgrLue; + char * entCur; + char deb[HEADER_LENGTH_TO_READ]; + + // First, compare HostByteOrder and NetworkByteOrder in order to + // determine if we shall need to swap bytes (i.e. the Endian type). + if (x==ntohs(x)) + net2host = true; + else + net2host = false; + + // The easiest case is the one of a DICOM header, since it possesses a + // file preamble where it suffice to look for the string "DICM". + lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, fp); + + entCur = deb + 128; + if(memcmp(entCur, "DICM", (size_t)4) == 0) { + dbg.Verbose(1, "gdcmHeader::CheckSwap:", "looks like DICOM Version3"); + // Next, determine the value representation (VR). Let's skip to the + // first element (0002, 0000) and check there if we find "UL" + // - or "OB" if the 1st one is (0002,0001) -, + // in which case we (almost) know it is explicit VR. + // WARNING: if it happens to be implicit VR then what we will read + // is the length of the group. If this ascii representation of this + // length happens to be "UL" then we shall believe it is explicit VR. + // FIXME: in order to fix the above warning, we could read the next + // element value (or a couple of elements values) in order to make + // sure we are not commiting a big mistake. + // We need to skip : + // * the 128 bytes of File Preamble (often padded with zeroes), + // * the 4 bytes of "DICM" string, + // * the 4 bytes of the first tag (0002, 0000),or (0002, 0001) + // i.e. a total of 136 bytes. + entCur = deb + 136; + // FIXME + // Use gdcmHeader::dicom_vr to test all the possibilities + // instead of just checking for UL, OB and UI !? + if( (memcmp(entCur, "UL", (size_t)2) == 0) || + (memcmp(entCur, "OB", (size_t)2) == 0) || + (memcmp(entCur, "UI", (size_t)2) == 0) ) + { + filetype = ExplicitVR; + dbg.Verbose(1, "gdcmHeader::CheckSwap:", + "explicit Value Representation"); + } else { + filetype = ImplicitVR; + dbg.Verbose(1, "gdcmHeader::CheckSwap:", + "not an explicit Value Representation"); + } + if (net2host) { + sw = 4321; + dbg.Verbose(1, "gdcmHeader::CheckSwap:", + "HostByteOrder != NetworkByteOrder"); + } else { + sw = 0; + dbg.Verbose(1, "gdcmHeader::CheckSwap:", + "HostByteOrder = NetworkByteOrder"); + } + + // Position the file position indicator at first tag (i.e. + // after the file preamble and the "DICM" string). + rewind(fp); + fseek (fp, 132L, SEEK_SET); + return; + } // End of DicomV3 + + // Alas, this is not a DicomV3 file and whatever happens there is no file + // preamble. We can reset the file position indicator to where the data + // is (i.e. the beginning of the file). + dbg.Verbose(1, "gdcmHeader::CheckSwap:", "not a DICOM Version3 file"); + rewind(fp); + + // Our next best chance would be to be considering a 'clean' ACR/NEMA file. + // By clean we mean that the length of the first tag is written down. + // If this is the case and since the length of the first group HAS to be + // four (bytes), then determining the proper swap code is straightforward. + + entCur = deb + 4; + // We assume the array of char we are considering contains the binary + // representation of a 32 bits integer. Hence the following dirty + // trick : + s = *((guint32 *)(entCur)); + + switch (s) { + case 0x00040000 : + sw = 3412; + filetype = ACR; + return; + case 0x04000000 : + sw = 4321; + filetype = ACR; + return; + case 0x00000400 : + sw = 2143; + filetype = ACR; + return; + case 0x00000004 : + sw = 0; + filetype = ACR; + return; + default : + dbg.Verbose(0, "gdcmHeader::CheckSwap:", + "ACR/NEMA unfound swap info (time to raise bets)"); + } + + // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file. + // It is time for despaired wild guesses. So, let's assume this file + // happens to be 'dirty' ACR/NEMA, i.e. the length of the group is + // not present. Then the only info we have is the net2host one. + filetype = Unknown; + if (! net2host ) + sw = 0; + else + sw = 4321; + return; +} + +/** + * \ingroup gdcmHeader + * \brief + */ +void gdcmHeader::SwitchSwapToBigEndian(void) { + dbg.Verbose(1, "gdcmHeader::SwitchSwapToBigEndian", + "Switching to BigEndian mode."); + if ( sw == 0 ) { + sw = 4321; + return; + } + if ( sw == 4321 ) { + sw = 0; + return; + } + if ( sw == 3412 ) { + sw = 2143; + return; + } + if ( sw == 2143 ) + sw = 3412; } /** * \ingroup gdcmHeader - * \brief recupere la longueur d'un champ DICOM. - * Preconditions: - * 1/ le fichier doit deja avoir ete ouvert, - * 2/ CheckSwap() doit avoir ete appele - * 3/ la partie 'group' ainsi que la partie 'elem' - * de l'acr_element doivent avoir ete lues. + * \brief Find the value representation of the current tag. + * @param ElVal + */ +void gdcmHeader::FindVR( gdcmElValue *ElVal) { + if (filetype != ExplicitVR) + return; + + char VR[3]; + std::string vr; + int lgrLue; + char msg[100]; // for sprintf. Sorry + + long PositionOnEntry = ftell(fp); + // Warning: we believe this is explicit VR (Value Representation) because + // we used a heuristic that found "UL" in the first tag. Alas this + // doesn't guarantee that all the tags will be in explicit VR. In some + // cases (see e-film filtered files) one finds implicit VR tags mixed + // within an explicit VR file. Hence we make sure the present tag + // is in explicit VR and try to fix things if it happens not to be + // the case. + bool RealExplicit = true; + + lgrLue=fread (&VR, (size_t)2,(size_t)1, fp); + VR[2]=0; + vr = std::string(VR); + + // Assume we are reading a falsely explicit VR file i.e. we reached + // a tag where we expect reading a VR but are in fact we read the + // first to bytes of the length. Then we will interogate (through find) + // the dicom_vr dictionary with oddities like "\004\0" which crashes + // both GCC and VC++ implementations of the STL map. Hence when the + // expected VR read happens to be non-ascii characters we consider + // we hit falsely explicit VR tag. + + if ( (!isalpha(VR[0])) && (!isalpha(VR[1])) ) + RealExplicit = false; + + // CLEANME searching the dicom_vr at each occurence is expensive. + // PostPone this test in an optional integrity check at the end + // of parsing or only in debug mode. + if ( RealExplicit && !dicom_vr->Count(vr) ) + RealExplicit= false; + + if ( RealExplicit ) { + if ( ElVal->IsVrUnknown() ) { + // When not a dictionary entry, we can safely overwrite the vr. + ElVal->SetVR(vr); + return; + } + if ( ElVal->GetVR() == vr ) { + // The vr we just read and the dictionary agree. Nothing to do. + return; + } + // The vr present in the file and the dictionary disagree. We assume + // the file writer knew best and use the vr of the file. Since it would + // be unwise to overwrite the vr of a dictionary (since it would + // compromise it's next user), we need to clone the actual DictEntry + // and change the vr for the read one. + gdcmDictEntry* NewTag = new gdcmDictEntry(ElVal->GetGroup(), + ElVal->GetElement(), + vr, + "FIXME", + ElVal->GetName()); + ElVal->SetDictEntry(NewTag); + return; + } + + // We thought this was explicit VR, but we end up with an + // implicit VR tag. Let's backtrack. + + sprintf(msg,"Falsely explicit vr file (%04x,%04x)\n", + ElVal->GetGroup(),ElVal->GetElement()); + dbg.Verbose(1, "gdcmHeader::FindVR: ",msg); + + fseek(fp, PositionOnEntry, SEEK_SET); + // When this element is known in the dictionary we shall use, e.g. for + // the semantics (see the usage of IsAnInteger), the vr proposed by the + // dictionary entry. Still we have to flag the element as implicit since + // we know now our assumption on expliciteness is not furfilled. + // avoid . + if ( ElVal->IsVrUnknown() ) + ElVal->SetVR("Implicit"); + ElVal->SetImplicitVr(); +} + +/** + * \ingroup gdcmHeader + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a ImplicitVRLittleEndian one. * - * ACR-NEMA : we allways get - * GroupNumber (2 Octets) - * ElementNumber (2 Octets) - * ElementSize (4 Octets) - * DICOM en implicit Value Representation : - * GroupNumber (2 Octets) - * ElementNumber (2 Octets) - * ElementSize (4 Octets) + * @return True when ImplicitVRLittleEndian found. False in all other cases. + */ +bool gdcmHeader::IsImplicitVRLittleEndianTransferSyntax(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2" ) + return true; + return false; +} + +/** + * \ingroup gdcmHeader + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a ExplicitVRLittleEndian one. * - * DICOM en explicit Value Representation : - * GroupNumber (2 Octets) - * ElementNumber (2 Octets) - * ValueRepresentation (2 Octets) - * ElementSize (2 Octets) + * @return True when ExplicitVRLittleEndian found. False in all other cases. + */ +bool gdcmHeader::IsExplicitVRLittleEndianTransferSyntax(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.1" ) + return true; + return false; +} + +/** + * \ingroup gdcmHeader + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a DeflatedExplicitVRLittleEndian one. * - * ATTENTION : dans le cas ou ValueRepresentation = OB, OW, SQ, UN - * GroupNumber (2 Octets) - * ElementNumber (2 Octets) - * ValueRepresentation (2 Octets) - * zone reservee (2 Octets) - * ElementSize (4 Octets) + * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases. + */ +bool gdcmHeader::IsDeflatedExplicitVRLittleEndianTransferSyntax(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.1.99" ) + return true; + return false; +} + +/** + * \ingroup gdcmHeader + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a Explicit VR Big Endian one. * - * @param sw code swap - * @param skippedLength pointeur sur nombre d'octets que l'on a saute qd - * la lecture est finie - * @param longueurLue pointeur sur longueur (en nombre d'octets) - * effectivement lue - * @return longueur retenue pour le champ - */ - -void gdcmHeader::FindVR( ElValue *ElVal) { - char VR[3]; - string vr; - int lgrLue; - long PositionOnEntry = ftell(fp); - // Warning: we believe this is explicit VR (Value Representation) because - // we used a heuristic that found "UL" in the first tag. Alas this - // doesn't guarantee that all the tags will be in explicit VR. In some - // cases (see e-film filtered files) one finds implicit VR tags mixed - // within an explicit VR file. Hence we make sure the present tag - // is in explicit VR and try to fix things if it happens not to be - // the case. - bool RealExplicit = true; - - if (filetype != ExplicitVR) - return; + * @return True when big endian found. False in all other cases. + */ +bool gdcmHeader::IsExplicitVRBigEndianTransferSyntax(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.2" ) //1.2.2 ??? A verifier ! + return true; + return false; +} - lgrLue=fread (&VR, (size_t)2,(size_t)1, fp); - VR[2]=0; - vr = string(VR); - - // Assume we are reading a falsely explicit VR file i.e. we reached - // a tag where we expect reading a VR but are in fact we read the - // first to bytes of the length. Then we will interogate (through find) - // the dicom_vr dictionary with oddities like "\004\0" which crashes - // both GCC and VC++ implentations of the STL map. Hence when the - // expected VR read happens to be non-ascii characters we consider - // we hit falsely explicit VR tag. - - if ( (!isalpha(VR[0])) && (!isalpha(VR[1])) ) - RealExplicit = false; - - // CLEANME searching the dicom_vr at each occurence is expensive. - // PostPone this test in an optional integrity check at the end - // of parsing or only in debug mode. - if ( RealExplicit && !dicom_vr->count(vr) ) - RealExplicit = false; - - if ( RealExplicit ) { - if ( ElVal->IsVrUnknown() ) - ElVal->SetVR(vr); - return; - } - - // We thought this was explicit VR, but we end up with an - // implicit VR tag. Let's backtrack. - dbg.Verbose(1, "gdcmHeader::FindVR:", "Falsely explicit vr file"); - fseek(fp, PositionOnEntry, SEEK_SET); - // When this element is known in the dictionary we shall use, e.g. for - // the semantics (see the usage of IsAnInteger), the vr proposed by the - // dictionary entry. Still we have to flag the element as implicit since - // we know now our assumption on expliciteness is not furfilled. - // avoid . - if ( ElVal->IsVrUnknown() ) - ElVal->SetVR("Implicit"); - ElVal->SetImplicitVr(); -} - -void gdcmHeader::FindLength( ElValue * ElVal) { - guint32 length32; - guint16 length16; - string vr = ElVal->GetVR(); - - if ( (filetype == ExplicitVR) && ! ElVal->IsImplicitVr() ) { - if ( (vr=="OB") || (vr=="OW") || (vr=="SQ") || (vr=="UN") ) { - - // The following two bytes are reserved, so we skip them, - // and we proceed on reading the length on 4 bytes. - fseek(fp, 2L,SEEK_CUR); - length32 = ReadInt32(); - - } else { - // Length is encoded on 2 bytes. - length16 = ReadInt16(); - - if ( length16 == 0xffff) { - length32 = 0; - } else { - length32 = length16; - } - } - } else { - // Either implicit VR or an explicit VR that (at least for this - // element) lied a little bit. Length is on 4 bytes. - length32 = ReadInt32(); - } - - // Traitement des curiosites sur la longueur - if ( length32 == 0xffffffff) - length32=0; - - ElVal->SetLength(length32); +/** + * \ingroup gdcmHeader + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a JPEGBaseLineProcess1 one. + * + * @return True when JPEGBaseLineProcess1found. False in all other cases. + */ +bool gdcmHeader::IsJPEGBaseLineProcess1TransferSyntax(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.4.50" ) + return true; + return false; +} + +/** + * \ingroup gdcmHeader + * \brief + * + * @return + */ +bool gdcmHeader::IsJPEGLossless(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + // faire qq chose d'intelligent a la place de ça + if ( !Element ) + return false; + LoadElementValueSafe(Element); + const char * Transfert = Element->GetValue().c_str(); + if ( memcmp(Transfert+strlen(Transfert)-2 ,"70",2)==0) return true; + if ( memcmp(Transfert+strlen(Transfert)-2 ,"55",2)==0) return true; + if (Element->GetValue() == "1.2.840.10008.1.2.4.57") return true; + + return false; } /** * \ingroup gdcmHeader - * \brief remet les octets dans un ordre compatible avec celui du processeur + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a JPEGExtendedProcess2-4 one. + * + * @return True when JPEGExtendedProcess2-4 found. False in all other cases. + */ +bool gdcmHeader::IsJPEGExtendedProcess2_4TransferSyntax(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.4.51" ) + return true; + return false; +} - * @return longueur retenue pour le champ +/** + * \ingroup gdcmHeader + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a JPEGExtendeProcess3-5 one. + * + * @return True when JPEGExtendedProcess3-5 found. False in all other cases. + */ +bool gdcmHeader::IsJPEGExtendedProcess3_5TransferSyntax(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.4.52" ) + return true; + return false; +} + +/** + * \ingroup gdcmHeader + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a JPEGSpectralSelectionProcess6-8 one. + * + * @return True when JPEGSpectralSelectionProcess6-8 found. False in all + * other cases. + */ +bool gdcmHeader::IsJPEGSpectralSelectionProcess6_8TransferSyntax(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.4.53" ) + return true; + return false; +} + +/** + * \ingroup gdcmHeader + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a RLE Lossless one. + * + * @return True when RLE Lossless found. False in all + * other cases. + */ +bool gdcmHeader::IsRLELossLessTransferSyntax(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.5" ) + return true; + return false; +} + +/** + * \ingroup gdcmHeader + * \brief Determines if the Transfer Syntax was already encountered + * and if it corresponds to a JPEG200 one.0 + * + * @return True when JPEG2000 (Lossly or LossLess) found. False in all + * other cases. + */ +bool gdcmHeader::IsJPEG2000(void) { + gdcmElValue* Element = PubElValSet.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + std::string Transfer = Element->GetValue(); + if ( (Transfer == "1.2.840.10008.1.2.4.90") + || (Transfer == "1.2.840.10008.1.2.4.91") ) + return true; + return false; +} + +/** + * \ingroup gdcmHeader + * \brief Predicate for dicom version 3 file. + * @return True when the file is a dicom version 3. + */ +bool gdcmHeader::IsDicomV3(void) { + if ( (filetype == ExplicitVR) + || (filetype == ImplicitVR) ) + return true; + return false; +} + +/** + * \ingroup gdcmHeader + * \brief When the length of an element value is obviously wrong (because + * the parser went Jabberwocky) one can hope improving things by + * applying this heuristic. + */ +void gdcmHeader::FixFoundLength(gdcmElValue * ElVal, guint32 FoundLength) { + if ( FoundLength == 0xffffffff) + FoundLength = 0; + + // Sorry for the patch! + // XMedCom did the trick to read some nasty GE images ... + else if (FoundLength == 13) { + // The following 'if' will be removed when there is no more + // images on Creatis HDs with a 13 length for Manufacturer... + if ( (ElVal->GetGroup() != 0x0008) || (ElVal->GetElement() != 0x0070)) { + // end of remove area + FoundLength =10; + } + } + // to fix some garbage 'Leonardo' Siemens images + // May be commented out to avoid overhead + else if ( (ElVal->GetGroup() == 0x0009) + && + ( (ElVal->GetElement() == 0x1113) || (ElVal->GetElement() == 0x1114) ) ) + FoundLength =4; + + ElVal->SetLength(FoundLength); +} + +/** + * \ingroup gdcmHeader + * \brief + * + * @return + */ + guint32 gdcmHeader::FindLengthOB(void) { + // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data. + guint16 g; + guint16 n; + long PositionOnEntry = ftell(fp); + bool FoundSequenceDelimiter = false; + guint32 TotalLength = 0; + guint32 ItemLength; + + while ( ! FoundSequenceDelimiter) { + g = ReadInt16(); + n = ReadInt16(); + if (errno == 1) + return 0; + TotalLength += 4; // We even have to decount the group and element + + if ( g != 0xfffe && g!=0xb00c ) /*for bogus header */ { + char msg[100]; // for sprintf. Sorry + sprintf(msg,"wrong group (%04x) for an item sequence (%04x,%04x)\n",g, g,n); + dbg.Verbose(1, "gdcmHeader::FindLengthOB: ",msg); + errno = 1; + return 0; + } + if ( n == 0xe0dd || ( g==0xb00c && n==0x0eb6 ) ) /* for bogus header */ + FoundSequenceDelimiter = true; + else if ( n != 0xe000 ){ + char msg[100]; // for sprintf. Sorry + sprintf(msg,"wrong element (%04x) for an item sequence (%04x,%04x)\n", + n, g,n); + dbg.Verbose(1, "gdcmHeader::FindLengthOB: ",msg); + errno = 1; + return 0; + } + ItemLength = ReadInt32(); + TotalLength += ItemLength + 4; // We add 4 bytes since we just read + // the ItemLength with ReadInt32 + SkipBytes(ItemLength); + } + fseek(fp, PositionOnEntry, SEEK_SET); + return TotalLength; +} + +/** + * \ingroup gdcmHeader + * \brief + * + * @return + */ + void gdcmHeader::FindLength (gdcmElValue * ElVal) { + guint16 element = ElVal->GetElement(); + guint16 group = ElVal->GetGroup(); + std::string vr = ElVal->GetVR(); + guint16 length16; + if( (element == 0x0010) && (group == 0x7fe0) ) { + dbg.SetDebug(0); + dbg.Verbose(2, "gdcmHeader::FindLength: ", + "we reached 7fe0 0010"); + } + + if ( (filetype == ExplicitVR) && ! ElVal->IsImplicitVr() ) { + if ( (vr=="OB") || (vr=="OW") || (vr=="SQ") || (vr=="UN") ) { + + // The following reserved two bytes (see PS 3.5-2001, section + // 7.1.2 Data element structure with explicit vr p27) must be + // skipped before proceeding on reading the length on 4 bytes. + fseek(fp, 2L, SEEK_CUR); + + guint32 length32 = ReadInt32(); + + if ( (vr == "OB") && (length32 == 0xffffffff) ) { + ElVal->SetLength(FindLengthOB()); + return; + } + FixFoundLength(ElVal, length32); + return; + } + + // Length is encoded on 2 bytes. + length16 = ReadInt16(); + + // We can tell the current file is encoded in big endian (like + // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag + // and it's value is the one of the encoding of a big endian file. + // In order to deal with such big endian encoded files, we have + // (at least) two strategies: + // * when we load the "Transfer Syntax" tag with value of big endian + // encoding, we raise the proper flags. Then we wait for the end + // of the META group (0x0002) among which is "Transfer Syntax", + // before switching the swap code to big endian. We have to postpone + // the switching of the swap code since the META group is fully encoded + // in little endian, and big endian coding only starts at the next + // group. The corresponding code can be hard to analyse and adds + // many additional unnecessary tests for regular tags. + // * the second strategy consists in waiting for trouble, that shall + // appear when we find the first group with big endian encoding. This + // is easy to detect since the length of a "Group Length" tag (the + // ones with zero as element number) has to be of 4 (0x0004). When we + // encouter 1024 (0x0400) chances are the encoding changed and we + // found a group with big endian encoding. + // We shall use this second strategy. In order to make sure that we + // can interpret the presence of an apparently big endian encoded + // length of a "Group Length" without committing a big mistake, we + // add an additional check: we look in the already parsed elements + // for the presence of a "Transfer Syntax" whose value has to be "big + // endian encoding". When this is the case, chances are we have got our + // hands on a big endian encoded file: we switch the swap code to + // big endian and proceed... + if ( (element == 0x0000) && (length16 == 0x0400) ) { + if ( ! IsExplicitVRBigEndianTransferSyntax() ) { + dbg.Verbose(0, "gdcmHeader::FindLength", "not explicit VR"); + errno = 1; + return; + } + length16 = 4; + SwitchSwapToBigEndian(); + // Restore the unproperly loaded values i.e. the group, the element + // and the dictionary entry depending on them. + guint16 CorrectGroup = SwapShort(ElVal->GetGroup()); + guint16 CorrectElem = SwapShort(ElVal->GetElement()); + gdcmDictEntry * NewTag = GetDictEntryByNumber(CorrectGroup, + CorrectElem); + if (!NewTag) { + // This correct tag is not in the dictionary. Create a new one. + NewTag = new gdcmDictEntry(CorrectGroup, CorrectElem); + } + // FIXME this can create a memory leaks on the old entry that be + // left unreferenced. + ElVal->SetDictEntry(NewTag); + } + + // Heuristic: well some files are really ill-formed. + if ( length16 == 0xffff) { + length16 = 0; + //dbg.Verbose(0, "gdcmHeader::FindLength", + // "Erroneous element length fixed."); + // Actually, length= 0xffff means that we deal with + // Unknown Sequence Length + } + + FixFoundLength(ElVal, (guint32)length16); + return; + } + + // Either implicit VR or a non DICOM conformal (see not below) explicit + // VR that ommited the VR of (at least) this element. Farts happen. + // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25 + // on Data elements "Implicit and Explicit VR Data Elements shall + // not coexist in a Data Set and Data Sets nested within it".] + // Length is on 4 bytes. + FixFoundLength(ElVal, ReadInt32()); +} + +/** + * \ingroup gdcmHeader + * \brief Swaps back the bytes of 4-byte long integer accordingly to + * processor order. + * @return The properly swaped 32 bits integer. */ guint32 gdcmHeader::SwapLong(guint32 a) { - // FIXME: il pourrait y avoir un pb pour les entiers negatifs ... - switch (sw) { - case 0 : - break; - case 4321 : - a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) | - ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) ); - break; - - case 3412 : - a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) ); - break; - - case 2143 : - a=( ((a<<8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) ); - break; - default : - dbg.Error(" gdcmHeader::SwapLong : unset swap code"); - a=0; - } - return(a); + switch (sw) { + case 0 : + break; + case 4321 : + a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) | + ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) ); + break; + + case 3412 : + a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) ); + break; + + case 2143 : + a=( ((a<<8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) ); + break; + default : + dbg.Error(" gdcmHeader::SwapLong : unset swap code"); + a=0; + } + return(a); } /** * \ingroup gdcmHeader * \brief Swaps the bytes so they agree with the processor order - - * @return longueur retenue pour le champ + * @return The properly swaped 16 bits integer. */ guint16 gdcmHeader::SwapShort(guint16 a) { - //FIXME how could sw be equal to 2143 since we never set it this way ? - if ( (sw==4321) || (sw==2143) ) - a =(((a<<8) & 0x0ff00) | ((a>>8)&0x00ff)); - return (a); + if ( (sw==4321) || (sw==2143) ) + a =(((a<<8) & 0x0ff00) | ((a>>8)&0x00ff)); + return (a); } -void gdcmHeader::SkipElementValue(ElValue * ElVal) { - //FIXME don't dump the returned value - (void)fseek(fp, (long)ElVal->GetLength(), SEEK_CUR); +/** + * \ingroup gdcmHeader + * \brief + * + * @return + */ + void gdcmHeader::SkipBytes(guint32 NBytes) { + //FIXME don't dump the returned value + (void)fseek(fp, (long)NBytes, SEEK_CUR); +} + +/** + * \ingroup gdcmHeader + * \brief + * @param ElVal + * @return + */ + void gdcmHeader::SkipElementValue(gdcmElValue * ElVal) { + SkipBytes(ElVal->GetLength()); +} + +/** + * \ingroup gdcmHeader + * \brief + * @param NewSize + * @return + */ + void gdcmHeader::SetMaxSizeLoadElementValue(long NewSize) { + if (NewSize < 0) + return; + if ((guint32)NewSize >= (guint32)0xffffffff) { + MaxSizeLoadElementValue = 0xffffffff; + return; + } + MaxSizeLoadElementValue = NewSize; } /** * \ingroup gdcmHeader - * \brief Loads the element if it's size is not to big. - * @param ElVal Element whose value shall be loaded. - * @param MaxSize Size treshold above which the element value is not - * loaded in memory. The element value is allways loaded - * when MaxSize is equal to UINT32_MAX. - * @return + * \brief Loads the element content if it's length is not bigger + * than the value specified with + * gdcmHeader::SetMaxSizeLoadElementValue() */ -void gdcmHeader::LoadElementValue(ElValue * ElVal) { - size_t item_read; - guint16 group = ElVal->GetGroup(); - guint16 elem = ElVal->GetElement(); - string vr = ElVal->GetVR(); - guint32 length = ElVal->GetLength(); +void gdcmHeader::LoadElementValue(gdcmElValue * ElVal) { + size_t item_read; + guint16 group = ElVal->GetGroup(); + std::string vr= ElVal->GetVR(); + guint32 length = ElVal->GetLength(); + bool SkipLoad = false; - fseek(fp, (long)ElVal->GetOffset(), SEEK_SET); - - // Sequences not treated yet ! - if( vr == "SQ" ) { - SkipElementValue(ElVal); - ElVal->SetLength(0); - return; - } - // A sequence "contains" a set of tags (called items). It looks like - // the last tag of a sequence (the one that terminates the sequence) - // has a group of 0xfffe (with a dummy length). - if( group == 0xfffe) { - SkipElementValue(ElVal); - ElVal->SetLength(0); - return; - } + fseek(fp, (long)ElVal->GetOffset(), SEEK_SET); + + // FIXME Sequences not treated yet ! + // + // Ne faudrait-il pas au contraire trouver immediatement + // une maniere 'propre' de traiter les sequences (vr = SQ) + // car commencer par les ignorer risque de conduire a qq chose + // qui pourrait ne pas etre generalisable + // Well, I'm expecting your code !!! + + if( vr == "SQ" ) + SkipLoad = true; + + // A sequence "contains" a set of Elements. + // (fffe e000) tells us an Element is beginning + // (fffe e00d) tells us an Element just ended + // (fffe e0dd) tells us the current SQuence just ended + + if( group == 0xfffe ) + SkipLoad = true; + + if ( SkipLoad ) { + ElVal->SetLength(0); + ElVal->SetValue("gdcm::Skipped"); + return; + } + + // When the length is zero things are easy: + if ( length == 0 ) { + ElVal->SetValue(""); + return; + } + + // The elements whose length is bigger than the specified upper bound + // are not loaded. Instead we leave a short notice of the offset of + // the element content and it's length. + if (length > MaxSizeLoadElementValue) { + std::ostringstream s; + s << "gdcm::NotLoaded."; + s << " Address:" << (long)ElVal->GetOffset(); + s << " Length:" << ElVal->GetLength(); + ElVal->SetValue(s.str()); + return; + } + + // When an integer is expected, read and convert the following two or + // four bytes properly i.e. as an integer as opposed to a string. - if ( IsAnInteger(group, elem, vr, length) ) { - guint32 NewInt; - if( length == 2 ) { - NewInt = ReadInt16(); - } else if( length == 4 ) { - NewInt = ReadInt32(); - } else - dbg.Error(true, "LoadElementValue: Inconsistency when reading Int."); + // pour les elements de Value Multiplicity > 1 + // on aura en fait une serie d'entiers + // on devrait pouvoir faire + compact (?) - //FIXME: make the following an util fonction - ostringstream s; - s << NewInt; - ElVal->SetValue(s.str()); - return; - } - - // FIXME The exact size should be length if we move to strings or whatever - char* NewValue = (char*)g_malloc(length+1); - if( !NewValue) { - dbg.Verbose(1, "LoadElementValue: Failed to allocate NewValue"); - return; - } - NewValue[length]= 0; - - // FIXME les elements trop long (seuil a fixer a la main) ne devraient - // pas etre charge's !!!! Voir TODO. - item_read = fread(NewValue, (size_t)length, (size_t)1, fp); - if ( item_read != 1 ) { - g_free(NewValue); - Error::FileReadError(fp, "gdcmHeader::LoadElementValue"); - ElVal->SetValue("gdcm::UnRead"); - return; - } - ElVal->SetValue(NewValue); + if ( IsAnInteger(ElVal) ) { + guint32 NewInt; + std::ostringstream s; + int nbInt; + if (vr == "US" || vr == "SS") { + nbInt = length / 2; + NewInt = ReadInt16(); + s << NewInt; + if (nbInt > 1) { + for (int i=1; i < nbInt; i++) { + s << '\\'; + NewInt = ReadInt16(); + s << NewInt; + } + } + + } else if (vr == "UL" || vr == "SL") { + nbInt = length / 4; + NewInt = ReadInt32(); + s << NewInt; + if (nbInt > 1) { + for (int i=1; i < nbInt; i++) { + s << '\\'; + NewInt = ReadInt32(); + s << NewInt; + } + } + } + ElVal->SetValue(s.str()); + return; + } + + // We need an additional byte for storing \0 that is not on disk + char* NewValue = (char*)malloc(length+1); + if( !NewValue) { + dbg.Verbose(1, "LoadElementValue: Failed to allocate NewValue"); + return; + } + NewValue[length]= 0; + + item_read = fread(NewValue, (size_t)length, (size_t)1, fp); + if ( item_read != 1 ) { + free(NewValue); + dbg.Verbose(1, "gdcmHeader::LoadElementValue","unread element value"); + ElVal->SetValue("gdcm::UnRead"); + return; + } + ElVal->SetValue(NewValue); + free(NewValue); } +/** + * \ingroup gdcmHeader + * \brief Loads the element while preserving the current + * underlying file position indicator as opposed to + * to LoadElementValue that modifies it. + * @param ElVal Element whose value shall be loaded. + * @return + */ +void gdcmHeader::LoadElementValueSafe(gdcmElValue * ElVal) { + long PositionOnEntry = ftell(fp); + LoadElementValue(ElVal); + fseek(fp, PositionOnEntry, SEEK_SET); +} +/** + * \ingroup gdcmHeader + * \brief Reads a supposed to be 16 Bits integer + * \ (swaps it depending on processor endianity) + * + * @return integer acts as a boolean + */ guint16 gdcmHeader::ReadInt16(void) { - guint16 g; - size_t item_read; - item_read = fread (&g, (size_t)2,(size_t)1, fp); - if ( item_read != 1 ) - throw Error::FileReadError(fp, "gdcmHeader::ReadInt16"); - g = SwapShort(g); - return g; + guint16 g; + size_t item_read; + item_read = fread (&g, (size_t)2,(size_t)1, fp); + if ( item_read != 1 ) { + // dbg.Verbose(0, "gdcmHeader::ReadInt16", " Failed to read :"); + // if(feof(fp)) + // dbg.Verbose(0, "gdcmHeader::ReadInt16", " End of File encountered"); + if(ferror(fp)) + dbg.Verbose(0, "gdcmHeader::ReadInt16", " File Error"); + errno = 1; + return 0; + } + errno = 0; + g = SwapShort(g); + return g; } +/** + * \ingroup gdcmHeader + * \brief Reads a supposed to be 32 Bits integer + * \ (swaps it depending on processor endianity) + * + * @return + */ guint32 gdcmHeader::ReadInt32(void) { - guint32 g; - size_t item_read; - item_read = fread (&g, (size_t)4,(size_t)1, fp); - if ( item_read != 1 ) - throw Error::FileReadError(fp, "gdcmHeader::ReadInt32"); - g = SwapLong(g); - return g; + guint32 g; + size_t item_read; + item_read = fread (&g, (size_t)4,(size_t)1, fp); + if ( item_read != 1 ) { + //dbg.Verbose(0, "gdcmHeader::ReadInt32", " Failed to read :"); + //if(feof(fp)) + // dbg.Verbose(0, "gdcmHeader::ReadInt32", " End of File encountered"); + if(ferror(fp)) + dbg.Verbose(0, "gdcmHeader::ReadInt32", " File Error"); + errno = 1; + return 0; + } + errno = 0; + g = SwapLong(g); + return g; } /** * \ingroup gdcmHeader - * \brief Read the next tag without loading it's value + * \brief + * + * @return + */ + gdcmElValue* gdcmHeader::GetElValueByNumber(guint16 Group, guint16 Elem) { + + gdcmElValue* elValue = PubElValSet.GetElementByNumber(Group, Elem); + if (!elValue) { + dbg.Verbose(1, "gdcmHeader::GetElValueByNumber", + "failed to Locate gdcmElValue"); + return (gdcmElValue*)0; + } + return elValue; +} + +/** + * \ingroup gdcmHeader + * \brief Build a new Element Value from all the low level arguments. + * Check for existence of dictionary entry, and build + * a default one when absent. + * @param Group group of the underlying DictEntry + * @param Elem element of the underlying DictEntry + */ +gdcmElValue* gdcmHeader::NewElValueByNumber(guint16 Group, guint16 Elem) { + // Find out if the tag we encountered is in the dictionaries: + gdcmDictEntry * NewTag = GetDictEntryByNumber(Group, Elem); + if (!NewTag) + NewTag = new gdcmDictEntry(Group, Elem); + + gdcmElValue* NewElVal = new gdcmElValue(NewTag); + if (!NewElVal) { + dbg.Verbose(1, "gdcmHeader::NewElValueByNumber", + "failed to allocate gdcmElValue"); + return (gdcmElValue*)0; + } + return NewElVal; +} + +/** + * \ingroup gdcmHeader + * \brief TODO + * @param Value + * @param Group + * @param Elem + * \return integer acts as a boolean + */ +int gdcmHeader::ReplaceOrCreateByNumber(std::string Value, + guint16 Group, guint16 Elem ) { + + // TODO : FIXME JPRx + // curieux, non ? + // on (je) cree une Elvalue ne contenant pas de valeur + // on l'ajoute au ElValSet + // on affecte une valeur a cette ElValue a l'interieur du ElValSet + // --> devrait pouvoir etre fait + simplement ??? + if (CheckIfExistByNumber(Group, Elem) == 0) { + gdcmElValue* a =NewElValueByNumber(Group, Elem); + if (a == NULL) + return 0; + PubElValSet.Add(a); + } + PubElValSet.SetElValueByNumber(Value, Group, Elem); + return(1); +} + + +/** + * \ingroup gdcmHeader + * \brief Modify (or Creates if not found) an element + * @param Value new value + * @param Group + * @param Elem + * \return integer acts as a boolean + * + */ +int gdcmHeader::ReplaceOrCreateByNumber(char* Value, guint16 Group, guint16 Elem ) { + + gdcmElValue* nvElValue=NewElValueByNumber(Group, Elem); + PubElValSet.Add(nvElValue); + std::string v = Value; + PubElValSet.SetElValueByNumber(v, Group, Elem); + return(1); +} + + +/** + * \ingroup gdcmHeader + * \brief Set a new value if the invoked element exists + * Seems to be useless !!! + * @param Value + * @param Group + * @param Elem + * \return integer acts as a boolean + */ +int gdcmHeader::ReplaceIfExistByNumber(char* Value, guint16 Group, guint16 Elem ) { + + //gdcmElValue* elValue = PubElValSet.GetElementByNumber(Group, Elem); + std::string v = Value; + PubElValSet.SetElValueByNumber(v, Group, Elem); + return 1; +} + + +/** + * \ingroup gdcmHeader + * \brief Checks if a given ElValue (group,number) + * \ exists in the Public ElValSet + * @param Group + * @param Elem + * @return integer acts as a boolean + */ + +int gdcmHeader::CheckIfExistByNumber(guint16 Group, guint16 Elem ) { + return (PubElValSet.CheckIfExistByNumber(Group, Elem)); + } + +/** + * \ingroup gdcmHeader + * \brief Build a new Element Value from all the low level arguments. + * Check for existence of dictionary entry, and build + * a default one when absent. + * @param Name Name of the underlying DictEntry + */ +gdcmElValue* gdcmHeader::NewElValueByName(std::string Name) { + + gdcmDictEntry * NewTag = GetDictEntryByName(Name); + if (!NewTag) + NewTag = new gdcmDictEntry(0xffff, 0xffff, "LO", "Unknown", Name); + + gdcmElValue* NewElVal = new gdcmElValue(NewTag); + if (!NewElVal) { + dbg.Verbose(1, "gdcmHeader::ObtainElValueByName", + "failed to allocate gdcmElValue"); + return (gdcmElValue*)0; + } + return NewElVal; +} + +/** + * \ingroup gdcmHeader + * \brief Read the next tag but WITHOUT loading it's value * @return On succes the newly created ElValue, NULL on failure. */ +gdcmElValue * gdcmHeader::ReadNextElement(void) { + + guint16 g,n; + gdcmElValue * NewElVal; + + g = ReadInt16(); + n = ReadInt16(); + + if (errno == 1) + // We reached the EOF (or an error occured) and header parsing + // has to be considered as finished. + return (gdcmElValue *)0; + + NewElVal = NewElValueByNumber(g, n); + FindVR(NewElVal); + FindLength(NewElVal); + if (errno == 1) { + // Call it quits + return (gdcmElValue *)0; + } + NewElVal->SetOffset(ftell(fp)); + //if ( (g==0x7fe0) && (n==0x0010) ) + return NewElVal; +} -ElValue * gdcmHeader::ReadNextElement(void) { - guint16 g; - guint16 n; - ElValue * NewElVal; - - try { - g = ReadInt16(); - n = ReadInt16(); - } - catch ( Error::FileReadError ) { - // We reached the EOF (or an error occured) and header parsing - // has to be considered as finished. - return (ElValue *)0; - } - - // Find out if the tag we encountered is in the dictionaries: - gdcmDictEntry * NewTag = IsInDicts(g, n); - if (!NewTag) - NewTag = new gdcmDictEntry(g, n); - - NewElVal = new ElValue(NewTag); - if (!NewElVal) { - dbg.Verbose(1, "ReadNextElement: failed to allocate ElValue"); - return (ElValue*)0; - } - - FindVR(NewElVal); - FindLength(NewElVal); - NewElVal->SetOffset(ftell(fp)); - return NewElVal; -} - -bool gdcmHeader::IsAnInteger(guint16 group, guint16 element, - string vr, guint32 length ) { - // When we have some semantics on the element we just read, and if we - // a priori know we are dealing with an integer, then we shall be - // able to swap it's element value properly. - if ( element == 0 ) { // This is the group length of the group - if (length != 4) - dbg.Error("gdcmHeader::ShouldBeSwaped", "should be four"); - return true; - } - - if ( group % 2 != 0 ) - // We only have some semantics on documented elements, which are - // the even ones. - return false; - - if ( (length != 4) && ( length != 2) ) - // Swapping only make sense on integers which are 2 or 4 bytes long. - return false; - - if ( (vr == "UL") || (vr == "US") || (vr == "SL") || (vr == "SS") ) - return true; - - if ( (group == 0x0028) && (element == 0x0005) ) - // This tag is retained from ACR/NEMA - // CHECKME Why should "Image Dimensions" be a single integer ? - return true; - - if ( (group == 0x0028) && (element == 0x0200) ) - // This tag is retained from ACR/NEMA - return true; - - return false; +/** + * \ingroup gdcmHeader + * \brief Apply some heuristics to predict wether the considered + * element value contains/represents an integer or not. + * @param ElVal The element value on which to apply the predicate. + * @return The result of the heuristical predicate. + */ +bool gdcmHeader::IsAnInteger(gdcmElValue * ElVal) { + guint16 element = ElVal->GetElement(); + std::string vr = ElVal->GetVR(); + guint32 length = ElVal->GetLength(); + + // When we have some semantics on the element we just read, and if we + // a priori know we are dealing with an integer, then we shall be + // able to swap it's element value properly. + if ( element == 0 ) { // This is the group length of the group + if (length == 4) + return true; + else { + dbg.Error("gdcmHeader::IsAnInteger", + "Erroneous Group Length element length."); + } + } + if ( (vr == "UL") || (vr == "US") || (vr == "SL") || (vr == "SS") ) + return true; + + return false; } /** @@ -574,74 +1269,476 @@ bool gdcmHeader::IsAnInteger(guint16 group, guint16 element, * \brief Recover the offset (from the beginning of the file) of the pixels. */ size_t gdcmHeader::GetPixelOffset(void) { - // If this file complies with the norm we should encounter the - // "Image Location" tag (0x0028, 0x0200). This tag contains the - // the group that contains the pixel data (hence the "Pixel Data" - // is found by indirection through the "Image Location"). - // Inside the group pointed by "Image Location" the searched element - // is conventionally the element 0x0010 (when the norm is respected). - // When the "Image Location" is absent we default to group 0x7fe0. - guint16 grPixel; - guint16 numPixel; - string ImageLocation = GetPubElValByName("Image Location"); - if ( ImageLocation == "UNFOUND" ) { - grPixel = 0x7FE0; - } else { - grPixel = (guint16) atoi( ImageLocation.c_str() ); - } - if (grPixel != 0x7fe0) - // FIXME is this still necessary ? - // Now, this looks like an old dirty fix for Philips imager - numPixel = 0x1010; - else - numPixel = 0x0010; - ElValue* PixelElement = PubElVals.GetElement(grPixel, numPixel); - if (PixelElement) - return PixelElement->GetOffset(); - else - return 0; -} - -gdcmDictEntry * gdcmHeader::IsInDicts(guint32 group, guint32 element) { - gdcmDictEntry * found = (gdcmDictEntry*)0; - if (!RefPubDict && !RefShaDict) { - //FIXME build a default dictionary ! - printf("FIXME in gdcmHeader::IsInDicts\n"); - } - if (RefPubDict) { - found = RefPubDict->GetTag(group, element); - if (found) - return found; - } - if (RefShaDict) { - found = RefShaDict->GetTag(group, element); - if (found) - return found; - } - return found; -} - -string gdcmHeader::GetPubElValByNumber(guint16 group, guint16 element) { - return PubElVals.GetElValue(group, element); -} - -string gdcmHeader::GetPubElValByName(string TagName) { - return PubElVals.GetElValue(TagName); -} - -/** - * \ingroup gdcmHeader - * \brief Parses the header of the file but does NOT load element values. - */ -void gdcmHeader::ParseHeader(void) { - ElValue * newElValue = (ElValue *)0; - - rewind(fp); - CheckSwap(); - while ( (newElValue = ReadNextElement()) ) { - SkipElementValue(newElValue); - PubElVals.Add(newElValue); - } + // If this file complies with the norm we should encounter the + // "Image Location" tag (0x0028, 0x0200). This tag contains the + // the group that contains the pixel data (hence the "Pixel Data" + // is found by indirection through the "Image Location"). + // Inside the group pointed by "Image Location" the searched element + // is conventionally the element 0x0010 (when the norm is respected). + // When the "Image Location" is absent we default to group 0x7fe0. + guint16 grPixel; + guint16 numPixel; + std::string ImageLocation = GetPubElValByName("Image Location"); + if ( ImageLocation == GDCM_UNFOUND ) { + grPixel = 0x7fe0; + } else { + grPixel = (guint16) atoi( ImageLocation.c_str() ); + } + if (grPixel != 0x7fe0) + // This is a kludge for old dirty Philips imager. + numPixel = 0x1010; + else + numPixel = 0x0010; + + gdcmElValue* PixelElement = PubElValSet.GetElementByNumber(grPixel, + numPixel); + if (PixelElement) + return PixelElement->GetOffset(); + else + return 0; +} + +/** + * \ingroup gdcmHeader + * \brief Recover the pixel area length (in Bytes) . + */ +size_t gdcmHeader::GetPixelAreaLength(void) { + // If this file complies with the norm we should encounter the + // "Image Location" tag (0x0028, 0x0200). This tag contains the + // the group that contains the pixel data (hence the "Pixel Data" + // is found by indirection through the "Image Location"). + // Inside the group pointed by "Image Location" the searched element + // is conventionally the element 0x0010 (when the norm is respected). + // When the "Image Location" is absent we default to group 0x7fe0. + guint16 grPixel; + guint16 numPixel; + std::string ImageLocation = GetPubElValByName("Image Location"); + if ( ImageLocation == GDCM_UNFOUND ) { + grPixel = 0x7fe0; + } else { + grPixel = (guint16) atoi( ImageLocation.c_str() ); + } + if (grPixel != 0x7fe0) + // This is a kludge for old dirty Philips imager. + numPixel = 0x1010; + else + numPixel = 0x0010; + + gdcmElValue* PixelElement = PubElValSet.GetElementByNumber(grPixel, + numPixel); + if (PixelElement) + return PixelElement->GetLength(); + else + return 0; +} + +/** + * \ingroup gdcmHeader + * \brief Searches both the public and the shadow dictionary (when they + * exist) for the presence of the DictEntry with given + * group and element. The public dictionary has precedence on the + * shadow one. + * @param group group of the searched DictEntry + * @param element element of the searched DictEntry + * @return Corresponding DictEntry when it exists, NULL otherwise. + */ +gdcmDictEntry * gdcmHeader::GetDictEntryByNumber(guint16 group, + guint16 element) { + gdcmDictEntry * found = (gdcmDictEntry*)0; + if (!RefPubDict && !RefShaDict) { + dbg.Verbose(0, "gdcmHeader::GetDictEntry", + "we SHOULD have a default dictionary"); + } + if (RefPubDict) { + found = RefPubDict->GetTagByNumber(group, element); + if (found) + return found; + } + if (RefShaDict) { + found = RefShaDict->GetTagByNumber(group, element); + if (found) + return found; + } + return found; +} + +/** + * \ingroup gdcmHeader + * \brief Searches both the public and the shadow dictionary (when they + * exist) for the presence of the DictEntry with given name. + * The public dictionary has precedence on the shadow one. + * @param Name name of the searched DictEntry + * @return Corresponding DictEntry when it exists, NULL otherwise. + */ +gdcmDictEntry * gdcmHeader::GetDictEntryByName(std::string Name) { + gdcmDictEntry * found = (gdcmDictEntry*)0; + if (!RefPubDict && !RefShaDict) { + dbg.Verbose(0, "gdcmHeader::GetDictEntry", + "we SHOULD have a default dictionary"); + } + if (RefPubDict) { + found = RefPubDict->GetTagByName(Name); + if (found) + return found; + } + if (RefShaDict) { + found = RefShaDict->GetTagByName(Name); + if (found) + return found; + } + return found; +} + +/** + * \ingroup gdcmHeader + * \brief Searches within the public dictionary for element value of + * a given tag. + * @param group Group of the researched tag. + * @param element Element of the researched tag. + * @return Corresponding element value when it exists, and the string + * GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetPubElValByNumber(guint16 group, guint16 element) { + return PubElValSet.GetElValueByNumber(group, element); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within the public dictionary for element value + * representation of a given tag. + * + * Obtaining the VR (Value Representation) might be needed by caller + * to convert the string typed content to caller's native type + * (think of C++ vs Python). The VR is actually of a higher level + * of semantics than just the native C++ type. + * @param group Group of the researched tag. + * @param element Element of the researched tag. + * @return Corresponding element value representation when it exists, + * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetPubElValRepByNumber(guint16 group, guint16 element) { + gdcmElValue* elem = PubElValSet.GetElementByNumber(group, element); + if ( !elem ) + return GDCM_UNFOUND; + return elem->GetVR(); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within the public dictionary for element value of + * a given tag. + * @param TagName name of the researched element. + * @return Corresponding element value when it exists, and the string + * GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetPubElValByName(std::string TagName) { + return PubElValSet.GetElValueByName(TagName); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within the elements parsed with the public dictionary for + * the element value representation of a given tag. + * + * Obtaining the VR (Value Representation) might be needed by caller + * to convert the string typed content to caller's native type + * (think of C++ vs Python). The VR is actually of a higher level + * of semantics than just the native C++ type. + * @param TagName name of the researched element. + * @return Corresponding element value representation when it exists, + * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetPubElValRepByName(std::string TagName) { + gdcmElValue* elem = PubElValSet.GetElementByName(TagName); + if ( !elem ) + return GDCM_UNFOUND; + return elem->GetVR(); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within elements parsed with the SHADOW dictionary + * for the element value of a given tag. + * @param group Group of the researched tag. + * @param element Element of the researched tag. + * @return Corresponding element value representation when it exists, + * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetShaElValByNumber(guint16 group, guint16 element) { + return ShaElValSet.GetElValueByNumber(group, element); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within the elements parsed with the SHADOW dictionary + * for the element value representation of a given tag. + * + * Obtaining the VR (Value Representation) might be needed by caller + * to convert the string typed content to caller's native type + * (think of C++ vs Python). The VR is actually of a higher level + * of semantics than just the native C++ type. + * @param group Group of the researched tag. + * @param element Element of the researched tag. + * @return Corresponding element value representation when it exists, + * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetShaElValRepByNumber(guint16 group, guint16 element) { + gdcmElValue* elem = ShaElValSet.GetElementByNumber(group, element); + if ( !elem ) + return GDCM_UNFOUND; + return elem->GetVR(); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within the elements parsed with the shadow dictionary + * for an element value of given tag. + * @param TagName name of the researched element. + * @return Corresponding element value when it exists, and the string + * GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetShaElValByName(std::string TagName) { + return ShaElValSet.GetElValueByName(TagName); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within the elements parsed with the shadow dictionary for + * the element value representation of a given tag. + * + * Obtaining the VR (Value Representation) might be needed by caller + * to convert the string typed content to caller's native type + * (think of C++ vs Python). The VR is actually of a higher level + * of semantics than just the native C++ type. + * @param TagName name of the researched element. + * @return Corresponding element value representation when it exists, + * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetShaElValRepByName(std::string TagName) { + gdcmElValue* elem = ShaElValSet.GetElementByName(TagName); + if ( !elem ) + return GDCM_UNFOUND; + return elem->GetVR(); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within elements parsed with the public dictionary + * and then within the elements parsed with the shadow dictionary + * for the element value of a given tag. + * @param group Group of the researched tag. + * @param element Element of the researched tag. + * @return Corresponding element value representation when it exists, + * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetElValByNumber(guint16 group, guint16 element) { + std::string pub = GetPubElValByNumber(group, element); + if (pub.length()) + return pub; + return GetShaElValByNumber(group, element); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within elements parsed with the public dictionary + * and then within the elements parsed with the shadow dictionary + * for the element value representation of a given tag. + * + * Obtaining the VR (Value Representation) might be needed by caller + * to convert the string typed content to caller's native type + * (think of C++ vs Python). The VR is actually of a higher level + * of semantics than just the native C++ type. + * @param group Group of the researched tag. + * @param element Element of the researched tag. + * @return Corresponding element value representation when it exists, + * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetElValRepByNumber(guint16 group, guint16 element) { + std::string pub = GetPubElValRepByNumber(group, element); + if (pub.length()) + return pub; + return GetShaElValRepByNumber(group, element); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within elements parsed with the public dictionary + * and then within the elements parsed with the shadow dictionary + * for the element value of a given tag. + * @param TagName name of the researched element. + * @return Corresponding element value when it exists, + * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetElValByName(std::string TagName) { + std::string pub = GetPubElValByName(TagName); + if (pub.length()) + return pub; + return GetShaElValByName(TagName); +} + +/** + * \ingroup gdcmHeader + * \brief Searches within elements parsed with the public dictionary + * and then within the elements parsed with the shadow dictionary + * for the element value representation of a given tag. + * + * Obtaining the VR (Value Representation) might be needed by caller + * to convert the string typed content to caller's native type + * (think of C++ vs Python). The VR is actually of a higher level + * of semantics than just the native C++ type. + * @param TagName name of the researched element. + * @return Corresponding element value representation when it exists, + * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise. + */ +std::string gdcmHeader::GetElValRepByName(std::string TagName) { + std::string pub = GetPubElValRepByName(TagName); + if (pub.length()) + return pub; + return GetShaElValRepByName(TagName); +} + +/** + * \ingroup gdcmHeader + * \brief Accesses an existing gdcmElValue in the PubElValSet of this instance + * through it's (group, element) and modifies it's content with + * the given value. + * @param content new value to substitute with + * @param group group of the ElVal to modify + * @param element element of the ElVal to modify + */ +int gdcmHeader::SetPubElValByNumber(std::string content, guint16 group, + guint16 element) + +//TODO : homogeneiser les noms : SetPubElValByNumber +// qui appelle PubElValSet.SetElValueByNumber +// pourquoi pas SetPubElValueByNumber ?? +{ + + return ( PubElValSet.SetElValueByNumber (content, group, element) ); +} + +/** + * \ingroup gdcmHeader + * \brief Accesses an existing gdcmElValue in the PubElValSet of this instance + * through tag name and modifies it's content with the given value. + * @param content new value to substitute with + * @param TagName name of the tag to be modified + */ +int gdcmHeader::SetPubElValByName(std::string content, std::string TagName) { + return ( PubElValSet.SetElValueByName (content, TagName) ); +} + +/** + * \ingroup gdcmHeader + * \brief Accesses an existing gdcmElValue in the PubElValSet of this instance + * through it's (group, element) and modifies it's length with + * the given value. + * \warning Use with extreme caution. + * @param length new length to substitute with + * @param group group of the ElVal to modify + * @param element element of the ElVal to modify + * @return 1 on success, 0 otherwise. + */ + +int gdcmHeader::SetPubElValLengthByNumber(guint32 length, guint16 group, + guint16 element) { + return ( PubElValSet.SetElValueLengthByNumber (length, group, element) ); +} + +/** + * \ingroup gdcmHeader + * \brief Accesses an existing gdcmElValue in the ShaElValSet of this instance + * through it's (group, element) and modifies it's content with + * the given value. + * @param content new value to substitute with + * @param group group of the ElVal to modify + * @param element element of the ElVal to modify + * @return 1 on success, 0 otherwise. + */ +int gdcmHeader::SetShaElValByNumber(std::string content, + guint16 group, guint16 element) { + return ( ShaElValSet.SetElValueByNumber (content, group, element) ); +} + +/** + * \ingroup gdcmHeader + * \brief Accesses an existing gdcmElValue in the ShaElValSet of this instance + * through tag name and modifies it's content with the given value. + * @param content new value to substitute with + * @param ShadowTagName name of the tag to be modified + */ +int gdcmHeader::SetShaElValByName(std::string content, std::string ShadowTagName) { + return ( ShaElValSet.SetElValueByName (content, ShadowTagName) ); +} + +/** + * \ingroup gdcmHeader + * \brief Parses the header of the file but WITHOUT loading element values. + */ +void gdcmHeader::ParseHeader(bool exception_on_error) throw(gdcmFormatError) { + gdcmElValue * newElValue = (gdcmElValue *)0; + + rewind(fp); + CheckSwap(); + while ( (newElValue = ReadNextElement()) ) { + SkipElementValue(newElValue); + PubElValSet.Add(newElValue); + } +} + +/** + * \ingroup gdcmHeader + * \brief This predicate, based on hopefully reasonnable heuristics, + * decides whether or not the current gdcmHeader was properly parsed + * and contains the mandatory information for being considered as + * a well formed and usable image. + * @return true when gdcmHeader is the one of a reasonable Dicom file, + * false otherwise. + */ +bool gdcmHeader::IsReadable(void) { + if ( GetElValByName("Image Dimensions") != GDCM_UNFOUND + && atoi(GetElValByName("Image Dimensions").c_str()) > 4 ) { + return false; + } + if ( GetElValByName("Bits Allocated") == GDCM_UNFOUND ) + return false; + if ( GetElValByName("Bits Stored") == GDCM_UNFOUND ) + return false; + if ( GetElValByName("High Bit") == GDCM_UNFOUND ) + return false; + if ( GetElValByName("Pixel Representation") == GDCM_UNFOUND ) + return false; + return true; +} + +/** + * \ingroup gdcmHeader + * \brief Small utility function that creates a new manually crafted + * (as opposed as read from the file) gdcmElValue with user + * specified name and adds it to the public tag hash table. + * \note A fake TagKey is generated so the PubDict can keep it's coherence. + * @param NewTagName The name to be given to this new tag. + * @param VR The Value Representation to be given to this new tag. + * @ return The newly hand crafted Element Value. + */ +gdcmElValue* gdcmHeader::NewManualElValToPubDict(std::string NewTagName, + std::string VR) { + gdcmElValue* NewElVal = (gdcmElValue*)0; + guint32 StuffGroup = 0xffff; // Group to be stuffed with additional info + guint32 FreeElem = 0; + gdcmDictEntry* NewEntry = (gdcmDictEntry*)0; + + FreeElem = PubElValSet.GenerateFreeTagKeyInGroup(StuffGroup); + if (FreeElem == UINT32_MAX) { + dbg.Verbose(1, "gdcmHeader::NewManualElValToPubDict", + "Group 0xffff in Public Dict is full"); + return (gdcmElValue*)0; + } + NewEntry = new gdcmDictEntry(StuffGroup, FreeElem, + VR, "GDCM", NewTagName); + NewElVal = new gdcmElValue(NewEntry); + PubElValSet.Add(NewElVal); + return NewElVal; } /** @@ -650,16 +1747,556 @@ void gdcmHeader::ParseHeader(void) { * public tag based hash table. */ void gdcmHeader::LoadElements(void) { - rewind(fp); - TagElValueHT ht = PubElVals.GetTagHt(); - for (TagElValueHT::iterator tag = ht.begin(); tag != ht.end(); ++tag) - LoadElementValue(tag->second); + rewind(fp); + TagElValueHT ht = PubElValSet.GetTagHt(); + for (TagElValueHT::iterator tag = ht.begin(); tag != ht.end(); ++tag) { + LoadElementValue(tag->second); + } + rewind(fp); + + // Load 'non string' values + std::string PhotometricInterpretation = GetPubElValByNumber(0x0028,0x0004); + if( PhotometricInterpretation == "PALETTE COLOR " ){ + LoadElementVoidArea(0x0028,0x1200); // gray LUT + LoadElementVoidArea(0x0028,0x1201); // R LUT + LoadElementVoidArea(0x0028,0x1202); // G LUT + LoadElementVoidArea(0x0028,0x1203); // B LUT + + LoadElementVoidArea(0x0028,0x1221); // Segmented Red Palette Color LUT Data + LoadElementVoidArea(0x0028,0x1222); // Segmented Green Palette Color LUT Data + LoadElementVoidArea(0x0028,0x1223); // Segmented Blue Palette Color LUT Data + } + + // -------------------------------------------------------------- + // Special Patch to allow gdcm to read ACR-LibIDO formated images + // + // if recognition code tells us we deal with a LibIDO image + // we switch lineNumber and columnNumber + // + std::string RecCode; + RecCode = GetPubElValByNumber(0x0008, 0x0010); + if (RecCode == "ACRNEMA_LIBIDO_1.1" || + RecCode == "CANRME_AILIBOD1_1." ) { + filetype = ACR_LIBIDO; + std::string rows = GetPubElValByNumber(0x0028, 0x0010); + std::string columns = GetPubElValByNumber(0x0028, 0x0011); + SetPubElValByNumber(columns, 0x0028, 0x0010); + SetPubElValByNumber(rows , 0x0028, 0x0011); + } + // ----------------- End of Special Patch ---------------- +} + +/** + * \ingroup gdcmHeader + * \brief + * @return + */ +void gdcmHeader::PrintPubElVal(std::ostream & os) { + PubElValSet.Print(os); +} + +/** + * \ingroup gdcmHeader + * \brief + * @return + */ +void gdcmHeader::PrintPubDict(std::ostream & os) { + RefPubDict->Print(os); +} + +/** + * \ingroup gdcmHeader + * \brief + * @return integer, acts as a Boolean + */ +int gdcmHeader::Write(FILE * fp, FileType type) { + + + // TODO : move the following lines (and a lot of others) + // to a future function CheckAndCorrectHeader + + if (type == ImplicitVR) { + std::string implicitVRTransfertSyntax = "1.2.840.10008.1.2"; + ReplaceOrCreateByNumber(implicitVRTransfertSyntax,0x0002, 0x0010); + + //FIXME Refer to standards on page 21, chapter 6.2 "Value representation": + // values with a VR of UI shall be padded with a single trailing null + // Dans le cas suivant on doit pader manuellement avec un 0 + + PubElValSet.SetElValueLengthByNumber(18, 0x0002, 0x0010); + } + + if (type == ExplicitVR) { + std::string explicitVRTransfertSyntax = "1.2.840.10008.1.2.1"; + ReplaceOrCreateByNumber(explicitVRTransfertSyntax,0x0002, 0x0010); + + //FIXME Refer to standards on page 21, chapter 6.2 "Value representation": + // values with a VR of UI shall be padded with a single trailing null + // Dans le cas suivant on doit pader manuellement avec un 0 + + PubElValSet.SetElValueLengthByNumber(20, 0x0002, 0x0010); + } + + return PubElValSet.Write(fp, type); +} + +// +// ------------------------ 'non string' elements related functions +// + +/** + * \ingroup gdcmHeader + * \brief Loads (from disk) the element content + * when a string is not suitable + */ +void * gdcmHeader::LoadElementVoidArea(guint16 Group, guint16 Elem) { + gdcmElValue * Element= PubElValSet.GetElementByNumber(Group, Elem); + if ( !Element ) + return NULL; + size_t o =(size_t)Element->GetOffset(); + fseek(fp, o, SEEK_SET); + int l=Element->GetLength(); + void * a = malloc(l); + if(!a) { + std::cout << "Big Broblem (LoadElementVoidArea, malloc) " + << std::hex << Group << " " << Elem << std::endl; + return NULL; + } + /* int res = */ PubElValSet.SetVoidAreaByNumber(a, Group, Elem); + // TODO check the result + size_t l2 = fread(a, 1, l ,fp); + if(l != l2) { + std::cout << "Big Broblem (LoadElementVoidArea, fread) " + << std::hex << Group << " " << Elem << std::endl; + free(a); + return NULL; + } + return a; +} + +/** + * \ingroup gdcmHeader + * \brief Gets (from Header) the offset of a 'non string' element value + * \ (LoadElementValue has already be executed) + * @param Group + * @param Elem + * @return File Offset of the Element Value + */ + size_t gdcmHeader::GetPubElValOffsetByNumber(guint16 Group, guint16 Elem) { + gdcmElValue* elValue = PubElValSet.GetElementByNumber(Group, Elem); + if (!elValue) { + dbg.Verbose(1, "gdcmHeader::GetElValueByNumber", + "failed to Locate gdcmElValue"); + return (size_t)0; + } + return elValue->GetOffset(); +} + +/** + * \ingroup gdcmHeader +* \brief Gets (from Header) a 'non string' element value + * \ (LoadElementValue has already be executed) + * @param Group + * @param Elem + * @return Pointer to the 'non string' area + + */ + void * gdcmHeader::GetPubElValVoidAreaByNumber(guint16 Group, guint16 Elem) { + gdcmElValue* elValue = PubElValSet.GetElementByNumber(Group, Elem); + if (!elValue) { + dbg.Verbose(1, "gdcmHeader::GetElValueByNumber", + "failed to Locate gdcmElValue"); + return (NULL); + } + return elValue->GetVoidArea(); +} + + +// +// ============================================================================= +// Heuristics based accessors +//============================================================================== +// + +// TODO : move to an other file. + + +/** + * \ingroup gdcmHeader + * \brief Retrieve the number of columns of image. + * @return The encountered size when found, 0 by default. + */ +int gdcmHeader::GetXSize(void) { + // We cannot check for "Columns" because the "Columns" tag is present + // both in IMG (0028,0011) and OLY (6000,0011) sections of the dictionary. + std::string StrSize = GetPubElValByNumber(0x0028,0x0011); + if (StrSize == GDCM_UNFOUND) + return 0; + return atoi(StrSize.c_str()); +} + +/** + * \ingroup gdcmHeader + * \brief Retrieve the number of lines of image. + * \warning The defaulted value is 1 as opposed to gdcmHeader::GetXSize() + * @return The encountered size when found, 1 by default. + */ +int gdcmHeader::GetYSize(void) { + // We cannot check for "Rows" because the "Rows" tag is present + // both in IMG (0028,0010) and OLY (6000,0010) sections of the dictionary. + std::string StrSize = GetPubElValByNumber(0x0028,0x0010); + if (StrSize != GDCM_UNFOUND) + return atoi(StrSize.c_str()); + if ( IsDicomV3() ) + return 0; + else + // The Rows (0028,0010) entry is optional for ACR/NEMA. It might + // hence be a signal (1d image). So we default to 1: + return 1; +} + +/** + * \ingroup gdcmHeader + * \brief Retrieve the number of planes of volume or the number + * of frames of a multiframe. + * \warning When present we consider the "Number of Frames" as the third + * dimension. When absent we consider the third dimension as + * being the "Planes" tag content. + * @return The encountered size when found, 1 by default. + */ +int gdcmHeader::GetZSize(void) { + // Both in DicomV3 and ACR/Nema the consider the "Number of Frames" + // as the third dimension. + std::string StrSize = GetPubElValByNumber(0x0028,0x0008); + if (StrSize != GDCM_UNFOUND) + return atoi(StrSize.c_str()); + + // We then consider the "Planes" entry as the third dimension [we + // cannot retrieve by name since "Planes tag is present both in + // IMG (0028,0012) and OLY (6000,0012) sections of the dictionary]. + StrSize = GetPubElValByNumber(0x0028,0x0012); + if (StrSize != GDCM_UNFOUND) + return atoi(StrSize.c_str()); + return 1; +} + +/** + * \ingroup gdcmHeader + * \brief Retrieve the number of Bits Stored + * (as opposite to number of Bits Allocated) + * + * @return The encountered number of Bits Stored, 0 by default. + */ +int gdcmHeader::GetBitsStored(void) { + std::string StrSize = GetPubElValByNumber(0x0028,0x0101); + if (StrSize == GDCM_UNFOUND) + return 1; + return atoi(StrSize.c_str()); +} + +/** + * \ingroup gdcmHeader + * \brief Retrieve the number of Bits Allocated + * (8, 12 -compacted ACR-NEMA files, 16, ...) + * + * @return The encountered number of Bits Allocated, 0 by default. + */ +int gdcmHeader::GetBitsAllocated(void) { + std::string StrSize = GetPubElValByNumber(0x0028,0x0100); + if (StrSize == GDCM_UNFOUND) + return 1; + return atoi(StrSize.c_str()); +} + +/** + * \ingroup gdcmHeader + * \brief Retrieve the number of Samples Per Pixel + * (1 : gray level, 3 : RGB) + * + * @return The encountered number of Samples Per Pixel, 1 by default. + */ +int gdcmHeader::GetSamplesPerPixel(void) { + std::string StrSize = GetPubElValByNumber(0x0028,0x0002); + if (StrSize == GDCM_UNFOUND) + return 1; // Well, it's supposed to be mandatory ... + return atoi(StrSize.c_str()); +} + +/** + * \ingroup gdcmHeader + * \brief Retrieve the Planar Configuration for RGB images + * (0 : RGB Pixels , 1 : R Plane + G Plane + B Plane) + * + * @return The encountered Planar Configuration, 0 by default. + */ +int gdcmHeader::GetPlanarConfiguration(void) { + std::string StrSize = GetPubElValByNumber(0x0028,0x0006); + if (StrSize == GDCM_UNFOUND) + return 0; + return atoi(StrSize.c_str()); } -void gdcmHeader::PrintPubElVal(ostream & os) { - PubElVals.Print(os); +/** + * \ingroup gdcmHeader + * \brief Return the size (in bytes) of a single pixel of data. + * @return The size in bytes of a single pixel of data. + * + */ +int gdcmHeader::GetPixelSize(void) { + std::string PixelType = GetPixelType(); + if (PixelType == "8U" || PixelType == "8S") + return 1; + if (PixelType == "16U" || PixelType == "16S") + return 2; + if (PixelType == "32U" || PixelType == "32S") + return 4; + dbg.Verbose(0, "gdcmHeader::GetPixelSize: Unknown pixel type"); + return 0; } -void gdcmHeader::PrintPubDict(ostream & os) { - RefPubDict->Print(os); +/** + * \ingroup gdcmHeader + * \brief Build the Pixel Type of the image. + * Possible values are: + * - 8U unsigned 8 bit, + * - 8S signed 8 bit, + * - 16U unsigned 16 bit, + * - 16S signed 16 bit, + * - 32U unsigned 32 bit, + * - 32S signed 32 bit, + * \warning 12 bit images appear as 16 bit. + * @return + */ +std::string gdcmHeader::GetPixelType(void) { + std::string BitsAlloc; + BitsAlloc = GetElValByName("Bits Allocated"); + if (BitsAlloc == GDCM_UNFOUND) { + dbg.Verbose(0, "gdcmHeader::GetPixelType: unfound Bits Allocated"); + BitsAlloc = std::string("16"); + } + if (BitsAlloc == "12") + BitsAlloc = std::string("16"); + + std::string Signed; + Signed = GetElValByName("Pixel Representation"); + if (Signed == GDCM_UNFOUND) { + dbg.Verbose(0, "gdcmHeader::GetPixelType: unfound Pixel Representation"); + BitsAlloc = std::string("0"); + } + if (Signed == "0") + Signed = std::string("U"); + else + Signed = std::string("S"); + + return( BitsAlloc + Signed); } + +/** + * \ingroup gdcmHeader + * \brief gets the info from 0002,0010 : Transfert Syntax + * \ else 1. + * @return Transfert Syntax Name (as oposite to Transfert Syntax UID) + */ +std::string gdcmHeader::GetTransferSyntaxName(void) { + std::string TransfertSyntax = GetPubElValByNumber(0x0002,0x0010); + if (TransfertSyntax == GDCM_UNFOUND) { + dbg.Verbose(0, "gdcmHeader::GetTransferSyntaxName: unfound Transfert Syntax (0002,0010)"); + return "Uncompressed ACR-NEMA"; + } + // we do it only when we need it + gdcmTS * ts = gdcmGlobal::GetTS(); + std::string tsName=ts->GetValue(TransfertSyntax); + //delete ts; // Seg Fault when deleted ?! + return tsName; +} + +// -------------------------------- Lookup Table related functions ------------ + +/** + * \ingroup gdcmHeader + * \brief tells us if LUT are used + * @return int acts as a Boolean + */ + +int gdcmHeader::HasLUT(void) { + + // Just hope checking the presence of the LUT Descriptors is enough + if (GetPubElValByNumber(0x0028,0x1101) == GDCM_UNFOUND) + return 0; + // LutDescriptorGreen + if (GetPubElValByNumber(0x0028,0x1102) == GDCM_UNFOUND) + return 0; + // LutDescriptorBlue + if (GetPubElValByNumber(0x0028,0x1103) == GDCM_UNFOUND) + return 0; + return 1; +} + +/** + * \ingroup gdcmHeader + * \brief gets the info from 0028,1101 : Lookup Table Desc-Red + * \ else 0 + * @return Lookup Table nBit + * \ when (0028,0004),Photometric Interpretation = [PALETTE COLOR ] + */ + +int gdcmHeader::GetLUTNbits(void) { + std::vector tokens; + //int LutLength; + //int LutDepth; + int LutNbits; + //Just hope Lookup Table Desc-Red = Lookup Table Desc-Red = Lookup Table Desc-Blue + // Consistency already checked in GetLUTLength + std::string LutDescription = GetPubElValByNumber(0x0028,0x1101); + if (LutDescription == GDCM_UNFOUND) + return 0; + tokens.erase(tokens.begin(),tokens.end()); // clean any previous value + Tokenize (LutDescription, tokens, "\\"); + //LutLength=atoi(tokens[0].c_str()); + //LutDepth=atoi(tokens[1].c_str()); + LutNbits=atoi(tokens[2].c_str()); + tokens.clear(); + return LutNbits; +} + +/** + * \ingroup gdcmHeader + * \brief builts Red/Green/Blue/Alpha LUT from Header + * \ when (0028,0004),Photometric Interpretation = [PALETTE COLOR ] + * \ and (0028,1101),(0028,1102),(0028,1102) + * \ - xxx Palette Color Lookup Table Descriptor - are found + * \ and (0028,1201),(0028,1202),(0028,1202) + * \ - xxx Palette Color Lookup Table Data - are found + * \warning does NOT deal with : + * \ 0028 1100 Gray Lookup Table Descriptor (Retired) + * \ 0028 1221 Segmented Red Palette Color Lookup Table Data + * \ 0028 1222 Segmented Green Palette Color Lookup Table Data + * \ 0028 1223 Segmented Blue Palette Color Lookup Table Data + * \ no known Dicom reader deails with them :-( + * @return Lookup Table RGBA + */ + +void * gdcmHeader::GetLUTRGBA(void) { +// Not so easy : see +// http://www.barre.nom.fr/medical/dicom2/limitations.html#Color%20Lookup%20Tables +// and OT-PAL-8-face.dcm + +// if Photometric Interpretation # PALETTE COLOR, no LUT to be done + + if (gdcmHeader::GetPubElValByNumber(0x0028,0x0004) != "PALETTE COLOR ") { + return NULL; + } + + void * LutR,*LutG,*LutB; + int l; + int lengthR, debR, nbitsR; + int lengthG, debG, nbitsG; + int lengthB, debB, nbitsB; + +// Get info from Lut Descriptors +// (the 3 LUT descriptors may be different) + + std::string LutDescriptionR = GetPubElValByNumber(0x0028,0x1101); + if (LutDescriptionR == GDCM_UNFOUND) + return NULL; + std::string LutDescriptionG = GetPubElValByNumber(0x0028,0x1102); + if (LutDescriptionG == GDCM_UNFOUND) + return NULL; + std::string LutDescriptionB = GetPubElValByNumber(0x0028,0x1103); + if (LutDescriptionB == GDCM_UNFOUND) + return NULL; + + std::vector tokens; + + tokens.erase(tokens.begin(),tokens.end()); // clean any previous value + Tokenize (LutDescriptionR, tokens, "\\"); + lengthR=atoi(tokens[0].c_str()); // Red LUT length in Bytes + debR =atoi(tokens[1].c_str()); // subscript of the first Lut Value + nbitsR =atoi(tokens[2].c_str()); // Lut item size (in Bits) + tokens.clear(); + + tokens.erase(tokens.begin(),tokens.end()); // clean any previous value + Tokenize (LutDescriptionG, tokens, "\\"); + lengthG=atoi(tokens[0].c_str()); // Green LUT length in Bytes + debG =atoi(tokens[1].c_str()); + nbitsG =atoi(tokens[2].c_str()); + tokens.clear(); + + tokens.erase(tokens.begin(),tokens.end()); // clean any previous value + Tokenize (LutDescriptionB, tokens, "\\"); + lengthB=atoi(tokens[0].c_str()); // Blue LUT length in Bytes + debB =atoi(tokens[1].c_str()); + nbitsB =atoi(tokens[2].c_str()); + tokens.clear(); + +// Load LUTs into memory, (as they were stored on disk) + + unsigned char *lutR =(unsigned char *) + GetPubElValVoidAreaByNumber(0x0028,0x1201); + unsigned char *lutG =(unsigned char *) + GetPubElValVoidAreaByNumber(0x0028,0x1202); + unsigned char *lutB =(unsigned char *) + GetPubElValVoidAreaByNumber(0x0028,0x1203); + + if (!lutR || !lutG || !lutB ) { + return NULL; + } + // forge the 4 * 8 Bits Red/Green/Blue/Alpha LUT + + char *LUTRGBA = (char *)calloc(1024,1); // 256 * 4 (R, G, B, Alpha) + if (!LUTRGBA) { + return NULL; + } + // Bits Allocated + int nb; + std::string str_nb = GetPubElValByNumber(0x0028,0x0100); + if (str_nb == GDCM_UNFOUND ) { + nb = 16; + } else { + nb = atoi(str_nb.c_str() ); + } + int mult; + + if (nbitsR==16 && nb==8) // when LUT item size is different than pixel size + mult=2; // high byte must be = low byte + else // See PS 3.3-2003 C.11.1.1.2 p 619 + mult=1; + + + // if we get a black image, let's just remove the '+1' + // from 'i*mult+1' and check again + // if it works, we shall have to check the 3 Palettes + // to see which byte is ==0 (first one, or second one) + // and fix the code + // We give up the checking to avoid some overhead + char *a; + a= LUTRGBA+debR; + for(int i=0;i