From: frog Date: Thu, 6 Mar 2003 14:04:34 +0000 (+0000) Subject: * src/gdcmHeader.h and gdcmHeader.cxx Coding style + doxigenation. X-Git-Tag: April2003~30 X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=commitdiff_plain;h=d4f7a5d74be3a4a332380d24631f3d0d96f98f5f;p=gdcm.git * src/gdcmHeader.h and gdcmHeader.cxx Coding style + doxigenation. * src/gdcm.h general comments moved to TODO, and README * src/gdcm.h should now be seen as a user commodity (including it should suffice to fully use gdcm). It is nowhere including within any of the kernel files src/*.cxx and src/*.h. --- diff --git a/ChangeLog b/ChangeLog index 302a1eca..38d0686c 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,10 @@ +2003-03-06 Eric Boix + * src/gdcmHeader.h and gdcmHeader.cxx Coding style + doxigenation. + * src/gdcm.h general comments moved to TODO, and README + * src/gdcm.h should now be seen as a user commodity (including it + should suffice to fully use gdcm). It is nowhere including within + any of the kernel files src/*.cxx and src/*.h. + 2003-03-05 Eric Boix * src/gdcm.h splitted in gdcmCommon.h, gdcmDict.h, gdcmDictEntry.h, gdcmDictSet.h, gdcmElValSet.h, gdcmElValue.h, gdcmFile.h, diff --git a/README b/README index d6743f63..cc409ad1 100644 --- a/README +++ b/README @@ -1,2 +1,14 @@ -Requires Swig version >= 1.3.17 for handling member overloading, and -static methods. +MANIFEST: + * gdcmlib is a library dedicated to reading and writing dicom files. + * LGPL for the license + * lightweigth as opposed to CTN or DCMTK which come bundled which try + to implement the full DICOM standard (networking...). gdcmlib concentrates + on reading and writing files (if you want DICOMNET, forget it). + * Formats: this lib should be able to read ACR-NEMA v1 and v2, Dicom v3 (as + stated in part10). [cf dcmtk/dcmdata/docs/datadict.txt] + * Targeted plateforms: Un*xes and Win32/VC++6.0 + +REQUIREMENTS: + * a C++ compiler with the and STL containers. + * the python wrappers require Swig version >= 1.3.17 for handling member + overloading, shadow classes and static methods. diff --git a/TODO b/TODO index 1b0dc98a..5ca3f531 100644 --- a/TODO +++ b/TODO @@ -1,3 +1,8 @@ +* The declarations commented out and starting with "TODO Swig" (try + grep "TODO Swig" *.h) needed to be temporarily removed for swig to + proceed correctly (in fact problems appears at loading of _gdcm.[so/dll]). + So, simply uncomment the declaration once you provided the definition of + the method... * As stated by the first lines of Test/ExceptionAndPython/README, it looks like we can move back to the exceptions and remove the errno stuff from src/gdcm* ! @@ -15,21 +20,13 @@ in the element values dictionary without doing something smarter (say, instead of storing the length store the group and the length so we can related a length to a group). - -GetPubElValByNumber doit faire la difference entre chaine vide -et chaine pas touve''. Eventuellement raiser une exception ? - -gdcmHeader::ReadNextElement: retarder le stockage en mem des gros elements - * etablir une taille limite sur les elements a ramener en memoire - (parametrisable en public dans la classe) - * ne pas ramener les elements au dela de ce seuil, mais les flaguer - comme a charger plus tard - * a l'acces (pour le groupe des pixels par exemple) retourner au fichier - pour trouver la bonne valeur (GetPixelData). - * fournir une method qui ne fait que lire les elements passes en arguments - sous forme d'une liste. - -grep str2num *.cxx: c'est une macro sans doute proprifiable - -gdcmHeader::CheckSwap() dans le cas ACR pas propre, degager tout de suite -si on a deduit que c'en est pas... +* GetPubElValByNumber doit faire la difference entre chaine vide + et chaine pas touve''. Eventuellement raiser une exception ? +* gdcmHeader::LoadElements only loads the element whose length is + below the specified size. When accessing the value of such a element + the content is unfound ! Find a decent way of loading the value on + explicit demand. +* fournir une method qui ne fait que lire les elements passes en arguments + sous forme d'une liste. +* gdcmHeader::CheckSwap() dans le cas ACR pas propre, degager tout de suite + si on a deduit que c'en est pas... diff --git a/src/gdcm.h b/src/gdcm.h index c3d5bda2..c9d6105b 100644 --- a/src/gdcm.h +++ b/src/gdcm.h @@ -1,44 +1,10 @@ // gdcm.h -// gdcmlib Intro: -// * gdcmlib is a library dedicated to reading and writing dicom files. -// * LGPL for the license -// * lightweigth as opposed to CTN or DCMTK which come bundled which try -// to implement the full DICOM standard (networking...). gdcmlib concentrates -// on reading and writing -// * Formats: this lib should be able to read ACR-NEMA v1 and v2, Dicom v3 (as -// stated in part10). [cf dcmtk/dcmdata/docs/datadict.txt] -// * Targeted plateforms: Un*xes and Win32/VC++6.0 -// -// -// TODO -// The declarations commented out and starting with "TODO Swig" needed -// to be temporarily removed for swig to proceed correctly (in fact -// problems appears at loading of _gdcm.[so/dll]). So, simply uncomment -// the declaration once you provided the definition of the method... - #ifndef GDCM_H #define GDCM_H using namespace std; -#include -#include // For size_t -#include // FIXME For FILE on GCC only -#include -#include -#include "gdcmException.h" - - -// The requirement for the hash table (or map) that we shall use: -// 1/ First, next, last (iterators) -// 2/ should be sortable (i.e. sorted by TagKey). This -// condition shall be droped since the Win32/VC++ -// implementation doesn't look a sorted one. Pffff.... -// 3/ Make sure we can setup some default size value, -// which should be around 4500 entries which is the -// average dictionary size (said JPR) -// // JPR CLEAN // En fait, je disais que dans LE Directory Dicom (dans son etat 2001) // il y a +/- 1600 entrees. @@ -64,6 +30,7 @@ using namespace std; // the | (pipe symbol) acts as a separator). Refer to // gdcmDictEntry::TranslateToKey for this conversion function. +#include "gdcmException.h" #include "gdcmCommon.h" #include "gdcmDictEntry.h" #include "gdcmDict.h" diff --git a/src/gdcmHeader.cxx b/src/gdcmHeader.cxx index a01595f8..bac92ce9 100644 --- a/src/gdcmHeader.cxx +++ b/src/gdcmHeader.cxx @@ -8,35 +8,26 @@ #else #include #endif -#include // for isalpha +#include // for isalpha #include #include "gdcmUtil.h" #include "gdcmHeader.h" -#define HEADER_LENGTH_TO_READ 256 // on ne lit plus que le debut -#define _MaxSizeLoadElementValue_ 1024 // longueur au dela de laquelle on ne charge plus les valeurs - -//FIXME: this looks dirty to me... - -#define str2num(str, typeNum) *((typeNum *)(str)) - -// str est un pointeur dans un tableau de caractères, qui doit contenir, -// à cet endroit la, la représentation binaire d'un entier (16 ou 32 bits) -// je veux récupérer ça ... dans un entier. -// s'il y a une autre solution, évitant des cast et les indirections, -// je suis preneur +// Refer to gdcmHeader::CheckSwap() +#define HEADER_LENGTH_TO_READ 256 +// Refer to gdcmHeader::SetMaxSizeLoadElementValue() +#define _MaxSizeLoadElementValue_ 1024 VRHT * gdcmHeader::dicom_vr = (VRHT*)0; void gdcmHeader::Initialise(void) { - if (!gdcmHeader::dicom_vr) - InitVRDict(); - Dicts = new gdcmDictSet(); - RefPubDict = Dicts->GetDefaultPubDict(); - RefShaDict = (gdcmDict*)0; + if (!gdcmHeader::dicom_vr) + InitVRDict(); + Dicts = new gdcmDictSet(); + RefPubDict = Dicts->GetDefaultPubDict(); + RefShaDict = (gdcmDict*)0; } - gdcmHeader::gdcmHeader(const char *InFilename, bool exception_on_error) throw(gdcmFileError) { SetMaxSizeLoadElementValue(_MaxSizeLoadElementValue_); @@ -55,43 +46,44 @@ gdcmHeader::gdcmHeader(const char *InFilename, bool exception_on_error) gdcmHeader::~gdcmHeader (void) { - fclose(fp); - return; + //FIXME obviously there is much to be done here ! + 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"; // At most 16 bytes - (*vr)["AS"] = "Age String"; // Exactly 4 bytes - (*vr)["AT"] = "Attribute Tag"; // 2 16-bit unsigned short integers - (*vr)["CS"] = "Code String"; // At most 16 bytes - (*vr)["DA"] = "Date"; // Exactly 8 bytes - (*vr)["DS"] = "Decimal String"; // At most 16 bytes - (*vr)["DT"] = "Date Time"; // At most 26 bytes - (*vr)["FL"] = "Floating Point Single"; // 32-bit IEEE 754:1985 float - (*vr)["FD"] = "Floating Point Double"; // 64-bit IEEE 754:1985 double - (*vr)["IS"] = "Integer String"; // At most 12 bytes - (*vr)["LO"] = "Long String"; // At most 64 chars - (*vr)["LT"] = "Long Text"; // At most 10240 chars - (*vr)["OB"] = "Other Byte String"; // String of bytes (vr independant) - (*vr)["OW"] = "Other Word String"; // String of 16-bit words (vr dep) - (*vr)["PN"] = "Person Name"; // At most 64 chars - (*vr)["SH"] = "Short String"; // At most 16 chars - (*vr)["SL"] = "Signed Long"; // Exactly 4 bytes - (*vr)["SQ"] = "Sequence of Items"; // Not Applicable - (*vr)["SS"] = "Signed Short"; // Exactly 2 bytes - (*vr)["ST"] = "Short Text"; // At most 1024 chars - (*vr)["TM"] = "Time"; // At most 16 bytes - (*vr)["UI"] = "Unique Identifier"; // At most 64 bytes - (*vr)["UL"] = "Unsigned Long "; // Exactly 4 bytes - (*vr)["UN"] = "Unknown"; // Any length of bytes - (*vr)["US"] = "Unsigned Short "; // Exactly 2 bytes - (*vr)["UT"] = "Unlimited Text"; // At most 2^32 -1 chars - dicom_vr = vr; + if (dicom_vr) { + dbg.Verbose(0, "gdcmHeader::InitVRDict:", "VR dictionary allready set"); + return; + } + VRHT *vr = new VRHT; + (*vr)["AE"] = "Application Entity"; // At most 16 bytes + (*vr)["AS"] = "Age String"; // Exactly 4 bytes + (*vr)["AT"] = "Attribute Tag"; // 2 16-bit unsigned short integers + (*vr)["CS"] = "Code String"; // At most 16 bytes + (*vr)["DA"] = "Date"; // Exactly 8 bytes + (*vr)["DS"] = "Decimal String"; // At most 16 bytes + (*vr)["DT"] = "Date Time"; // At most 26 bytes + (*vr)["FL"] = "Floating Point Single"; // 32-bit IEEE 754:1985 float + (*vr)["FD"] = "Floating Point Double"; // 64-bit IEEE 754:1985 double + (*vr)["IS"] = "Integer String"; // At most 12 bytes + (*vr)["LO"] = "Long String"; // At most 64 chars + (*vr)["LT"] = "Long Text"; // At most 10240 chars + (*vr)["OB"] = "Other Byte String"; // String of bytes (vr independant) + (*vr)["OW"] = "Other Word String"; // String of 16-bit words (vr dep) + (*vr)["PN"] = "Person Name"; // At most 64 chars + (*vr)["SH"] = "Short String"; // At most 16 chars + (*vr)["SL"] = "Signed Long"; // Exactly 4 bytes + (*vr)["SQ"] = "Sequence of Items"; // Not Applicable + (*vr)["SS"] = "Signed Short"; // Exactly 2 bytes + (*vr)["ST"] = "Short Text"; // At most 1024 chars + (*vr)["TM"] = "Time"; // At most 16 bytes + (*vr)["UI"] = "Unique Identifier"; // At most 64 bytes + (*vr)["UL"] = "Unsigned Long "; // Exactly 4 bytes + (*vr)["UN"] = "Unknown"; // Any length of bytes + (*vr)["US"] = "Unsigned Short "; // Exactly 2 bytes + (*vr)["UT"] = "Unlimited Text"; // At most 2^32 -1 chars + dicom_vr = vr; } /** @@ -102,151 +94,154 @@ void gdcmHeader::InitVRDict (void) { */ void gdcmHeader::CheckSwap() { - // 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 : pour 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) { - 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, guint32); - - 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. - 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 : pour 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) { + 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; + // 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. + if (! net2host ) + sw = 0; + else + sw = 4321; + return; } 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; + 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; } void gdcmHeader::GetPixels(size_t lgrTotale, void* _Pixels) { - size_t pixelsOffset; - pixelsOffset = GetPixelOffset(); - fseek(fp, pixelsOffset, SEEK_SET); - fread(_Pixels, 1, lgrTotale, fp); + size_t pixelsOffset; + pixelsOffset = GetPixelOffset(); + fseek(fp, pixelsOffset, SEEK_SET); + fread(_Pixels, 1, lgrTotale, fp); } @@ -256,79 +251,79 @@ void gdcmHeader::GetPixels(size_t lgrTotale, void* _Pixels) { * \brief Find the value representation of the current tag. */ void gdcmHeader::FindVR( ElValue *ElVal) { - if (filetype != ExplicitVR) - return; - - 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; - - 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++ 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. - 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(); + if (filetype != ExplicitVR) + return; + + 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; + + 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++ 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. + 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(); } /** @@ -339,14 +334,14 @@ void gdcmHeader::FindVR( ElValue *ElVal) { * @return True when ImplicitVRLittleEndian found. False in all other cases. */ bool gdcmHeader::IsImplicitVRLittleEndianTransferSyntax(void) { - ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); - if ( !Element ) - return false; - LoadElementValueSafe(Element); - string Transfer = Element->GetValue(); - if ( Transfer == "1.2.840.10008.1.2" ) - return true; - return false; + ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2" ) + return true; + return false; } /** @@ -357,14 +352,14 @@ bool gdcmHeader::IsImplicitVRLittleEndianTransferSyntax(void) { * @return True when ExplicitVRLittleEndian found. False in all other cases. */ bool gdcmHeader::IsExplicitVRLittleEndianTransferSyntax(void) { - ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); - if ( !Element ) - return false; - LoadElementValueSafe(Element); - string Transfer = Element->GetValue(); - if ( Transfer == "1.2.840.10008.1.2.1" ) - return true; - return false; + ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.1" ) + return true; + return false; } /** @@ -375,17 +370,16 @@ bool gdcmHeader::IsExplicitVRLittleEndianTransferSyntax(void) { * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases. */ bool gdcmHeader::IsDeflatedExplicitVRLittleEndianTransferSyntax(void) { - ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); - if ( !Element ) - return false; - LoadElementValueSafe(Element); - string Transfer = Element->GetValue(); - if ( Transfer == "1.2.840.10008.1.2.1.99" ) - return true; - return false; + ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + 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 allready encountered @@ -394,17 +388,16 @@ bool gdcmHeader::IsDeflatedExplicitVRLittleEndianTransferSyntax(void) { * @return True when big endian found. False in all other cases. */ bool gdcmHeader::IsExplicitVRBigEndianTransferSyntax(void) { - ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); - if ( !Element ) - return false; - LoadElementValueSafe(Element); - string Transfer = Element->GetValue(); - if ( Transfer == "1.2.840.10008.1.2.2" ) - return true; - return false; + ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.2" ) + return true; + return false; } - /** * \ingroup gdcmHeader * \brief Determines if the Transfer Syntax was allready encountered @@ -413,14 +406,14 @@ bool gdcmHeader::IsExplicitVRBigEndianTransferSyntax(void) { * @return True when JPEGBaseLineProcess1found. False in all other cases. */ bool gdcmHeader::IsJPEGBaseLineProcess1TransferSyntax(void) { - ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); - if ( !Element ) - return false; - LoadElementValueSafe(Element); - string Transfer = Element->GetValue(); - if ( Transfer == "1.2.840.10008.1.2.4.50" ) - return true; - return false; + ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.4.50" ) + return true; + return false; } /** @@ -431,17 +424,16 @@ bool gdcmHeader::IsJPEGBaseLineProcess1TransferSyntax(void) { * @return True when JPEGExtendedProcess2-4 found. False in all other cases. */ bool gdcmHeader::IsJPEGExtendedProcess2_4TransferSyntax(void) { - ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); - if ( !Element ) - return false; - LoadElementValueSafe(Element); - string Transfer = Element->GetValue(); - if ( Transfer == "1.2.840.10008.1.2.4.51" ) - return true; - return false; + ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.4.51" ) + return true; + return false; } - /** * \ingroup gdcmHeader * \brief Determines if the Transfer Syntax was allready encountered @@ -450,14 +442,14 @@ bool gdcmHeader::IsJPEGExtendedProcess2_4TransferSyntax(void) { * @return True when JPEGExtendedProcess3-5 found. False in all other cases. */ bool gdcmHeader::IsJPEGExtendedProcess3_5TransferSyntax(void) { - ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); - if ( !Element ) - return false; - LoadElementValueSafe(Element); - string Transfer = Element->GetValue(); - if ( Transfer == "1.2.840.10008.1.2.4.52" ) - return true; - return false; + ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.4.52" ) + return true; + return false; } /** @@ -465,163 +457,160 @@ bool gdcmHeader::IsJPEGExtendedProcess3_5TransferSyntax(void) { * \brief Determines if the Transfer Syntax was allready encountered * and if it corresponds to a JPEGSpectralSelectionProcess6-8 one. * - * @return True when JPEGSpectralSelectionProcess6-8 found. False in all other cases. + * @return True when JPEGSpectralSelectionProcess6-8 found. False in all + * other cases. */ bool gdcmHeader::IsJPEGSpectralSelectionProcess6_8TransferSyntax(void) { - ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); - if ( !Element ) - return false; - LoadElementValueSafe(Element); - string Transfer = Element->GetValue(); - if ( Transfer == "1.2.840.10008.1.2.4.53" ) - return true; - return false; + ElValue* Element = PubElVals.GetElementByNumber(0x0002, 0x0010); + if ( !Element ) + return false; + LoadElementValueSafe(Element); + string Transfer = Element->GetValue(); + if ( Transfer == "1.2.840.10008.1.2.4.53" ) + return true; + return false; } -// -// Euhhhhhhh -// Il y en a encore DIX-SEPT, comme ça. -// Il faudrait trouver qq chose + rusé ... -// -// --> probablement TOUS les supprimer (Eric dixit) -// - - +/** + * \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(ElValue * ElVal, guint32 FoundLength) { - // Heuristic: a final fix. - if ( FoundLength == 0xffffffff) - FoundLength = 0; - ElVal->SetLength(FoundLength); + if ( FoundLength == 0xffffffff) + FoundLength = 0; + ElVal->SetLength(FoundLength); } 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 ) { - dbg.Verbose(1, "gdcmHeader::FindLengthOB: ", - "wrong group for an item sequence."); - errno = 1; - return 0; - } - if ( n == 0xe0dd ) - FoundSequenceDelimiter = true; - else if ( n != 0xe000) { - dbg.Verbose(1, "gdcmHeader::FindLengthOB: ", - "wrong element for an item sequence."); - 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; + // 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 ) { + dbg.Verbose(1, "gdcmHeader::FindLengthOB: ", + "wrong group for an item sequence."); + errno = 1; + return 0; + } + if ( n == 0xe0dd ) + FoundSequenceDelimiter = true; + else if ( n != 0xe000) { + dbg.Verbose(1, "gdcmHeader::FindLengthOB: ", + "wrong element for an item sequence."); + 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; } void gdcmHeader::FindLength(ElValue * ElVal) { - guint16 element = ElVal->GetElement(); - string vr = ElVal->GetVR(); - guint16 length16; - - 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 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 allready 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 got our - // hands on a big endian encoded file: we switch the swap code to - // big endian and proceed... - if ( (element == 0x000) && (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 = GetDictEntryByKey(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."); - } - 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()); + guint16 element = ElVal->GetElement(); + string vr = ElVal->GetVR(); + guint16 length16; + + 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 allready 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 == 0x000) && (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 = GetDictEntryByKey(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."); + } + 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()); } /** @@ -632,27 +621,27 @@ void gdcmHeader::FindLength(ElValue * ElVal) { * @return The suggested 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); + // 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); } /** @@ -661,28 +650,28 @@ guint32 gdcmHeader::SwapLong(guint32 a) { * @return The properly swaped 16 bits integer. */ guint16 gdcmHeader::SwapShort(guint16 a) { - 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::SkipBytes(guint32 NBytes) { - //FIXME don't dump the returned value - (void)fseek(fp, (long)NBytes, SEEK_CUR); + //FIXME don't dump the returned value + (void)fseek(fp, (long)NBytes, SEEK_CUR); } void gdcmHeader::SkipElementValue(ElValue * ElVal) { - SkipBytes(ElVal->GetLength()); + SkipBytes(ElVal->GetLength()); } void gdcmHeader::SetMaxSizeLoadElementValue(long NewSize) { - if (NewSize < 0) - return; - if ((guint32)NewSize >= (guint32)0xffffffff) { - MaxSizeLoadElementValue = 0xffffffff; - return; - } - MaxSizeLoadElementValue = NewSize; + if (NewSize < 0) + return; + if ((guint32)NewSize >= (guint32)0xffffffff) { + MaxSizeLoadElementValue = 0xffffffff; + return; + } + MaxSizeLoadElementValue = NewSize; } /** @@ -692,98 +681,98 @@ void gdcmHeader::SetMaxSizeLoadElementValue(long NewSize) { * 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(); - bool SkipLoad = false; - - fseek(fp, (long)ElVal->GetOffset(), SEEK_SET); - - // 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 - // - if( vr == "SQ" ) - SkipLoad = true; - - // Heuristic : 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 ) - SkipLoad = true; - - // The group length doesn't represent data to be loaded in memory, since - // each element of the group shall be loaded individualy. - if( elem == 0 ) - //SkipLoad = true; // modif sauvage JPR - // On charge la longueur du groupe - // quand l'element 0x0000 est présent ! - - if ( SkipLoad ) { - // FIXME the following skip is not necessary - SkipElementValue(ElVal); - 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) { - 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(ElVal) ) { - guint32 NewInt; - if( length == 2 ) { - NewInt = ReadInt16(); - } else if( length == 4 ) { - NewInt = ReadInt32(); - } else - dbg.Error(true, "LoadElementValue: Inconsistency when reading Int."); - - //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*)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); + size_t item_read; + guint16 group = ElVal->GetGroup(); + guint16 elem = ElVal->GetElement(); + string vr = ElVal->GetVR(); + guint32 length = ElVal->GetLength(); + bool SkipLoad = false; + + fseek(fp, (long)ElVal->GetOffset(), SEEK_SET); + + // 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 + // + if( vr == "SQ" ) + SkipLoad = true; + + // Heuristic : 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 ) + SkipLoad = true; + + // The group length doesn't represent data to be loaded in memory, since + // each element of the group shall be loaded individualy. + if( elem == 0 ) + //SkipLoad = true; // modif sauvage JPR + // On charge la longueur du groupe + // quand l'element 0x0000 est présent ! + + if ( SkipLoad ) { + // FIXME the following skip is not necessary + SkipElementValue(ElVal); + 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) { + 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(ElVal) ) { + guint32 NewInt; + if( length == 2 ) { + NewInt = ReadInt16(); + } else if( length == 4 ) { + NewInt = ReadInt32(); + } else + dbg.Error(true, "LoadElementValue: Inconsistency when reading Int."); + + //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*)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); } /** @@ -795,38 +784,38 @@ void gdcmHeader::LoadElementValue(ElValue * ElVal) { * @return */ void gdcmHeader::LoadElementValueSafe(ElValue * ElVal) { - long PositionOnEntry = ftell(fp); - LoadElementValue(ElVal); - fseek(fp, PositionOnEntry, SEEK_SET); + long PositionOnEntry = ftell(fp); + LoadElementValue(ElVal); + fseek(fp, PositionOnEntry, SEEK_SET); } guint16 gdcmHeader::ReadInt16(void) { - guint16 g; - size_t item_read; - item_read = fread (&g, (size_t)2,(size_t)1, fp); - errno = 0; - if ( item_read != 1 ) { - dbg.Verbose(1, "gdcmHeader::ReadInt16", " File read error"); - errno = 1; - return 0; - } - g = SwapShort(g); - return g; + guint16 g; + size_t item_read; + item_read = fread (&g, (size_t)2,(size_t)1, fp); + errno = 0; + if ( item_read != 1 ) { + dbg.Verbose(1, "gdcmHeader::ReadInt16", " File read error"); + errno = 1; + return 0; + } + g = SwapShort(g); + return g; } guint32 gdcmHeader::ReadInt32(void) { - guint32 g; - size_t item_read; - item_read = fread (&g, (size_t)4,(size_t)1, fp); - errno = 0; - if ( item_read != 1 ) { - dbg.Verbose(1, "gdcmHeader::ReadInt32", " File read error"); - errno = 1; - return 0; - } - g = SwapLong(g); - return g; + guint32 g; + size_t item_read; + item_read = fread (&g, (size_t)4,(size_t)1, fp); + errno = 0; + if ( item_read != 1 ) { + dbg.Verbose(1, "gdcmHeader::ReadInt32", " File read error"); + errno = 1; + return 0; + } + g = SwapLong(g); + return g; } /** @@ -838,17 +827,17 @@ guint32 gdcmHeader::ReadInt32(void) { * @param Elem element of the underlying DictEntry */ ElValue* gdcmHeader::NewElValueByKey(guint16 Group, guint16 Elem) { - // Find out if the tag we encountered is in the dictionaries: - gdcmDictEntry * NewTag = GetDictEntryByKey(Group, Elem); - if (!NewTag) - NewTag = new gdcmDictEntry(Group, Elem); - - ElValue* NewElVal = new ElValue(NewTag); - if (!NewElVal) { - dbg.Verbose(1, "gdcmHeader::NewElValueByKey", - "failed to allocate ElValue"); - return (ElValue*)0; - } + // Find out if the tag we encountered is in the dictionaries: + gdcmDictEntry * NewTag = GetDictEntryByKey(Group, Elem); + if (!NewTag) + NewTag = new gdcmDictEntry(Group, Elem); + + ElValue* NewElVal = new ElValue(NewTag); + if (!NewElVal) { + dbg.Verbose(1, "gdcmHeader::NewElValueByKey", + "failed to allocate ElValue"); + return (ElValue*)0; + } return NewElVal; } @@ -874,75 +863,80 @@ ElValue* gdcmHeader::NewElValueByName(string Name) { return NewElVal; } - /** * \ingroup gdcmHeader - * \brief Read the next tag without loading it's value + * \brief Read the next tag but WITHOUT loading it's value * @return On succes the newly created ElValue, NULL on failure. */ - ElValue * gdcmHeader::ReadNextElement(void) { - guint16 g; - guint16 n; - ElValue * 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 (ElValue *)0; - - NewElVal = NewElValueByKey(g, n); - FindVR(NewElVal); - FindLength(NewElVal); - if (errno == 1) - // Call it quits - return (ElValue *)0; - NewElVal->SetOffset(ftell(fp)); - return NewElVal; + guint16 g; + guint16 n; + ElValue * 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 (ElValue *)0; + + NewElVal = NewElValueByKey(g, n); + FindVR(NewElVal); + FindLength(NewElVal); + if (errno == 1) + // Call it quits + return (ElValue *)0; + NewElVal->SetOffset(ftell(fp)); + return NewElVal; } +/** + * \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(ElValue * ElVal) { - guint16 group = ElVal->GetGroup(); - guint16 element = ElVal->GetElement(); - 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 ( 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) ) - // The "Image Dimensions" tag is retained from ACR/NEMA and contains - // the number of dimensions of the contained object (1 for Signal, - // 2 for Image, 3 for Volume, 4 for Sequence). - return true; - - if ( (group == 0x0028) && (element == 0x0200) ) - // This tag is retained from ACR/NEMA - return true; - - return false; + guint16 group = ElVal->GetGroup(); + guint16 element = ElVal->GetElement(); + 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 ( 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) ) + // The "Image Dimensions" tag is retained from ACR/NEMA and contains + // the number of dimensions of the contained object (1 for Signal, + // 2 for Image, 3 for Volume, 4 for Sequence). + return true; + + if ( (group == 0x0028) && (element == 0x0200) ) + // This tag is retained from ACR/NEMA + return true; + + return false; } /** @@ -950,32 +944,32 @@ bool gdcmHeader::IsAnInteger(ElValue * ElVal) { * \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 == "gdcm::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.GetElementByNumber(grPixel, numPixel); - if (PixelElement) - return PixelElement->GetOffset(); - else - return 0; + // 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 == "gdcm::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.GetElementByNumber(grPixel, numPixel); + if (PixelElement) + return PixelElement->GetOffset(); + else + return 0; } /** @@ -989,22 +983,22 @@ size_t gdcmHeader::GetPixelOffset(void) { * @return Corresponding DictEntry when it exists, NULL otherwise. */ gdcmDictEntry * gdcmHeader::GetDictEntryByKey(guint16 group, guint16 element) { - gdcmDictEntry * found = (gdcmDictEntry*)0; - if (!RefPubDict && !RefShaDict) { - dbg.Verbose(0, "FIXME in gdcmHeader::GetDictEntry", + gdcmDictEntry * found = (gdcmDictEntry*)0; + if (!RefPubDict && !RefShaDict) { + dbg.Verbose(0, "FIXME in gdcmHeader::GetDictEntry", "we SHOULD have a default dictionary"); - } - if (RefPubDict) { - found = RefPubDict->GetTagByKey(group, element); - if (found) - return found; - } - if (RefShaDict) { - found = RefShaDict->GetTagByKey(group, element); - if (found) - return found; - } - return found; + } + if (RefPubDict) { + found = RefPubDict->GetTagByKey(group, element); + if (found) + return found; + } + if (RefShaDict) { + found = RefShaDict->GetTagByKey(group, element); + if (found) + return found; + } + return found; } /** @@ -1012,98 +1006,234 @@ gdcmDictEntry * gdcmHeader::GetDictEntryByKey(guint16 group, guint16 element) { * \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. - * @earam Name name of the searched DictEntry + * @param Name name of the searched DictEntry * @return Corresponding DictEntry when it exists, NULL otherwise. */ gdcmDictEntry * gdcmHeader::GetDictEntryByName(string Name) { - gdcmDictEntry * found = (gdcmDictEntry*)0; - if (!RefPubDict && !RefShaDict) { - dbg.Verbose(0, "FIXME in gdcmHeader::GetDictEntry", + gdcmDictEntry * found = (gdcmDictEntry*)0; + if (!RefPubDict && !RefShaDict) { + dbg.Verbose(0, "FIXME in 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; + } + 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" otherwise. + */ string gdcmHeader::GetPubElValByNumber(guint16 group, guint16 element) { - return PubElVals.GetElValueByNumber(group, element); + return PubElVals.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" otherwise. + */ string gdcmHeader::GetPubElValRepByNumber(guint16 group, guint16 element) { - ElValue* elem = PubElVals.GetElementByNumber(group, element); - if ( !elem ) - return "gdcm::Unfound"; - return elem->GetVR(); + ElValue* elem = PubElVals.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" otherwise. + */ string gdcmHeader::GetPubElValByName(string TagName) { - return PubElVals.GetElValueByName(TagName); + return PubElVals.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" otherwise. + */ string gdcmHeader::GetPubElValRepByName(string TagName) { - ElValue* elem = PubElVals.GetElementByName(TagName); - if ( !elem ) - return "gdcm::Unfound"; - return elem->GetVR(); + ElValue* elem = PubElVals.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" otherwise. + */ string gdcmHeader::GetShaElValByNumber(guint16 group, guint16 element) { - return ShaElVals.GetElValueByNumber(group, element); + return ShaElVals.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" otherwise. + */ string gdcmHeader::GetShaElValRepByNumber(guint16 group, guint16 element) { - ElValue* elem = ShaElVals.GetElementByNumber(group, element); - if ( !elem ) - return "gdcm::Unfound"; - return elem->GetVR(); + ElValue* elem = ShaElVals.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" otherwise. + */ string gdcmHeader::GetShaElValByName(string TagName) { - return ShaElVals.GetElValueByName(TagName); + return ShaElVals.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" otherwise. + */ string gdcmHeader::GetShaElValRepByName(string TagName) { - ElValue* elem = ShaElVals.GetElementByName(TagName); - if ( !elem ) - return "gdcm::Unfound"; - return elem->GetVR(); + ElValue* elem = ShaElVals.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" otherwise. + */ string gdcmHeader::GetElValByNumber(guint16 group, guint16 element) { - string pub = GetPubElValByNumber(group, element); - if (pub.length()) - return pub; - return GetShaElValByNumber(group, element); + 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" otherwise. + */ string gdcmHeader::GetElValRepByNumber(guint16 group, guint16 element) { - string pub = GetPubElValRepByNumber(group, element); - if (pub.length()) - return pub; - return GetShaElValRepByNumber(group, element); + 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" otherwise. + */ string gdcmHeader::GetElValByName(string TagName) { - string pub = GetPubElValByName(TagName); - if (pub.length()) - return pub; - return GetShaElValByName(TagName); + 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" otherwise. + */ string gdcmHeader::GetElValRepByName(string TagName) { - string pub = GetPubElValRepByName(TagName); - if (pub.length()) - return pub; - return GetShaElValRepByName(TagName); + string pub = GetPubElValRepByName(TagName); + if (pub.length()) + return pub; + return GetShaElValRepByName(TagName); } /** @@ -1118,9 +1248,7 @@ string gdcmHeader::GetElValRepByName(string TagName) { int gdcmHeader::SetPubElValByNumber(string content, guint16 group, guint16 element) { - //CLEANME TagKey key = gdcmDictEntry::TranslateToKey(group, element); - //CLEANME PubElVals.tagHt[key]->SetValue(content); - return ( PubElVals.SetElValueByNumber (content, group, element) ); + return ( PubElVals.SetElValueByNumber (content, group, element) ); } /** @@ -1131,9 +1259,7 @@ int gdcmHeader::SetPubElValByNumber(string content, guint16 group, * @param TagName name of the tag to be modified */ int gdcmHeader::SetPubElValByName(string content, string TagName) { - //CLEANME TagKey key = gdcmDictEntry::TranslateToKey(group, element); - //CLEANME PubElVals.tagHt[key]->SetValue(content); - return ( PubElVals.SetElValueByName (content, TagName) ); + return ( PubElVals.SetElValueByName (content, TagName) ); } /** @@ -1148,7 +1274,7 @@ int gdcmHeader::SetPubElValByName(string content, string TagName) { int gdcmHeader::SetShaElValByNumber(string content, guint16 group, guint16 element) { - return ( ShaElVals.SetElValueByNumber (content, group, element) ); + return ( ShaElVals.SetElValueByNumber (content, group, element) ); } /** @@ -1159,7 +1285,7 @@ int gdcmHeader::SetShaElValByNumber(string content, * @param TagName name of the tag to be modified */ int gdcmHeader::SetShaElValByName(string content, string TagName) { - return ( ShaElVals.SetElValueByName (content, TagName) ); + return ( ShaElVals.SetElValueByName (content, TagName) ); } /** @@ -1167,14 +1293,14 @@ int gdcmHeader::SetShaElValByName(string content, string TagName) { * \brief Parses the header of the file but WITHOUT loading element values. */ void gdcmHeader::ParseHeader(bool exception_on_error) throw(gdcmFormatError) { - ElValue * newElValue = (ElValue *)0; - - rewind(fp); - CheckSwap(); - while ( (newElValue = ReadNextElement()) ) { - SkipElementValue(newElValue); - PubElVals.Add(newElValue); - } + ElValue * newElValue = (ElValue *)0; + + rewind(fp); + CheckSwap(); + while ( (newElValue = ReadNextElement()) ) { + SkipElementValue(newElValue); + PubElVals.Add(newElValue); + } } /** @@ -1190,19 +1316,19 @@ void gdcmHeader::ParseHeader(bool exception_on_error) throw(gdcmFormatError) { * the ones of the official DICOM fields Rows, Columns and Planes. */ void gdcmHeader::AddAndDefaultElements(void) { - ElValue* NewEntry = (ElValue*)0; + ElValue* NewEntry = (ElValue*)0; - NewEntry = NewElValueByName("gdcmXSize"); - NewEntry->SetValue(GetElValByName("Rows")); - PubElVals.Add(NewEntry); + NewEntry = NewElValueByName("gdcmXSize"); + NewEntry->SetValue(GetElValByName("Rows")); + PubElVals.Add(NewEntry); - NewEntry = NewElValueByName("gdcmYSize"); - NewEntry->SetValue(GetElValByName("Columns")); - PubElVals.Add(NewEntry); + NewEntry = NewElValueByName("gdcmYSize"); + NewEntry->SetValue(GetElValByName("Columns")); + PubElVals.Add(NewEntry); - NewEntry = NewElValueByName("gdcmZSize"); - NewEntry->SetValue(GetElValByName("Planes")); - PubElVals.Add(NewEntry); + NewEntry = NewElValueByName("gdcmZSize"); + NewEntry->SetValue(GetElValByName("Planes")); + PubElVals.Add(NewEntry); } /** @@ -1211,17 +1337,17 @@ void gdcmHeader::AddAndDefaultElements(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 = PubElVals.GetTagHt(); + for (TagElValueHT::iterator tag = ht.begin(); tag != ht.end(); ++tag) { + LoadElementValue(tag->second); + } } void gdcmHeader::PrintPubElVal(ostream & os) { - PubElVals.Print(os); + PubElVals.Print(os); } void gdcmHeader::PrintPubDict(ostream & os) { - RefPubDict->Print(os); + RefPubDict->Print(os); } diff --git a/src/gdcmHeader.h b/src/gdcmHeader.h index 0f1c8f32..f46a12c6 100644 --- a/src/gdcmHeader.h +++ b/src/gdcmHeader.h @@ -10,160 +10,156 @@ #include "gdcmElValue.h" #include "gdcmElValSet.h" -//////////////////////////////////////////////////////////////////////////// -// The purpous of an instance of gdcmHeader is to act as a container of -// all the elements and their corresponding values (and additionaly the -// corresponding DICOM dictionary entry) of the header of a DICOM file. -// -// The typical usage of instances of class gdcmHeader is to classify a set of -// dicom files according to header information e.g. to create a file hierarchy -// reflecting the Patient/Study/Serie informations, or extracting a given -// SerieId. Accesing the content (image[s] or volume[s]) is beyond the -// functionality of this class and belong to gdmcFile (see below). -// Notes: -// * the various entries of the explicit value representation (VR) shall -// be managed within a dictionary which is shared by all gdcmHeader instances -// * the gdcmHeader::Set*Tag* family members cannot be defined as protected -// (Swig limitations for as Has_a dependency between gdcmFile and gdcmHeader) - typedef string VRKey; typedef string VRAtr; typedef map VRHT; // Value Representation Hash Table +/// \brief The purpous of an instance of gdcmHeader is to act as a container of +/// all the DICOM elements and their corresponding values (and +/// additionaly the corresponding DICOM dictionary entry) of the header +/// of a DICOM file. +/// +/// The typical usage of instances of class gdcmHeader is to classify a set of +/// dicom files according to header information e.g. to create a file hierarchy +/// reflecting the Patient/Study/Serie informations, or extracting a given +/// SerieId. Accesing the content (image[s] or volume[s]) is beyond the +/// functionality of this class and belongs to gdmcFile. +/// \note The various entries of the explicit value representation (VR) shall +/// be managed within a dictionary which is shared by all gdcmHeader +/// instances. +/// \note The gdcmHeader::Set*Tag* family members cannot be defined as +/// protected due to Swig limitations for as Has_a dependency between +/// gdcmFile and gdcmHeader. + class GDCM_EXPORT gdcmHeader { - void SkipBytes(guint32); + void SkipBytes(guint32); private: - static VRHT *dicom_vr; - // Dictionaries of data elements: - - gdcmDictSet* Dicts; // global dictionary container - gdcmDict* RefPubDict; // public dictionary - gdcmDict* RefShaDict; // shadow dictionary (optional) - // Parsed element values: - ElValSet PubElVals; // parsed with Public Dictionary - ElValSet ShaElVals; // parsed with Shadow Dictionary - string filename; // refering underlying file - FILE * fp; - guint16 grPixel; - guint16 numPixel; - // Ne faudrait-il pas une indication sur la presence ou non - // du 'groupe des pixels' dans l'entete? - // (voir pb du DICOMDIR) - - // Swap code (little, big, bad-big, bad-little endian): this code is not fixed - // during parsing.FIXME sw should be an enum e.g. - //enum EndianType { - //LittleEndian, - //BadLittleEndian, - //BigEndian, - //BadBigEndian}; - int sw; - - // Only the elements whose size is below this bound will be loaded. - // By default, this upper bound is limited to 1024 (which looks reasonable - // when one considers the definition of the various VR contents). - guint32 MaxSizeLoadElementValue; - - guint16 ReadInt16(void); - guint32 ReadInt32(void); - guint16 SwapShort(guint16); - guint32 SwapLong(guint32); - guint32 FindLengthOB(void); - void Initialise(void); - void CheckSwap(void); - void InitVRDict(void); - void SwitchSwapToBigEndian(void); - void AddAndDefaultElements(void); - void SetMaxSizeLoadElementValue(long); - - gdcmDictEntry * GetDictEntryByKey(guint16, guint16); - gdcmDictEntry * GetDictEntryByName(string name); - - // ElValue related utilities - ElValue * ReadNextElement(void); - ElValue * NewElValueByKey(guint16 group, guint16 element); - ElValue * NewElValueByName(string name); - void FindLength(ElValue *); - void FindVR(ElValue *); - void LoadElementValue(ElValue *); - void LoadElementValueSafe(ElValue *); - void SkipElementValue(ElValue *); - void FixFoundLength(ElValue*, guint32); - bool IsAnInteger(ElValue *); - - bool IsImplicitVRLittleEndianTransferSyntax(void); - bool IsExplicitVRLittleEndianTransferSyntax(void); - bool IsDeflatedExplicitVRLittleEndianTransferSyntax(void); - bool IsExplicitVRBigEndianTransferSyntax(void); - bool IsJPEGBaseLineProcess1TransferSyntax(void); - bool IsJPEGExtendedProcess2_4TransferSyntax(void); - bool IsJPEGExtendedProcess3_5TransferSyntax(void); - bool IsJPEGSpectralSelectionProcess6_8TransferSyntax(void); - + /// Pointer to the Value Representation Hash Table which contains all + /// the VR of the DICOM version3 public dictionary. + static VRHT *dicom_vr; + + /// Global dictionary container + gdcmDictSet* Dicts; + /// Public dictionary used to parse this header + gdcmDict* RefPubDict; + /// Optional "shadow dictionary" (private elements) used to parse this header + gdcmDict* RefShaDict; + + /// ELement VALueS parsed with the PUBlic dictionary. + ElValSet PubElVals; + /// ELement VALueS parsed with the SHAdow dictionary. + ElValSet ShaElVals; + /// Refering underlying filename. + string filename; + FILE * fp; + + // FIXME sw should be an enum e.g. + //enum EndianType { + //LittleEndian, + //BadLittleEndian, + //BigEndian, + //BadBigEndian}; + /// Swap code e.g. little, big, bad-big, bad-little endian). Warning: + /// this code is not fixed during header parsing. + int sw; + + /// Size treshold above which an element value will NOT be loaded in + /// memory (to avoid loading the image/volume itself). By default, + /// this upper bound is fixed to 1024 bytes (which might look reasonable + /// when one considers the definition of the various VR contents). + guint32 MaxSizeLoadElementValue; + + guint16 ReadInt16(void); + guint32 ReadInt32(void); + guint16 SwapShort(guint16); + guint32 SwapLong(guint32); + guint32 FindLengthOB(void); + void Initialise(void); + void CheckSwap(void); + void InitVRDict(void); + void SwitchSwapToBigEndian(void); + void AddAndDefaultElements(void); + void SetMaxSizeLoadElementValue(long); + + gdcmDictEntry * GetDictEntryByKey(guint16, guint16); + gdcmDictEntry * GetDictEntryByName(string name); + + // ElValue related utilities + ElValue * ReadNextElement(void); + ElValue * NewElValueByKey(guint16 group, guint16 element); + ElValue * NewElValueByName(string name); + void FindLength(ElValue *); + void FindVR(ElValue *); + void LoadElementValue(ElValue *); + void LoadElementValueSafe(ElValue *); + void SkipElementValue(ElValue *); + void FixFoundLength(ElValue*, guint32); + bool IsAnInteger(ElValue *); + + bool IsImplicitVRLittleEndianTransferSyntax(void); + bool IsExplicitVRLittleEndianTransferSyntax(void); + bool IsDeflatedExplicitVRLittleEndianTransferSyntax(void); + bool IsExplicitVRBigEndianTransferSyntax(void); + bool IsJPEGBaseLineProcess1TransferSyntax(void); + bool IsJPEGExtendedProcess2_4TransferSyntax(void); + bool IsJPEGExtendedProcess3_5TransferSyntax(void); + bool IsJPEGSpectralSelectionProcess6_8TransferSyntax(void); + protected: - enum FileType { - Unknown = 0, - TrueDicom, - ExplicitVR, - ImplicitVR, - ACR, - ACR_LIBIDO}; // CLEANME - FileType filetype; - int write(ostream&); - int anonymize(ostream&); // FIXME : anonymize should be a friend ? + // FIXME: is this enum still necessary ?? + enum FileType { + Unknown = 0, + TrueDicom, + ExplicitVR, + ImplicitVR, + ACR, + ACR_LIBIDO}; // CLEANME + FileType filetype; + int write(ostream&); + int anonymize(ostream&); // FIXME : anonymize should be a friend ? public: - void LoadElements(void); - virtual void ParseHeader(bool exception_on_error = false) - throw(gdcmFormatError); - gdcmHeader(const char *filename, bool exception_on_error = false) - throw(gdcmFileError); - virtual ~gdcmHeader(); - - size_t GetPixelOffset(void); - void GetPixels(size_t, void *); - int GetSwapCode(void) { return sw; } - - // TODO Swig int SetPubDict(string filename); - // When some proprietary shadow groups are disclosed, we can set up - // an additional specific dictionary to access extra information. - // TODO Swig int SetShaDict(string filename); - - // Get the element values themselves: - string GetPubElValByName(string TagName); - string GetPubElValByNumber(guint16 group, guint16 element); - - // Getting the element value representation (VR) might be needed by caller - // to convert the string typed content to caller's native type - // (think of C/C++ vs Python). - - string GetPubElValRepByName(string TagName); - string GetPubElValRepByNumber(guint16 group, guint16 element); - - TagElValueHT & GetPubElVal(void) { return PubElVals.GetTagHt(); }; - void PrintPubElVal(ostream & os = cout); - void PrintPubDict(ostream & os = cout); - - // Same thing with the shadow : - // TODO Swig string* GetShaTagNames(); - string GetShaElValByName(string TagName); - string GetShaElValByNumber(guint16 group, guint16 element); - string GetShaElValRepByName(string TagName); - string GetShaElValRepByNumber(guint16 group, guint16 element); - - // Wrappers of the above (public is privileged over shadow) to avoid - // bugging the caller with knowing if ElVal is from the public or shadow - // dictionary. - string GetElValByName(string TagName); - string GetElValByNumber(guint16 group, guint16 element); - string GetElValRepByName(string TagName); - string GetElValRepByNumber(guint16 group, guint16 element); - - int SetPubElValByName(string content, string TagName); - int SetPubElValByNumber(string content, guint16 group, guint16 element); - int SetShaElValByName(string content, string ShadowTagName); - int SetShaElValByNumber(string content, guint16 group, guint16 element); - - ElValSet GetPubElVals() { return(PubElVals); } + void LoadElements(void); + virtual void ParseHeader(bool exception_on_error = false) + throw(gdcmFormatError); + gdcmHeader(const char *filename, bool exception_on_error = false) + throw(gdcmFileError); + virtual ~gdcmHeader(); + + size_t GetPixelOffset(void); + void GetPixels(size_t, void *); + int GetSwapCode(void) { return sw; } + + // TODO Swig int SetPubDict(string filename); + // When some proprietary shadow groups are disclosed, we can set up + // an additional specific dictionary to access extra information. + // TODO Swig int SetShaDict(string filename); + + string GetPubElValByName(string TagName); + string GetPubElValByNumber(guint16 group, guint16 element); + string GetPubElValRepByName(string TagName); + string GetPubElValRepByNumber(guint16 group, guint16 element); + + TagElValueHT & GetPubElVal(void) { return PubElVals.GetTagHt(); }; + void PrintPubElVal(ostream & os = cout); + void PrintPubDict(ostream & os = cout); + + // TODO Swig string* GetShaTagNames(); + string GetShaElValByName(string TagName); + string GetShaElValByNumber(guint16 group, guint16 element); + string GetShaElValRepByName(string TagName); + string GetShaElValRepByNumber(guint16 group, guint16 element); + + string GetElValByName(string TagName); + string GetElValByNumber(guint16 group, guint16 element); + string GetElValRepByName(string TagName); + string GetElValRepByNumber(guint16 group, guint16 element); + + int SetPubElValByName(string content, string TagName); + int SetPubElValByNumber(string content, guint16 group, guint16 element); + int SetShaElValByName(string content, string ShadowTagName); + int SetShaElValByNumber(string content, guint16 group, guint16 element); + + ElValSet GetPubElVals() { return(PubElVals); } }; #endif