X-Git-Url: https://git.creatis.insa-lyon.fr/pubgit/?a=blobdiff_plain;f=TODO;h=fdc27d9d7a8670545f94edf1482bfc3f96de2356;hb=cf1c0279551125013bbb1bb5407dac134fc5ceae;hp=f7a7dd6021bc1204ed67e6f7d6380515da5d7c3e;hpb=70e24c6c61481f9836e26b6b44a9670d92a4f43b;p=gdcm.git diff --git a/TODO b/TODO index f7a7dd60..fdc27d9d 100644 --- a/TODO +++ b/TODO @@ -1,17 +1,367 @@ -GetPubElValByNumber doit faire la difference entre chaine vide -et chaine pas touve''. Eventuellement raiser une exception ? +----------------------------------------------------------------------------- +Proposed Template of an entry in this TODO: +(Note: Date is the date of registering of first demand.) -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. +------------- +Description: +Date: +Attributed: +Details: +Comments: +------------- -grep str2num *.c: c'est une macro sans doute proprifiable +---------------------------------------------------------------------------- +Description: Rewrite gdcmDebug so that it works +Date: 2004 Dec 07 +Attributed: Mathieu +Details: +Comments: +---------------------------------------------------------------------------- +Description: Need to fix cygwin + wrong path (passing UNIX path when windows patch expected) +Date: 2004 Nov 16 +Attributed: Mathieu +Details: +Comments: +---------------------------------------------------------------------------- +Description: Need a test for gdcmHeaderHelper (iterating over gdcmData) for instance +Date: 2004 Nov 16 +Attributed: Mathieu +Details: +Comments: It used to seg fault when a README file was found +---------------------------------------------------------------------------- +Description: gdcmJpeg8 is strictly a copy/paste of gdcmJpeg12.cxx. +Date: 2004 Oct 13 +Attributed: +Details: +We should write the code in a common place, then include this 'cxx' file so +the define from gdcm_mangle redefine to the proper one. +Comments: +This will be usefull since I may need in the future a 16bits version of this +reading +----------------------------------------------------------------------------- +Description: Change jpeg 'exit' call to standard c++ exception using the jpeg + error +handler +Date: 2004 Oct 8 +Attributed: +Details: +Comments: +----------------------------------------------------------------------------- +Description: Add kwsys as a subdir somewhere in gdcm +Date: 2004 Oct 8 +Attributed: Mathieu +Details: +Comments: +----------------------------------------------------------------------------- +Description: ljpeg +Date: 2004 Oct 8 +Attributed: Mathieu +Details: +Comments: ljpeg was rip from medcon and not the official one. medcon tried to +optimised function using MACRO (doh!), so it make its very unreadable and very +hard to fix warnings. Should go back to official source, copy proper copyright +and fix warnings on dashboard +13/10: update apparently no dicom toolkit use this lib as it is too buggy. +We should use the ls-patch for ijg instead. Thus we can safely get rid of that +lib. +14/10: PHILIPS_Gyroscan-12-MONO2-Jpeg_Lossless.dcm prove that I was right the +old Cornwell lib is buggy and does not read anything. +----------------------------------------------------------------------------- +Description: Extent reading support +Date: 2004 Oct 8 +Attributed: Mathieu +Details: +Comments: All ITK/VTK readers support selecting extent. gdcm should support +selecting extent before being inserted into ITK +----------------------------------------------------------------------------- +Description: Generate new UID each time we write DICOM +Date: 2004 Oct 8 +Attributed: Mathieu +Details: +Comments: According to DICOM ref a new UID should be created each we write a +DICOM images. I guess it should be an option so that we can still use md5sum to +check dicom file. The proposed way was: +http://www.creatis.insa-lyon.fr/pipermail/dcmlib/2004-September/000611.html + +Bah, comme Win32 pose encore pb: + echo "gdcm" | od -b + 0000000 147 144 143 155 012 +et si on prenait: + radical + 147.144.143.155 + IP + time() + +----------------------------------------------------------------------------- +Description: gdcmDicomDir and SQItem creation +Date: 2004 Nov 16 +Attributed: +Details: DicomDir creates some SQItem (by new). After that, it creates + the corresponding DicomDirPatient, etc. using the content of the SQItem + (the content is composed with some DocEntry's that can't be destroyed). + So, if the SQItem is deleted, then it's content is deleted to. But the + DicomDirPatient, etc. use the content of the SQItem. Then, the SQItem can't + be deleted, and when have memory leaks + +----------------------------------------------------------------------------- +Description: [BUG] Better handling of unfound Dicom dictionary. + When gdcm doesn't find the Dicom dictionary (because it's + path to the directory of dictionary is uncorrect, either + because the install relative layout was broken after file moves + or because the environnement variable GDCM_DICT_PATH is + unpropely set), gdcm will: + 1/ print a warning + 2/ throw an exception (that is internaly UNcatched by gdcm) + that in most cases provoques the caller application to + exit uncleanly (uncatched excpetions in fine call abort() ). + Additionaly on Win32 the warning print isn't displayed because + exiting occurs prior to cerr or cout is flushed properly. +Date: 2004 Oct 15 +Attributed: +Details: fixes (from dirty to clean) + 1/ force Win32 to flush it's buffer so at least the user gets some + reason why it's application exited (when called in command + environement). Note: it looks like the "cerr << flush" fails. Sigh. + 2/ within gdcm catch the exception, display a decent warning, and + return to caller. + 3/ see the comment below on how to enhance the API and fix things + really cleanly. +Comments: ENH proposal: + The caller migth not be aware of the path to the dictionaries + on invocation of gdcm (think this path is set by the Interface + because the caller wants to skip the default gdcm dictionary in order + to provide his own ones e.g. another language based one). + Hence, gdcm should postpone the parsing of the default dictionary + instead of doing it on library entry. + This would enable two things: + - It would give a chance to the caller to set the path to + the dictionaries he whishes to use, through a call to + newly created DictSet::SetDictionaryPath( string ). + - It would avoid the burden of using the GDCM_DICT_PATH + environnement variable and enable GDCM CONTROL FROM WITHIN + THE API. Optionaly, if the caller didn't use the API to + provide his prefered path, gdcm could still default to + GDCM_DICT_PATH... +----------------------------------------------------------------------------- +Description: vtk/vtkGdcmReader doesn't positionate the normal to the image +Date: 2004 Oct 1 +Attributed: +Details: vtkGdcmReader::CheckFileCoherence() sets the DataOrigin[i] + but doesn't set the plane (image seen in 3D) normal (is it + possible any how). This plane normal could be extracted from + the "orientation" info of the gdcmHeader ( refer to + grep "Orientation" Dicts/dicomV3.dic). + Problem exhibiting this defect: cine loop on a pile of images + whose Origin is correct, but whose normal is not set will + plainly suck ! +Comments: + * vtkGdcmReader inherits from vtkImageReader which aggregates + a vtkTranform. vtkGdcmReader could store (when the user requires + it, see below) the origin/normal taken from the Dicom Header + within this vtkTransform (looks like a natural place to store + this spacial information). + * Both settings of the origin and/OR the normal of the plane (image) + should be an option defined with a flag (On/Off) in the + vtkGdcmReader... +----------------------------------------------------------------------------- +Description: Fix the Python wrappers +Date: 2004 Sep 24 +Attributed: no +Details: +Comments: +----------------------------------------------------------------------------- +Description: clean up gdcmFile/gdcmHeader relationship +Date: 2004 Sep 24 +Attributed: +Details: + * simplify the API for the user (no need to call GetImageData() before + calling Write(). + * avoid memory leaks with with Pixel_Data. +Comments: +----------------------------------------------------------------------------- +Description: remove all autotools references +Date: 2004 Sep 24 +Attributed: no +Details: +Comments: +----------------------------------------------------------------------------- +Description: introduce namespace "gdcm" +Date: 2004 Jul 30 +Attributed: +Details: +Comments: + 1/ Problem: using enum with name like 'Unknow' on .Net, or LP on cygwin + causes problems. + 2/ Question: when introducing the namespace, should we remove the gdcm + prefix from classes or keep it ? +----------------------------------------------------------------------------- +Description: complete the doxygen Documentation +Date: 2004 Sep 24 +Attributed: +Details: +Comments: +----------------------------------------------------------------------------- +Description: test the private dictionary part. +Date: 2004 Sep 24 +Attributed: +Details: +Comments: + * Frog: where can we obtain such a private/dictionary and the corresponding + Dicom file ? Any examples on-line ? +----------------------------------------------------------------------------- +Description: fix definitively the memory leaks problems. +Date: 2004 Sep 24 +Attributed: +Details: +Comments: There is a nightly dashboard that run valgrind every night +(zorglub | GDCM-Linux-g++) +----------------------------------------------------------------------------- +Description: test gdcm on a big endian OS. +Date: 2004 Sep 24 +Attributed: +Details: +Comments: There is a nightly dashboard that run on MacOSX each nite +(midworld.kitwarein | GDCM-DarwinG5-g++ ) +----------------------------------------------------------------------------- +Description: More tests ! +Date: 2004 Sep 24 +Attributed: +Details: + * an example of new test could be to clone a Dicom image by + copying gdcmDocEntry one after the other +Comments: + * look at traversal used in Example/FlatHashTablePrint.cxx +----------------------------------------------------------------------------- +Description: revoir la gestion des resources, win32 permet d'inclure des + fichiers texte (=dicomV3.dic) dans une dll ou quelquechose du genre. +Date: 2004 Sep 24 +Attributed: Mathieu +Details: +Comments: + * Frog: no comprendo !? De plus est-ce portable ? + * To improve load time it could be usefull to have the dictionary directly in + 'c++' code or in a more binary format. + * This will also solve some issues where /dummy/ user did nor set + GDCM_DICT_PATH properly neither 'make install' +----------------------------------------------------------------------------- +Description: Convert the C-like IO to C++ IO: +Date: 2004 Jul 30 +Attributed: +Details: + remove all the C-oriented IO references like FILE*, fread... + and replace them with the C++ fstream notation. + Provide overload of operators << and >> for any gdcm class using file IO. +Comments: + * Question: the underlying jpeg libraries (written in C) use the FILE* + notation. Is there a way to still use fstream in gdcm, and + build or pass the proper FILE* to jpeg libs ? + * Binary IO references are available at + http://www.angelfire.com/country/aldev0/cpphowto/cpp_BinaryFileIO.html + * The internal API should be rewritten so that gdcm speaking to jpeg lib is + done with stream/string and not directly opened FILE* + * No stdio.h anymore anywhere tolerated ! +----------------------------------------------------------------------------- +Description: generate methods based on VM. +Date: 2004 Jul 30 +Attributed: +Details: + * cmake should parse le DICOM dictionary to generate methods like + gdcm???::SetImagePosition(int, int) + { + //generated content do not edit + ... + } + gdcm???::SetImageNumber(int) + { + //generated content do not edit + ... + } +Comments: + * Regrain: a dicom dictionary entry name is NOT UNIQUE [this means + two tags=(group, element) can share the same name]. + What should the wrapper do in such a case !? + * Frog: what does VM stand for ? + * VM = Value Multiplicity +----------------------------------------------------------------------------- +Description: Add information on supported imagers (constructor/model) +Date: 2004 9 7 +Attributed: +Details: in order to promote gdcm make a list (on the web pages) + of images successfully parsed based on a constructor/model ordering +Comments: * frog: gdcmData only lists pathological images. How to collect + the ones gdcm works smoothly with (hopefully gdcmData is a small + subset of what we would like). +----------------------------------------------------------------------------- +Description: Add a GetVersion() global function. +Date: 2003 july 7 +Attributed: +Details: This is to be used for version assertion with gdcmPython +Comments: +----------------------------------------------------------------------------- +----------------------------------------------------------------------------- +* vtk/vtkGdcmHeader.cxx: if speed becomes a concern some changes can + be made at the cost of memory consumption (refer to header of + vtk/vtkGdcmHeader.cxx) +----------------------------------------------------------------------------- +* gdcmElValSet::SetElValueLengthByNumber IMNSHO should be trashed. + It's only purpose is a onliner substitute to calling GetElValueByNumber + and then SetLength. This only obfuscates the caller code more than + clarifying it. + Besides the definition of gdcmElValSet::SetElValueLengthByNumber itself + it quite poor since it is a almost exact copy of + gdcmElValSet::GetElValueByNumber except for the returned code. + gdcmHeader::SetPubElValLengthByNumber (which is based on + gdcmElValSet::SetElValueLengthByNumber) is used nowhere... +----------------------------------------------------------------------------- +* Fix the bug in Test/bug1.cxx (see first comment line): Win32 only. +----------------------------------------------------------------------------- +* All (or at least many of) the methods of gdcmHeader whose only arguments + are an ElValue* (e.g. FindLength, FindVR, LoadElementValue...) can + be moved away to ElValue class on condition of transmitting the + gdcmHeader.fp attribute. This change should be considered since it + would allow those method to avoid artificial calls to ElValue::GetElement(), + ElValue::GetVR()... +----------------------------------------------------------------------------- +* Eat leading_trailing_whitespace (found in python/gdcmPython/gdcmi) should + be used when parsing the dictionary in C++ ! +----------------------------------------------------------------------------- +* Group length is not a unique tag in a file. Hence avoid putting it + 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::LoadElements only loads the element whose length is + below the specified size. When accessing the value of such an element + the content is unfound ! Find a decent way of loading the value on + explicit demand. +----------------------------------------------------------------------------- +* JPR: fournir une method qui ne fait que lire les elements passes en arguments + sous forme d'une liste. +----------------------------------------------------------------------------- +* JPR: gdcmHeader::CheckSwap() dans le cas ACR pas propre, degager tout de + suite si on a deduit que c'en est pas... +----------------------------------------------------------------------------- +* python /usr/lib/python2.2/site-packages/DaVaW/demo/dvwDcmReader.py + and load image /home/frog/cvs/DCMlib/Data/CT-MONO2-16-ankle.dcm + will yield wrong coloring scheme as opposed to + affim filein=/home/frog/cvs/DCMlib/Data/CT-MONO2-16-ankle.dcm +----------------------------------------------------------------------------- +* gdcmFile should implement the following API: + gdcmFile WriteDicom; + WriteDicom.SetFileName("MyDicomFile.dcm"); + string * AllTags = gdcmHeader.GetDcmTagNames(); + WriteDicom.SetDcmTag(AllTags[5], "253"); + WriteDicom.SetDcmTag("Patient Name", "bozo"); + WriteDicom.SetDcmTag("Patient Name", "bozo"); + WriteDicom.SetImageData(Image); + WriteDicom.Write(); + + Anonymize(ostream& output) { + a = gdcmFile("toto1"); + a.SetPubValueByName("Patient Name", ""); + a.SetPubValueByName("Date", ""); + a.SetPubValueByName("Study Date", ""); + a.write(output); + } +----------------------------------------------------------------------------- -gdcmHeader::CheckSwap() dans le cas ACR pas propre, degager tout de suite -si on a deduit que c'en est pas...