jpr [Wed, 26 Jan 2005 15:03:32 +0000 (15:03 +0000)]
Contructors and Destructor are now protected,
in order to be sure no user will try to use them.
(at user level, only the children classes are meaningfull)
regrain [Tue, 25 Jan 2005 15:44:21 +0000 (15:44 +0000)]
* src/ : rename some methods on Entry (SetXxx, InsertXxx) to have a better
use with python. Some methods ard moved from Document to ElementSet or
DocEntrySet. The equivalent methods to ElementSet have been added to
SQItem and are virtual in DocEntrySet. Methods that may have the same code
in SQItem and ElementSet is directly placed in DocEntrySet
-- BeNours
jpr [Tue, 25 Jan 2005 15:21:20 +0000 (15:21 +0000)]
In ordrer to avoid missbehaviour :
gdcm::DocEntry is now specialized in
- gdcm::SeqEntry (Dicom Sequence) as opposed to :
- gdcm::ContentEntry (Dicom entry with a 'content')
gdcm::ValContent is now specialized in
- gdcm::ValEntry (Dicom entry with a std::string representable value, i.e.
characters value, or integers)
- gdcm::BinEntry (Dicom entry with a non strig representable value
(user hasn't any longer to thing about the precedence of his dynamic_casts)
regrain [Tue, 25 Jan 2005 11:11:57 +0000 (11:11 +0000)]
* src/gdcmDocument.cxx : bug fix on potential memory leak
* src/gdcmSQItem.cxx : initialize the SQItemNumber value in the constructor
* src/gdcmSQItem.[h|cxx], gdcmElementSet.[h|cxx], gdcmDicomDir*.[h|cxx] :
Add ClearXxx methods to remove all the content of the list or map
* gdcmDocEntrySet.h : add GetFirst/NextEntry in pure virtual
-- BeNours
jpr [Mon, 24 Jan 2005 16:10:49 +0000 (16:10 +0000)]
Stage 3 of normalization :
in order to avoid any missunderstanding while reading method names :
- All the methods xxxEntry now deal with dicom Entries
- Methods GetEntry splitted into
GetEntryValue
GetEntryBinArea
- Method SetEntry renamed as SetEntryValue
- Method SetEntryBinArea let as it was
malaterre [Mon, 24 Jan 2005 14:52:49 +0000 (14:52 +0000)]
ENH: Grealty simplify the JPEg decompression, no need to differenciate based on the JPEG possible type (with/without fragment, with/without multiple frame). Everything is done throught the JPEG IO Suspension mechanism in IJG
regrain [Mon, 24 Jan 2005 14:14:07 +0000 (14:14 +0000)]
* src/gdcmDicomDirElement.h, gdcmDict.h : rename AddNewXxx methods to AddXxx.
* src/gdcmDict.h : rename GetDictEntry moethod to GetEntry, to be coherent
with other method names
* gdcmPython/gdcm.i : bug fix. Add the typemaps for the TagName type
* Test/TestDicomDirElement.cxx : add new test
-- BeNours
jpr [Sun, 23 Jan 2005 10:12:31 +0000 (10:12 +0000)]
Some normalizations :
- Move some methods to parent class
- Position correctly the entry points for JPEG2000 and JPEG-LS
(JPG-LS in NOT jpeg 'lossless')
- Remove deprecated and useless methods
- Update comments
- Update Doxygenation
malaterre [Sun, 23 Jan 2005 01:00:17 +0000 (01:00 +0000)]
BUG: ostream opeartor << is only defined for __int?? but not unsigned __int??. Fallback to previous implementation. According to implementation char/short/int have fixed lenght in Microsoft world. For more info look for /error C2679 __int64/
malaterre [Sat, 22 Jan 2005 22:21:19 +0000 (22:21 +0000)]
BUG: there was a problem with uint64_t. C99 clearly specify that vendor compiler should provide a stdint.h file. If not include inttypes (HPUX, SunOS). And on broekn platerform Win32 provide the proper typedef (it used to be broekn on Win64).
malaterre [Sat, 22 Jan 2005 01:40:41 +0000 (01:40 +0000)]
BUG: was leading zeroes in numeric components. This is illegal, according to PS 3.8 Annex F and PS 3.5 Section 6.1 and ISO 8824. Also remove the union this is a bad idea when working on little and big endian machine, the result was not constitant across endianess
malaterre [Fri, 21 Jan 2005 16:06:21 +0000 (16:06 +0000)]
ENH: Add two new methods: GetCurrentThreadID and ProcessID. This could be usefull to generate the DICOM UID. Also reimplement GetCurrentDate / GetCurrentTime in only one call since we could be right at the edge of midnight and the two function call would not be consistent. Also this simplify the implementation of GetCurrentDateTime and add millisecond precision (yes we write DICOM quite fast)
jpr [Fri, 21 Jan 2005 15:28:18 +0000 (15:28 +0000)]
Begin of 'friend' classe hunting.
We miss GetLastEntry and GetPriorEntry methods to be able to remove
any direct reference to ElementSet::TagHT by DicomDir
jpr [Fri, 21 Jan 2005 11:40:52 +0000 (11:40 +0000)]
Stage 2 of names normalization :
- A gdcm::Document is now either a gdcm::DicomDir or a gdcm::File
(easy to understand ...)
- gdcm::FileHelper replaces old class gdcm::File
regrain [Thu, 20 Jan 2005 17:15:53 +0000 (17:15 +0000)]
* gdcmDicomDirElement.[h|cxx] : change the AddEntry method. Now, it gets a
type and not a string. So, there remove problems on the format of the
string.
-- BeNours
regrain [Thu, 20 Jan 2005 16:16:41 +0000 (16:16 +0000)]
* src/gdcmDicomDirObject.h : set the FillObject method in protected.
Thus, DicomDirXXX are created with a flad that indicates if the created
object is empty or not (when not empty, the DicomDirElement is used to
fill the object)
-- BeNours
jpr [Thu, 20 Jan 2005 13:28:23 +0000 (13:28 +0000)]
Final bugg fix.
TestPapyrus - reads a Papyrus file (gdcm::Header compliant),
- gets the pixels of all the images, using low level iterators
- writes an multiframes Dicom image.
(Just to show it's *possible* to do that)
Uses SQItems accessors (we could use it as a 'Test' program, to improve test
coverage ...)
regrain [Thu, 20 Jan 2005 11:07:06 +0000 (11:07 +0000)]
* src/gdcmDebug.h : remove the Debug test in the gdcmErrorMacro
* src/gdcmDocument.[h|cxx] : rename the Initialise method to Initialize, to
be complient with all the gdcm source code
-- BeNours
regrain [Wed, 19 Jan 2005 14:56:09 +0000 (14:56 +0000)]
* gdcmPython/ : change GDCM_DATA_PATH to GDCM_DATA_ROOT to be complient with
all in gdcm
* Remove some useless files in the Python part. Files now generated by CMake
-- BeNours
regrain [Wed, 19 Jan 2005 10:43:00 +0000 (10:43 +0000)]
* Test/ : rename VTK tests to have a best name for the tests
* gdcmPython/demo/*.py.in : Change the InitTraversal/GetNextEntry use to
GetFirstEntry/GetNextEntry use
-- BeNours
regrain [Tue, 18 Jan 2005 18:24:06 +0000 (18:24 +0000)]
* Test/ShowDicom.cxx, ShowDicomSeq.cxx, TestWriteWithVTK.cxx : change the
test. Now it's on image content and not on the resulting visualisation...
might solve the problem on linux when using the OpenGL drivers
-- BeNours
jpr [Tue, 18 Jan 2005 11:56:52 +0000 (11:56 +0000)]
In SerieHeader::ImagePositionPatientOrdering()
// a Dicom 'Serie' may contain scout views
// and images may have differents directions.
// -> More than one may have the same 'pos'
// Sorting has then NO meaning !