Program: gdcm
Module: $RCSfile: gdcmDocument.cxx,v $
Language: C++
- Date: $Date: 2006/10/19 10:30:45 $
- Version: $Revision: 1.354 $
+ Date: $Date: 2008/09/15 15:49:22 $
+ Version: $Revision: 1.381 $
Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de
l'Image). All rights reserved. See Doc/License.txt or
#include <mem.h> // for memset
#endif
-namespace gdcm
+namespace GDCM_NAME_SPACE
{
//-----------------------------------------------------------------------------
/**
* \brief This default constructor neither loads nor parses the file.
- * You should then invoke \ref Document::Load.
+ * You should then invoke Document::Load.
*
*/
Document::Document()
SwapCode = 1234;
Filetype = ExplicitVR;
CurrentOffsetPosition = 0;
+ OffsetOfPreviousParseDES =0;
// Load will set it to true if sucessfull
Group0002Parsed = false;
IsDocumentAlreadyLoaded = false;
LoadMode = LD_ALL; // default : load everything, later
SetFileName("");
+ changeFromUN=false;
+ UnexpectedEOF=false;
}
/**
// Recursive call.
// Loading is done during parsing
+ OffsetOfPreviousParseDES = beg;
ParseDES( this, beg, lgt, false); // delim_mode is first defaulted to false
if ( IsEmpty() )
// Force Loading some more elements if user asked to.
- gdcm::DocEntry *d;
+ GDCM_NAME_SPACE::DocEntry *d;
for (ListElements::iterator it = UserForceLoadList.begin();
it != UserForceLoadList.end();
++it)
//
std::string RecCode;
RecCode = GetEntryString(0x0008, 0x0010); // recognition code (RET)
- if (RecCode == "ACRNEMA_LIBIDO_1.1" ||
- RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares
- // with "little-endian strings"
+
+
+ if(RecCode.find("ACRNEMA_LIBIDO") == 0 || // any version
+ RecCode.find("CANRME_AILIBOD") == 0) // for brain-damaged softwares
+ // with "little-endian strings"
{
+
Filetype = ACR_LIBIDO;
std::string rows = GetEntryString(0x0028, 0x0010);
std::string columns = GetEntryString(0x0028, 0x0011);
//
// -----------------File I/O ---------------
/**
- * \brief Tries to open the file \ref Document::Filename and
- * checks the preamble when existing.
- * @return The FILE pointer on success.
+ * \brief Tries to open the file Document::Filename and
+ * checks the preamble when existing,
+ * or if the file starts with an ACR-NEMA look-like element.
+ * @return The FILE pointer on success, 0 on failure.
*/
std::ifstream *Document::OpenFile()
{
CloseFile();
return 0;
}
+
if ( memcmp(dicm, "DICM", 4) == 0 )
{
HasDCMPreamble = true;
* @param filetype Type of the File to be written
* (ACR-NEMA, ExplicitVR, ImplicitVR)
*/
-void Document::WriteContent(std::ofstream *fp, FileType filetype)
+void Document::WriteContent(std::ofstream *fp, FileType filetype, bool, bool)
{
// Skip if user wants to write an ACR-NEMA file
* --> was too much tricky / we were [in a hurry / too lazy]
* --> We don't write the element 0x0000 (group length)
*/
-
- ElementSet::WriteContent(fp, filetype); // This one is recursive
+ // This one is recursive
+ // false : outside MetaElements
+ // false : outside Sequence
+ ElementSet::WriteContent(fp, filetype, false, false);
}
// -----------------------------------------
//}
/**
- * \brief Compares two documents, according to \ref DicomDir rules
+ * \brief Compares two documents, according to DicomDir rules
* \warning Does NOT work with ACR-NEMA files
* \todo Find a trick to solve the pb (use RET fields ?)
* @param document to compare with current one
/**
* \brief Reads a given length of bytes
- * (in order to avoid to many CPU time consuming fread-s)
+ * (in order to avoid to many CPU time-consuming fread-s)
* @param l length to read
*/
void Document::ReadBegBuffer(size_t l)
vr = entry->GetVR();
//if ( (vr == "OB")||(vr == "OW")||(vr == "UT")||(vr == "SQ"))
- // (no SQ, OW, UT in group 0x0002;)
+ // (no SQ, OW, OL, UT in group 0x0002;)
if ( vr == "OB" )
{
- // explicit VR AND (OB, OW, SQ, UT) : 4 more bytes
+ // explicit VR AND (OB, OW, OL, SQ, UT, UN) : 4 more bytes
groupLength += 4;
}
groupLength += 2 + 2 + 4 + entry->GetLength();
{
///\todo FIXME : On 64 bits processors, tellg gives unexpected results after a while ?
- /// Probabely a bug in gdcm code somwhere (some memory erased ?)
+ /// Probabely a bug in gdcm code somewhere (some memory erased ?)
// Uncomment to track the bug
-/*
+
if( Debug::GetDebugFlag() )
std::cout << std::dec <<"(long)(Fp->tellg()) " << (long)(Fp->tellg()) // in Debug mode
<< std::hex << " 0x(" <<(long)(Fp->tellg()) << ")" << std::endl;
- */
+
// if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max) // Once per DocEntry
if ( !delim_mode ) // 'and then' doesn't exist in C++ :-(
if ( !first && newDocEntry->IsItemStarter() )
{
// Debug message within the method !
- newDocEntry = Backtrack(newDocEntry);
+ newDocEntry = Backtrack(newDocEntry, set);
}
else
{
{
/////////////////////// SeqEntry : VR = "SQ"
- unsigned long l = newDocEntry->GetReadLength();
+ unsigned long l = newDocEntry->GetReadLength();
if ( l != 0 ) // don't mess the delim_mode for 'zero-length sequence'
{
if ( l == 0xffffffff )
<< " at offset 0x(" << std::hex
<< newDocEntry->GetOffset() << ")");
- ParseSQ( newSeqEntry,
- newDocEntry->GetOffset(),
- l, delim_mode_intern);
+ bool res = ParseSQ( newSeqEntry,
+ newDocEntry->GetOffset(),
+ l, delim_mode_intern);
- gdcmDebugMacro( "Exit from ParseSQ, delim " << delim_mode_intern);
+ gdcmDebugMacro( "Exit from ParseSQ, delim " << delim_mode_intern << " -->return : " << res);
}
if ( !set->AddEntry( newSeqEntry ) )
{
newDocEntry->Delete();
}
first = false;
+
+ if (UnexpectedEOF) // some terminator was missing
+ {
+ break;
+ }
} // end While
gdcmDebugMacro( "Exit from ParseDES, delim-mode " << delim_mode );
}
/**
* \brief Parses a Sequence ( SeqEntry after SeqEntry)
- * @return parsed length for this level
+ * @return false if expected fff0,e000 not found
*/
-void Document::ParseSQ( SeqEntry *seqEntry,
+bool Document::ParseSQ( SeqEntry *seqEntry,
long offset, long l_max, bool delim_mode)
{
int SQItemNumber = 0;
DocEntry *newDocEntry = ReadNextDocEntry();
if ( !newDocEntry )
- {
+ {
+ // The most frequent is when a SQ terminator is missing (?!?)
gdcmWarningMacro("in ParseSQ : should never get here!");
- break;
+ UnexpectedEOF = true;
+ return false;
}
if ( delim_mode )
{
dlm_mod = false;
}
+ // avoid infinite loop when Bad assumption was made on illegal 'unknown length' UN //JPRx
+
+ if (offsetStartCurrentSQItem <= OffsetOfPreviousParseDES)
+ {
+ gdcmWarningMacro("Bad assumption was made on illegal 'unknown length' UN!" << std::endl <<
+ "OffsetOfPreviousParseDES " << std::hex << OffsetOfPreviousParseDES
+ << " offsetStartCurrentSQItem " << offsetStartCurrentSQItem);
+ /// \todo when "Bad assumption (SQ) on illegal 'unknown length' UN", Backtrack again + try OB
+ return false;
+ }
+ else
+ {
+ OffsetOfPreviousParseDES = offsetStartCurrentSQItem;
+ }
+
// fill up the current SQItem, starting at the beginning of fff0,e000
-
Fp->seekg(offsetStartCurrentSQItem, std::ios::beg); // Once per SQItem
ParseDES(itemSQ, offsetStartCurrentSQItem, l+8, dlm_mod);
offsetStartCurrentSQItem = Fp->tellg(); // Once per SQItem
break;
}
}
+ return true;
}
/**
* \brief When a private Sequence + Implicit VR is encountered
* we cannot guess it's a Sequence till we find the first
* Item Starter. We then backtrack to do the job.
- * @param docEntry Item Starter that warned us
+ * @param docEntry Item Starter that warned us
+ * @param set DocEntrySet (ElementSet/SQItem) the DocEntry will belong
*/
-DocEntry *Document::Backtrack(DocEntry *docEntry)
+DocEntry *Document::Backtrack(DocEntry *docEntry, DocEntrySet *set)
{
// delete the Item Starter, built erroneously out of any Sequence
// it's not yet in the HTable/chained list
gdcmDebugMacro( "Backtrack :" << std::hex << group
<< "|" << elem
<< " at offset 0x(" <<offset << ")" );
- RemoveEntry( PreviousDocEntry );
+
+ set->RemoveEntry( PreviousDocEntry );
// forge the Seq Entry
DocEntry *newEntry = NewSeqEntry(group, elem);
// Move back to the beginning of the Sequence
Fp->seekg(offset, std::ios::beg); // Only for Shadow Implicit VR SQ
- return newEntry;
+ return newEntry; // It will added where it has to be!
}
/**
{
const VRKey &vr = entry->GetVR();
uint16_t length16;
-
if ( Filetype == ExplicitVR && !entry->IsImplicitVR() )
{
- if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UT"
- || vr == "UN" )
+
+ // WARNING :
+ //
+ // For some images, length of UN elements is coded on 2 bytes (instead of 4)
+ // There are *not* readable !
+ // You can make a quick and dirty patch, commenting out
+ //| vr == "UN"
+ // in the following line.
+ // (the 'straight' images will no longer be readable ...)
+
+ if ( vr == "OB" || vr == "OW" || vr == "OL" || vr == "SQ" || vr == "UT"
+ || vr == "UN" || changeFromUN == true)
{
+ changeFromUN = false;
// The following reserved two bytes (see PS 3.5-2003, section
// "7.1.2 Data element structure with explicit vr", p 27) must be
// skipped before proceeding on reading the length on 4 bytes.
- //Fp->seekg( 2L, std::ios::cur); // Once per OW,OB,SQ DocEntry
- uint32_t length32 = ReadInt32(); // Once per OW,OB,SQ DocEntry
+ //Fp->seekg( 2L, std::ios::cur); // Once per OB,OW,OL,UT,UN,SQ DocEntry
+ uint32_t length32 = ReadInt32(); // Once per OB,OW,OL,UT,UN,SQ DocEntry
CurrentOffsetPosition+=4;
- if ( (vr == "OB" || vr == "OW") && length32 == 0xffffffff )
+ if ( (vr == "OB" || vr == "OW" || vr == "OL") && length32 == 0xffffffff )
{
uint32_t lengthOB;
try
FixDocEntryFoundLength(entry, length32);
return;
}
-
// Length is encoded on 2 bytes.
//length16 = ReadInt16();
length16 = GetInt16();
-
// 0xffff means that we deal with 'No Length' Sequence
// or 'No Length' SQItem
if ( length16 == 0xffff)
uint16_t elem;
try
- {
+ { ///\todo make sure there is never OL encoded pixel data!
+
//group = ReadInt16(); // Once per fragment (if any) of OB,OW DataElements
//elem = ReadInt16(); // Once per fragment (if any) of OB,OW DataElements
ReadBegBuffer(4); // Once per fragment (if any) of OB,OW DataElements
{
gdcmWarningMacro(
"Neither an Item tag nor a Sequence delimiter tag on :"
- << std::hex << group << " , " << elem
- << ")" );
+ << std::hex << group << "|" << elem << ") Pos. on entry was 0x(" <<positionOnEntry<< ") "
+ );
Fp->seekg(positionOnEntry, std::ios::beg); // Once per fragment (if any) of OB,OW DataElements
throw FormatUnexpected(
//if ( !CheckDocEntryVR(vr) ) // avoid useless function call
if ( !Global::GetVR()->IsValidVR(vr) )
- {
-/*
-// std::cout << "================================================================Unknown VR"
- << std::hex << "0x("
- << (unsigned int)vr[0] << "|" << (unsigned int)vr[1]
- << ")" << "for : " << CurrentGroup
- << " at offset : 0x(" << positionOnEntry << ")"
- << std::endl;
-*/
- gdcmWarningMacro( "Unknown VR " << std::hex << "0x("
- << (unsigned int)vr[0] << "|" << (unsigned int)vr[1]
- << ")"
- << " at offset : 0x(" << CurrentOffsetPosition-4<< ") for group " << CurrentGroup
- );
+ {
+ gdcmWarningMacro( "Unknown VR " << vr.GetHexaRepresentation() << std::hex
+ << " at offset : 0x(" << CurrentOffsetPosition-4
+ << ") for group " << std::hex << CurrentGroup );
+
//Fp->seekg(positionOnEntry, std::ios::beg); //JPRx
//Fp->seekg((long)-2, std::ios::cur);// only for unrecognized VR (?!?)
//see :MR_Philips_Intera_PrivateSequenceExplicitVR.dcm
void Document::SkipToNextDocEntry(DocEntry *currentDocEntry)
{
long l = currentDocEntry->GetReadLength();
- if ( l == -1 ) // length = 0xffff shouldn't appear here ...
+ if ( (uint32_t) l == (uint32_t)-1 ) // length = 0xffff shouldn't appear here ...
// ... but PMS imagers happen !
return;
Fp->seekg((size_t)(currentDocEntry->GetOffset()), std::ios::beg); //FIXME :each DocEntry
if ( foundLength % 2)
{
- gdcmWarningMacro( "Warning : Tag with uneven length " << foundLength
- << " in x(" << std::hex << gr << "," << elem <<")");
+ gdcmWarningMacro( "Warning : Tag (" << std::hex << gr << "|" << elem << ") with uneven length "
+ << std::dec << foundLength << " 0x(" << std::hex << foundLength << ") "
+ //<< " at offset x(" << offset << ")"
+ );
}
-
+
//////// Fix for some naughty General Electric images.
// Allthough not recent many such GE corrupted images are still present
// on Creatis hard disks. Hence this fix shall remain when such images
// 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.
+ // able to swap its element value properly.
if ( elem == 0 ) // This is the group length of the group
{
if ( length == 4 )
// i.e. a total of 136 bytes.
entCur = deb + 136;
- // group 0x0002 *is always* Explicit VR Sometimes ,
+ // group 0x0002 *is always* Explicit VR Sometimes,
// even if elem 0002,0010 (Transfer Syntax) tells us the file is
// *Implicit* VR (see former 'gdcmData/icone.dcm')
<< "Looks like a bugged Header!");
}
+ // Here, we assume that the file IS kosher Dicom !
+ // (The meta elements - group 0x0002 - ARE little endian !)
if ( net2host )
{
SwapCode = 4321;
s16 = *((uint16_t *)(deb));
gdcmDebugMacro("not a DicomV3 nor a 'clean' ACR/NEMA;"
- << " (->despaired wild guesses !)");
-
+ << " (->despaired wild guesses !)");
switch ( s16 )
{
case 0x0001 :
Filetype = ACR;
return true;
default :
- gdcmWarningMacro("ACR/NEMA unfound swap info (Hopeless !)");
- Filetype = Unknown;
- return false;
+
+ s16 = *((uint16_t *)(deb));
+ if (s16 != 0x0000)
+ return false;
+ s16 = *((uint16_t *)(deb+2));
+
+ Fp->seekg ( 0L, std::ios::beg); // Once per Document
+ CurrentOffsetPosition = 0;
+ switch(s16) // try an other trick!
+ // -> to be able to decode 0029|1010 DataElement
+ // -> and be not less cleaver than dcmdump ;-)
+ {
+ case 0x0004 :
+ SwapCode = 1234;
+ break;
+ case 0x0400 :
+ SwapCode = 3412;
+ break;
+ default:
+ gdcmWarningMacro("ACR/NEMA unfound swap info (Hopeless !)");
+ Filetype = Unknown;
+ return false;
+ }
+ // Check if next 2 bytes are a VR
+ // Probabely something more time-consuming exists with std::string
+ const char VRvalues[] = "AEASATCSDADTFLFDISLOLTPNSHSLSSSTTMUIULUSUTOBOWOLOFATUNSQRT";
+ int nbVal = 29;
+ const char *pt = VRvalues;
+ for (int i=0;i<nbVal;i++)
+ {
+ if(*(deb+4) == *pt++) {
+ if(*(deb+5) == *pt++) {
+ Filetype = ExplicitVR;
+ return true;
+ }
+ else {
+ pt++;
+ }
+ }
+ }
+ Filetype = ImplicitVR;
+ return true;
}
}
}
return 0;
}
+ changeFromUN = false;
CurrentGroup = GetInt16();
CurrentElem = GetInt16();
-
+
// In 'true DICOM' files Group 0002 is always little endian
if ( HasDCMPreamble )
{
realVR = "UL";
}
- // Commented out in order not to generate 'Shadow Groups' where some
+ // Was commented out in order not to generate 'Shadow Groups' where some
// Data Elements are Explicit VR and some other ones Implicit VR
- // (Stupid MatLab DICOM Reader couldn't read gdcm-written images)
- /*
- else if (CurrentGroup%2 == 1 &&
- (CurrentElem >= 0x0010 && CurrentElem <=0x00ff ))
- {
- // DICOM PS 3-5 7.8.1 a) states that those
- // (gggg-0010->00FF where gggg is odd) attributes have to be LO
- realVR = "LO";
+ // -> Better we fix the problem at Write time
+
+ else if (CurrentGroup%2 == 1 )
+ {
+ if (CurrentElem >= 0x0010 && CurrentElem <=0x00ff )
+ {
+ // DICOM PS 3-5 7.8.1 a) states that :
+ // Private Creator Data Elements numbered (gggg,0010-00FF) (gggg is odd)
+ // attributes have to be LO (Long String) and the VM shall be equal to 1
+ realVR = "LO";
+
+ // Seems not to be true
+ // Still in gdcmtk, David Clunnie disagrees, Marco Eichelberg says it's OK ...
+ // We let it for a while?
+ //(We should check length==4, for more security, but we don't have it yet !)
+ }
+ else if ( CurrentElem == 0x0001)
+ {
+ realVR = "UL"; // Private Group Length To End
+ }
+ else // check the private dictionary for shadow elements when Implicit VR!
+ {
+ DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem);
+ if ( dictEntry )
+ {
+ realVR = dictEntry->GetVR();
+ dictEntry->Unregister(); // GetDictEntry registered it
+ }
+ }
}
- */
+
else
{
DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem);//only when ImplicitVR
}
}
+ // if UN found, let's check the dictionary, and trust it!
+ // (maybe a private dictionary exists?)
+ else if (vr == "UN")
+ {
+ DictEntry *dictEntry = GetDictEntry(CurrentGroup,CurrentElem);
+ if ( dictEntry )
+ {
+ realVR = dictEntry->GetVR();
+ dictEntry->Unregister(); // GetDictEntry registered it
+
+ // for VR = "UN", length is always stored on 4 bytes.
+ // remember this info, in order not to crash later
+ changeFromUN=true;
+ /// \todo : fixme If inside a vr = "UN" DataElement (but SQ according to a private dictionnary)
+ /// there is some more vr = "UN" DataElements, it will probabely fail.
+ /// --> find a -non time consuming- trick to store changeFromUN info at DataElement level,
+ /// not at the Document level.
+ /// --> ?!? JPR
+
+ }
+ }
+
DocEntry *newEntry;
//if ( Global::GetVR()->IsVROfSequence(realVR) )
if (realVR == "SQ")
int offset = Fp->tellg();//Only when heuristic for Explicit/Implicit was wrong
gdcmWarningMacro("Entry (" << newEntry->GetKey() << ") at x("
- << offset << ") should be Explicit VR");
+ << std::hex << offset << ") should be Explicit VR");
}
}
newEntry->SetImplicitVR();
//
// --> Probabely normal, since we considered we never have
// to trust manufacturers.
- // (we find very often 'Implicit VR' tag,
+ // (we often find 'Implicit VR' tag,
// even when Transfer Syntax tells us it's Explicit ...
+ // NEVER trust the meta elements!
+ // (see what ezDICOM does ...)
+
+ /*
if ( s == TS::ExplicitVRBigEndian )
{
gdcmDebugMacro("Transfer Syntax Name = ["
group = SwapShort(group);
elem = SwapShort(elem);
}
-
+ */
+ //-- Broken ACR may start with a Shadow Group --
+ // worse : some ACR-NEMA like files start 00028 group ?!?
+ if ( !( (group >= 0x0001 && group <= 0x0008) || group == 0x0028 ) )
+ {
+ // We trust what we see.
+ SwitchByteSwapCode();
+ group = SwapShort(group);
+ elem = SwapShort(elem);
+ // not what we where told (by meta elements) !
+ gdcmDebugMacro("Transfer Syntax Name = ["
+ << GetTransferSyntaxName() << "]" );
+ }
+
/// \todo find a trick to warn user and stop processing
if ( s == TS::DeflatedExplicitVRLittleEndian)
git://git.creatis.insa-lyon.fr / gdcm.git / blobdiff