Program: gdcm
Module: $RCSfile: gdcmDocument.cxx,v $
Language: C++
- Date: $Date: 2006/07/06 12:38:06 $
- Version: $Revision: 1.352 $
+ Date: $Date: 2007/09/17 12:16:02 $
+ Version: $Revision: 1.369 $
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()
Initialize();
SwapCode = 1234;
Filetype = ExplicitVR;
+ CurrentOffsetPosition = 0;
// Load will set it to true if sucessfull
Group0002Parsed = false;
IsDocumentAlreadyLoaded = false;
IsDocumentModified = true;
LoadMode = LD_ALL; // default : load everything, later
+
SetFileName("");
+ changeFromUN=false;
+ UnexpectedEOF=false;
}
/**
// 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)
//
// -----------------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()
{
return 0;
}
- //-- Broken ACR or DICOM with no Preamble; may start with a Shadow Group --
- // FIXME : We cannot be sure the preable is only zeroes..
- // (see ACUSON-24-YBR_FULL-RLE.dcm )
- if (
- zero == 0x0001 || zero == 0x0100 || zero == 0x0002 || zero == 0x0200 ||
- zero == 0x0003 || zero == 0x0300 || zero == 0x0004 || zero == 0x0400 ||
- zero == 0x0005 || zero == 0x0500 || zero == 0x0006 || zero == 0x0600 ||
- zero == 0x0007 || zero == 0x0700 || zero == 0x0008 || zero == 0x0800 )
- {
- std::string msg = Util::Format(
- "ACR/DICOM starting by 0x(%04x) at the beginning of the file\n", zero);
- // FIXME : is it a Warning message, or a Debug message?
- gdcmWarningMacro( msg.c_str() );
- return Fp;
- }
-
//-- DICOM --
Fp->seekg(126L, std::ios::cur); // Once per Document
char dicm[4]; // = {' ',' ',' ',' '};
CloseFile();
return 0;
}
+
if ( memcmp(dicm, "DICM", 4) == 0 )
{
HasDCMPreamble = true;
return Fp;
}
+ //-- Broken ACR or DICOM (?) with no Preamble; may start with a Shadow Group --
+ // FIXME : We cannot be sure the preable is only zeroes..
+ // (see ACUSON-24-YBR_FULL-RLE.dcm )
+ if (
+ zero == 0x0001 || zero == 0x0100 || zero == 0x0002 || zero == 0x0200 ||
+ zero == 0x0003 || zero == 0x0300 || zero == 0x0004 || zero == 0x0400 ||
+ zero == 0x0005 || zero == 0x0500 || zero == 0x0006 || zero == 0x0600 ||
+ zero == 0x0007 || zero == 0x0700 || zero == 0x0008 || zero == 0x0800 ||
+ zero == 0x0028 || 0x2800 // worse : some ACR-NEMA like files
+ // start 00028 group ?!?
+ )
+ {
+ std::string msg = Util::Format(
+ "ACR/DICOM starting by 0x(%04x) at the beginning of the file\n", zero);
+ // FIXME : is it a Warning message, or a Debug message?
+ gdcmWarningMacro( msg.c_str() );
+ return Fp;
+ }
+
// -- Neither ACR/No Preamble Dicom nor DICOMV3 file
CloseFile();
// Don't user Warning nor Error, not to pollute the output
* --> 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
//-----------------------------------------------------------------------------
// Protected
+
+/**
+ * \brief Reads a given length of bytes
+ * (in order to avoid to many CPU time-consuming fread-s)
+ * @param l length to read
+ */
+void Document::ReadBegBuffer(size_t l)
+ throw( FormatError )
+{
+ Fp->read (BegBuffer, (size_t)l);
+ if ( Fp->fail() )
+ {
+ throw FormatError( "Document::ReadBegBuffer()", " file error." );
+ }
+ if ( Fp->eof() )
+ {
+ throw FormatError( "Document::ReadBegBuffer()", "EOF." );
+ }
+ PtrBegBuffer = BegBuffer;
+ CurrentOffsetPosition+=l;
+}
/**
* \brief Reads a supposed to be 16 Bits integer
* (swaps it depending on processor endianness)
return g;
}
+/**
+ * \brief Gets from BegBuffer a supposed to be 16 Bits integer
+ * (swaps it depending on processor endianness)
+ * @return read value
+ */
+uint16_t Document::GetInt16()
+{
+ uint16_t g = *((uint16_t*)PtrBegBuffer);
+ g = SwapShort(g);
+ PtrBegBuffer+=2;
+ return g;
+}
/**
* \brief Reads a supposed to be 32 Bits integer
* (swaps it depending on processor endianness)
return g;
}
+/**
+ * \brief Gets from BegBuffer a supposed to be 32 Bits integer
+ * (swaps it depending on processor endianness)
+ * @return read value
+ */
+uint32_t Document::GetInt32()
+{
+ uint32_t g = *((uint32_t*)PtrBegBuffer);
+ g = SwapLong(g);
+ PtrBegBuffer+=4;
+ return g;
+}
+
/**
* \brief Re-computes the length of the Dicom group 0002.
*/
<< " 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 )
{
break;
}
}
+ return true;
}
/**
{
const VRKey &vr = entry->GetVR();
uint16_t length16;
-
if ( Filetype == ExplicitVR && !entry->IsImplicitVR() )
{
+
+ // 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 == "SQ" || vr == "UT"
- || vr == "UN" )
+ || 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();
-
+ //Fp->seekg( 2L, std::ios::cur); // Once per OW,OB,SQ DocEntry
+ uint32_t length32 = ReadInt32(); // Once per OW,OB,SQ DocEntry
+ CurrentOffsetPosition+=4;
if ( (vr == "OB" || vr == "OW") && length32 == 0xffffffff )
{
uint32_t lengthOB;
FixDocEntryFoundLength(entry, length32);
return;
}
-
// Length is encoded on 2 bytes.
- length16 = ReadInt16();
-
+ //length16 = ReadInt16();
+ length16 = GetInt16();
// 0xffff means that we deal with 'No Length' Sequence
// or 'No Length' SQItem
if ( length16 == 0xffff)
// even if Transfer Syntax is 'Implicit VR ...'
// --> Except for 'Implicit VR Big Endian Transfer Syntax GE Private'
// where Group 0x0002 is *also* encoded in Implicit VR !
-
- FixDocEntryFoundLength( entry, ReadInt32() );
+
+ FixDocEntryFoundLength( entry, GetInt32() /*ReadInt32()*/ );
return;
}
}
throw( FormatUnexpected )
{
// See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
+
long positionOnEntry = Fp->tellg(); // Only for OB,OW DataElements
bool foundSequenceDelimiter = false;
{
uint16_t group;
uint16_t elem;
+
try
{
- group = ReadInt16();
- elem = ReadInt16();
+ //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
}
catch ( FormatError )
{
throw FormatError("Unexpected end of file encountered during ",
"Document::FindDocEntryLengthOBOrOW()");
}
+ group = GetInt16();
+ elem = GetInt16();
+
// We have to decount the group and element we just read
totalLength += 4;
if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) )
{
foundSequenceDelimiter = true;
}
- uint32_t itemLength = ReadInt32();
+ uint32_t itemLength = ReadInt32(); // Once per fragment (if any) of OB,OW DataElements
// We add 4 bytes since we just read the ItemLength with ReadInt32
totalLength += itemLength + 4;
SkipBytes(itemLength);
break;
}
}
- Fp->seekg( positionOnEntry, std::ios::beg); // Only for OB,OW DataElements
+ Fp->seekg( positionOnEntry, std::ios::beg); // Only once for OB,OW DataElements
return totalLength;
}
if ( CurrentGroup == 0xfffe )
return GDCM_VRUNKNOWN;
- long positionOnEntry;
- if( Debug::GetWarningFlag() )
- positionOnEntry = Fp->tellg(); // Only in Warning Mode
+ //long positionOnEntry;
+ //if( Debug::GetWarningFlag() )
+ // positionOnEntry = Fp->tellg(); // Only in Warning Mode
// Warning: we believe this is explicit VR (Value Representation) because
// we used a heuristic that found "UL" in the first tag and/or
// if it happens not to be the case.
VRKey vr;
- Fp->read(&(vr[0]),(size_t)2);
-
- if ( !CheckDocEntryVR(vr) )
+ //Fp->read(&(vr[0]),(size_t)2);
+ vr[0] = *PtrBegBuffer++;
+ vr[1] = *PtrBegBuffer++;
+
+ //if ( !CheckDocEntryVR(vr) ) // avoid useless function call
+ if ( !Global::GetVR()->IsValidVR(vr) )
{
/*
// std::cout << "================================================================Unknown VR"
gdcmWarningMacro( "Unknown VR " << std::hex << "0x("
<< (unsigned int)vr[0] << "|" << (unsigned int)vr[1]
<< ")"
- << " at offset : 0x(" << positionOnEntry<< ") for group " << CurrentGroup
+ << " at offset : 0x(" << CurrentOffsetPosition-4<< ") for group " << CurrentGroup
);
//Fp->seekg(positionOnEntry, std::ios::beg); //JPRx
- Fp->seekg((long)-2, std::ios::cur);// only for unrecognized VR (?!?)
+ //Fp->seekg((long)-2, std::ios::cur);// only for unrecognized VR (?!?)
//see :MR_Philips_Intera_PrivateSequenceExplicitVR.dcm
+ PtrBegBuffer-=2;
return GDCM_VRUNKNOWN;
}
return vr;
// 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;
// (i.e. after the file preamble and the "DICM" string).
Fp->seekg ( 132L, std::ios::beg); // Once per Document
+ CurrentOffsetPosition = 132;
return true;
} // ------------------------------- End of DicomV3 ----------------
gdcmWarningMacro( "Not a Kosher DICOM Version3 file (no preamble)");
Fp->seekg(0, std::ios::beg); // Once per ACR-NEMA Document
-
+ CurrentOffsetPosition = 0;
// Let's check 'No Preamble Dicom File' :
// Should start with group 0x0002
// and be Explicit Value Representation
s16 = *((uint16_t *)(deb));
gdcmDebugMacro("not a DicomV3 nor a 'clean' ACR/NEMA;"
- << " (->despaired wild guesses !)");
-
+ << " (->despaired wild guesses !)");
switch ( s16 )
{
case 0x0001 :
case 0x0006 :
case 0x0007 :
case 0x0008 :
+ case 0x0028 :
SwapCode = 1234;
Filetype = ACR;
return true;
case 0x0600 :
case 0x0700 :
case 0x0800 :
+ case 0x2800 :
SwapCode = 4321;
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[] = "AEASCSDADSFLFDISLOLTPNSHSLSSSTTMUIULUSUTOBOWOFATUNSQ";
+ int nbVal = 26;
+ const char *pt = VRvalues;
+ for (int i=0;i<nbVal;i++)
+ {
+ if(*(deb+4) == *pt++)
+ if(*(deb+5) == *pt++) {
+ Filetype = ExplicitVR;
+ return true;
+ }
+
+ }
+ Filetype = ImplicitVR;
+ return true;
}
}
}
{
try
{
- CurrentGroup = ReadInt16();
- CurrentElem = ReadInt16();
+ ReadBegBuffer(8); // Avoid to many time consuming freads
+ //CurrentGroup = ReadInt16();
+ //CurrentElem = ReadInt16();
}
catch ( FormatError )
{
// header parsing has to be considered as finished.
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
{
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.
+ changeFromUN=true;
+ /// \todo : fixme If inside a supposed to be UN DataElement (but SQ according to a private dictionnary)
+ /// there is some more supposed to be UN DataElements, it will probabely fail.
+ /// --> find a -non time consuming- trick to store changeFromUN info at DataElement level,
+ /// not at the Document level.
+ }
+ }
+
+
DocEntry *newEntry;
//if ( Global::GetVR()->IsVROfSequence(realVR) )
if (realVR == "SQ")
//
// --> 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)