Program: gdcm
Module: $RCSfile: gdcmDocument.cxx,v $
Language: C++
- Date: $Date: 2005/01/26 16:28:58 $
- Version: $Revision: 1.213 $
+ Date: $Date: 2005/04/26 16:18:23 $
+ Version: $Revision: 1.237 $
Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de
l'Image). All rights reserved. See Doc/License.txt or
#include "gdcmDebug.h"
#include "gdcmTS.h"
#include "gdcmDictSet.h"
-#include "gdcmRLEFramesInfo.h"
-#include "gdcmJPEGFragmentsInfo.h"
#include "gdcmDocEntrySet.h"
#include "gdcmSQItem.h"
#include <vector>
#include <iomanip>
-
-// For nthos:
-#if defined(_MSC_VER) || defined(__BORLANDC__) || defined(__MINGW32__)
- #include <winsock.h>
-#endif
-
-#ifdef CMAKE_HAVE_NETINET_IN_H
- #include <netinet/in.h>
-#endif
+#include <fstream>
namespace gdcm
{
-
//-----------------------------------------------------------------------------
// Refer to Document::CheckSwap()
//const unsigned int Document::HEADER_LENGTH_TO_READ = 256;
// Refer to Document::SetMaxSizeLoadEntry()
const unsigned int Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096
-const unsigned int Document::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff;
//-----------------------------------------------------------------------------
// Constructor / Destructor
+// Constructors and destructors are protected to avoid user to invoke directly
+
+/**
+ * \brief This default constructor neither loads nor parses the file.
+ * You should then invoke \ref Document::Load.
+ *
+ */
+Document::Document()
+ :ElementSet(-1)
+{
+ Fp = 0;
+
+ SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
+ Initialize();
+ SwapCode = 1234;
+ Filetype = ExplicitVR;
+ // Load will set it to true if sucessfull
+ Group0002Parsed = false;
+ IsDocumentAlreadyLoaded = false;
+ LoadMode = 0x00000000; // default : load everything, later
+}
/**
- * \brief constructor
+ * \brief Constructor (not to break the API)
* @param filename 'Document' (File or DicomDir) to be opened for parsing
*/
-Document::Document( std::string const &filename ) : ElementSet(-1)
+Document::Document( std::string const &filename )
+ :ElementSet(-1)
{
- RLEInfo = NULL;
- JPEGInfo = NULL;
+ Fp = 0;
- SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
- Filename = filename;
+ SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
Initialize();
+ SwapCode = 1234;
+ Filetype = ExplicitVR;
+ Group0002Parsed = false;
+ LoadMode = 0x00000000; // Load everything
+
+ // Load will set it to true if sucessfull
+ IsDocumentAlreadyLoaded = false;
+
+ Load(filename);
+}
+/**
+ * \brief Canonical destructor.
+ */
+Document::~Document ()
+{
+ RefPubDict = NULL;
+ RefShaDict = NULL;
+}
+
+//-----------------------------------------------------------------------------
+// Public
+
+/**
+ * \brief Loader
+ * @param filename 'Document' (File or DicomDir) to be opened for parsing
+ */
+void Document::Load( std::string const &filename )
+{
+ // We should clean out anything that already exists.
+ // Check IsDocumentAlreadyLoaded to be sure.
+ if( IsDocumentAlreadyLoaded )
+ {
+ gdcmWarningMacro( "A file was already parsed inside this " <<
+ "gdcm::Document (previous name was: "
+ << Filename.c_str() << ". New name is :"
+ << filename );
+ // todo : clean out the 'Document'
+ // We should call ClearEntry() on the parent object ?!?
+ }
+
+ Filename = filename;
Fp = 0;
if ( !OpenFile() )
{
+ // warning already performed in OpenFile()
+ //gdcmWarningMacro( "Unable to open as an ACR/DICOM file: "
+ // << Filename.c_str() );
+ Filetype = Unknown;
return;
}
Group0002Parsed = false;
- gdcmVerboseMacro( "Starting parsing of file: " << Filename.c_str());
- // Fp->seekg( 0, std::ios::beg);
-
- Fp->seekg(0, std::ios::end);
- long lgt = Fp->tellg();
-
- Fp->seekg( 0, std::ios::beg);
-
- CheckSwap();
- long beg = Fp->tellg();
- lgt -= beg;
-
+ gdcmWarningMacro( "Starting parsing of file: " << Filename.c_str());
+
+ Fp->seekg(0, std::ios::end);
+ long lgt = Fp->tellg(); // total length of the file
+
+ Fp->seekg(0, std::ios::beg);
+
+ // CheckSwap returns a boolean
+ // (false if no swap info of any kind was found)
+ if (! CheckSwap() )
+ {
+ gdcmWarningMacro( "Neither a DICOM V3 nor an ACR-NEMA file: "
+ << Filename.c_str());
+ CloseFile();
+ return ;
+ }
+
+ long beg = Fp->tellg(); // just after DICOM preamble (if any)
+
+ lgt -= beg; // remaining length to parse
+
ParseDES( this, beg, lgt, false); // Loading is done during parsing
- Fp->seekg( 0, std::ios::beg);
+ if ( IsEmpty() )
+ {
+ gdcmWarningMacro( "No tag in internal hash table for: "
+ << Filename.c_str());
+ CloseFile();
+ return ;
+ }
+ IsDocumentAlreadyLoaded = true;
+
+ Fp->seekg( 0, std::ios::beg);
// Load 'non string' values
CloseFile();
- // --------------------------------------------------------------
+ // ----------------------------
// Specific code to allow gdcm to read ACR-LibIDO formated images
// Note: ACR-LibIDO is an extension of the ACR standard that was
// used at CREATIS. For the time being (say a couple years)
- // we keep this kludge to allow a smooth move to gdcm for
- // CREATIS developpers (sorry folks).
+ // we keep this kludge to allow CREATIS users
+ // reading their old images.
//
// if recognition code tells us we deal with a LibIDO image
// we switch lineNumber and columnNumber
SetValEntry(columns, 0x0028, 0x0010);
SetValEntry(rows , 0x0028, 0x0011);
}
- // ----------------- End of ACR-LibIDO kludge ------------------
-}
-
-/**
- * \brief This default constructor doesn't parse the file. You should
- * then invoke \ref Document::SetFileName and then the parsing.
- */
-Document::Document() : ElementSet(-1)
-{
- RLEInfo = NULL;
- JPEGInfo = NULL;
-
- Fp = 0;
-
- SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
- Initialize();
- SwapCode = 1234;
- Filetype = ExplicitVR;
- Group0002Parsed = false;
-}
-
-/**
- * \brief Canonical destructor.
- */
-Document::~Document ()
-{
- RefPubDict = NULL;
- RefShaDict = NULL;
-
- if( RLEInfo )
- delete RLEInfo;
- if( JPEGInfo )
- delete JPEGInfo;
-}
-
-//-----------------------------------------------------------------------------
-// Print
-
-/**
- * \brief Prints The Dict Entries of THE public Dicom Dictionary
- * @param os ostream to print to
- * @return
- */
-void Document::PrintPubDict(std::ostream &os)
-{
- RefPubDict->SetPrintLevel(PrintLevel);
- RefPubDict->Print(os);
-}
-
-/**
- * \brief Prints The Dict Entries of THE shadow Dicom Dictionary
- * @param os ostream to print to
- * @return
- */
-void Document::PrintShaDict(std::ostream &os)
-{
- RefShaDict->SetPrintLevel(PrintLevel);
- RefShaDict->Print(os);
+ // --- End of ACR-LibIDO kludge ---
}
-//-----------------------------------------------------------------------------
-// Public
/**
* \brief Get the public dictionary used
*/
}
/**
- * \brief This predicate, based on hopefully reasonable heuristics,
- * decides whether or not the current Document was properly parsed
- * and contains the mandatory information for being considered as
- * a well formed and usable Dicom/Acr File.
- * @return true when Document is the one of a reasonable Dicom/Acr file,
- * false otherwise.
+ * \brief This predicate tells us whether or not the current Document
+ * was properly parsed and contains at least *one* Dicom Element
+ * (and nothing more, sorry).
+ * @return false when we're 150 % sure it's NOT a Dicom/Acr file,
+ * true otherwise.
*/
bool Document::IsReadable()
{
if( Filetype == Unknown)
{
- gdcmVerboseMacro( "Wrong filetype");
+ gdcmWarningMacro( "Wrong filetype");
return false;
}
if ( IsEmpty() )
{
- gdcmVerboseMacro( "No tag in internal hash table.");
+ gdcmWarningMacro( "No tag in internal hash table.");
return false;
}
return true;
}
+/**
+ * \brief Predicate for dicom version 3 file.
+ * @return True when the file is a dicom version 3.
+ */
+bool Document::IsDicomV3()
+{
+ // Checking if Transfer Syntax exists is enough
+ // Anyway, it's too late check if the 'Preamble' was found ...
+ // And ... would it be a rich idea to check ?
+ // (some 'no Preamble' DICOM images exist !)
+ return GetDocEntry(0x0002, 0x0010) != NULL;
+}
+
+/**
+ * \brief Predicate for Papyrus file
+ * Dedicated to whomsoever it may concern
+ * @return True when the file is a Papyrus file.
+ */
+bool Document::IsPapyrus()
+{
+ // check for Papyrus private Sequence
+ DocEntry *e = GetDocEntry(0x0041, 0x1050);
+ if ( !e )
+ return false;
+ // check if it's actually a Sequence
+ if ( !dynamic_cast<SeqEntry*>(e) )
+ return false;
+ return true;
+}
+
+/**
+ * \brief returns the File Type
+ * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
+ * @return the FileType code
+ */
+FileType Document::GetFileType()
+{
+ return Filetype;
+}
+
/**
* \brief Accessor to the Transfer Syntax (when present) of the
* current document (it internally handles reading the
}
/**
- * \brief Predicate for dicom version 3 file.
- * @return True when the file is a dicom version 3.
+ * \brief Accesses the info from 0002,0010 : Transfer Syntax and TS
+ * @return The full Transfer Syntax Name (as opposed to Transfer Syntax UID)
*/
-bool Document::IsDicomV3()
+std::string Document::GetTransferSyntaxName()
{
- // Checking if Transfer Syntax exists is enough
- // Anyway, it's to late check if the 'Preamble' was found ...
- // And ... would it be a rich idea to check ?
- // (some 'no Preamble' DICOM images exist !)
- return GetDocEntry(0x0002, 0x0010) != NULL;
-}
+ // use the TS (TS : Transfer Syntax)
+ std::string transferSyntax = GetEntryValue(0x0002,0x0010);
+
+ if ( (transferSyntax.find(GDCM_NOTLOADED) < transferSyntax.length()) )
+ {
+ gdcmErrorMacro( "Transfer Syntax not loaded. " << std::endl
+ << "Better you increase MAX_SIZE_LOAD_ELEMENT_VALUE" );
+ return "Uncompressed ACR-NEMA";
+ }
+ if ( transferSyntax == GDCM_UNFOUND )
+ {
+ gdcmWarningMacro( "Unfound Transfer Syntax (0002,0010)");
+ return "Uncompressed ACR-NEMA";
+ }
+
+ // we do it only when we need it
+ const TSKey &tsName = Global::GetTS()->GetValue( transferSyntax );
+ // Global::GetTS() is a global static you shall never try to delete it!
+ return tsName;
+}
+//
+// --------------- Swap Code ------------------
/**
- * \brief Predicate for Papyrus file
- * Dedicated to whomsoever it may concern
- * @return True when the file is a Papyrus file.
+ * \brief Swaps the bytes so they agree with the processor order
+ * @return The properly swaped 16 bits integer.
*/
-bool Document::IsPapyrus()
+uint16_t Document::SwapShort(uint16_t a)
{
- // check for Papyrus private Sequence
- DocEntry *e = GetDocEntry(0x0041, 0x1050);
- if ( !e )
- return false;
- // check if it's actually a Sequence
- if ( !dynamic_cast<SeqEntry*>(e) )
- return false;
- return true;
+ if ( SwapCode == 4321 || SwapCode == 2143 )
+ {
+ //a = ((( a << 8 ) & 0xff00 ) | (( a >> 8 ) & 0x00ff ) );
+ // Save CPU time
+ a = ( a << 8 ) | ( a >> 8 );
+ }
+ return a;
}
/**
- * \brief returns the File Type
- * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
- * @return the FileType code
+ * \brief Swaps back the bytes of 4-byte long integer accordingly to
+ * processor order.
+ * @return The properly swaped 32 bits integer.
*/
-FileType Document::GetFileType()
+uint32_t Document::SwapLong(uint32_t a)
{
- return Filetype;
-}
+ switch (SwapCode)
+ {
+ case 1234 :
+ break;
+ case 4321 :
+// a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) |
+// ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) );
+// save CPU time
+ a=( ( a<<24) | ((a<<8) & 0x00ff0000) |
+ ((a>>8) & 0x0000ff00) | (a>>24) );
+ break;
+ case 3412 :
+// a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
+ a=( (a<<16) | (a>>16) );
+ break;
+ case 2143 :
+ a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
+ break;
+ default :
+ gdcmErrorMacro( "Unset swap code:" << SwapCode );
+ a = 0;
+ }
+ return a;
+}
+//
+// -----------------File I/O ---------------
/**
* \brief Tries to open the file \ref Document::Filename and
* checks the preamble when existing.
*/
std::ifstream *Document::OpenFile()
{
-
HasDCMPreamble = false;
if (Filename.length() == 0)
{
if(Fp)
{
- gdcmVerboseMacro( "File already open: " << Filename.c_str());
+ gdcmWarningMacro( "File already open: " << Filename.c_str());
+ CloseFile();
}
Fp = new std::ifstream(Filename.c_str(), std::ios::in | std::ios::binary);
if( ! *Fp )
{
- gdcmDebugMacro( "Cannot open file: " << Filename.c_str());
+ gdcmErrorMacro( "Cannot open file: " << Filename.c_str());
delete Fp;
Fp = 0;
return 0;
+ //exit(1); // No function is allowed to leave the application instead
+ // of warning the caller
}
- uint16_t zero;
+ uint16_t zero = 0;
Fp->read((char*)&zero, (size_t)2);
if( Fp->eof() )
{
return 0;
}
- //ACR -- or DICOM with no Preamble; may start with a Shadow Group --
+ //-- ACR or DICOM with no Preamble; may start with a Shadow Group --
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 with no preamble: (%04x)\n", zero);
- gdcmVerboseMacro( msg.c_str() );
+ std::string msg = Util::Format(
+ "ACR/DICOM starting at the begining of the file:(%04x)\n", zero);
+ gdcmWarningMacro( msg.c_str() );
return Fp;
}
- //DICOM
+ //-- DICOM --
Fp->seekg(126L, std::ios::cur);
- char dicm[4];
+ char dicm[4] = {' ',' ',' ',' '};
Fp->read(dicm, (size_t)4);
if( Fp->eof() )
{
HasDCMPreamble = true;
return Fp;
}
-
+
+ // -- Neither ACR/No Preamble Dicom nor DICOMV3 file
CloseFile();
- gdcmVerboseMacro( "Not DICOM/ACR (missing preamble)" << Filename.c_str());
-
+ gdcmWarningMacro( "Neither ACR/No Preamble Dicom nor DICOMV3 file: "
+ << Filename.c_str());
return 0;
}
delete Fp;
Fp = 0;
}
- return true; //FIXME how do we detect a non-closed ifstream ?
+ return true;
}
/**
fp->write("DICM", 4);
}
-/*
- * \todo rewrite later, if really usefull
- * - 'Group Length' element is optional in DICOM
- * - but un-updated odd groups lengthes can causes pb
- * (xmedcon breaker)
- *
- * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) )
- * UpdateGroupLength(false,filetype);
- * if ( filetype == ACR)
- * UpdateGroupLength(true,ACR);
- */
-
+ /*
+ * \todo rewrite later, if really usefull
+ * - 'Group Length' element is optional in DICOM
+ * - but un-updated odd groups lengthes can causes pb
+ * (xmedcon breaker)
+ *
+ * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) )
+ * UpdateGroupLength(false,filetype);
+ * if ( filetype == ACR)
+ * UpdateGroupLength(true,ACR);
+ */
+
ElementSet::WriteContent(fp, filetype); // This one is recursive
}
-//-----------------------------------------------------------------------------
-// Protected
+// -----------------------------------------
+// Content entries
/**
* \brief Loads (from disk) the element content
* when a string is not suitable
uint8_t *a = new uint8_t[l];
if( !a )
{
- gdcmVerboseMacro( "Cannot allocate BinEntry content");
+ gdcmWarningMacro( "Cannot allocate BinEntry content");
return;
}
- /// \todo check the result
Fp->read((char*)a, l);
if( Fp->fail() || Fp->eof())
{
CloseFile();
}
-/**
- * \brief Sets a 'non string' value to a given Dicom Element
- * @param area area containing the 'non string' value
- * @param group Group number of the searched Dicom Element
- * @param elem Element number of the searched Dicom Element
- * @return
- */
-/*bool Document::SetEntryBinArea(uint8_t *area,
- uint16_t group, uint16_t elem)
-{
- DocEntry *currentEntry = GetDocEntry(group, elem);
- if ( !currentEntry )
- {
- return false;
- }
-
- if ( BinEntry *binEntry = dynamic_cast<BinEntry*>(currentEntry) )
- {
- binEntry->SetBinArea( area );
- return true;
- }
-
- return false;
-}*/
-
-
/**
* \brief Loads the element while preserving the current
* underlying file position indicator as opposed to
* LoadDocEntry that modifies it.
- * @param entry Header Entry whose value will be loaded.
- * @return
+ * @param entry DocEntry whose value will be loaded.
*/
void Document::LoadDocEntrySafe(DocEntry *entry)
{
}
/**
- * \brief Swaps back the bytes of 4-byte long integer accordingly to
- * processor order.
- * @return The properly swaped 32 bits integer.
+ * \brief Compares two documents, according to \ref 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
+ * @return true if 'smaller'
*/
-uint32_t Document::SwapLong(uint32_t a)
+bool Document::operator<(Document &document)
{
- switch (SwapCode)
+ // Patient Name
+ std::string s1 = GetEntryValue(0x0010,0x0010);
+ std::string s2 = document.GetEntryValue(0x0010,0x0010);
+ if(s1 < s2)
{
- case 1234 :
- 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 :
- gdcmErrorMacro( "Unset swap code:" << SwapCode );
- a = 0;
+ return true;
}
- return a;
-}
+ else if( s1 > s2 )
+ {
+ return false;
+ }
+ else
+ {
+ // Patient ID
+ s1 = GetEntryValue(0x0010,0x0020);
+ s2 = document.GetEntryValue(0x0010,0x0020);
+ if ( s1 < s2 )
+ {
+ return true;
+ }
+ else if ( s1 > s2 )
+ {
+ return false;
+ }
+ else
+ {
+ // Study Instance UID
+ s1 = GetEntryValue(0x0020,0x000d);
+ s2 = document.GetEntryValue(0x0020,0x000d);
+ if ( s1 < s2 )
+ {
+ return true;
+ }
+ else if( s1 > s2 )
+ {
+ return false;
+ }
+ else
+ {
+ // Serie Instance UID
+ s1 = GetEntryValue(0x0020,0x000e);
+ s2 = document.GetEntryValue(0x0020,0x000e);
+ if ( s1 < s2 )
+ {
+ return true;
+ }
+ else if( s1 > s2 )
+ {
+ return false;
+ }
+ }
+ }
+ }
+ return false;
+}
+//-----------------------------------------------------------------------------
+// Protected
/**
- * \brief Unswaps back the bytes of 4-byte long integer accordingly to
- * processor order.
- * @return The properly unswaped 32 bits integer.
+ * \brief Reads a supposed to be 16 Bits integer
+ * (swaps it depending on processor endianness)
+ * @return read value
*/
-uint32_t Document::UnswapLong(uint32_t a)
+uint16_t Document::ReadInt16()
+ throw( FormatError )
{
- return SwapLong(a);
+ uint16_t g;
+ Fp->read ((char*)&g, (size_t)2);
+ if ( Fp->fail() )
+ {
+ throw FormatError( "Document::ReadInt16()", " file error." );
+ }
+ if( Fp->eof() )
+ {
+ throw FormatError( "Document::ReadInt16()", "EOF." );
+ }
+ g = SwapShort(g);
+ return g;
}
/**
- * \brief Swaps the bytes so they agree with the processor order
- * @return The properly swaped 16 bits integer.
+ * \brief Reads a supposed to be 32 Bits integer
+ * (swaps it depending on processor endianness)
+ * @return read value
*/
-uint16_t Document::SwapShort(uint16_t a)
+uint32_t Document::ReadInt32()
+ throw( FormatError )
{
- if ( SwapCode == 4321 || SwapCode == 2143 )
+ uint32_t g;
+ Fp->read ((char*)&g, (size_t)4);
+ if ( Fp->fail() )
{
- a = ((( a << 8 ) & 0x0ff00 ) | (( a >> 8 ) & 0x00ff ) );
+ throw FormatError( "Document::ReadInt32()", " file error." );
}
- return a;
+ if( Fp->eof() )
+ {
+ throw FormatError( "Document::ReadInt32()", "EOF." );
+ }
+ g = SwapLong(g);
+ return g;
}
/**
- * \brief Unswaps the bytes so they agree with the processor order
- * @return The properly unswaped 16 bits integer.
+ * \brief skips bytes inside the source file
+ * \warning NOT end user intended method !
+ * @return
+ */
+void Document::SkipBytes(uint32_t nBytes)
+{
+ //FIXME don't dump the returned value
+ Fp->seekg((long)nBytes, std::ios::cur);
+}
+
+/**
+ * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader
+ * @param filetype Type of the File to be written
*/
-uint16_t Document::UnswapShort(uint16_t a)
+int Document::ComputeGroup0002Length( FileType filetype )
{
- return SwapShort(a);
+ uint16_t gr;
+ std::string vr;
+
+ int groupLength = 0;
+ bool found0002 = false;
+
+ // for each zero-level Tag in the DCM Header
+ DocEntry *entry = GetFirstEntry();
+ while( entry )
+ {
+ gr = entry->GetGroup();
+
+ if( gr == 0x0002 )
+ {
+ found0002 = true;
+
+ if( entry->GetElement() != 0x0000 )
+ {
+ vr = entry->GetVR();
+
+ if( filetype == ExplicitVR )
+ {
+ if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") )
+ {
+ // explicit VR AND OB, OW, SQ : 4 more bytes
+ groupLength += 4;
+ }
+ }
+ groupLength += 2 + 2 + 4 + entry->GetLength();
+ }
+ }
+ else if (found0002 )
+ break;
+
+ entry = GetNextEntry();
+ }
+ return groupLength;
}
//-----------------------------------------------------------------------------
// Private
+/**
+ * \brief Loads all the needed Dictionaries
+ * \warning NOT end user intended method !
+ */
+void Document::Initialize()
+{
+ RefPubDict = Global::GetDicts()->GetDefaultPubDict();
+ RefShaDict = NULL;
+ Filetype = Unknown;
+}
/**
* \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries)
- * @return length of the parsed set.
+ * @param set DocEntrySet we are going to parse ('zero level' or a SQItem)
+ * @param offset start of parsing
+ * @param l_max length to parse
+ * @param delim_mode : whether we are in 'delimitor mode' (l=0xffffff) or not
*/
void Document::ParseDES(DocEntrySet *set, long offset,
long l_max, bool delim_mode)
{
- DocEntry *newDocEntry = 0;
+ DocEntry *newDocEntry;
ValEntry *newValEntry;
BinEntry *newBinEntry;
SeqEntry *newSeqEntry;
VRKey vr;
- bool used = false;
+ bool used;
while (true)
{
!Global::GetVR()->IsVROfBinaryRepresentable(vr) )
{
////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR
- gdcmVerboseMacro( std::hex << newDocEntry->GetGroup()
+ gdcmWarningMacro( std::hex << newDocEntry->GetGroup()
<< "|" << newDocEntry->GetElement()
<< " : Neither Valentry, nor BinEntry."
"Probably unknown VR.");
else
{
/////////////////////// ValEntry
+
// When "set" is a Document, then we are at the top of the
// hierarchy and the Key is simply of the form ( group, elem )...
if ( dynamic_cast< Document* > ( set ) )
LoadDocEntry( newValEntry );
bool delimitor=newValEntry->IsItemDelimitor();
+
+ if ( LoadMode & NO_SHADOW ) // User asked to skip, if possible,
+ // shadow groups ( if possible :
+ // whether element 0x0000 exits)
+ {
+ if ( newValEntry->GetGroup()%2 != 0 )
+ {
+ if ( newValEntry->GetElement() == 0x0000 )
+ {
+ std::string strLgrGroup = newValEntry->GetValue();
+ int lgrGroup;
+ if ( strLgrGroup != GDCM_UNFOUND)
+ {
+ lgrGroup = atoi(strLgrGroup.c_str());
+ Fp->seekg(lgrGroup , std::ios::cur);
+ used = false;
+ continue;
+ }
+ }
+ }
+ }
+
if( !set->AddEntry( newValEntry ) )
{
// If here expect big troubles
- //delete newValEntry; //otherwise mem leak
+ // delete newValEntry; //otherwise mem leak
used=false;
}
}
}
- if ( ( newDocEntry->GetGroup() == 0x7fe0 )
- && ( newDocEntry->GetElement() == 0x0010 ) )
- {
- std::string ts = GetTransferSyntax();
- if ( Global::GetTS()->IsRLELossless(ts) )
- {
- long positionOnEntry = Fp->tellg();
- Fp->seekg( newDocEntry->GetOffset(), std::ios::beg );
- ComputeRLEInfo();
- Fp->seekg( positionOnEntry, std::ios::beg );
- }
- else if ( Global::GetTS()->IsJPEG(ts) )
- {
- long positionOnEntry = Fp->tellg();
- Fp->seekg( newDocEntry->GetOffset(), std::ios::beg );
- ComputeJPEGFragmentInfo();
- Fp->seekg( positionOnEntry, std::ios::beg );
- }
- }
-
// Just to make sure we are at the beginning of next entry.
SkipToNextDocEntry(newDocEntry);
}
else
{
// 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 )
delim_mode = false;
}
}
+
+ if ( (LoadMode & NO_SEQ) && ! delim_mode ) // User asked to skip SQ
+ {
+ Fp->seekg( l, std::ios::cur);
+ used = false;
+ continue;
+ }
+
// no other way to create it ...
newSeqEntry->SetDelimitorMode( delim_mode );
// is a Document, then we are building the first depth level.
// Hence the SeqEntry we are building simply has a depth
// level of one:
- if (/*Document *dummy =*/ dynamic_cast< Document* > ( set ) )
+ if ( dynamic_cast< Document* > ( set ) )
{
- //(void)dummy;
newSeqEntry->SetDepthLevel( 1 );
newSeqEntry->SetKey( newSeqEntry->GetKey() );
}
{
used = false;
}
+
if ( !delim_mode && ((long)(Fp->tellg())-offset) >= l_max)
{
if( !used )
{
dlm_mod = false;
}
- // FIXME, TODO
- // when we're here, element fffe,e000 is already passed.
- // it's lost for the SQItem we're going to process !!
-
- //ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod);
- //delete newDocEntry; // FIXME well ... it's too late to use it !
// Let's try :------------
// remove fff0,e000, created out of the SQItem
}
/**
- * \brief Loads the element content if its length doesn't exceed
- * the value specified with Document::SetMaxSizeLoadEntry()
- * @param entry Header Entry (Dicom Element) to be dealt with
+ * \brief Loads the element content if its length doesn't exceed
+ * the value specified with Document::SetMaxSizeLoadEntry()
+ * @param entry Header Entry (Dicom Element) to be dealt with
*/
void Document::LoadDocEntry(DocEntry *entry)
{
{
//s << "gdcm::NotLoaded (BinEntry)";
s << GDCM_NOTLOADED;
- s << " Address:" << (long)entry->GetOffset();
- s << " Length:" << entry->GetLength();
+ s << " Ad.:" << (long)entry->GetOffset();
+ s << " x(" << std::hex << entry->GetOffset() << ")";
+ s << std::dec;
+ s << " Lgt:" << entry->GetLength();
s << " x(" << std::hex << entry->GetLength() << ")";
binEntryPtr->SetValue(s.str());
}
return;
}
- /// \todo Any compacter code suggested (?)
if ( IsDocEntryAnInteger(entry) )
{
uint32_t NewInt;
if( length % 2 )
{
newValue = Util::DicomString(str, length+1);
- gdcmVerboseMacro("Warning: bad length: " << length <<
+ gdcmWarningMacro("Warning: bad length: " << length <<
",For string :" << newValue.c_str());
// Since we change the length of string update it length
//entry->SetReadLength(length+1);
{
if ( Fp->fail() || Fp->eof())
{
- gdcmVerboseMacro("Unread element value");
+ gdcmWarningMacro("Unread element value");
valEntry->SetValue(GDCM_UNREAD);
return;
}
}
}
-
/**
* \brief Find the value Length of the passed Header Entry
* @param entry Header Entry whose length of the value shall be loaded.
// Length is encoded on 2 bytes.
length16 = ReadInt16();
-
- // FIXME : This heuristic supposes that the first group following
- // group 0002 *has* and element 0000.
- // BUT ... Element 0000 is optionnal :-(
-
-
- // Fixed using : HandleOutOfGroup0002()
- // (first hereafter strategy ...)
-
- // 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
- // encounter 1024 (0x0400) chances are the encoding changed and we
- // found a group with big endian encoding.
- //---> Unfortunately, element 0000 is optional.
- //---> This will not work when missing!
- // 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 already 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 == 0x0000 && length16 == 0x0400 )
-// {
-// std::string ts = GetTransferSyntax();
-// if ( Global::GetTS()->GetSpecialTransferSyntax(ts)
-// != TS::ExplicitVRBigEndian )
-// {
-// throw FormatError( "Document::FindDocEntryLength()",
-// " not explicit VR." );
-// return;
-// }
-// length16 = 4;
-// SwitchByteSwapCode();
-//
-// // Restore the unproperly loaded values i.e. the group, the element
-// // and the dictionary entry depending on them.
-// uint16_t correctGroup = SwapShort( entry->GetGroup() );
-// uint16_t correctElem = SwapShort( entry->GetElement() );
-// DictEntry *newTag = GetDictEntry( correctGroup, correctElem );
-// if ( !newTag )
-// {
-// // This correct tag is not in the dictionary. Create a new one.
-// newTag = NewVirtualDictEntry(correctGroup, correctElem);
-// }
-// // FIXME this can create a memory leaks on the old entry that be
-// // left unreferenced.
-// entry->SetDictEntry( newTag );
-// }
// 0xffff means that we deal with 'No Length' Sequence
// or 'No Length' SQItem
}
}
+/**
+ * \brief Find the Length till the next sequence delimiter
+ * \warning NOT end user intended method !
+ * @return
+ */
+uint32_t Document::FindDocEntryLengthOBOrOW()
+ throw( FormatUnexpected )
+{
+ // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
+ long positionOnEntry = Fp->tellg();
+ bool foundSequenceDelimiter = false;
+ uint32_t totalLength = 0;
+
+ while ( !foundSequenceDelimiter )
+ {
+ uint16_t group;
+ uint16_t elem;
+ try
+ {
+ group = ReadInt16();
+ elem = ReadInt16();
+ }
+ catch ( FormatError )
+ {
+ throw FormatError("Unexpected end of file encountered during ",
+ "Document::FindDocEntryLengthOBOrOW()");
+ }
+ // We have to decount the group and element we just read
+ totalLength += 4;
+ if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) )
+ {
+ long filePosition = Fp->tellg();
+ gdcmWarningMacro(
+ "Neither an Item tag nor a Sequence delimiter tag on :"
+ << std::hex << group << " , " << elem
+ << ") -before- position x(" << filePosition << ")" );
+
+ Fp->seekg(positionOnEntry, std::ios::beg);
+ throw FormatUnexpected(
+ "Neither an Item tag nor a Sequence delimiter tag.");
+ }
+ if ( elem == 0xe0dd )
+ {
+ foundSequenceDelimiter = true;
+ }
+ uint32_t itemLength = ReadInt32();
+ // We add 4 bytes since we just read the ItemLength with ReadInt32
+ totalLength += itemLength + 4;
+ SkipBytes(itemLength);
+
+ if ( foundSequenceDelimiter )
+ {
+ break;
+ }
+ }
+ Fp->seekg( positionOnEntry, std::ios::beg);
+ return totalLength;
+}
+
/**
* \brief Find the Value Representation of the current Dicom Element.
* @return Value Representation of the current Entry
*/
bool Document::CheckDocEntryVR(VRKey vr)
{
- // 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 ( !Global::GetVR()->IsValidVR(vr) )
return false;
#endif //GDCM_NO_ANSI_STRING_STREAM
return s.str();
}
-
return ((ValEntry *)entry)->GetValue();
}
}
/**
- * \brief Skips to the begining of the next Header Entry
+ * \brief Skips to the beginning of the next Header Entry
* \warning NOT end user intended method !
* @param currentDocEntry entry to skip
*/
if ( foundLength % 2)
{
- gdcmVerboseMacro( "Warning : Tag with uneven length " << foundLength
+ gdcmWarningMacro( "Warning : Tag with uneven length " << foundLength
<< " in x(" << std::hex << gr << "," << elem <<")");
}
if ( gr != 0x0008 || ( elem != 0x0070 && elem != 0x0080 ) )
{
foundLength = 10;
- entry->SetReadLength(10); /// \todo a bug is to be fixed !?
+ entry->SetReadLength(10); // a bug is to be fixed !?
}
}
else if ( gr == 0x0009 && ( elem == 0x1113 || elem == 0x1114 ) )
{
foundLength = 4;
- entry->SetReadLength(4); /// \todo a bug is to be fixed !?
+ entry->SetReadLength(4); // a bug is to be fixed !?
}
else if ( entry->GetVR() == "SQ" )
{
foundLength = 0;
}
- }
-
+ }
entry->SetLength(foundLength);
}
*/
bool Document::IsDocEntryAnInteger(DocEntry *entry)
{
- uint16_t elem = entry->GetElement();
- uint16_t group = entry->GetGroup();
- const std::string &vr = entry->GetVR();
- uint32_t length = entry->GetLength();
+ uint16_t elem = entry->GetElement();
+ uint16_t group = entry->GetGroup();
+ const std::string &vr = entry->GetVR();
+ uint32_t length = entry->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
// encounter such an ill-formed image, we simply display a warning
// message and proceed on parsing (while crossing fingers).
long filePosition = Fp->tellg();
- gdcmVerboseMacro( "Erroneous Group Length element length on : ("
+ gdcmWarningMacro( "Erroneous Group Length element length on : ("
<< std::hex << group << " , " << elem
<< ") -before- position x(" << filePosition << ")"
<< "lgt : " << length );
}
/**
- * \brief Find the Length till the next sequence delimiter
- * \warning NOT end user intended method !
- * @return
+ * \brief Discover what the swap code is (among little endian, big endian,
+ * bad little endian, bad big endian).
+ * sw is set
+ * @return false when we are absolutely sure
+ * it's neither ACR-NEMA nor DICOM
+ * true when we hope ours assuptions are OK
*/
-
-uint32_t Document::FindDocEntryLengthOBOrOW()
- throw( FormatUnexpected )
-{
- // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
- long positionOnEntry = Fp->tellg();
- bool foundSequenceDelimiter = false;
- uint32_t totalLength = 0;
-
- while ( !foundSequenceDelimiter )
- {
- uint16_t group;
- uint16_t elem;
- try
- {
- group = ReadInt16();
- elem = ReadInt16();
- }
- catch ( FormatError )
- {
- throw FormatError("Unexpected end of file encountered during ",
- "Document::FindDocEntryLengthOBOrOW()");
- }
-
- // We have to decount the group and element we just read
- totalLength += 4;
-
- if ( group != 0xfffe || ( ( elem != 0xe0dd ) && ( elem != 0xe000 ) ) )
- {
- long filePosition = Fp->tellg();
- gdcmVerboseMacro( "Neither an Item tag nor a Sequence delimiter tag on :"
- << std::hex << group << " , " << elem
- << ") -before- position x(" << filePosition << ")" );
-
- Fp->seekg(positionOnEntry, std::ios::beg);
- throw FormatUnexpected( "Neither an Item tag nor a Sequence delimiter tag.");
- }
-
- if ( elem == 0xe0dd )
- {
- foundSequenceDelimiter = true;
- }
-
- uint32_t itemLength = ReadInt32();
- // We add 4 bytes since we just read the ItemLength with ReadInt32
- totalLength += itemLength + 4;
- SkipBytes(itemLength);
-
- if ( foundSequenceDelimiter )
- {
- break;
- }
- }
- Fp->seekg( positionOnEntry, std::ios::beg);
- return totalLength;
-}
-
-/**
- * \brief Reads a supposed to be 16 Bits integer
- * (swaps it depending on processor endianity)
- * @return read value
- */
-uint16_t Document::ReadInt16()
- throw( FormatError )
-{
- uint16_t g;
- Fp->read ((char*)&g, (size_t)2);
- if ( Fp->fail() )
- {
- throw FormatError( "Document::ReadInt16()", " file error." );
- }
- if( Fp->eof() )
- {
- throw FormatError( "Document::ReadInt16()", "EOF." );
- }
- g = SwapShort(g);
- return g;
-}
-
-/**
- * \brief Reads a supposed to be 32 Bits integer
- * (swaps it depending on processor endianity)
- * @return read value
- */
-uint32_t Document::ReadInt32()
- throw( FormatError )
-{
- uint32_t g;
- Fp->read ((char*)&g, (size_t)4);
- if ( Fp->fail() )
- {
- throw FormatError( "Document::ReadInt32()", " file error." );
- }
- if( Fp->eof() )
- {
- throw FormatError( "Document::ReadInt32()", "EOF." );
- }
- g = SwapLong(g);
- return g;
-}
-
-/**
- * \brief skips bytes inside the source file
- * \warning NOT end user intended method !
- * @return
- */
-void Document::SkipBytes(uint32_t nBytes)
-{
- //FIXME don't dump the returned value
- Fp->seekg((long)nBytes, std::ios::cur);
-}
-
-/**
- * \brief Loads all the needed Dictionaries
- * \warning NOT end user intended method !
- */
-void Document::Initialize()
-{
- RefPubDict = Global::GetDicts()->GetDefaultPubDict();
- RefShaDict = NULL;
- RLEInfo = new RLEFramesInfo;
- JPEGInfo = new JPEGFragmentsInfo;
- Filetype = Unknown;
-}
-
-/**
- * \brief Discover what the swap code is (among little endian, big endian,
- * bad little endian, bad big endian).
- * sw is set
- * @return false when we are absolutely sure
- * it's neither ACR-NEMA nor DICOM
- * true when we hope ours assuptions are OK
- */
-bool Document::CheckSwap()
+bool Document::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...
- uint32_t x = 4; // x : for ntohs
- bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
uint32_t s32;
uint16_t s16;
// 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;
- }
+ bool net2host = Util::IsCurrentProcessorBigEndian();
// The easiest case is the one of a 'true' DICOM header, we just have
// to look for the string "DICM" inside the file preamble.
char *entCur = deb + 128;
if( memcmp(entCur, "DICM", (size_t)4) == 0 )
{
- gdcmVerboseMacro( "Looks like DICOM Version3 (preamble + DCM)" );
+ gdcmWarningMacro( "Looks like DICOM Version3 (preamble + DCM)" );
// Group 0002 should always be VR, and the first element 0000
// Let's be carefull (so many wrong headers ...)
// instead of just checking for UL, OB and UI !? group 0000
{
Filetype = ExplicitVR;
- gdcmVerboseMacro( "Group 0002 : Explicit Value Representation");
+ gdcmWarningMacro( "Group 0002 : Explicit Value Representation");
}
else
{
Filetype = ImplicitVR;
- gdcmVerboseMacro( "Group 0002 :Not an explicit Value Representation;"
+ gdcmWarningMacro( "Group 0002 :Not an explicit Value Representation;"
<< "Looks like a bugged Header!");
}
if ( net2host )
{
SwapCode = 4321;
- gdcmVerboseMacro( "HostByteOrder != NetworkByteOrder");
+ gdcmWarningMacro( "HostByteOrder != NetworkByteOrder");
}
else
{
SwapCode = 1234;
- gdcmVerboseMacro( "HostByteOrder = NetworkByteOrder");
+ gdcmWarningMacro( "HostByteOrder = NetworkByteOrder");
}
// Position the file position indicator at first tag
Fp->seekg(0, std::ios::beg);
Fp->seekg ( 132L, std::ios::beg);
return true;
- } // End of DicomV3
+ } // ------------------------------- End of DicomV3 ----------------
// 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).
- gdcmVerboseMacro( "Not a DICOM Version3 file");
+
+ gdcmWarningMacro( "Not a DICOM Version3 file");
+
Fp->seekg(0, std::ios::beg);
// 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.
+ // By clean we mean that the length of the first group 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.
// representation of a 32 bits integer. Hence the following dirty
// trick :
s32 = *((uint32_t *)(entCur));
-
switch( s32 )
{
case 0x00040000 :
Filetype = ACR;
return true;
default :
- gdcmVerboseMacro( "ACR/NEMA unfound swap info (Really hopeless !)");
+ gdcmWarningMacro( "ACR/NEMA unfound swap info (Really hopeless !)");
Filetype = Unknown;
return false;
}
*/
void Document::SwitchByteSwapCode()
{
- gdcmVerboseMacro( "Switching Byte Swap code from "<< SwapCode);
+ gdcmWarningMacro( "Switching Byte Swap code from "<< SwapCode);
if ( SwapCode == 1234 )
{
SwapCode = 4321;
/**
* \brief during parsing, Header Elements too long are not loaded in memory
- * @param newSize
+ * @param newSize new size
*/
void Document::SetMaxSizeLoadEntry(long newSize)
{
MaxSizeLoadEntry = newSize;
}
-
-/**
- * \brief Header Elements too long will not be printed
- * \todo See comments of \ref Document::MAX_SIZE_PRINT_ELEMENT_VALUE
- * @param newSize
- */
-void Document::SetMaxSizePrintEntry(long newSize)
-{
- if ( newSize < 0 )
- {
- return;
- }
- if ((uint32_t)newSize >= (uint32_t)0xffffffff )
- {
- MaxSizePrintEntry = 0xffffffff;
- return;
- }
- MaxSizePrintEntry = newSize;
-}
-
-
-
-/**
- * \brief Handle broken private tag from Philips NTSCAN
- * where the endianess is being switch to BigEndian for no
- * apparent reason
- * @return no return
- */
-void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem)
-{
- // Endian reversion. Some files contain groups of tags with reversed endianess.
- static int reversedEndian = 0;
- // try to fix endian switching in the middle of headers
- if ((group == 0xfeff) && (elem == 0x00e0))
- {
- // start endian swap mark for group found
- reversedEndian++;
- SwitchByteSwapCode();
- // fix the tag
- group = 0xfffe;
- elem = 0xe000;
- }
- else if (group == 0xfffe && elem == 0xe00d && reversedEndian)
- {
- // end of reversed endian group
- reversedEndian--;
- SwitchByteSwapCode();
- }
-}
-
-/**
- * \brief Accesses the info from 0002,0010 : Transfer Syntax and TS
- * @return The full Transfer Syntax Name (as opposed to Transfer Syntax UID)
- */
-std::string Document::GetTransferSyntaxName()
-{
- // use the TS (TS : Transfer Syntax)
- std::string transferSyntax = GetEntryValue(0x0002,0x0010);
-
- if ( (transferSyntax.find(GDCM_NOTLOADED) < transferSyntax.length()) )
- {
- gdcmErrorMacro( "Transfer Syntax not loaded. " << std::endl
- << "Better you increase MAX_SIZE_LOAD_ELEMENT_VALUE" );
- return "Uncompressed ACR-NEMA";
- }
- if ( transferSyntax == GDCM_UNFOUND )
- {
- gdcmVerboseMacro( "Unfound Transfer Syntax (0002,0010)");
- return "Uncompressed ACR-NEMA";
- }
-
- // we do it only when we need it
- const TSKey &tsName = Global::GetTS()->GetValue( transferSyntax );
-
- // Global::GetTS() is a global static you shall never try to delete it!
- return tsName;
-}
-
-/**
- * \brief Group 0002 is always coded Little Endian
- * whatever Transfer Syntax is
- * @return no return
- */
-void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem)
-{
- // Endian reversion. Some files contain groups of tags with reversed endianess.
- if ( !Group0002Parsed && group != 0x0002)
- {
- Group0002Parsed = true;
- // we just came out of group 0002
- // if Transfer syntax is Big Endian we have to change CheckSwap
-
- std::string ts = GetTransferSyntax();
- if ( !Global::GetTS()->IsTransferSyntax(ts) )
- {
- gdcmVerboseMacro("True DICOM File, with NO Tansfer Syntax: " << ts );
- return;
- }
-
- // Group 0002 is always 'Explicit ...' enven when Transfer Syntax says 'Implicit ..."
-
- if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ImplicitVRLittleEndian )
- {
- Filetype = ImplicitVR;
- }
-
- // FIXME Strangely, this works with
- //'Implicit VR Transfer Syntax (GE Private)
- if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ExplicitVRBigEndian )
- {
- gdcmVerboseMacro("Transfer Syntax Name = ["
- << GetTransferSyntaxName() << "]" );
- SwitchByteSwapCode();
- group = SwapShort(group);
- elem = SwapShort(elem);
- }
- }
-}
-
/**
* \brief Read the next tag but WITHOUT loading it's value
* (read the 'Group Number', the 'Element Number',
{
// We reached the EOF (or an error occured) therefore
// header parsing has to be considered as finished.
- //std::cout << e;
return 0;
}
// Sometimes file contains groups of tags with reversed endianess.
HandleBrokenEndian(group, elem);
-// In 'true DICOM' files Group 0002 is always little endian
+ // In 'true DICOM' files Group 0002 is always little endian
if ( HasDCMPreamble )
HandleOutOfGroup0002(group, elem);
if( vr == GDCM_UNKNOWN)
{
- DictEntry *dictEntry = GetDictEntry(group,elem);
- if( dictEntry )
- realVR = dictEntry->GetVR();
+ if ( elem == 0x0000 ) // Group Length
+ realVR = "UL"; // must be UL
+ else
+ {
+ DictEntry *dictEntry = GetDictEntry(group,elem);
+ if( dictEntry )
+ realVR = dictEntry->GetVR();
+ }
}
DocEntry *newEntry;
std::string msg;
msg = Util::Format("Entry (%04x,%04x) should be Explicit VR\n",
newEntry->GetGroup(), newEntry->GetElement());
- gdcmVerboseMacro( msg.c_str() );
+ gdcmWarningMacro( msg.c_str() );
}
}
newEntry->SetImplicitVR();
catch ( FormatError e )
{
// Call it quits
- //std::cout << e;
delete newEntry;
return 0;
}
return newEntry;
}
-//GenerateFreeTagKeyInGroup? What was it designed for ?!?
-/**
- * \brief Generate a free TagKey i.e. a TagKey that is not present
- * in the TagHt dictionary.
- * @param group The generated tag must belong to this group.
- * @return The element of tag with given group which is fee.
- */
-//uint32_t Document::GenerateFreeTagKeyInGroup(uint16_t group)
-//{
-// for (uint32_t elem = 0; elem < UINT32_MAX; elem++)
-// {
-// TagKey key = DictEntry::TranslateToKey(group, elem);
-// if (TagHT.count(key) == 0)
-// {
-// return elem;
-// }
-// }
-// return UINT32_MAX;
-//}
-
-/**
- * \brief Assuming the internal file pointer \ref Document::Fp
- * is placed at the beginning of a tag check whether this
- * tag is (TestGroup, TestElement).
- * \warning On success the internal file pointer \ref Document::Fp
- * is modified to point after the tag.
- * On failure (i.e. when the tag wasn't the expected tag
- * (TestGroup, TestElement) the internal file pointer
- * \ref Document::Fp is restored to it's original position.
- * @param testGroup The expected group of the tag.
- * @param testElement The expected Element of the tag.
- * @return True on success, false otherwise.
- */
-bool Document::ReadTag(uint16_t testGroup, uint16_t testElement)
-{
- long positionOnEntry = Fp->tellg();
- long currentPosition = Fp->tellg(); // On debugging purposes
-
- //// Read the Item Tag group and element, and make
- // sure they are what we expected:
- uint16_t itemTagGroup;
- uint16_t itemTagElement;
- try
- {
- itemTagGroup = ReadInt16();
- itemTagElement = ReadInt16();
- }
- catch ( FormatError e )
- {
- //std::cerr << e << std::endl;
- return false;
- }
- if ( itemTagGroup != testGroup || itemTagElement != testElement )
- {
- gdcmVerboseMacro( "Wrong Item Tag found:"
- << " We should have found tag ("
- << std::hex << testGroup << "," << testElement << ")" << std::endl
- << " but instead we encountered tag ("
- << std::hex << itemTagGroup << "," << itemTagElement << ")"
- << " at address: " << " 0x(" << (unsigned int)currentPosition << ")"
- ) ;
- Fp->seekg(positionOnEntry, std::ios::beg);
-
- return false;
- }
- return true;
-}
-
-/**
- * \brief Assuming the internal file pointer \ref Document::Fp
- * is placed at the beginning of a tag (TestGroup, TestElement),
- * read the length associated to the Tag.
- * \warning On success the internal file pointer \ref Document::Fp
- * is modified to point after the tag and it's length.
- * On failure (i.e. when the tag wasn't the expected tag
- * (TestGroup, TestElement) the internal file pointer
- * \ref Document::Fp is restored to it's original position.
- * @param testGroup The expected group of the tag.
- * @param testElement The expected Element of the tag.
- * @return On success returns the length associated to the tag. On failure
- * returns 0.
- */
-uint32_t Document::ReadTagLength(uint16_t testGroup, uint16_t testElement)
-{
-
- if ( !ReadTag(testGroup, testElement) )
- {
- return 0;
- }
-
- //// Then read the associated Item Length
- long currentPosition = Fp->tellg();
- uint32_t itemLength = ReadInt32();
- {
- gdcmVerboseMacro( "Basic Item Length is: "
- << itemLength << std::endl
- << " at address: " << std::hex << (unsigned int)currentPosition);
- }
- return itemLength;
-}
-
/**
- * \brief When parsing the Pixel Data of an encapsulated file, read
- * the basic offset table (when present, and BTW dump it).
- */
-void Document::ReadAndSkipEncapsulatedBasicOffsetTable()
-{
- //// Read the Basic Offset Table Item Tag length...
- uint32_t itemLength = ReadTagLength(0xfffe, 0xe000);
-
- // When present, read the basic offset table itself.
- // Notes: - since the presence of this basic offset table is optional
- // we can't rely on it for the implementation, and we will simply
- // trash it's content (when present).
- // - still, when present, we could add some further checks on the
- // lengths, but we won't bother with such fuses for the time being.
- if ( itemLength != 0 )
- {
- char *basicOffsetTableItemValue = new char[itemLength + 1];
- Fp->read(basicOffsetTableItemValue, itemLength);
-
-#ifdef GDCM_DEBUG
- for (unsigned int i=0; i < itemLength; i += 4 )
- {
- uint32_t individualLength = str2num( &basicOffsetTableItemValue[i],
- uint32_t);
- gdcmVerboseMacro( "Read one length: " <<
- std::hex << individualLength );
- }
-#endif //GDCM_DEBUG
-
- delete[] basicOffsetTableItemValue;
- }
-}
-
-/**
- * \brief Parse pixel data from disk of [multi-]fragment RLE encoding.
- * Compute the RLE extra information and store it in \ref RLEInfo
- * for later pixel retrieval usage.
+ * \brief Handle broken private tag from Philips NTSCAN
+ * where the endianess is being switch to BigEndian for no
+ * apparent reason
+ * @return no return
*/
-void Document::ComputeRLEInfo()
+void Document::HandleBrokenEndian(uint16_t &group, uint16_t &elem)
{
- std::string ts = GetTransferSyntax();
- if ( !Global::GetTS()->IsRLELossless(ts) )
+ // Endian reversion. Some files contain groups of tags with reversed endianess.
+ static int reversedEndian = 0;
+ // try to fix endian switching in the middle of headers
+ if ((group == 0xfeff) && (elem == 0x00e0))
{
- return;
- }
-
- // Encoded pixel data: for the time being we are only concerned with
- // Jpeg or RLE Pixel data encodings.
- // As stated in PS 3.5-2003, section 8.2 p44:
- // "If sent in Encapsulated Format (i.e. other than the Native Format) the
- // value representation OB is used".
- // Hence we expect an OB value representation. Concerning OB VR,
- // the section PS 3.5-2003, section A.4.c p 58-59, states:
- // "For the Value Representations OB and OW, the encoding shall meet the
- // following specifications depending on the Data element tag:"
- // [...snip...]
- // - the first item in the sequence of items before the encoded pixel
- // data stream shall be basic offset table item. The basic offset table
- // item value, however, is not required to be present"
-
- ReadAndSkipEncapsulatedBasicOffsetTable();
-
- // Encapsulated RLE Compressed Images (see PS 3.5-2003, Annex G)
- // Loop on the individual frame[s] and store the information
- // on the RLE fragments in a RLEFramesInfo.
- // Note: - when only a single frame is present, this is a
- // classical image.
- // - when more than one frame are present, then we are in
- // the case of a multi-frame image.
- long frameLength;
- while ( (frameLength = ReadTagLength(0xfffe, 0xe000)) )
- {
- // Parse the RLE Header and store the corresponding RLE Segment
- // Offset Table information on fragments of this current Frame.
- // Note that the fragment pixels themselves are not loaded
- // (but just skipped).
- long frameOffset = Fp->tellg();
-
- uint32_t nbRleSegments = ReadInt32();
- if ( nbRleSegments > 16 )
- {
- // There should be at most 15 segments (refer to RLEFrame class)
- gdcmVerboseMacro( "Too many segments.");
- }
-
- uint32_t rleSegmentOffsetTable[16];
- for( int k = 1; k <= 15; k++ )
- {
- rleSegmentOffsetTable[k] = ReadInt32();
- }
-
- // Deduce from both the RLE Header and the frameLength the
- // fragment length, and again store this info in a
- // RLEFramesInfo.
- long rleSegmentLength[15];
- // skipping (not reading) RLE Segments
- if ( nbRleSegments > 1)
- {
- for(unsigned int k = 1; k <= nbRleSegments-1; k++)
- {
- rleSegmentLength[k] = rleSegmentOffsetTable[k+1]
- - rleSegmentOffsetTable[k];
- SkipBytes(rleSegmentLength[k]);
- }
- }
-
- rleSegmentLength[nbRleSegments] = frameLength
- - rleSegmentOffsetTable[nbRleSegments];
- SkipBytes(rleSegmentLength[nbRleSegments]);
-
- // Store the collected info
- RLEFrame *newFrame = new RLEFrame;
- newFrame->SetNumberOfFragments(nbRleSegments);
- for( unsigned int uk = 1; uk <= nbRleSegments; uk++ )
- {
- newFrame->SetOffset(uk,frameOffset + rleSegmentOffsetTable[uk]);
- newFrame->SetLength(uk,rleSegmentLength[uk]);
- }
- RLEInfo->AddFrame(newFrame);
- }
-
- // Make sure that at the end of the item we encounter a 'Sequence
- // Delimiter Item':
- if ( !ReadTag(0xfffe, 0xe0dd) )
+ // start endian swap mark for group found
+ reversedEndian++;
+ SwitchByteSwapCode();
+ // fix the tag
+ group = 0xfffe;
+ elem = 0xe000;
+ }
+ else if (group == 0xfffe && elem == 0xe00d && reversedEndian)
{
- gdcmVerboseMacro( "No sequence delimiter item at end of RLE item sequence");
+ // end of reversed endian group
+ reversedEndian--;
+ SwitchByteSwapCode();
}
}
/**
- * \brief Parse pixel data from disk of [multi-]fragment Jpeg encoding.
- * Compute the jpeg extra information (fragment[s] offset[s] and
- * length) and store it[them] in \ref JPEGInfo for later pixel
- * retrieval usage.
+ * \brief Group 0002 is always coded Little Endian
+ * whatever Transfer Syntax is
+ * @return no return
*/
-void Document::ComputeJPEGFragmentInfo()
+void Document::HandleOutOfGroup0002(uint16_t &group, uint16_t &elem)
{
- // If you need to, look for comments of ComputeRLEInfo().
- std::string ts = GetTransferSyntax();
- if ( ! Global::GetTS()->IsJPEG(ts) )
- {
- return;
- }
-
- ReadAndSkipEncapsulatedBasicOffsetTable();
-
- // Loop on the fragments[s] and store the parsed information in a
- // JPEGInfo.
- long fragmentLength;
- while ( (fragmentLength = ReadTagLength(0xfffe, 0xe000)) )
- {
- long fragmentOffset = Fp->tellg();
-
- // Store the collected info
- JPEGFragment *newFragment = new JPEGFragment;
- newFragment->SetOffset(fragmentOffset);
- newFragment->SetLength(fragmentLength);
- JPEGInfo->AddFragment(newFragment);
-
- SkipBytes(fragmentLength);
- }
-
- // Make sure that at the end of the item we encounter a 'Sequence
- // Delimiter Item':
- if ( !ReadTag(0xfffe, 0xe0dd) )
+ // Endian reversion. Some files contain groups of tags with reversed endianess.
+ if ( !Group0002Parsed && group != 0x0002)
{
- gdcmVerboseMacro( "No sequence delimiter item at end of JPEG item sequence");
- }
-}
-
-/*
- * \brief Walk recursively the given \ref DocEntrySet, and feed
- * the given hash table (\ref TagDocEntryHT) with all the
- * \ref DocEntry (Dicom entries) encountered.
- * This method does the job for \ref BuildFlatHashTable.
- * @param builtHT Where to collect all the \ref DocEntry encountered
- * when recursively walking the given set.
- * @param set The structure to be traversed (recursively).
- */
-/*void Document::BuildFlatHashTableRecurse( TagDocEntryHT &builtHT,
- DocEntrySet *set )
-{
- if (ElementSet *elementSet = dynamic_cast< ElementSet* > ( set ) )
- {
- TagDocEntryHT const ¤tHT = elementSet->GetTagHT();
- for( TagDocEntryHT::const_iterator i = currentHT.begin();
- i != currentHT.end();
- ++i)
- {
- DocEntry *entry = i->second;
- if ( SeqEntry *seqEntry = dynamic_cast<SeqEntry*>(entry) )
- {
- const ListSQItem& items = seqEntry->GetSQItems();
- for( ListSQItem::const_iterator item = items.begin();
- item != items.end();
- ++item)
- {
- BuildFlatHashTableRecurse( builtHT, *item );
- }
- continue;
- }
- builtHT[entry->GetKey()] = entry;
- }
- return;
- }
+ Group0002Parsed = true;
+ // we just came out of group 0002
+ // if Transfer syntax is Big Endian we have to change CheckSwap
- if (SQItem *SQItemSet = dynamic_cast< SQItem* > ( set ) )
- {
- const ListDocEntry& currentList = SQItemSet->GetDocEntries();
- for (ListDocEntry::const_iterator i = currentList.begin();
- i != currentList.end();
- ++i)
+ std::string ts = GetTransferSyntax();
+ if ( !Global::GetTS()->IsTransferSyntax(ts) )
{
- DocEntry *entry = *i;
- if ( SeqEntry *seqEntry = dynamic_cast<SeqEntry*>(entry) )
- {
- const ListSQItem& items = seqEntry->GetSQItems();
- for( ListSQItem::const_iterator item = items.begin();
- item != items.end();
- ++item)
- {
- BuildFlatHashTableRecurse( builtHT, *item );
- }
- continue;
- }
- builtHT[entry->GetKey()] = entry;
+ gdcmWarningMacro("True DICOM File, with NO Tansfer Syntax: " << ts );
+ return;
}
- }
-}*/
-
-/*
- * \brief Build a \ref TagDocEntryHT (i.e. a std::map<>) from the current
- * Document.
- *
- * The structure used by a Document (through \ref ElementSet),
- * in order to hold the parsed entries of a Dicom header, is a recursive
- * one. This is due to the fact that the sequences (when present)
- * can be nested. Additionaly, the sequence items (represented in
- * gdcm as \ref SQItem) add an extra complexity to the data
- * structure. Hence, a gdcm user whishing to visit all the entries of
- * a Dicom header will need to dig in the gdcm internals (which
- * implies exposing all the internal data structures to the API).
- * In order to avoid this burden to the user, \ref BuildFlatHashTable
- * recursively builds a temporary hash table, which holds all the
- * Dicom entries in a flat structure (a \ref TagDocEntryHT i.e. a
- * std::map<>).
- * \warning Of course there is NO integrity constrain between the
- * returned \ref TagDocEntryHT and the \ref ElementSet used
- * to build it. Hence if the underlying \ref ElementSet is
- * altered, then it is the caller responsability to invoke
- * \ref BuildFlatHashTable again...
- * @return The flat std::map<> we juste build.
- */
-/*TagDocEntryHT *Document::BuildFlatHashTable()
-{
- TagDocEntryHT *FlatHT = new TagDocEntryHT;
- BuildFlatHashTableRecurse( *FlatHT, this );
- return FlatHT;
-}*/
-
-
+ // Group 0002 is always 'Explicit ...' enven when Transfer Syntax says 'Implicit ..."
-/**
- * \brief Compares two documents, according to \ref DicomDir rules
- * \warning Does NOT work with ACR-NEMA files
- * \todo Find a trick to solve the pb (use RET fields ?)
- * @param document
- * @return true if 'smaller'
- */
-bool Document::operator<(Document &document)
-{
- // Patient Name
- std::string s1 = GetEntryValue(0x0010,0x0010);
- std::string s2 = document.GetEntryValue(0x0010,0x0010);
- if(s1 < s2)
- {
- return true;
- }
- else if( s1 > s2 )
- {
- return false;
- }
- else
- {
- // Patient ID
- s1 = GetEntryValue(0x0010,0x0020);
- s2 = document.GetEntryValue(0x0010,0x0020);
- if ( s1 < s2 )
- {
- return true;
- }
- else if ( s1 > s2 )
- {
- return false;
- }
- else
- {
- // Study Instance UID
- s1 = GetEntryValue(0x0020,0x000d);
- s2 = document.GetEntryValue(0x0020,0x000d);
- if ( s1 < s2 )
- {
- return true;
- }
- else if( s1 > s2 )
- {
- return false;
- }
- else
+ if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ImplicitVRLittleEndian )
{
- // Serie Instance UID
- s1 = GetEntryValue(0x0020,0x000e);
- s2 = document.GetEntryValue(0x0020,0x000e);
- if ( s1 < s2 )
- {
- return true;
- }
- else if( s1 > s2 )
- {
- return false;
- }
+ Filetype = ImplicitVR;
}
- }
- }
- return false;
-}
-
-/**
- * \brief Re-computes the length of a ACR-NEMA/Dicom group from a DcmHeader
- * @param filetype Type of the File to be written
- */
-int Document::ComputeGroup0002Length( FileType filetype )
-{
- uint16_t gr;
- std::string vr;
-
- int groupLength = 0;
- bool found0002 = false;
-
- // for each zero-level Tag in the DCM Header
- DocEntry *entry = GetFirstEntry();
- while(entry)
- {
- gr = entry->GetGroup();
-
- if (gr == 0x0002)
+
+ // FIXME Strangely, this works with
+ //'Implicit VR Transfer Syntax (GE Private)
+ if ( Global::GetTS()->GetSpecialTransferSyntax(ts) == TS::ExplicitVRBigEndian )
{
- found0002 = true;
- vr = entry->GetVR();
-
- if (filetype == ExplicitVR)
- {
- if ( (vr == "OB") || (vr == "OW") || (vr == "SQ") )
- {
- groupLength += 4; // explicit VR AND OB, OW, SQ : 4 more bytes
- }
- }
- groupLength += 2 + 2 + 4 + entry->GetLength();
+ gdcmWarningMacro("Transfer Syntax Name = ["
+ << GetTransferSyntaxName() << "]" );
+ SwitchByteSwapCode();
+ group = SwapShort(group);
+ elem = SwapShort(elem);
}
- else if (found0002 )
- break;
-
- entry = GetNextEntry();
}
- return groupLength;
}
-} // end namespace gdcm
+//-----------------------------------------------------------------------------
+// Print
//-----------------------------------------------------------------------------
+} // end namespace gdcm
git://git.creatis.insa-lyon.fr / gdcm.git / blobdiff