Example/SplitIntoDirectories.cxx

   1 /*=========================================================================
   2
   3   Program:   gdcm
   4   Module:    $RCSfile: SplitIntoDirectories.cxx,v $
   5   Language:  C++
   6   Date:      $Date: 2007/09/26 08:14:27 $
   7   Version:   $Revision: 1.1 $
   8
   9   Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de
  10   l'Image). All rights reserved. See Doc/License.txt or
  11   http://www.creatis.insa-lyon.fr/Public/Gdcm/License.html for details.
  12
  13      This software is distributed WITHOUT ANY WARRANTY; without even
  14      the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
  15      PURPOSE.  See the above copyright notices for more information.
  16
  17 =========================================================================*/
  18 #include "gdcmDocEntry.h"
  19 #include "gdcmDicomDir.h"
  20 #include "gdcmDicomDirPatient.h"
  21 #include "gdcmFile.h"
  22 #include "gdcmFileHelper.h"
  23 #include "gdcmDirList.h"
  24 #include "gdcmDebug.h"
  25 #include "gdcmArgMgr.h"
  26 #include "gdcmUtil.h"
  27 #include "gdcmSerieHelper.h"
  28
  29 #include <iostream>
  30
  31 /**
  32   * \brief
  33   *          - explores recursively the given directory
  34   *          - keeps the requested series
  35   *          - orders the gdcm-readable found Files
  36   *            according to their Patient/Study/Serie/Image characteristics
  37   */
  38
  39 typedef std::map<std::string, GDCM_NAME_SPACE::File*> SortedFiles;
  40
  41 int main(int argc, char *argv[])
  42 {
  43    START_USAGE(usage)
  44    " \n SplitIntoDirectories :\n                                              ",
  45    " - explores recursively the given directory,                              ",
  46    " - keeps the requested series / drops the unrequested series              ",
  47    " - orders the gdcm-readable found Files according to their                ",
  48    "           (0x0010, 0x0010) Patient's Name                                ",
  49    "           (0x0020, 0x000d) Study Instance UID                            ",
  50    "           (0x0020, 0x000e) Series Instance UID                           ",
  51    " - fills a tree-like structure of directories as :                        ",
  52    "        - Patient                                                         ",
  53    "        -- Study                                                          ",
  54    "        --- Serie                                                         ",
  55    "                                                                          ",
  56    " usage:                                                                   ",
  57    " -----                                                                    ",
  58    " SplitIntoDirectories                                                     ",
  59    "                  dirin=rootDirectoryName                                 ",
  60    "                  dirout=outputDirectoryName                              ",
  61    "                  {  [keep= list of seriesNumber to process]              ",
  62    "                   | [drop= list of seriesNumber to ignore] }             ",
  63    "                  [listonly]                                              ",
  64    "                  [noshadowseq][noshadow][noseq] [verbose] [debug]        ",
  65    "                                                                          ",
  66    " dirout : will be created if doesn't exist                                ",
  67    " keep : if user wants to process a limited number of series               ",
  68    "            he gives the list of 'SeriesNumber' (tag 0020|0011)           ",
  69    " drop : if user wants to ignore a limited number of series                ",
  70    "            he gives the list of 'SeriesNumber' (tag 0020|0011)           ",
  71    "        SeriesNumber are short enough to be human readable                ",
  72    "        e.g : 1030,1035,1043                                              ",
  73    " noshadowseq: user doesn't want to load Private Sequences                 ",
  74    " noshadow : user doesn't want to load Private groups (odd number)         ",
  75    " noseq    : user doesn't want to load Sequences                           ",
  76    " verbose  : user wants to run the program in 'verbose mode'               ",
  77    " debug    : *developer*  wants to run the program in 'debug mode'         ",
  78    FINISH_USAGE
  79
  80
  81    enum Index
  82    {
  83       IND_PatientName,
  84       IND_StudyInstanceUID,
  85       IND_SerieInstanceUID
  86    };
  87
  88    std::cout << "... inside " << argv[0] << std::endl;
  89
  90    // ----- Initialize Arguments Manager ------
  91
  92    GDCM_NAME_SPACE::ArgMgr *am = new GDCM_NAME_SPACE::ArgMgr(argc, argv);
  93
  94    if (argc == 1 || am->ArgMgrDefined("usage"))
  95    {
  96       am->ArgMgrUsage(usage); // Display 'usage'
  97       delete am;
  98       return 0;
  99    }
 100
 101    const char *dirNamein;
 102    dirNamein  = am->ArgMgrGetString("dirin",".");
 103
 104    const char *dirNameout;
 105    dirNameout  = am->ArgMgrGetString("dirout",".");
 106
 107    int loadMode = GDCM_NAME_SPACE::LD_ALL;
 108    if ( am->ArgMgrDefined("noshadowseq") )
 109       loadMode |= GDCM_NAME_SPACE::LD_NOSHADOWSEQ;
 110    else
 111    {
 112    if ( am->ArgMgrDefined("noshadow") )
 113          loadMode |= GDCM_NAME_SPACE::LD_NOSHADOW;
 114       if ( am->ArgMgrDefined("noseq") )
 115          loadMode |= GDCM_NAME_SPACE::LD_NOSEQ;
 116    }
 117
 118    if (am->ArgMgrDefined("debug"))
 119       GDCM_NAME_SPACE::Debug::DebugOn();
 120
 121    bool verbose  = ( 0 != am->ArgMgrDefined("verbose") );
 122    bool listonly = ( 0 != am->ArgMgrDefined("listonly") );
 123
 124    int nbSeriesToKeep;
 125    int *seriesToKeep = am->ArgMgrGetListOfInt("keep", &nbSeriesToKeep);
 126    int nbSeriesToDrop;
 127    int *seriesToDrop = am->ArgMgrGetListOfInt("drop", &nbSeriesToDrop);
 128
 129    if ( nbSeriesToKeep!=0 && nbSeriesToDrop!=0)
 130    {
 131       std::cout << "KEEP and DROP are mutually exclusive !" << std::endl;
 132       delete am;
 133       return 0;
 134    }
 135
 136    int hasSkel = am->ArgMgrDefined("skel");
 137    const char *skel;
 138    if (hasSkel)
 139       skel = am->ArgMgrGetString("skel");
 140
 141    const char *input   = am->ArgMgrGetString("input","DCM");
 142
 143    // if unused Param we give up
 144    if ( am->ArgMgrPrintUnusedLabels() )
 145    {
 146       am->ArgMgrUsage(usage);
 147       delete am;
 148       return 0;
 149    }
 150    delete am;  // we don't need Argument Manager any longer
 151
 152    // ----- Begin Processing -----
 153
 154
 155    // --> Check supposed-to-be-directory names
 156
 157    if ( ! GDCM_NAME_SPACE::DirList::IsDirectory(dirNamein) )
 158    {
 159       std::cout << "KO : [" << dirNamein << "] is not a Directory."
 160                 << std::endl;
 161       return 0;
 162
 163    }
 164    else
 165    {
 166       std::cout << "OK : [" << dirNamein << "] is a Directory." << std::endl;
 167    }
 168
 169    std::string systemCommand;
 170
 171    std::cout << "Check for output directory :[" << dirNameout << "]."
 172              <<std::endl;
 173    if ( ! GDCM_NAME_SPACE::DirList::IsDirectory(dirNameout) )    // dirout not found
 174    {
 175       std::string strDirNameout(dirNameout);          // to please gcc 4
 176       systemCommand = "mkdir " +strDirNameout;        // create it!
 177       if (verbose)
 178          std::cout << systemCommand << std::endl;
 179       system (systemCommand.c_str());
 180       if ( ! GDCM_NAME_SPACE::DirList::IsDirectory(dirNameout) ) // be sure it worked
 181       {
 182           std::cout << "KO : not a dir : [" << dirNameout << "] (creation failure ?)"
 183                     << std::endl;
 184       return 0;
 185
 186       }
 187       else
 188       {
 189         std::cout << "Directory [" << dirNameout << "] created." << std::endl;
 190       }
 191    }
 192    else
 193    {
 194        std::cout << "Output Directory [" << dirNameout
 195                  << "] already exists; Used as is."
 196                  << std::endl;
 197    }
 198    // --> End of checking supposed-to-be-directory names
 199
 200    std::string strDirNamein(dirNamein);
 201    // true ; get recursively the list of files
 202    GDCM_NAME_SPACE::DirList dirList(strDirNamein, true);
 203
 204    if (listonly)
 205    {
 206       std::cout << "------------List of found files ------------" << std::endl;
 207       dirList.Print();
 208       std::cout << std::endl;
 209    }
 210
 211
 212 // ======================================= The job starts here =========================
 213
 214    GDCM_NAME_SPACE::DirListType fileNames;
 215    fileNames = dirList.GetFilenames();
 216
 217    GDCM_NAME_SPACE::SerieHelper *s;     // Needed to use SerieHelper::AddSeriesDetail()
 218    s = GDCM_NAME_SPACE::SerieHelper::New();
 219
 220    std::string token = "%%%"; // Hope it's enough!
 221
 222    GDCM_NAME_SPACE::File *f;
 223    std::vector<std::string> tokens;
 224    std::vector<std::string> tokensForFileName;
 225
 226    if (verbose)
 227       std::cout << "------------------Print Break levels-----------------" << std::endl;
 228
 229    std::string userFileIdentifier;
 230    SortedFiles sf;
 231
 232    s->AddSeriesDetail(0x0010, 0x0010, false); // Patient's Name (false : no convert)
 233
 234    // You may prefer 0020 0010  Study ID
 235    // use :
 236    // s->AddSeriesDetail(0x0020, 0x0010, true);
 237    // Avoid using 0008 0020 Study Date,
 238    // since you may have more than one study, for a given Patient, at a given Date!
 239    // or the field may be empty!
 240    s->AddSeriesDetail(0x0020, 0x000d, false); // Study Instance UID (false : no convert)
 241
 242
 243    // You may prefer 0020 0011 Series Number
 244    // use :
 245    // s->AddSeriesDetail(0x0020, 0x0011, true);
 246    s->AddSeriesDetail(0x0020, 0x000e, false); // Series Instance UID (false : no convert)
 247
 248    // Feel free to add more fields, if they can help a suitable (for you)
 249    // image sorting
 250
 251 // Loop on all the gdcm-readable files
 252    for (GDCM_NAME_SPACE::DirListType::iterator it = fileNames.begin();
 253                                     it != fileNames.end();
 254                                   ++it)
 255    {
 256       f = GDCM_NAME_SPACE::File::New();
 257       f->SetLoadMode(loadMode);
 258       f->SetFileName( *it );
 259       f->Load();
 260
 261       std::string strSeriesNumber;
 262       int seriesNumber;
 263       int j;
 264
 265       // keep only requested Series
 266       bool keep = false;
 267       if (nbSeriesToKeep != 0)
 268       {
 269          strSeriesNumber = f->GetEntryString(0x0020, 0x0011 );
 270          seriesNumber = atoi( strSeriesNumber.c_str() );
 271          for (j=0; j<nbSeriesToKeep; j++)
 272          {
 273             if(seriesNumber == seriesToKeep[j])
 274             {
 275                keep = true;
 276                break;
 277             }
 278          }
 279          if ( !keep)
 280          {
 281             f->Delete();
 282             continue;
 283          }
 284       }
 285       // drop all unrequested Series
 286       bool drop = false;
 287       if (nbSeriesToDrop != 0)
 288       {
 289          strSeriesNumber = f->GetEntryString(0x0020, 0x0011 );
 290          seriesNumber = atoi( strSeriesNumber.c_str() );
 291          for (j=0;j<nbSeriesToDrop; j++)
 292          {
 293             if(seriesNumber == seriesToDrop[j])
 294             {
 295                drop = true;
 296                break;
 297             }
 298         }
 299         if (drop)
 300         {
 301            f->Delete();
 302            continue;
 303         }
 304       }
 305
 306       userFileIdentifier=s->CreateUserDefinedFileIdentifier(f);
 307       tokens.clear();
 308       GDCM_NAME_SPACE::Util::Tokenize (userFileIdentifier, tokens, token);
 309
 310       int imageNum; // Within FileName
 311       char newName[1024];
 312
 313
 314          // Patient's Name
 315          // Study Instance UID
 316          // Series Instance UID
 317
 318       userFileIdentifier = tokens[IND_PatientName]      + token +
 319                            tokens[IND_StudyInstanceUID] + token +
 320                            tokens[IND_SerieInstanceUID] + token;
 321
 322       if (verbose)
 323          std::cout << "[" << userFileIdentifier  << "] : " << *it << std::endl;
 324
 325       // storing in a map ensures automatic sorting !
 326       sf[userFileIdentifier] = f;
 327    }
 328
 329    if (verbose)
 330       std::cout << "  " << std::endl;
 331
 332    std::string fullFilename, lastFilename;
 333    std::string previousPatientName, currentPatientName;
 334    std::string previousStudyInstanceUID, currentStudyInstanceUID;
 335    std::string previousSerieInstanceUID, currentSerieInstanceUID;
 336
 337
 338    std::string writeDir, currentWriteDir;
 339    std::string currentPatientWriteDir;
 340    std::string currentStudyWriteDir;
 341    std::string currentSerieWriteDir;
 342
 343    std::string fullWriteFilename;
 344
 345    writeDir = GDCM_NAME_SPACE::Util::NormalizePath(dirNameout);
 346    SortedFiles::iterator it2;
 347
 348    previousPatientName            = "";
 349    previousStudyInstanceUID       = "";
 350    previousSerieInstanceUID       = "";
 351
 352    GDCM_NAME_SPACE::File *currentFile;
 353
 354    for (it2 = sf.begin() ; it2 != sf.end(); ++it2)
 355    {
 356       currentFile = it2->second;
 357
 358       fullFilename =  currentFile->GetFileName();
 359       lastFilename =  GDCM_NAME_SPACE::Util::GetName( fullFilename );
 360       if (verbose)
 361       std::cout <<" ------------------------------------------------------------------------------"
 362                 << std::endl << " Deal with [" << it2->first << "] : [" <<fullFilename << "]"
 363                 << std::endl;
 364
 365       tokens.clear();
 366       GDCM_NAME_SPACE::Util::Tokenize (it2->first, tokens, token);
 367
 368       currentPatientName            = tokens[IND_PatientName];
 369       currentStudyInstanceUID       = tokens[IND_StudyInstanceUID];
 370       currentSerieInstanceUID       = tokens[IND_SerieInstanceUID];
 371
 372       if (previousPatientName != currentPatientName)
 373       {
 374          previousPatientName = currentPatientName;
 375          if (verbose)
 376             std::cout << "==== new Patient  [" << currentPatientName  << "]" << std::endl;
 377
 378          previousPatientName            = currentPatientName;
 379          previousStudyInstanceUID       = "";
 380          previousSerieInstanceUID       = "";
 381
 382          currentPatientWriteDir = writeDir + currentPatientName;
 383
 384          systemCommand   = "mkdir " + currentPatientWriteDir;
 385          if (verbose)
 386             std::cout << systemCommand << std::endl;
 387
 388          system ( systemCommand.c_str() );
 389       }
 390
 391       if (previousStudyInstanceUID != currentStudyInstanceUID)
 392       {
 393          if (verbose)
 394             std::cout << "==== === new Study [" << currentStudyInstanceUID << "]"
 395                       << std::endl;
 396
 397          currentStudyWriteDir  = currentPatientWriteDir + GDCM_NAME_SPACE::GDCM_FILESEPARATOR
 398                              + currentStudyInstanceUID;
 399          systemCommand   = "mkdir " + currentStudyWriteDir;
 400          system (systemCommand.c_str());
 401
 402          previousStudyInstanceUID       = currentStudyInstanceUID;
 403       }
 404
 405       if (previousSerieInstanceUID != currentSerieInstanceUID)
 406       {
 407          if (verbose)
 408             std::cout << "=== ==== === new Serie [" << currentSerieInstanceUID << "]"
 409                       << std::endl;
 410
 411          currentSerieWriteDir  = currentStudyWriteDir + GDCM_NAME_SPACE::GDCM_FILESEPARATOR
 412                              + currentSerieInstanceUID;
 413          systemCommand   = "mkdir " + currentSerieWriteDir;
 414          system (systemCommand.c_str());
 415
 416          previousSerieInstanceUID       = currentSerieInstanceUID;
 417       }
 418
 419       if ( GDCM_NAME_SPACE::Debug::GetDebugFlag())
 420          std::cout << "--- --- --- --- --- " << it2->first << "  "
 421                    << (it2->second)->GetFileName() << " "
 422                    << GDCM_NAME_SPACE::Util::GetName( fullFilename ) << std::endl;
 423
 424       // If you want to create file names of your own, here is the place!
 425       // Just replace 'lastFilename' by anything that's better for you.
 426       fullWriteFilename = currentSerieWriteDir + GDCM_NAME_SPACE::GDCM_FILESEPARATOR
 427                                          + lastFilename;
 428
 429       systemCommand   = "cp " + fullFilename + " " + fullWriteFilename;
 430       system ( systemCommand.c_str());
 431
 432    }
 433    return 0;
 434  }
 435