sffinfocommand.cpp

   1 /*
   2  *  sffinfocommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 7/7/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "sffinfocommand.h"
  11 #include "endiannessmacros.h"
  12 #include "trimoligos.h"
  13 #include "sequence.hpp"
  14 #include "qualityscores.h"
  15
  16 //**********************************************************************************************************************
  17 vector<string> SffInfoCommand::setParameters(){
  18         try {
  19                 CommandParameter psff("sff", "InputTypes", "", "", "none", "none", "none","",false,false,true); parameters.push_back(psff);
  20         CommandParameter poligos("oligos", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(poligos);
  21                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(paccnos);
  22                 CommandParameter psfftxt("sfftxt", "String", "", "", "", "", "","",false,false); parameters.push_back(psfftxt);
  23                 CommandParameter pflow("flow", "Boolean", "", "T", "", "", "","flow",false,false); parameters.push_back(pflow);
  24                 CommandParameter ptrim("trim", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(ptrim);
  25                 CommandParameter pfasta("fasta", "Boolean", "", "T", "", "", "","fasta",false,false); parameters.push_back(pfasta);
  26                 CommandParameter pqfile("qfile", "Boolean", "", "T", "", "", "","qfile",false,false); parameters.push_back(pqfile);
  27         CommandParameter ppdiffs("pdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(ppdiffs);
  28                 CommandParameter pbdiffs("bdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pbdiffs);
  29         CommandParameter pldiffs("ldiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pldiffs);
  30                 CommandParameter psdiffs("sdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(psdiffs);
  31         CommandParameter ptdiffs("tdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(ptdiffs);
  32                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  33                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  34
  35                 vector<string> myArray;
  36                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  37                 return myArray;
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "SffInfoCommand", "setParameters");
  41                 exit(1);
  42         }
  43 }
  44 //**********************************************************************************************************************
  45 string SffInfoCommand::getHelpString(){
  46         try {
  47                 string helpString = "";
  48                 helpString += "The sffinfo command reads a sff file and extracts the sequence data, or you can use it to parse a sfftxt file.\n";
  49                 helpString += "The sffinfo command parameters are sff, fasta, qfile, accnos, flow, sfftxt, oligos, bdiffs, tdiffs, ldiffs, sdiffs, pdiffs and trim. sff is required. \n";
  50                 helpString += "The sff parameter allows you to enter the sff file you would like to extract data from.  You may enter multiple files by separating them by -'s.\n";
  51                 helpString += "The fasta parameter allows you to indicate if you would like a fasta formatted file generated.  Default=True. \n";
  52                 helpString += "The qfile parameter allows you to indicate if you would like a quality file generated.  Default=True. \n";
  53         helpString += "The oligos parameter allows you to provide an oligos file to split your sff file into separate sff files by barcode. \n";
  54         helpString += "The tdiffs parameter is used to specify the total number of differences allowed in the sequence. The default is pdiffs + bdiffs + sdiffs + ldiffs.\n";
  55                 helpString += "The bdiffs parameter is used to specify the number of differences allowed in the barcode. The default is 0.\n";
  56                 helpString += "The pdiffs parameter is used to specify the number of differences allowed in the primer. The default is 0.\n";
  57         helpString += "The ldiffs parameter is used to specify the number of differences allowed in the linker. The default is 0.\n";
  58                 helpString += "The sdiffs parameter is used to specify the number of differences allowed in the spacer. The default is 0.\n";
  59                 helpString += "The flow parameter allows you to indicate if you would like a flowgram file generated.  Default=True. \n";
  60                 helpString += "The sfftxt parameter allows you to indicate if you would like a sff.txt file generated.  Default=False. \n";
  61                 helpString += "If you want to parse an existing sfftxt file into flow, fasta and quality file, enter the file name using the sfftxt parameter. \n";
  62                 helpString += "The trim parameter allows you to indicate if you would like a sequences and quality scores trimmed to the clipQualLeft and clipQualRight values.  Default=True. \n";
  63                 helpString += "The accnos parameter allows you to provide a accnos file containing the names of the sequences you would like extracted. You may enter multiple files by separating them by -'s. \n";
  64                 helpString += "Example sffinfo(sff=mySffFile.sff, trim=F).\n";
  65                 helpString += "Note: No spaces between parameter labels (i.e. sff), '=' and parameters (i.e.yourSffFileName).\n";
  66                 return helpString;
  67         }
  68         catch(exception& e) {
  69                 m->errorOut(e, "SffInfoCommand", "getHelpString");
  70                 exit(1);
  71         }
  72 }
  73
  74 //**********************************************************************************************************************
  75 string SffInfoCommand::getOutputPattern(string type) {
  76     try {
  77         string pattern = "";
  78
  79         if (type == "fasta")            {   pattern =  "[filename],fasta-[filename],[tag],fasta";   }
  80         else if (type == "flow")    {   pattern =  "[filename],flow";   }
  81         else if (type == "sfftxt")        {   pattern =  "[filename],sff.txt";   }
  82         else if (type == "sff")        {   pattern =  "[filename],[group],sff";   }
  83         else if (type == "qfile")       {   pattern =  "[filename],qual-[filename],[tag],qual";   }
  84         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  85
  86         return pattern;
  87     }
  88     catch(exception& e) {
  89         m->errorOut(e, "SffInfoCommand", "getOutputPattern");
  90         exit(1);
  91     }
  92 }
  93 //**********************************************************************************************************************
  94 SffInfoCommand::SffInfoCommand(){
  95         try {
  96                 abort = true; calledHelp = true;
  97                 setParameters();
  98                 vector<string> tempOutNames;
  99                 outputTypes["fasta"] = tempOutNames;
 100                 outputTypes["flow"] = tempOutNames;
 101                 outputTypes["sfftxt"] = tempOutNames;
 102                 outputTypes["qfile"] = tempOutNames;
 103         outputTypes["sff"] = tempOutNames;
 104         }
 105         catch(exception& e) {
 106                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
 107                 exit(1);
 108         }
 109 }
 110 //**********************************************************************************************************************
 111
 112 SffInfoCommand::SffInfoCommand(string option)  {
 113         try {
 114                 abort = false; calledHelp = false;
 115                 hasAccnos = false; hasOligos = false;
 116         split = 1;
 117
 118                 //allow user to run help
 119                 if(option == "help") { help(); abort = true; calledHelp = true; }
 120                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 121
 122                 else {
 123                         //valid paramters for this command
 124                         vector<string> myArray = setParameters();
 125
 126                         OptionParser parser(option);
 127                         map<string, string> parameters = parser.getParameters();
 128
 129                         ValidParameters validParameter;
 130                         //check to make sure all parameters are valid for command
 131                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 132                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 133                         }
 134
 135                         //initialize outputTypes
 136                         vector<string> tempOutNames;
 137                         outputTypes["fasta"] = tempOutNames;
 138                         outputTypes["flow"] = tempOutNames;
 139                         outputTypes["sfftxt"] = tempOutNames;
 140                         outputTypes["qfile"] = tempOutNames;
 141             outputTypes["sff"] = tempOutNames;
 142
 143                         //if the user changes the output directory command factory will send this info to us in the output parameter
 144                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 145
 146                         //if the user changes the input directory command factory will send this info to us in the output parameter
 147                         string inputDir = validParameter.validFile(parameters, "inputdir", false);        if (inputDir == "not found"){ inputDir = "";          }
 148
 149                         sffFilename = validParameter.validFile(parameters, "sff", false);
 150                         if (sffFilename == "not found") { sffFilename = "";  }
 151                         else {
 152                                 m->splitAtDash(sffFilename, filenames);
 153
 154                                 //go through files and make sure they are good, if not, then disregard them
 155                                 for (int i = 0; i < filenames.size(); i++) {
 156                                         bool ignore = false;
 157                                         if (filenames[i] == "current") {
 158                                                 filenames[i] = m->getSFFFile();
 159                                                 if (filenames[i] != "") {  m->mothurOut("Using " + filenames[i] + " as input file for the sff parameter where you had given current."); m->mothurOutEndLine(); }
 160                                                 else {
 161                                                         m->mothurOut("You have no current sfffile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 162                                                         //erase from file list
 163                                                         filenames.erase(filenames.begin()+i);
 164                                                         i--;
 165                                                 }
 166                                         }
 167
 168                                         if (!ignore) {
 169                                                 if (inputDir != "") {
 170                                                         string path = m->hasPath(filenames[i]);
 171                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 172                                                         if (path == "") {       filenames[i] = inputDir + filenames[i];         }
 173                                                 }
 174
 175                                                 ifstream in;
 176                                                 int ableToOpen = m->openInputFile(filenames[i], in, "noerror");
 177
 178                                                 //if you can't open it, try default location
 179                                                 if (ableToOpen == 1) {
 180                                                         if (m->getDefaultPath() != "") { //default path is set
 181                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(filenames[i]);
 182                                                                 m->mothurOut("Unable to open " + filenames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 183                                                                 ifstream in2;
 184                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 185                                                                 in2.close();
 186                                                                 filenames[i] = tryPath;
 187                                                         }
 188                                                 }
 189
 190                                                 //if you can't open it, try default location
 191                                                 if (ableToOpen == 1) {
 192                                                         if (m->getOutputDir() != "") { //default path is set
 193                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(filenames[i]);
 194                                                                 m->mothurOut("Unable to open " + filenames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 195                                                                 ifstream in2;
 196                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 197                                                                 in2.close();
 198                                                                 filenames[i] = tryPath;
 199                                                         }
 200                                                 }
 201
 202                                                 in.close();
 203
 204                                                 if (ableToOpen == 1) {
 205                                                         m->mothurOut("Unable to open " + filenames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 206                                                         //erase from file list
 207                                                         filenames.erase(filenames.begin()+i);
 208                                                         i--;
 209                                                 }else { m->setSFFFile(filenames[i]); }
 210                                         }
 211                                 }
 212
 213                                 //make sure there is at least one valid file left
 214                                 if (filenames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 215                         }
 216
 217                         accnosName = validParameter.validFile(parameters, "accnos", false);
 218                         if (accnosName == "not found") { accnosName = "";  }
 219                         else {
 220                                 hasAccnos = true;
 221                                 m->splitAtDash(accnosName, accnosFileNames);
 222
 223                                 //go through files and make sure they are good, if not, then disregard them
 224                                 for (int i = 0; i < accnosFileNames.size(); i++) {
 225                                         bool ignore = false;
 226                                         if (accnosFileNames[i] == "current") {
 227                                                 accnosFileNames[i] = m->getAccnosFile();
 228                                                 if (accnosFileNames[i] != "") {  m->mothurOut("Using " + accnosFileNames[i] + " as input file for the accnos parameter where you had given current."); m->mothurOutEndLine(); }
 229                                                 else {
 230                                                         m->mothurOut("You have no current accnosfile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 231                                                         //erase from file list
 232                                                         accnosFileNames.erase(accnosFileNames.begin()+i);
 233                                                         i--;
 234                                                 }
 235                                         }
 236
 237                                         if (!ignore) {
 238
 239                                                 if (inputDir != "") {
 240                                                         string path = m->hasPath(accnosFileNames[i]);
 241                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 242                                                         if (path == "") {       accnosFileNames[i] = inputDir + accnosFileNames[i];             }
 243                                                 }
 244
 245                                                 ifstream in;
 246                                                 int ableToOpen = m->openInputFile(accnosFileNames[i], in, "noerror");
 247
 248                                                 //if you can't open it, try default location
 249                                                 if (ableToOpen == 1) {
 250                                                         if (m->getDefaultPath() != "") { //default path is set
 251                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(accnosFileNames[i]);
 252                                                                 m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 253                                                                 ifstream in2;
 254                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 255                                                                 in2.close();
 256                                                                 accnosFileNames[i] = tryPath;
 257                                                         }
 258                                                 }
 259                                                 //if you can't open it, try default location
 260                                                 if (ableToOpen == 1) {
 261                                                         if (m->getOutputDir() != "") { //default path is set
 262                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(accnosFileNames[i]);
 263                                                                 m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 264                                                                 ifstream in2;
 265                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 266                                                                 in2.close();
 267                                                                 accnosFileNames[i] = tryPath;
 268                                                         }
 269                                                 }
 270                                                 in.close();
 271
 272                                                 if (ableToOpen == 1) {
 273                                                         m->mothurOut("Unable to open " + accnosFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 274                                                         //erase from file list
 275                                                         accnosFileNames.erase(accnosFileNames.begin()+i);
 276                                                         i--;
 277                                                 }
 278                                         }
 279                                 }
 280
 281                                 //make sure there is at least one valid file left
 282                                 if (accnosFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 283                         }
 284
 285             oligosfile = validParameter.validFile(parameters, "oligos", false);
 286                         if (oligosfile == "not found") { oligosfile = "";  }
 287                         else {
 288                                 hasOligos = true;
 289                                 m->splitAtDash(oligosfile, oligosFileNames);
 290
 291                                 //go through files and make sure they are good, if not, then disregard them
 292                                 for (int i = 0; i < oligosFileNames.size(); i++) {
 293                                         bool ignore = false;
 294                                         if (oligosFileNames[i] == "current") {
 295                                                 oligosFileNames[i] = m->getOligosFile();
 296                                                 if (oligosFileNames[i] != "") {  m->mothurOut("Using " + oligosFileNames[i] + " as input file for the accnos parameter where you had given current."); m->mothurOutEndLine(); }
 297                                                 else {
 298                                                         m->mothurOut("You have no current oligosfile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 299                                                         //erase from file list
 300                                                         oligosFileNames.erase(oligosFileNames.begin()+i);
 301                                                         i--;
 302                                                 }
 303                                         }
 304
 305                                         if (!ignore) {
 306
 307                                                 if (inputDir != "") {
 308                                                         string path = m->hasPath(oligosFileNames[i]);
 309                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 310                                                         if (path == "") {       oligosFileNames[i] = inputDir + oligosFileNames[i];             }
 311                                                 }
 312
 313                                                 ifstream in;
 314                                                 int ableToOpen = m->openInputFile(oligosFileNames[i], in, "noerror");
 315
 316                                                 //if you can't open it, try default location
 317                                                 if (ableToOpen == 1) {
 318                                                         if (m->getDefaultPath() != "") { //default path is set
 319                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(oligosFileNames[i]);
 320                                                                 m->mothurOut("Unable to open " + oligosFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 321                                                                 ifstream in2;
 322                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 323                                                                 in2.close();
 324                                                                 oligosFileNames[i] = tryPath;
 325                                                         }
 326                                                 }
 327                                                 //if you can't open it, try default location
 328                                                 if (ableToOpen == 1) {
 329                                                         if (m->getOutputDir() != "") { //default path is set
 330                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(oligosFileNames[i]);
 331                                                                 m->mothurOut("Unable to open " + oligosFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 332                                                                 ifstream in2;
 333                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 334                                                                 in2.close();
 335                                                                 oligosFileNames[i] = tryPath;
 336                                                         }
 337                                                 }
 338                                                 in.close();
 339
 340                                                 if (ableToOpen == 1) {
 341                                                         m->mothurOut("Unable to open " + oligosFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 342                                                         //erase from file list
 343                                                         oligosFileNames.erase(oligosFileNames.begin()+i);
 344                                                         i--;
 345                                                 }
 346                                         }
 347                                 }
 348
 349                                 //make sure there is at least one valid file left
 350                                 if (oligosFileNames.size() == 0) { m->mothurOut("no valid oligos files."); m->mothurOutEndLine(); abort = true; }
 351                         }
 352
 353                         if (hasOligos) {
 354                 split = 2;
 355                                 if (oligosFileNames.size() != filenames.size()) { abort = true; m->mothurOut("If you provide a oligos file, you must have one for each sff file."); m->mothurOutEndLine(); }
 356                         }
 357
 358                         if (hasAccnos) {
 359                                 if (accnosFileNames.size() != filenames.size()) { abort = true; m->mothurOut("If you provide a accnos file, you must have one for each sff file."); m->mothurOutEndLine(); }
 360                         }
 361
 362                         string temp = validParameter.validFile(parameters, "qfile", false);                     if (temp == "not found"){       temp = "T";                             }
 363                         qual = m->isTrue(temp);
 364
 365                         temp = validParameter.validFile(parameters, "fasta", false);                            if (temp == "not found"){       temp = "T";                             }
 366                         fasta = m->isTrue(temp);
 367
 368                         temp = validParameter.validFile(parameters, "flow", false);                                     if (temp == "not found"){       temp = "T";                             }
 369                         flow = m->isTrue(temp);
 370
 371                         temp = validParameter.validFile(parameters, "trim", false);                                     if (temp == "not found"){       temp = "T";                             }
 372                         trim = m->isTrue(temp);
 373
 374             temp = validParameter.validFile(parameters, "bdiffs", false);               if (temp == "not found") { temp = "0"; }
 375                         m->mothurConvert(temp, bdiffs);
 376
 377                         temp = validParameter.validFile(parameters, "pdiffs", false);           if (temp == "not found") { temp = "0"; }
 378                         m->mothurConvert(temp, pdiffs);
 379
 380             temp = validParameter.validFile(parameters, "ldiffs", false);               if (temp == "not found") { temp = "0"; }
 381                         m->mothurConvert(temp, ldiffs);
 382
 383             temp = validParameter.validFile(parameters, "sdiffs", false);               if (temp == "not found") { temp = "0"; }
 384                         m->mothurConvert(temp, sdiffs);
 385
 386                         temp = validParameter.validFile(parameters, "tdiffs", false);           if (temp == "not found") { int tempTotal = pdiffs + bdiffs + ldiffs + sdiffs;  temp = toString(tempTotal); }
 387                         m->mothurConvert(temp, tdiffs);
 388
 389                         if(tdiffs == 0){        tdiffs = bdiffs + pdiffs + ldiffs + sdiffs;     }
 390
 391                         temp = validParameter.validFile(parameters, "sfftxt", false);
 392                         if (temp == "not found")        {       temp = "F";      sfftxt = false; sfftxtFilename = "";           }
 393                         else if (m->isTrue(temp))       {       sfftxt = true;          sfftxtFilename = "";                            }
 394                         else {
 395                                 //you are a filename
 396                                 if (inputDir != "") {
 397                                         map<string,string>::iterator it = parameters.find("sfftxt");
 398                                         //user has given a template file
 399                                         if(it != parameters.end()){
 400                                                 string path = m->hasPath(it->second);
 401                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 402                                                 if (path == "") {       parameters["sfftxt"] = inputDir + it->second;           }
 403                                         }
 404                                 }
 405
 406                                 sfftxtFilename = validParameter.validFile(parameters, "sfftxt", true);
 407                                 if (sfftxtFilename == "not found") { sfftxtFilename = "";  }
 408                                 else if (sfftxtFilename == "not open") { sfftxtFilename = "";  }
 409                         }
 410
 411                         if ((sfftxtFilename == "") && (filenames.size() == 0)) {
 412                                 //if there is a current sff file, use it
 413                                 string filename = m->getSFFFile();
 414                                 if (filename != "") { filenames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the sff parameter."); m->mothurOutEndLine(); }
 415                                 else {  m->mothurOut("[ERROR]: you must provide a valid sff or sfftxt file."); m->mothurOutEndLine(); abort=true;  }
 416                         }
 417
 418
 419                 }
 420         }
 421         catch(exception& e) {
 422                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
 423                 exit(1);
 424         }
 425 }
 426 //**********************************************************************************************************************
 427 int SffInfoCommand::execute(){
 428         try {
 429                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 430
 431                 for (int s = 0; s < filenames.size(); s++) {
 432
 433                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } return 0; }
 434
 435                         int start = time(NULL);
 436
 437             filenames[s] = m->getFullPathName(filenames[s]);
 438                         m->mothurOut("Extracting info from " + filenames[s] + " ..." ); m->mothurOutEndLine();
 439
 440                         string accnos = "";
 441                         if (hasAccnos) { accnos = accnosFileNames[s]; }
 442
 443             string oligos = "";
 444             if (hasOligos) { oligos = oligosFileNames[s]; }
 445
 446                         int numReads = extractSffInfo(filenames[s], accnos, oligos);
 447
 448                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to extract " + toString(numReads) + ".");
 449                 }
 450
 451                 if (sfftxtFilename != "") {  parseSffTxt(); }
 452
 453                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } return 0; }
 454
 455                 //set fasta file as new current fastafile
 456                 string current = "";
 457                 itTypes = outputTypes.find("fasta");
 458                 if (itTypes != outputTypes.end()) {
 459                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 460                 }
 461
 462                 itTypes = outputTypes.find("qfile");
 463                 if (itTypes != outputTypes.end()) {
 464                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 465                 }
 466
 467                 itTypes = outputTypes.find("flow");
 468                 if (itTypes != outputTypes.end()) {
 469                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFlowFile(current); }
 470                 }
 471
 472                 //report output filenames
 473                 m->mothurOutEndLine();
 474                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 475                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 476                 m->mothurOutEndLine();
 477
 478                 return 0;
 479         }
 480         catch(exception& e) {
 481                 m->errorOut(e, "SffInfoCommand", "execute");
 482                 exit(1);
 483         }
 484 }
 485 //**********************************************************************************************************************
 486 int SffInfoCommand::extractSffInfo(string input, string accnos, string oligos){
 487         try {
 488                 currentFileName = input;
 489                 if (outputDir == "") {  outputDir += m->hasPath(input); }
 490
 491                 if (accnos != "")       {  readAccnosFile(accnos);  }
 492                 else                            {       seqNames.clear();               }
 493
 494         if (oligos != "")   {   readOligos(oligos);  split = 2;   }
 495
 496                 ofstream outSfftxt, outFasta, outQual, outFlow;
 497                 string outFastaFileName, outQualFileName;
 498         string rootName = outputDir + m->getRootName(m->getSimpleName(input));
 499         if(rootName.find_last_of(".") == rootName.npos){ rootName += "."; }
 500
 501         map<string, string> variables;
 502                 variables["[filename]"] = rootName;
 503                 string sfftxtFileName = getOutputFileName("sfftxt",variables);
 504                 string outFlowFileName = getOutputFileName("flow",variables);
 505                 if (!trim) { variables["[tag]"] = "raw"; }
 506                 outFastaFileName = getOutputFileName("fasta",variables);
 507         outQualFileName = getOutputFileName("qfile",variables);
 508
 509                 if (sfftxt) { m->openOutputFile(sfftxtFileName, outSfftxt); outSfftxt.setf(ios::fixed, ios::floatfield); outSfftxt.setf(ios::showpoint);  outputNames.push_back(sfftxtFileName);  outputTypes["sfftxt"].push_back(sfftxtFileName); }
 510                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
 511                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qfile"].push_back(outQualFileName);  }
 512                 if (flow)       { m->openOutputFile(outFlowFileName, outFlow);          outputNames.push_back(outFlowFileName);  outFlow.setf(ios::fixed, ios::floatfield); outFlow.setf(ios::showpoint); outputTypes["flow"].push_back(outFlowFileName);  }
 513
 514                 ifstream in;
 515                 in.open(input.c_str(), ios::binary);
 516
 517                 CommonHeader header;
 518                 readCommonHeader(in, header);
 519
 520                 int count = 0;
 521                 mycount = 0;
 522
 523                 //check magic number and version
 524                 if (header.magicNumber != 779314790) { m->mothurOut("Magic Number is not correct, not a valid .sff file"); m->mothurOutEndLine(); return count; }
 525                 if (header.version != "0001") { m->mothurOut("Version is not supported, only support version 0001."); m->mothurOutEndLine(); return count; }
 526
 527                 //print common header
 528                 if (sfftxt) {   printCommonHeader(outSfftxt, header);           }
 529                 if (flow)       {       outFlow << header.numFlowsPerRead << endl;      }
 530
 531                 //read through the sff file
 532                 while (!in.eof()) {
 533
 534                         bool print = true;
 535
 536                         //read data
 537                         seqRead read;  Header readheader;
 538                         readSeqData(in, read, header.numFlowsPerRead, readheader);
 539             bool okay = sanityCheck(readheader, read);
 540             if (!okay) { break; }
 541
 542                         //if you have provided an accosfile and this seq is not in it, then dont print
 543                         if (seqNames.size() != 0) {   if (seqNames.count(readheader.name) == 0) { print = false; }  }
 544
 545                         //print
 546                         if (print) {
 547                                 if (sfftxt) { printHeader(outSfftxt, readheader); printSffTxtSeqData(outSfftxt, read, readheader); }
 548                                 if (fasta)      {       printFastaSeqData(outFasta, read, readheader);  }
 549                                 if (qual)       {       printQualSeqData(outQual, read, readheader);    }
 550                                 if (flow)       {       printFlowSeqData(outFlow, read, readheader);    }
 551                         }
 552
 553                         count++;
 554                         mycount++;
 555
 556                         //report progress
 557                         if((count+1) % 10000 == 0){     m->mothurOut(toString(count+1)); m->mothurOutEndLine();         }
 558
 559                         if (m->control_pressed) { count = 0; break;   }
 560
 561                         if (count >= header.numReads) { break; }
 562                 }
 563
 564                 //report progress
 565                 if (!m->control_pressed) {   if((count) % 10000 != 0){  m->mothurOut(toString(count)); m->mothurOutEndLine();           }  }
 566
 567                 in.close();
 568
 569                 if (sfftxt) {  outSfftxt.close();       }
 570                 if (fasta)      {  outFasta.close();    }
 571                 if (qual)       {  outQual.close();             }
 572                 if (flow)       {  outFlow.close();             }
 573
 574         if (split > 1) {
 575             //create new common headers for each file with the correct number of reads
 576             adjustCommonHeader(header);
 577
 578                         map<string, string>::iterator it;
 579                         set<string> namesToRemove;
 580                         for(int i=0;i<filehandles.size();i++){
 581                                 for(int j=0;j<filehandles[0].size();j++){
 582                                         if (filehandles[i][j] != "") {
 583                                                 if (namesToRemove.count(filehandles[i][j]) == 0) {
 584                                                         if(m->isBlank(filehandles[i][j])){
 585                                                                 m->mothurRemove(filehandles[i][j]);
 586                                 m->mothurRemove(filehandlesHeaders[i][j]);
 587                                                                 namesToRemove.insert(filehandles[i][j]);
 588                             }
 589                                                 }
 590                                         }
 591                                 }
 592                         }
 593
 594             //append new header to reads
 595             for (int i = 0; i < filehandles.size(); i++) {
 596                 for (int j = 0; j < filehandles[i].size(); j++) {
 597                     m->appendFiles(filehandles[i][j], filehandlesHeaders[i][j]);
 598                     m->renameFile(filehandlesHeaders[i][j], filehandles[i][j]);
 599                     m->mothurRemove(filehandlesHeaders[i][j]);
 600                     if (numSplitReads[i][j] == 0) { m->mothurRemove(filehandles[i][j]); }
 601                 }
 602             }
 603
 604                         //remove names for outputFileNames, just cleans up the output
 605                         for(int i = 0; i < outputNames.size(); i++) {
 606                 if (namesToRemove.count(outputNames[i]) != 0) {
 607                     outputNames.erase(outputNames.begin()+i);
 608                     i--;
 609                 }
 610             }
 611
 612             if(m->isBlank(noMatchFile)){  m->mothurRemove(noMatchFile); }
 613             else { outputNames.push_back(noMatchFile); outputTypes["sff"].push_back(noMatchFile); }
 614         }
 615
 616                 return count;
 617         }
 618         catch(exception& e) {
 619                 m->errorOut(e, "SffInfoCommand", "extractSffInfo");
 620                 exit(1);
 621         }
 622 }
 623 //**********************************************************************************************************************
 624 int SffInfoCommand::readCommonHeader(ifstream& in, CommonHeader& header){
 625         try {
 626
 627                 if (!in.eof()) {
 628
 629                         //read magic number
 630                         char buffer[4];
 631                         in.read(buffer, 4);
 632                         header.magicNumber = be_int4(*(unsigned int *)(&buffer));
 633
 634                         //read version
 635                         char buffer9[4];
 636                         in.read(buffer9, 4);
 637                         header.version = "";
 638                         for (int i = 0; i < 4; i++) {  header.version += toString((int)(buffer9[i]));  }
 639
 640                         //read offset
 641                         char buffer2 [8];
 642                         in.read(buffer2, 8);
 643                         header.indexOffset =  be_int8(*(unsigned long long *)(&buffer2));
 644
 645                         //read index length
 646                         char buffer3 [4];
 647                         in.read(buffer3, 4);
 648                         header.indexLength =  be_int4(*(unsigned int *)(&buffer3));
 649
 650                         //read num reads
 651                         char buffer4 [4];
 652                         in.read(buffer4, 4);
 653                         header.numReads =  be_int4(*(unsigned int *)(&buffer4));
 654
 655                         //read header length
 656                         char buffer5 [2];
 657                         in.read(buffer5, 2);
 658                         header.headerLength =  be_int2(*(unsigned short *)(&buffer5));
 659
 660                         //read key length
 661                         char buffer6 [2];
 662                         in.read(buffer6, 2);
 663                         header.keyLength = be_int2(*(unsigned short *)(&buffer6));
 664
 665                         //read number of flow reads
 666                         char buffer7 [2];
 667                         in.read(buffer7, 2);
 668                         header.numFlowsPerRead =  be_int2(*(unsigned short *)(&buffer7));
 669
 670                         //read format code
 671                         char buffer8 [1];
 672                         in.read(buffer8, 1);
 673                         header.flogramFormatCode = (int)(buffer8[0]);
 674
 675                         //read flow chars
 676                         char* tempBuffer = new char[header.numFlowsPerRead];
 677                         in.read(&(*tempBuffer), header.numFlowsPerRead);
 678                         header.flowChars = tempBuffer;
 679                         if (header.flowChars.length() > header.numFlowsPerRead) { header.flowChars = header.flowChars.substr(0, header.numFlowsPerRead);  }
 680                         delete[] tempBuffer;
 681
 682                         //read key
 683                         char* tempBuffer2 = new char[header.keyLength];
 684                         in.read(&(*tempBuffer2), header.keyLength);
 685                         header.keySequence = tempBuffer2;
 686                         if (header.keySequence.length() > header.keyLength) { header.keySequence = header.keySequence.substr(0, header.keyLength);  }
 687                         delete[] tempBuffer2;
 688
 689                         /* Pad to 8 chars */
 690                         unsigned long long spotInFile = in.tellg();
 691                         unsigned long long spot = (spotInFile + 7)& ~7;  // ~ inverts
 692                         in.seekg(spot);
 693
 694         }else{
 695                         m->mothurOut("Error reading sff common header."); m->mothurOutEndLine();
 696                 }
 697
 698                 return 0;
 699
 700         }
 701         catch(exception& e) {
 702                 m->errorOut(e, "SffInfoCommand", "readCommonHeader");
 703                 exit(1);
 704         }
 705 }
 706 //**********************************************************************************************************************
 707 int SffInfoCommand::adjustCommonHeader(CommonHeader header){
 708         try {
 709
 710         char* mybuffer = new char[4];
 711         ifstream in;
 712         in.open(currentFileName.c_str(), ios::binary);
 713
 714         //magic number
 715         in.read(mybuffer,4);
 716         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 717             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 718                 ofstream out;
 719                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 720                 out.write(mybuffer, in.gcount());
 721                 out.close();
 722             }
 723         }
 724         delete[] mybuffer;
 725
 726         //version
 727         mybuffer = new char[4];
 728         in.read(mybuffer,4);
 729         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 730             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 731                 ofstream out;
 732                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 733                 out.write(mybuffer, in.gcount());
 734                 out.close();
 735             }
 736         }
 737         delete[] mybuffer;
 738
 739         //offset
 740         mybuffer = new char[8];
 741         in.read(mybuffer,8);
 742         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 743             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 744                 unsigned long long offset = 0;
 745                 char* thisbuffer = new char[8];
 746                 thisbuffer[0] = (offset >> 56) & 0xFF;
 747                 thisbuffer[1] = (offset >> 48) & 0xFF;
 748                 thisbuffer[2] = (offset >> 40) & 0xFF;
 749                 thisbuffer[3] = (offset >> 32) & 0xFF;
 750                 thisbuffer[4] = (offset >> 24) & 0xFF;
 751                 thisbuffer[5] = (offset >> 16) & 0xFF;
 752                 thisbuffer[6] = (offset >> 8) & 0xFF;
 753                 thisbuffer[7] = offset & 0xFF;
 754                 ofstream out;
 755                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 756                 out.write(thisbuffer, 8);
 757                 out.close();
 758             }
 759         }
 760         delete[] mybuffer;
 761
 762
 763         //read index length
 764                 mybuffer = new char[4];
 765         in.read(mybuffer,4);
 766         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 767             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 768                 ofstream out;
 769                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 770                 int offset = 0;
 771                 char* thisbuffer = new char[4];
 772                 thisbuffer[0] = (offset >> 24) & 0xFF;
 773                 thisbuffer[1] = (offset >> 16) & 0xFF;
 774                 thisbuffer[2] = (offset >> 8) & 0xFF;
 775                 thisbuffer[3] = offset & 0xFF;
 776                 out.write(thisbuffer, 4);
 777                 out.close();
 778             }
 779         }
 780         delete[] mybuffer;
 781
 782         //change num reads
 783         mybuffer = new char[4];
 784         in.read(mybuffer,4);
 785         delete[] mybuffer;
 786         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 787             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 788                 ofstream out;
 789                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 790                 //convert number of reads to 4 byte char*
 791                 char* thisbuffer = new char[4];
 792                 thisbuffer[0] = (numSplitReads[i][j] >> 24) & 0xFF;
 793                 thisbuffer[1] = (numSplitReads[i][j] >> 16) & 0xFF;
 794                 thisbuffer[2] = (numSplitReads[i][j] >> 8) & 0xFF;
 795                 thisbuffer[3] = numSplitReads[i][j] & 0xFF;
 796                 out.write(thisbuffer, 4);
 797                 out.close();
 798                 delete[] thisbuffer;
 799             }
 800         }
 801
 802         //read header length
 803         mybuffer = new char[2];
 804         in.read(mybuffer,2);
 805         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 806             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 807                 ofstream out;
 808                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 809                 out.write(mybuffer, in.gcount());
 810                 out.close();
 811             }
 812         }
 813         delete[] mybuffer;
 814
 815         //read key length
 816         mybuffer = new char[2];
 817         in.read(mybuffer,2);
 818         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 819             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 820                 ofstream out;
 821                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 822                 out.write(mybuffer, in.gcount());
 823                 out.close();
 824             }
 825         }
 826         delete[] mybuffer;
 827
 828         //read number of flow reads
 829         mybuffer = new char[2];
 830         in.read(mybuffer,2);
 831         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 832             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 833                 ofstream out;
 834                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 835                 out.write(mybuffer, in.gcount());
 836                 out.close();
 837             }
 838         }
 839         delete[] mybuffer;
 840
 841         //read format code
 842         mybuffer = new char[1];
 843         in.read(mybuffer,1);
 844         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 845             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 846                 ofstream out;
 847                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 848                 out.write(mybuffer, in.gcount());
 849                 out.close();
 850             }
 851         }
 852         delete[] mybuffer;
 853
 854         //read flow chars
 855         mybuffer = new char[header.numFlowsPerRead];
 856         in.read(mybuffer,header.numFlowsPerRead);
 857         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 858             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 859                 ofstream out;
 860                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 861                 out.write(mybuffer, in.gcount());
 862                 out.close();
 863             }
 864         }
 865         delete[] mybuffer;
 866
 867         //read key
 868         mybuffer = new char[header.keyLength];
 869         in.read(mybuffer,header.keyLength);
 870         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 871             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 872                 ofstream out;
 873                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 874                 out.write(mybuffer, in.gcount());
 875                 out.close();
 876             }
 877         }
 878         delete[] mybuffer;
 879
 880
 881         /* Pad to 8 chars */
 882         unsigned long long spotInFile = in.tellg();
 883         unsigned long long spot = (spotInFile + 7)& ~7;  // ~ inverts
 884         in.seekg(spot);
 885
 886         mybuffer = new char[spot-spotInFile];
 887         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 888             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 889                 ofstream out;
 890                 m->openOutputFileBinaryAppend(filehandlesHeaders[i][j], out);
 891                 out.write(mybuffer, spot-spotInFile);
 892                 out.close();
 893             }
 894         }
 895         delete[] mybuffer;
 896         in.close();
 897                 return 0;
 898
 899         }
 900         catch(exception& e) {
 901                 m->errorOut(e, "SffInfoCommand", "adjustCommonHeader");
 902                 exit(1);
 903         }
 904 }
 905 //**********************************************************************************************************************
 906 int SffInfoCommand::readSeqData(ifstream& in, seqRead& read, int numFlowReads, Header& header){
 907         try {
 908         unsigned long long startSpotInFile = in.tellg();
 909                 if (!in.eof()) {
 910
 911             /*****************************************/
 912             //read header
 913
 914             //read header length
 915                         char buffer [2];
 916                         in.read(buffer, 2);
 917                         header.headerLength = be_int2(*(unsigned short *)(&buffer));
 918
 919                         //read name length
 920                         char buffer2 [2];
 921                         in.read(buffer2, 2);
 922                         header.nameLength = be_int2(*(unsigned short *)(&buffer2));
 923
 924                         //read num bases
 925                         char buffer3 [4];
 926                         in.read(buffer3, 4);
 927                         header.numBases =  be_int4(*(unsigned int *)(&buffer3));
 928
 929                         //read clip qual left
 930                         char buffer4 [2];
 931                         in.read(buffer4, 2);
 932                         header.clipQualLeft =  be_int2(*(unsigned short *)(&buffer4));
 933                         header.clipQualLeft = 5;
 934
 935                         //read clip qual right
 936                         char buffer5 [2];
 937                         in.read(buffer5, 2);
 938                         header.clipQualRight =  be_int2(*(unsigned short *)(&buffer5));
 939
 940                         //read clipAdapterLeft
 941                         char buffer6 [2];
 942                         in.read(buffer6, 2);
 943                         header.clipAdapterLeft = be_int2(*(unsigned short *)(&buffer6));
 944
 945                         //read clipAdapterRight
 946                         char buffer7 [2];
 947                         in.read(buffer7, 2);
 948                         header.clipAdapterRight = be_int2(*(unsigned short *)(&buffer7));
 949
 950                         //read name
 951                         char* tempBuffer = new char[header.nameLength];
 952                         in.read(&(*tempBuffer), header.nameLength);
 953                         header.name = tempBuffer;
 954                         if (header.name.length() > header.nameLength) { header.name = header.name.substr(0, header.nameLength);  }
 955                         delete[] tempBuffer;
 956
 957                         //extract info from name
 958                         decodeName(header.timestamp, header.region, header.xy, header.name);
 959
 960                         /* Pad to 8 chars */
 961                         unsigned long long spotInFile = in.tellg();
 962                         unsigned long long spot = (spotInFile + 7)& ~7;
 963                         in.seekg(spot);
 964
 965             /*****************************************/
 966             //sequence read
 967
 968                         //read flowgram
 969                         read.flowgram.resize(numFlowReads);
 970                         for (int i = 0; i < numFlowReads; i++) {
 971                                 char buffer [2];
 972                                 in.read(buffer, 2);
 973                                 read.flowgram[i] = be_int2(*(unsigned short *)(&buffer));
 974                         }
 975
 976                         //read flowIndex
 977                         read.flowIndex.resize(header.numBases);
 978                         for (int i = 0; i < header.numBases; i++) {
 979                                 char temp[1];
 980                                 in.read(temp, 1);
 981                                 read.flowIndex[i] = be_int1(*(unsigned char *)(&temp));
 982                         }
 983
 984                         //read bases
 985                         char* tempBuffer6 = new char[header.numBases];
 986                         in.read(&(*tempBuffer6), header.numBases);
 987                         read.bases = tempBuffer6;
 988                         if (read.bases.length() > header.numBases) { read.bases = read.bases.substr(0, header.numBases);  }
 989                         delete[] tempBuffer6;
 990
 991                         //read qual scores
 992                         read.qualScores.resize(header.numBases);
 993                         for (int i = 0; i < header.numBases; i++) {
 994                                 char temp[1];
 995                                 in.read(temp, 1);
 996                                 read.qualScores[i] = be_int1(*(unsigned char *)(&temp));
 997                         }
 998
 999                         /* Pad to 8 chars */
1000                         spotInFile = in.tellg();
1001                         spot = (spotInFile + 7)& ~7;
1002                         in.seekg(spot);
1003
1004             if (split > 1) {
1005                 char * mybuffer;
1006                 mybuffer = new char [spot-startSpotInFile];
1007                 ifstream in2;
1008                 in2.open(currentFileName.c_str(), ios::binary);
1009                 in2.seekg(startSpotInFile);
1010                 in2.read(mybuffer,spot-startSpotInFile);
1011                 in2.close();
1012
1013                 int barcodeIndex, primerIndex;
1014                 int trashCodeLength = findGroup(header, read, barcodeIndex, primerIndex);
1015
1016                 if(trashCodeLength == 0){
1017                     ofstream out;
1018                     m->openOutputFileBinaryAppend(filehandles[barcodeIndex][primerIndex], out);
1019                     out.write(mybuffer, in2.gcount());
1020                     out.close();
1021                     delete[] mybuffer;
1022                     numSplitReads[barcodeIndex][primerIndex]++;
1023                                 }
1024                                 else{
1025                                         ofstream out;
1026                     m->openOutputFileBinaryAppend(noMatchFile, out);
1027                     out.write(mybuffer, in2.gcount());
1028                     out.close();
1029                     delete[] mybuffer;
1030                                 }
1031
1032                         }
1033                 }else{
1034                         m->mothurOut("Error reading."); m->mothurOutEndLine();
1035                 }
1036
1037                 return 0;
1038         }
1039         catch(exception& e) {
1040                 m->errorOut(e, "SffInfoCommand", "readSeqData");
1041                 exit(1);
1042         }
1043 }
1044 //**********************************************************************************************************************
1045 int SffInfoCommand::findGroup(Header header, seqRead read, int& barcode, int& primer) {
1046         try {
1047         //find group read belongs to
1048         TrimOligos trimOligos(pdiffs, bdiffs, ldiffs, sdiffs, primers, barcodes, revPrimer, linker, spacer);
1049
1050         int success = 1;
1051         string trashCode = "";
1052         int currentSeqsDiffs = 0;
1053
1054         string seq = read.bases;
1055
1056         if (trim) {
1057             if(header.clipQualRight < header.clipQualLeft){
1058                 if (header.clipQualRight == 0) { //don't trim right
1059                     seq = seq.substr(header.clipQualLeft-1);
1060                 }else {
1061                     seq = "NNNN";
1062                 }
1063             }
1064             else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
1065                 seq = seq.substr((header.clipQualLeft-1), (header.clipQualRight-header.clipQualLeft));
1066             }
1067             else {
1068                 seq = seq.substr(header.clipQualLeft-1);
1069             }
1070         }else{
1071             //if you wanted the sfftxt then you already converted the bases to the right case
1072             if (!sfftxt) {
1073                 int endValue = header.clipQualRight;
1074                 //make the bases you want to clip lowercase and the bases you want to keep upper case
1075                 if(endValue == 0){      endValue = seq.length();        }
1076                 for (int i = 0; i < (header.clipQualLeft-1); i++) { seq[i] = tolower(seq[i]);  }
1077                 for (int i = (header.clipQualLeft-1); i < (endValue-1); i++)  {   seq[i] = toupper(seq[i]);  }
1078                 for (int i = (endValue-1); i < seq.length(); i++) {   seq[i] = tolower(seq[i]);  }
1079             }
1080         }
1081
1082         Sequence currSeq(header.name, seq);
1083         QualityScores currQual;
1084
1085         if(numLinkers != 0){
1086             success = trimOligos.stripLinker(currSeq, currQual);
1087             if(success > ldiffs)                {       trashCode += 'k';       }
1088             else{ currentSeqsDiffs += success;  }
1089
1090         }
1091
1092         if(barcodes.size() != 0){
1093             success = trimOligos.stripBarcode(currSeq, currQual, barcode);
1094             if(success > bdiffs)                {       trashCode += 'b';       }
1095             else{ currentSeqsDiffs += success;  }
1096         }
1097
1098         if(numSpacers != 0){
1099             success = trimOligos.stripSpacer(currSeq, currQual);
1100             if(success > sdiffs)                {       trashCode += 's';       }
1101             else{ currentSeqsDiffs += success;  }
1102
1103         }
1104
1105         if(numFPrimers != 0){
1106             success = trimOligos.stripForward(currSeq, currQual, primer, true);
1107             if(success > pdiffs)                {       trashCode += 'f';       }
1108             else{ currentSeqsDiffs += success;  }
1109         }
1110
1111         if (currentSeqsDiffs > tdiffs)  {       trashCode += 't';   }
1112
1113         if(revPrimer.size() != 0){
1114             success = trimOligos.stripReverse(currSeq, currQual);
1115             if(!success)                                {       trashCode += 'r';       }
1116         }
1117
1118
1119         return trashCode.length();
1120     }
1121         catch(exception& e) {
1122                 m->errorOut(e, "SffInfoCommand", "findGroup");
1123                 exit(1);
1124         }
1125 }
1126 //**********************************************************************************************************************
1127 int SffInfoCommand::decodeName(string& timestamp, string& region, string& xy, string name) {
1128         try {
1129
1130                 if (name.length() >= 6) {
1131                         string time = name.substr(0, 6);
1132                         unsigned int timeNum = m->fromBase36(time);
1133
1134                         int q1 = timeNum / 60;
1135                         int sec = timeNum - 60 * q1;
1136                         int q2 = q1 / 60;
1137                         int minute = q1 - 60 * q2;
1138                         int q3 = q2 / 24;
1139                         int hr = q2 - 24 * q3;
1140                         int q4 = q3 / 32;
1141                         int day = q3 - 32 * q4;
1142                         int q5 = q4 / 13;
1143                         int mon = q4 - 13 * q5;
1144                         int year = 2000 + q5;
1145
1146                         timestamp = toString(year) + "_" + toString(mon) + "_" + toString(day) + "_" + toString(hr) + "_" + toString(minute) + "_" + toString(sec);
1147                 }
1148
1149                 if (name.length() >= 9) {
1150                         region = name.substr(7, 2);
1151
1152                         string xyNum = name.substr(9);
1153                         unsigned int myXy = m->fromBase36(xyNum);
1154                         int x = myXy >> 12;
1155                         int y = myXy & 4095;
1156
1157                         xy = toString(x) + "_" + toString(y);
1158                 }
1159
1160                 return 0;
1161         }
1162         catch(exception& e) {
1163                 m->errorOut(e, "SffInfoCommand", "decodeName");
1164                 exit(1);
1165         }
1166 }
1167 //**********************************************************************************************************************
1168 int SffInfoCommand::printCommonHeader(ofstream& out, CommonHeader& header) {
1169         try {
1170
1171                 out << "Common Header:\nMagic Number: " << header.magicNumber << endl;
1172                 out << "Version: " << header.version << endl;
1173                 out << "Index Offset: " << header.indexOffset << endl;
1174                 out << "Index Length: " << header.indexLength << endl;
1175                 out << "Number of Reads: " << header.numReads << endl;
1176                 out << "Header Length: " << header.headerLength << endl;
1177                 out << "Key Length: " << header.keyLength << endl;
1178                 out << "Number of Flows: " << header.numFlowsPerRead << endl;
1179                 out << "Format Code: " << header.flogramFormatCode << endl;
1180                 out << "Flow Chars: " << header.flowChars << endl;
1181                 out << "Key Sequence: " << header.keySequence << endl << endl;
1182
1183                 return 0;
1184         }
1185         catch(exception& e) {
1186                 m->errorOut(e, "SffInfoCommand", "printCommonHeader");
1187                 exit(1);
1188         }
1189 }
1190 //**********************************************************************************************************************
1191 int SffInfoCommand::printHeader(ofstream& out, Header& header) {
1192         try {
1193
1194                 out << ">" << header.name << endl;
1195                 out << "Run Prefix: " << header.timestamp << endl;
1196                 out << "Region #:  " << header.region << endl;
1197                 out << "XY Location: " << header.xy << endl << endl;
1198
1199                 out << "Run Name:  " << endl;
1200                 out << "Analysis Name:  " << endl;
1201                 out << "Full Path: " << endl << endl;
1202
1203                 out << "Read Header Len: " << header.headerLength << endl;
1204                 out << "Name Length: " << header.nameLength << endl;
1205                 out << "# of Bases: " << header.numBases << endl;
1206                 out << "Clip Qual Left: " << header.clipQualLeft << endl;
1207                 out << "Clip Qual Right: " << header.clipQualRight << endl;
1208                 out << "Clip Adap Left: " << header.clipAdapterLeft << endl;
1209                 out << "Clip Adap Right: " << header.clipAdapterRight << endl << endl;
1210
1211                 return 0;
1212         }
1213         catch(exception& e) {
1214                 m->errorOut(e, "SffInfoCommand", "printHeader");
1215                 exit(1);
1216         }
1217 }
1218 //**********************************************************************************************************************
1219 bool SffInfoCommand::sanityCheck(Header& header, seqRead& read) {
1220         try {
1221         bool okay = true;
1222         string message = "[WARNING]: Your sff file may be corrupted! Sequence: " + header.name + "\n";
1223
1224         if (header.clipQualLeft > read.bases.length()) {
1225             okay = false; message += "Clip Qual Left = " + toString(header.clipQualLeft) + ", but we only read " + toString(read.bases.length()) + " bases.\n";
1226         }
1227         if (header.clipQualRight > read.bases.length()) {
1228             okay = false; message += "Clip Qual Right = " + toString(header.clipQualRight) + ", but we only read " + toString(read.bases.length()) + " bases.\n";
1229         }
1230         if (header.clipQualLeft > read.qualScores.size()) {
1231             okay = false; message += "Clip Qual Left = " + toString(header.clipQualLeft) + ", but we only read " + toString(read.qualScores.size()) + " quality scores.\n";
1232         }
1233         if (header.clipQualRight > read.qualScores.size()) {
1234             okay = false; message += "Clip Qual Right = " + toString(header.clipQualRight) + ", but we only read " + toString(read.qualScores.size()) + " quality scores.\n";
1235         }
1236
1237         if (okay == false) {
1238             m->mothurOut(message); m->mothurOutEndLine();
1239         }
1240
1241                 return okay;
1242         }
1243         catch(exception& e) {
1244                 m->errorOut(e, "SffInfoCommand", "sanityCheck");
1245                 exit(1);
1246         }
1247 }
1248 //**********************************************************************************************************************
1249 int SffInfoCommand::printSffTxtSeqData(ofstream& out, seqRead& read, Header& header) {
1250         try {
1251                 out << "Flowgram: ";
1252                 for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << (read.flowgram[i]/(float)100) << '\t';  }
1253
1254                 out << endl <<  "Flow Indexes: ";
1255                 int sum = 0;
1256                 for (int i = 0; i < read.flowIndex.size(); i++) {  sum +=  read.flowIndex[i];  out << sum << '\t'; }
1257
1258                 //make the bases you want to clip lowercase and the bases you want to keep upper case
1259         int endValue = header.clipQualRight;
1260                 if(endValue == 0){      endValue = read.bases.length(); }
1261                 for (int i = 0; i < (header.clipQualLeft-1); i++) { read.bases[i] = tolower(read.bases[i]); }
1262                 for (int i = (header.clipQualLeft-1); i < (endValue-1); i++) {   read.bases[i] = toupper(read.bases[i]);  }
1263                 for (int i = (endValue-1); i < read.bases.length(); i++) {   read.bases[i] = tolower(read.bases[i]);  }
1264
1265                 out << endl <<  "Bases: " << read.bases << endl << "Quality Scores: ";
1266                 for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
1267
1268
1269                 out << endl << endl;
1270
1271                 return 0;
1272         }
1273         catch(exception& e) {
1274                 m->errorOut(e, "SffInfoCommand", "printSffTxtSeqData");
1275                 exit(1);
1276         }
1277 }
1278 //**********************************************************************************************************************
1279 int SffInfoCommand::printFastaSeqData(ofstream& out, seqRead& read, Header& header) {
1280         try {
1281                 string seq = read.bases;
1282
1283         if (trim) {
1284                         if(header.clipQualRight < header.clipQualLeft){
1285                                 if (header.clipQualRight == 0) { //don't trim right
1286                     seq = seq.substr(header.clipQualLeft-1);
1287                 }else {
1288                     seq = "NNNN";
1289                 }
1290                         }
1291                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
1292                                 seq = seq.substr((header.clipQualLeft-1), (header.clipQualRight-header.clipQualLeft));
1293                         }
1294                         else {
1295                                 seq = seq.substr(header.clipQualLeft-1);
1296                         }
1297                 }else{
1298                         //if you wanted the sfftxt then you already converted the bases to the right case
1299                         if (!sfftxt) {
1300                 int endValue = header.clipQualRight;
1301                                 //make the bases you want to clip lowercase and the bases you want to keep upper case
1302                                 if(endValue == 0){      endValue = seq.length();        }
1303                                 for (int i = 0; i < (header.clipQualLeft-1); i++) { seq[i] = tolower(seq[i]);  }
1304                                 for (int i = (header.clipQualLeft-1); i < (endValue-1); i++)  {   seq[i] = toupper(seq[i]);  }
1305                                 for (int i = (endValue-1); i < seq.length(); i++) {   seq[i] = tolower(seq[i]);  }
1306                         }
1307                 }
1308
1309                 out << ">" << header.name  << " xy=" << header.xy << endl;
1310                 out << seq << endl;
1311
1312                 return 0;
1313         }
1314         catch(exception& e) {
1315                 m->errorOut(e, "SffInfoCommand", "printFastaSeqData");
1316                 exit(1);
1317         }
1318 }
1319
1320 //**********************************************************************************************************************
1321 int SffInfoCommand::printQualSeqData(ofstream& out, seqRead& read, Header& header) {
1322         try {
1323
1324                 if (trim) {
1325                         if(header.clipQualRight < header.clipQualLeft){
1326                 if (header.clipQualRight == 0) { //don't trim right
1327                     out << ">" << header.name << " xy=" << header.xy << " length=" << (read.qualScores.size()-header.clipQualLeft) << endl;
1328                     for (int i = (header.clipQualLeft-1); i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';       }
1329                 }else {
1330                     out << ">" << header.name << " xy=" << header.xy << endl;
1331                     out << "0\t0\t0\t0";
1332                 }
1333                         }
1334                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
1335                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
1336                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   out << read.qualScores[i] << '\t'; }
1337                         }
1338                         else{
1339                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
1340                                 for (int i = (header.clipQualLeft-1); i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';   }
1341                         }
1342                 }else{
1343                         out << ">" << header.name << " xy=" << header.xy << " length=" << read.qualScores.size() << endl;
1344                         for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
1345                 }
1346
1347                 out << endl;
1348
1349                 return 0;
1350         }
1351         catch(exception& e) {
1352                 m->errorOut(e, "SffInfoCommand", "printQualSeqData");
1353                 exit(1);
1354         }
1355 }
1356
1357 //**********************************************************************************************************************
1358 int SffInfoCommand::printFlowSeqData(ofstream& out, seqRead& read, Header& header) {
1359         try {
1360
1361         int endValue = header.clipQualRight;
1362         if (header.clipQualRight == 0) {
1363             endValue = read.flowIndex.size();
1364             if (m->debug) { m->mothurOut("[DEBUG]: " + header.name + " has clipQualRight=0.\n"); }
1365         }
1366         if(endValue > header.clipQualLeft){
1367
1368             int rightIndex = 0;
1369             for (int i = 0; i < endValue; i++) {  rightIndex +=  read.flowIndex[i];      }
1370
1371             out << header.name << ' ' << rightIndex;
1372             for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << ' ' << (read.flowgram[i]/(float)100);  }
1373             out << endl;
1374         }
1375
1376
1377                 return 0;
1378         }
1379         catch(exception& e) {
1380                 m->errorOut(e, "SffInfoCommand", "printFlowSeqData");
1381                 exit(1);
1382         }
1383 }
1384 //**********************************************************************************************************************
1385 int SffInfoCommand::readAccnosFile(string filename) {
1386         try {
1387                 //remove old names
1388                 seqNames.clear();
1389
1390                 ifstream in;
1391                 m->openInputFile(filename, in);
1392                 string name;
1393
1394                 while(!in.eof()){
1395                         in >> name; m->gobble(in);
1396
1397                         seqNames.insert(name);
1398
1399                         if (m->control_pressed) { seqNames.clear(); break; }
1400                 }
1401                 in.close();
1402
1403                 return 0;
1404         }
1405         catch(exception& e) {
1406                 m->errorOut(e, "SffInfoCommand", "readAccnosFile");
1407                 exit(1);
1408         }
1409 }
1410 //**********************************************************************************************************************
1411 int SffInfoCommand::parseSffTxt() {
1412         try {
1413
1414                 ifstream inSFF;
1415                 m->openInputFile(sfftxtFilename, inSFF);
1416
1417                 if (outputDir == "") {  outputDir += m->hasPath(sfftxtFilename); }
1418
1419                 //output file names
1420                 ofstream outFasta, outQual, outFlow;
1421                 string outFastaFileName, outQualFileName;
1422                 string fileRoot = m->getRootName(m->getSimpleName(sfftxtFilename));
1423                 if (fileRoot.length() > 0) {
1424                         //rip off last .
1425                         fileRoot = fileRoot.substr(0, fileRoot.length()-1);
1426                         fileRoot = m->getRootName(fileRoot);
1427                 }
1428
1429         map<string, string> variables;
1430                 variables["[filename]"] = fileRoot;
1431                 string sfftxtFileName = getOutputFileName("sfftxt",variables);
1432                 string outFlowFileName = getOutputFileName("flow",variables);
1433                 if (!trim) { variables["[tag]"] = "raw"; }
1434                 outFastaFileName = getOutputFileName("fasta",variables);
1435         outQualFileName = getOutputFileName("qfile",variables);
1436
1437                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
1438                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qfile"].push_back(outQualFileName);  }
1439                 if (flow)       { m->openOutputFile(outFlowFileName, outFlow);          outputNames.push_back(outFlowFileName);  outFlow.setf(ios::fixed, ios::floatfield); outFlow.setf(ios::showpoint); outputTypes["flow"].push_back(outFlowFileName);  }
1440
1441                 //read common header
1442                 string commonHeader = m->getline(inSFF);
1443                 string magicNumber = m->getline(inSFF);
1444                 string version = m->getline(inSFF);
1445                 string indexOffset = m->getline(inSFF);
1446                 string indexLength = m->getline(inSFF);
1447                 int numReads = parseHeaderLineToInt(inSFF);
1448                 string headerLength = m->getline(inSFF);
1449                 string keyLength = m->getline(inSFF);
1450                 int numFlows = parseHeaderLineToInt(inSFF);
1451                 string flowgramCode = m->getline(inSFF);
1452                 string flowChars = m->getline(inSFF);
1453                 string keySequence = m->getline(inSFF);
1454                 m->gobble(inSFF);
1455
1456                 string seqName;
1457
1458                 if (flow)       {       outFlow << numFlows << endl;    }
1459
1460                 for(int i=0;i<numReads;i++){
1461
1462                         //sanity check
1463                         if (inSFF.eof()) { m->mothurOut("[ERROR]: Expected " + toString(numReads) + " but reached end of file at " + toString(i+1) + "."); m->mothurOutEndLine(); break; }
1464
1465                         Header header;
1466
1467                         //parse read header
1468                         inSFF >> seqName;
1469                         seqName = seqName.substr(1);
1470                         m->gobble(inSFF);
1471                         header.name = seqName;
1472
1473                         string runPrefix = parseHeaderLineToString(inSFF);              header.timestamp = runPrefix;
1474                         string regionNumber = parseHeaderLineToString(inSFF);   header.region = regionNumber;
1475                         string xyLocation = parseHeaderLineToString(inSFF);             header.xy = xyLocation;
1476                         m->gobble(inSFF);
1477
1478                         string runName = parseHeaderLineToString(inSFF);
1479                         string analysisName = parseHeaderLineToString(inSFF);
1480                         string fullPath = parseHeaderLineToString(inSFF);
1481                         m->gobble(inSFF);
1482
1483                         string readHeaderLen = parseHeaderLineToString(inSFF);  convert(readHeaderLen, header.headerLength);
1484                         string nameLength = parseHeaderLineToString(inSFF);             convert(nameLength, header.nameLength);
1485                         int numBases = parseHeaderLineToInt(inSFF);                             header.numBases = numBases;
1486                         string clipQualLeft = parseHeaderLineToString(inSFF);   convert(clipQualLeft, header.clipQualLeft);
1487                         int clipQualRight = parseHeaderLineToInt(inSFF);                header.clipQualRight = clipQualRight;
1488                         string clipAdapLeft = parseHeaderLineToString(inSFF);   convert(clipAdapLeft, header.clipAdapterLeft);
1489                         string clipAdapRight = parseHeaderLineToString(inSFF);  convert(clipAdapRight, header.clipAdapterRight);
1490                         m->gobble(inSFF);
1491
1492                         seqRead read;
1493
1494                         //parse read
1495                         vector<unsigned short> flowVector = parseHeaderLineToFloatVector(inSFF, numFlows);      read.flowgram = flowVector;
1496                         vector<unsigned int> flowIndices = parseHeaderLineToIntVector(inSFF, numBases);
1497
1498                         //adjust for print
1499                         vector<unsigned int> flowIndicesAdjusted; flowIndicesAdjusted.push_back(flowIndices[0]);
1500                         for (int j = 1; j < flowIndices.size(); j++) {   flowIndicesAdjusted.push_back(flowIndices[j] - flowIndices[j-1]);   }
1501                         read.flowIndex = flowIndicesAdjusted;
1502
1503                         string bases = parseHeaderLineToString(inSFF);                                                                          read.bases = bases;
1504                         vector<unsigned int> qualityScores = parseHeaderLineToIntVector(inSFF, numBases);       read.qualScores = qualityScores;
1505                         m->gobble(inSFF);
1506
1507                         //if you have provided an accosfile and this seq is not in it, then dont print
1508                         bool print = true;
1509                         if (seqNames.size() != 0) {   if (seqNames.count(header.name) == 0) { print = false; }  }
1510
1511                         //print
1512                         if (print) {
1513                                 if (fasta)      {       printFastaSeqData(outFasta, read, header);      }
1514                                 if (qual)       {       printQualSeqData(outQual, read, header);        }
1515                                 if (flow)       {       printFlowSeqData(outFlow, read, header);        }
1516                         }
1517
1518                         //report progress
1519                         if((i+1) % 10000 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine();             }
1520
1521                         if (m->control_pressed) {  break;  }
1522                 }
1523
1524                 //report progress
1525                 if (!m->control_pressed) {   if((numReads) % 10000 != 0){       m->mothurOut(toString(numReads)); m->mothurOutEndLine();                }  }
1526
1527                 inSFF.close();
1528
1529                 if (fasta)      {  outFasta.close();    }
1530                 if (qual)       {  outQual.close();             }
1531                 if (flow)       {  outFlow.close();             }
1532
1533                 return 0;
1534         }
1535         catch(exception& e) {
1536                 m->errorOut(e, "SffInfoCommand", "parseSffTxt");
1537                 exit(1);
1538         }
1539 }
1540 //**********************************************************************************************************************
1541
1542 int SffInfoCommand::parseHeaderLineToInt(ifstream& file){
1543         try {
1544                 int number;
1545
1546                 while (!file.eof())     {
1547
1548                         char c = file.get();
1549                         if (c == ':'){
1550                                 file >> number;
1551                                 break;
1552                         }
1553
1554                 }
1555                 m->gobble(file);
1556                 return number;
1557         }
1558         catch(exception& e) {
1559                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToInt");
1560                 exit(1);
1561         }
1562
1563 }
1564
1565 //**********************************************************************************************************************
1566
1567 string SffInfoCommand::parseHeaderLineToString(ifstream& file){
1568         try {
1569                 string text;
1570
1571                 while (!file.eof())     {
1572                         char c = file.get();
1573
1574                         if (c == ':'){
1575                                 //m->gobble(file);
1576                                 //text = m->getline(file);
1577                                 file >> text;
1578                                 break;
1579                         }
1580                 }
1581                 m->gobble(file);
1582
1583                 return text;
1584         }
1585         catch(exception& e) {
1586                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToString");
1587                 exit(1);
1588         }
1589 }
1590
1591 //**********************************************************************************************************************
1592
1593 vector<unsigned short> SffInfoCommand::parseHeaderLineToFloatVector(ifstream& file, int length){
1594         try {
1595                 vector<unsigned short> floatVector(length);
1596
1597                 while (!file.eof())     {
1598                         char c = file.get();
1599                         if (c == ':'){
1600                                 float temp;
1601                                 for(int i=0;i<length;i++){
1602                                         file >> temp;
1603                                         floatVector[i] = temp * 100;
1604                                 }
1605                                 break;
1606                         }
1607                 }
1608                 m->gobble(file);
1609                 return floatVector;
1610         }
1611         catch(exception& e) {
1612                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToFloatVector");
1613                 exit(1);
1614         }
1615 }
1616
1617 //**********************************************************************************************************************
1618
1619 vector<unsigned int> SffInfoCommand::parseHeaderLineToIntVector(ifstream& file, int length){
1620         try {
1621                 vector<unsigned int> intVector(length);
1622
1623                 while (!file.eof())     {
1624                         char c = file.get();
1625                         if (c == ':'){
1626                                 for(int i=0;i<length;i++){
1627                                         file >> intVector[i];
1628                                 }
1629                                 break;
1630                         }
1631                 }
1632                 m->gobble(file);
1633                 return intVector;
1634         }
1635         catch(exception& e) {
1636                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToIntVector");
1637                 exit(1);
1638         }
1639 }
1640 //***************************************************************************************************************
1641
1642 bool SffInfoCommand::readOligos(string oligoFile){
1643         try {
1644         filehandles.clear();
1645         numSplitReads.clear();
1646         filehandlesHeaders.clear();
1647
1648                 ifstream inOligos;
1649                 m->openInputFile(oligoFile, inOligos);
1650
1651                 string type, oligo, group;
1652
1653                 int indexPrimer = 0;
1654                 int indexBarcode = 0;
1655
1656                 while(!inOligos.eof()){
1657
1658                         inOligos >> type;
1659
1660                         if(type[0] == '#'){
1661                                 while (!inOligos.eof()) {       char c = inOligos.get();  if (c == 10 || c == 13){      break;  }       } // get rest of line if there's any crap there
1662                                 m->gobble(inOligos);
1663                         }
1664                         else{
1665                                 m->gobble(inOligos);
1666                                 //make type case insensitive
1667                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }
1668
1669                                 inOligos >> oligo;
1670
1671                                 for(int i=0;i<oligo.length();i++){
1672                                         oligo[i] = toupper(oligo[i]);
1673                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
1674                                 }
1675
1676                                 if(type == "FORWARD"){
1677                                         group = "";
1678
1679                                         // get rest of line in case there is a primer name
1680                                         while (!inOligos.eof()) {
1681                                                 char c = inOligos.get();
1682                                                 if (c == 10 || c == 13 || c == -1){     break;  }
1683                                                 else if (c == 32 || c == 9){;} //space or tab
1684                                                 else {  group += c;  }
1685                                         }
1686
1687                                         //check for repeat barcodes
1688                                         map<string, int>::iterator itPrime = primers.find(oligo);
1689                                         if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
1690
1691                                         primers[oligo]=indexPrimer; indexPrimer++;
1692                                         primerNameVector.push_back(group);
1693                                 }else if(type == "REVERSE"){
1694                                         //Sequence oligoRC("reverse", oligo);
1695                                         //oligoRC.reverseComplement();
1696                     string oligoRC = reverseOligo(oligo);
1697                                         revPrimer.push_back(oligoRC);
1698                                 }
1699                                 else if(type == "BARCODE"){
1700                                         inOligos >> group;
1701
1702                                         //check for repeat barcodes
1703                                         map<string, int>::iterator itBar = barcodes.find(oligo);
1704                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
1705
1706                                         barcodes[oligo]=indexBarcode; indexBarcode++;
1707                                         barcodeNameVector.push_back(group);
1708                                 }else if(type == "LINKER"){
1709                                         linker.push_back(oligo);
1710                                 }else if(type == "SPACER"){
1711                                         spacer.push_back(oligo);
1712                                 }
1713                                 else{   m->mothurOut("[WARNING]: " + type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine(); }
1714                         }
1715                         m->gobble(inOligos);
1716                 }
1717                 inOligos.close();
1718
1719                 if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ split = 1;      }
1720
1721                 //add in potential combos
1722                 if(barcodeNameVector.size() == 0){
1723                         barcodes[""] = 0;
1724                         barcodeNameVector.push_back("");
1725                 }
1726
1727                 if(primerNameVector.size() == 0){
1728                         primers[""] = 0;
1729                         primerNameVector.push_back("");
1730                 }
1731
1732                 filehandles.resize(barcodeNameVector.size());
1733                 for(int i=0;i<filehandles.size();i++){
1734                         filehandles[i].assign(primerNameVector.size(), "");
1735                 }
1736
1737                 if(split > 1){
1738                         set<string> uniqueNames; //used to cleanup outputFileNames
1739                         for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
1740                                 for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
1741
1742                                         string primerName = primerNameVector[itPrimer->second];
1743                                         string barcodeName = barcodeNameVector[itBar->second];
1744
1745                                         string comboGroupName = "";
1746                                         string fastaFileName = "";
1747                                         string qualFileName = "";
1748                                         string nameFileName = "";
1749
1750                                         if(primerName == ""){
1751                                                 comboGroupName = barcodeNameVector[itBar->second];
1752                                         }
1753                                         else{
1754                                                 if(barcodeName == ""){
1755                                                         comboGroupName = primerNameVector[itPrimer->second];
1756                                                 }
1757                                                 else{
1758                                                         comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
1759                                                 }
1760                                         }
1761
1762                                         ofstream temp;
1763                     map<string, string> variables;
1764                     variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(currentFileName));
1765                     variables["[group]"] = comboGroupName;
1766                                         string thisFilename = getOutputFileName("sff",variables);
1767                                         if (uniqueNames.count(thisFilename) == 0) {
1768                                                 outputNames.push_back(thisFilename);
1769                                                 outputTypes["sff"].push_back(thisFilename);
1770                                                 uniqueNames.insert(thisFilename);
1771                                         }
1772
1773                                         filehandles[itBar->second][itPrimer->second] = thisFilename;
1774                                         m->openOutputFile(thisFilename, temp);          temp.close();
1775                                 }
1776                         }
1777                 }
1778                 numFPrimers = primers.size();
1779         numLinkers = linker.size();
1780         numSpacers = spacer.size();
1781         map<string, string> variables;
1782         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(currentFileName));
1783         variables["[group]"] = "scrap";
1784                 noMatchFile = getOutputFileName("sff",variables);
1785         m->mothurRemove(noMatchFile);
1786
1787                 bool allBlank = true;
1788                 for (int i = 0; i < barcodeNameVector.size(); i++) {
1789                         if (barcodeNameVector[i] != "") {
1790                                 allBlank = false;
1791                                 break;
1792                         }
1793                 }
1794                 for (int i = 0; i < primerNameVector.size(); i++) {
1795                         if (primerNameVector[i] != "") {
1796                                 allBlank = false;
1797                                 break;
1798                         }
1799                 }
1800
1801         filehandlesHeaders.resize(filehandles.size());
1802         numSplitReads.resize(filehandles.size());
1803         for (int i = 0; i < filehandles.size(); i++) {
1804             numSplitReads[i].resize(filehandles[i].size(), 0);
1805             for (int j = 0; j < filehandles[i].size(); j++) {
1806                 filehandlesHeaders[i].push_back(filehandles[i][j]+"headers");
1807             }
1808         }
1809
1810                 if (allBlank) {
1811                         m->mothurOut("[WARNING]: your oligos file does not contain any group names.  mothur will not create a split the sff file."); m->mothurOutEndLine();
1812                         split = 1;
1813                         return false;
1814                 }
1815
1816                 return true;
1817
1818         }
1819         catch(exception& e) {
1820                 m->errorOut(e, "SffInfoCommand", "readOligos");
1821                 exit(1);
1822         }
1823 }
1824 //********************************************************************/
1825 string SffInfoCommand::reverseOligo(string oligo){
1826         try {
1827         string reverse = "";
1828
1829         for(int i=oligo.length()-1;i>=0;i--){
1830
1831             if(oligo[i] == 'A')         {       reverse += 'T'; }
1832             else if(oligo[i] == 'T'){   reverse += 'A'; }
1833             else if(oligo[i] == 'U'){   reverse += 'A'; }
1834
1835             else if(oligo[i] == 'G'){   reverse += 'C'; }
1836             else if(oligo[i] == 'C'){   reverse += 'G'; }
1837
1838             else if(oligo[i] == 'R'){   reverse += 'Y'; }
1839             else if(oligo[i] == 'Y'){   reverse += 'R'; }
1840
1841             else if(oligo[i] == 'M'){   reverse += 'K'; }
1842             else if(oligo[i] == 'K'){   reverse += 'M'; }
1843
1844             else if(oligo[i] == 'W'){   reverse += 'W'; }
1845             else if(oligo[i] == 'S'){   reverse += 'S'; }
1846
1847             else if(oligo[i] == 'B'){   reverse += 'V'; }
1848             else if(oligo[i] == 'V'){   reverse += 'B'; }
1849
1850             else if(oligo[i] == 'D'){   reverse += 'H'; }
1851             else if(oligo[i] == 'H'){   reverse += 'D'; }
1852
1853             else                                                {       reverse += 'N'; }
1854         }
1855
1856
1857         return reverse;
1858     }
1859         catch(exception& e) {
1860                 m->errorOut(e, "SffInfoCommand", "reverseOligo");
1861                 exit(1);
1862         }
1863 }
1864
1865 //**********************************************************************************************************************
1866
1867
1868
1869