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
  13 //**********************************************************************************************************************
  14 vector<string> SffInfoCommand::setParameters(){
  15         try {
  16                 CommandParameter psff("sff", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(psff);
  17                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
  18                 CommandParameter psfftxt("sfftxt", "String", "", "", "", "", "",false,false); parameters.push_back(psfftxt);
  19                 CommandParameter pflow("flow", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pflow);
  20                 CommandParameter ptrim("trim", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(ptrim);
  21                 CommandParameter pfasta("fasta", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pfasta);
  22                 CommandParameter pqfile("name", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pqfile);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "SffInfoCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string SffInfoCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The sffinfo command reads a sff file and extracts the sequence data, or you can use it to parse a sfftxt file.\n";
  40                 helpString += "The sffinfo command parameters are sff, fasta, qfile, accnos, flow, sfftxt, and trim. sff is required. \n";
  41                 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";
  42                 helpString += "The fasta parameter allows you to indicate if you would like a fasta formatted file generated.  Default=True. \n";
  43                 helpString += "The qfile parameter allows you to indicate if you would like a quality file generated.  Default=True. \n";
  44                 helpString += "The flow parameter allows you to indicate if you would like a flowgram file generated.  Default=True. \n";
  45                 helpString += "The sfftxt parameter allows you to indicate if you would like a sff.txt file generated.  Default=False. \n";
  46                 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";
  47                 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";
  48                 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";
  49                 helpString += "Example sffinfo(sff=mySffFile.sff, trim=F).\n";
  50                 helpString += "Note: No spaces between parameter labels (i.e. sff), '=' and parameters (i.e.yourSffFileName).\n";
  51                 return helpString;
  52         }
  53         catch(exception& e) {
  54                 m->errorOut(e, "SffInfoCommand", "getHelpString");
  55                 exit(1);
  56         }
  57 }
  58 //**********************************************************************************************************************
  59 string SffInfoCommand::getOutputFileNameTag(string type, string inputName=""){
  60         try {
  61         string outputFileName = "";
  62                 map<string, vector<string> >::iterator it;
  63
  64         //is this a type this command creates
  65         it = outputTypes.find(type);
  66         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  67         else {
  68             if (type == "fasta")            {   outputFileName =  "fasta";   }
  69             else if (type == "flow")    {   outputFileName =  "flow";   }
  70             else if (type == "sfftxt")        {   outputFileName =  "sff.txt";   }
  71             else if (type == "qfile")       {   outputFileName =  "qual";   }
  72              else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  73         }
  74         return outputFileName;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "SffInfoCommand", "getOutputFileNameTag");
  78                 exit(1);
  79         }
  80 }
  81
  82
  83 //**********************************************************************************************************************
  84 SffInfoCommand::SffInfoCommand(){
  85         try {
  86                 abort = true; calledHelp = true;
  87                 setParameters();
  88                 vector<string> tempOutNames;
  89                 outputTypes["fasta"] = tempOutNames;
  90                 outputTypes["flow"] = tempOutNames;
  91                 outputTypes["sfftxt"] = tempOutNames;
  92                 outputTypes["qfile"] = tempOutNames;
  93         }
  94         catch(exception& e) {
  95                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
  96                 exit(1);
  97         }
  98 }
  99 //**********************************************************************************************************************
 100
 101 SffInfoCommand::SffInfoCommand(string option)  {
 102         try {
 103                 abort = false; calledHelp = false;
 104                 hasAccnos = false;
 105
 106                 //allow user to run help
 107                 if(option == "help") { help(); abort = true; calledHelp = true; }
 108                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 109
 110                 else {
 111                         //valid paramters for this command
 112                         vector<string> myArray = setParameters();
 113
 114                         OptionParser parser(option);
 115                         map<string, string> parameters = parser.getParameters();
 116
 117                         ValidParameters validParameter;
 118                         //check to make sure all parameters are valid for command
 119                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 120                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 121                         }
 122
 123                         //initialize outputTypes
 124                         vector<string> tempOutNames;
 125                         outputTypes["fasta"] = tempOutNames;
 126                         outputTypes["flow"] = tempOutNames;
 127                         outputTypes["sfftxt"] = tempOutNames;
 128                         outputTypes["qfile"] = tempOutNames;
 129
 130                         //if the user changes the output directory command factory will send this info to us in the output parameter
 131                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 132
 133                         //if the user changes the input directory command factory will send this info to us in the output parameter
 134                         string inputDir = validParameter.validFile(parameters, "inputdir", false);        if (inputDir == "not found"){ inputDir = "";          }
 135
 136                         sffFilename = validParameter.validFile(parameters, "sff", false);
 137                         if (sffFilename == "not found") { sffFilename = "";  }
 138                         else {
 139                                 m->splitAtDash(sffFilename, filenames);
 140
 141                                 //go through files and make sure they are good, if not, then disregard them
 142                                 for (int i = 0; i < filenames.size(); i++) {
 143                                         bool ignore = false;
 144                                         if (filenames[i] == "current") {
 145                                                 filenames[i] = m->getSFFFile();
 146                                                 if (filenames[i] != "") {  m->mothurOut("Using " + filenames[i] + " as input file for the sff parameter where you had given current."); m->mothurOutEndLine(); }
 147                                                 else {
 148                                                         m->mothurOut("You have no current sfffile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 149                                                         //erase from file list
 150                                                         filenames.erase(filenames.begin()+i);
 151                                                         i--;
 152                                                 }
 153                                         }
 154
 155                                         if (!ignore) {
 156                                                 if (inputDir != "") {
 157                                                         string path = m->hasPath(filenames[i]);
 158                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 159                                                         if (path == "") {       filenames[i] = inputDir + filenames[i];         }
 160                                                 }
 161
 162                                                 ifstream in;
 163                                                 int ableToOpen = m->openInputFile(filenames[i], in, "noerror");
 164
 165                                                 //if you can't open it, try default location
 166                                                 if (ableToOpen == 1) {
 167                                                         if (m->getDefaultPath() != "") { //default path is set
 168                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(filenames[i]);
 169                                                                 m->mothurOut("Unable to open " + filenames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 170                                                                 ifstream in2;
 171                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 172                                                                 in2.close();
 173                                                                 filenames[i] = tryPath;
 174                                                         }
 175                                                 }
 176
 177                                                 //if you can't open it, try default location
 178                                                 if (ableToOpen == 1) {
 179                                                         if (m->getOutputDir() != "") { //default path is set
 180                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(filenames[i]);
 181                                                                 m->mothurOut("Unable to open " + filenames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 182                                                                 ifstream in2;
 183                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 184                                                                 in2.close();
 185                                                                 filenames[i] = tryPath;
 186                                                         }
 187                                                 }
 188
 189                                                 in.close();
 190
 191                                                 if (ableToOpen == 1) {
 192                                                         m->mothurOut("Unable to open " + filenames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 193                                                         //erase from file list
 194                                                         filenames.erase(filenames.begin()+i);
 195                                                         i--;
 196                                                 }else { m->setSFFFile(filenames[i]); }
 197                                         }
 198                                 }
 199
 200                                 //make sure there is at least one valid file left
 201                                 if (filenames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 202                         }
 203
 204                         accnosName = validParameter.validFile(parameters, "accnos", false);
 205                         if (accnosName == "not found") { accnosName = "";  }
 206                         else {
 207                                 hasAccnos = true;
 208                                 m->splitAtDash(accnosName, accnosFileNames);
 209
 210                                 //go through files and make sure they are good, if not, then disregard them
 211                                 for (int i = 0; i < accnosFileNames.size(); i++) {
 212                                         bool ignore = false;
 213                                         if (accnosFileNames[i] == "current") {
 214                                                 accnosFileNames[i] = m->getAccnosFile();
 215                                                 if (accnosFileNames[i] != "") {  m->mothurOut("Using " + accnosFileNames[i] + " as input file for the accnos parameter where you had given current."); m->mothurOutEndLine(); }
 216                                                 else {
 217                                                         m->mothurOut("You have no current accnosfile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 218                                                         //erase from file list
 219                                                         accnosFileNames.erase(accnosFileNames.begin()+i);
 220                                                         i--;
 221                                                 }
 222                                         }
 223
 224                                         if (!ignore) {
 225
 226                                                 if (inputDir != "") {
 227                                                         string path = m->hasPath(accnosFileNames[i]);
 228                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 229                                                         if (path == "") {       accnosFileNames[i] = inputDir + accnosFileNames[i];             }
 230                                                 }
 231
 232                                                 ifstream in;
 233                                                 int ableToOpen = m->openInputFile(accnosFileNames[i], in, "noerror");
 234
 235                                                 //if you can't open it, try default location
 236                                                 if (ableToOpen == 1) {
 237                                                         if (m->getDefaultPath() != "") { //default path is set
 238                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(accnosFileNames[i]);
 239                                                                 m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 240                                                                 ifstream in2;
 241                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 242                                                                 in2.close();
 243                                                                 accnosFileNames[i] = tryPath;
 244                                                         }
 245                                                 }
 246                                                 //if you can't open it, try default location
 247                                                 if (ableToOpen == 1) {
 248                                                         if (m->getOutputDir() != "") { //default path is set
 249                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(accnosFileNames[i]);
 250                                                                 m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 251                                                                 ifstream in2;
 252                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 253                                                                 in2.close();
 254                                                                 accnosFileNames[i] = tryPath;
 255                                                         }
 256                                                 }
 257                                                 in.close();
 258
 259                                                 if (ableToOpen == 1) {
 260                                                         m->mothurOut("Unable to open " + accnosFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 261                                                         //erase from file list
 262                                                         accnosFileNames.erase(accnosFileNames.begin()+i);
 263                                                         i--;
 264                                                 }
 265                                         }
 266                                 }
 267
 268                                 //make sure there is at least one valid file left
 269                                 if (accnosFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 270                         }
 271
 272                         if (hasAccnos) {
 273                                 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(); }
 274                         }
 275
 276                         string temp = validParameter.validFile(parameters, "qfile", false);                     if (temp == "not found"){       temp = "T";                             }
 277                         qual = m->isTrue(temp);
 278
 279                         temp = validParameter.validFile(parameters, "fasta", false);                            if (temp == "not found"){       temp = "T";                             }
 280                         fasta = m->isTrue(temp);
 281
 282                         temp = validParameter.validFile(parameters, "flow", false);                                     if (temp == "not found"){       temp = "T";                             }
 283                         flow = m->isTrue(temp);
 284
 285                         temp = validParameter.validFile(parameters, "trim", false);                                     if (temp == "not found"){       temp = "T";                             }
 286                         trim = m->isTrue(temp);
 287
 288                         temp = validParameter.validFile(parameters, "sfftxt", false);
 289                         if (temp == "not found")        {       temp = "F";      sfftxt = false; sfftxtFilename = "";           }
 290                         else if (m->isTrue(temp))       {       sfftxt = true;          sfftxtFilename = "";                            }
 291                         else {
 292                                 //you are a filename
 293                                 if (inputDir != "") {
 294                                         map<string,string>::iterator it = parameters.find("sfftxt");
 295                                         //user has given a template file
 296                                         if(it != parameters.end()){
 297                                                 string path = m->hasPath(it->second);
 298                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 299                                                 if (path == "") {       parameters["sfftxt"] = inputDir + it->second;           }
 300                                         }
 301                                 }
 302
 303                                 sfftxtFilename = validParameter.validFile(parameters, "sfftxt", true);
 304                                 if (sfftxtFilename == "not found") { sfftxtFilename = "";  }
 305                                 else if (sfftxtFilename == "not open") { sfftxtFilename = "";  }
 306                         }
 307
 308                         if ((sfftxtFilename == "") && (filenames.size() == 0)) {
 309                                 //if there is a current sff file, use it
 310                                 string filename = m->getSFFFile();
 311                                 if (filename != "") { filenames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the sff parameter."); m->mothurOutEndLine(); }
 312                                 else {  m->mothurOut("[ERROR]: you must provide a valid sff or sfftxt file."); m->mothurOutEndLine(); abort=true;  }
 313                         }
 314                 }
 315         }
 316         catch(exception& e) {
 317                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
 318                 exit(1);
 319         }
 320 }
 321 //**********************************************************************************************************************
 322 int SffInfoCommand::execute(){
 323         try {
 324                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 325
 326                 for (int s = 0; s < filenames.size(); s++) {
 327
 328                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } return 0; }
 329
 330                         int start = time(NULL);
 331
 332             filenames[s] = m->getFullPathName(filenames[s]);
 333                         m->mothurOut("Extracting info from " + filenames[s] + " ..." ); m->mothurOutEndLine();
 334
 335                         string accnos = "";
 336                         if (hasAccnos) { accnos = accnosFileNames[s]; }
 337
 338                         int numReads = extractSffInfo(filenames[s], accnos);
 339
 340                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to extract " + toString(numReads) + ".");
 341                 }
 342
 343                 if (sfftxtFilename != "") {  parseSffTxt(); }
 344
 345                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } return 0; }
 346
 347                 //set fasta file as new current fastafile
 348                 string current = "";
 349                 itTypes = outputTypes.find("fasta");
 350                 if (itTypes != outputTypes.end()) {
 351                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 352                 }
 353
 354                 itTypes = outputTypes.find("qfile");
 355                 if (itTypes != outputTypes.end()) {
 356                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 357                 }
 358
 359                 itTypes = outputTypes.find("flow");
 360                 if (itTypes != outputTypes.end()) {
 361                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFlowFile(current); }
 362                 }
 363
 364                 //report output filenames
 365                 m->mothurOutEndLine();
 366                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 367                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 368                 m->mothurOutEndLine();
 369
 370                 return 0;
 371         }
 372         catch(exception& e) {
 373                 m->errorOut(e, "SffInfoCommand", "execute");
 374                 exit(1);
 375         }
 376 }
 377 //**********************************************************************************************************************
 378 int SffInfoCommand::extractSffInfo(string input, string accnos){
 379         try {
 380
 381                 if (outputDir == "") {  outputDir += m->hasPath(input); }
 382
 383                 if (accnos != "")       {  readAccnosFile(accnos);  }
 384                 else                            {       seqNames.clear();               }
 385
 386                 ofstream outSfftxt, outFasta, outQual, outFlow;
 387                 string outFastaFileName, outQualFileName;
 388         string rootName = outputDir + m->getRootName(m->getSimpleName(input));
 389         if(rootName.find_last_of(".") == rootName.npos){ rootName += "."; }
 390
 391                 string sfftxtFileName = outputDir + m->getRootName(m->getSimpleName(input)) + getOutputFileNameTag("sfftxt");
 392                 string outFlowFileName = outputDir + m->getRootName(m->getSimpleName(input)) + getOutputFileNameTag("flow");
 393                 if (trim) {
 394                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + getOutputFileNameTag("fasta");
 395                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + getOutputFileNameTag("qfile");
 396                 }else{
 397                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw." + getOutputFileNameTag("fasta");
 398                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw." + getOutputFileNameTag("qfile");
 399                 }
 400
 401                 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); }
 402                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
 403                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qfile"].push_back(outQualFileName);  }
 404                 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);  }
 405
 406                 ifstream in;
 407                 in.open(input.c_str(), ios::binary);
 408
 409                 CommonHeader header;
 410                 readCommonHeader(in, header);
 411
 412                 int count = 0;
 413                 mycount = 0;
 414
 415                 //check magic number and version
 416                 if (header.magicNumber != 779314790) { m->mothurOut("Magic Number is not correct, not a valid .sff file"); m->mothurOutEndLine(); return count; }
 417                 if (header.version != "0001") { m->mothurOut("Version is not supported, only support version 0001."); m->mothurOutEndLine(); return count; }
 418
 419                 //print common header
 420                 if (sfftxt) {   printCommonHeader(outSfftxt, header);           }
 421                 if (flow)       {       outFlow << header.numFlowsPerRead << endl;      }
 422
 423                 //read through the sff file
 424                 while (!in.eof()) {
 425
 426                         bool print = true;
 427
 428                         //read header
 429                         Header readheader;
 430                         readHeader(in, readheader);
 431
 432                         //read data
 433                         seqRead read;
 434                         readSeqData(in, read, header.numFlowsPerRead, readheader.numBases);
 435             bool okay = sanityCheck(readheader, read);
 436             if (!okay) { break; }
 437
 438                         //if you have provided an accosfile and this seq is not in it, then dont print
 439                         if (seqNames.size() != 0) {   if (seqNames.count(readheader.name) == 0) { print = false; }  }
 440
 441                         //print
 442                         if (print) {
 443                                 if (sfftxt) { printHeader(outSfftxt, readheader); printSffTxtSeqData(outSfftxt, read, readheader); }
 444                                 if (fasta)      {       printFastaSeqData(outFasta, read, readheader);  }
 445                                 if (qual)       {       printQualSeqData(outQual, read, readheader);    }
 446                                 if (flow)       {       printFlowSeqData(outFlow, read, readheader);    }
 447                         }
 448
 449                         count++;
 450                         mycount++;
 451
 452                         //report progress
 453                         if((count+1) % 10000 == 0){     m->mothurOut(toString(count+1)); m->mothurOutEndLine();         }
 454
 455                         if (m->control_pressed) { count = 0; break;   }
 456
 457                         if (count >= header.numReads) { break; }
 458                 }
 459
 460                 //report progress
 461                 if (!m->control_pressed) {   if((count) % 10000 != 0){  m->mothurOut(toString(count)); m->mothurOutEndLine();           }  }
 462
 463                 in.close();
 464
 465                 if (sfftxt) {  outSfftxt.close();       }
 466                 if (fasta)      {  outFasta.close();    }
 467                 if (qual)       {  outQual.close();             }
 468                 if (flow)       {  outFlow.close();             }
 469
 470                 return count;
 471         }
 472         catch(exception& e) {
 473                 m->errorOut(e, "SffInfoCommand", "extractSffInfo");
 474                 exit(1);
 475         }
 476 }
 477 //**********************************************************************************************************************
 478 int SffInfoCommand::readCommonHeader(ifstream& in, CommonHeader& header){
 479         try {
 480
 481                 if (!in.eof()) {
 482
 483                         //read magic number
 484                         char buffer[4];
 485                         in.read(buffer, 4);
 486                         header.magicNumber = be_int4(*(unsigned int *)(&buffer));
 487
 488                         //read version
 489                         char buffer9[4];
 490                         in.read(buffer9, 4);
 491                         header.version = "";
 492                         for (int i = 0; i < 4; i++) {  header.version += toString((int)(buffer9[i])); }
 493
 494                         //read offset
 495                         char buffer2 [8];
 496                         in.read(buffer2, 8);
 497                         header.indexOffset =  be_int8(*(unsigned long long *)(&buffer2));
 498
 499                         //read index length
 500                         char buffer3 [4];
 501                         in.read(buffer3, 4);
 502                         header.indexLength =  be_int4(*(unsigned int *)(&buffer3));
 503
 504                         //read num reads
 505                         char buffer4 [4];
 506                         in.read(buffer4, 4);
 507                         header.numReads =  be_int4(*(unsigned int *)(&buffer4));
 508
 509                         //read header length
 510                         char buffer5 [2];
 511                         in.read(buffer5, 2);
 512                         header.headerLength =  be_int2(*(unsigned short *)(&buffer5));
 513
 514                         //read key length
 515                         char buffer6 [2];
 516                         in.read(buffer6, 2);
 517                         header.keyLength = be_int2(*(unsigned short *)(&buffer6));
 518
 519                         //read number of flow reads
 520                         char buffer7 [2];
 521                         in.read(buffer7, 2);
 522                         header.numFlowsPerRead =  be_int2(*(unsigned short *)(&buffer7));
 523
 524                         //read format code
 525                         char buffer8 [1];
 526                         in.read(buffer8, 1);
 527                         header.flogramFormatCode = (int)(buffer8[0]);
 528
 529                         //read flow chars
 530                         char* tempBuffer = new char[header.numFlowsPerRead];
 531                         in.read(&(*tempBuffer), header.numFlowsPerRead);
 532                         header.flowChars = tempBuffer;
 533                         if (header.flowChars.length() > header.numFlowsPerRead) { header.flowChars = header.flowChars.substr(0, header.numFlowsPerRead);  }
 534                         delete[] tempBuffer;
 535
 536                         //read key
 537                         char* tempBuffer2 = new char[header.keyLength];
 538                         in.read(&(*tempBuffer2), header.keyLength);
 539                         header.keySequence = tempBuffer2;
 540                         if (header.keySequence.length() > header.keyLength) { header.keySequence = header.keySequence.substr(0, header.keyLength);  }
 541                         delete[] tempBuffer2;
 542
 543                         /* Pad to 8 chars */
 544                         unsigned long long spotInFile = in.tellg();
 545                         unsigned long long spot = (spotInFile + 7)& ~7;  // ~ inverts
 546                         in.seekg(spot);
 547
 548                 }else{
 549                         m->mothurOut("Error reading sff common header."); m->mothurOutEndLine();
 550                 }
 551
 552                 return 0;
 553         }
 554         catch(exception& e) {
 555                 m->errorOut(e, "SffInfoCommand", "readCommonHeader");
 556                 exit(1);
 557         }
 558 }
 559 //**********************************************************************************************************************
 560 int SffInfoCommand::readHeader(ifstream& in, Header& header){
 561         try {
 562
 563                 if (!in.eof()) {
 564
 565                         //read header length
 566                         char buffer [2];
 567                         in.read(buffer, 2);
 568                         header.headerLength = be_int2(*(unsigned short *)(&buffer));
 569
 570                         //read name length
 571                         char buffer2 [2];
 572                         in.read(buffer2, 2);
 573                         header.nameLength = be_int2(*(unsigned short *)(&buffer2));
 574
 575                         //read num bases
 576                         char buffer3 [4];
 577                         in.read(buffer3, 4);
 578                         header.numBases =  be_int4(*(unsigned int *)(&buffer3));
 579
 580                         //read clip qual left
 581                         char buffer4 [2];
 582                         in.read(buffer4, 2);
 583                         header.clipQualLeft =  be_int2(*(unsigned short *)(&buffer4));
 584                         header.clipQualLeft = 5;
 585
 586                         //read clip qual right
 587                         char buffer5 [2];
 588                         in.read(buffer5, 2);
 589                         header.clipQualRight =  be_int2(*(unsigned short *)(&buffer5));
 590
 591                         //read clipAdapterLeft
 592                         char buffer6 [2];
 593                         in.read(buffer6, 2);
 594                         header.clipAdapterLeft = be_int2(*(unsigned short *)(&buffer6));
 595
 596                         //read clipAdapterRight
 597                         char buffer7 [2];
 598                         in.read(buffer7, 2);
 599                         header.clipAdapterRight = be_int2(*(unsigned short *)(&buffer7));
 600
 601                         //read name
 602                         char* tempBuffer = new char[header.nameLength];
 603                         in.read(&(*tempBuffer), header.nameLength);
 604                         header.name = tempBuffer;
 605                         if (header.name.length() > header.nameLength) { header.name = header.name.substr(0, header.nameLength);  }
 606                         delete[] tempBuffer;
 607
 608                         //extract info from name
 609                         decodeName(header.timestamp, header.region, header.xy, header.name);
 610
 611                         /* Pad to 8 chars */
 612                         unsigned long long spotInFile = in.tellg();
 613                         unsigned long long spot = (spotInFile + 7)& ~7;
 614                         in.seekg(spot);
 615
 616                 }else{
 617                         m->mothurOut("Error reading sff header info."); m->mothurOutEndLine();
 618                 }
 619
 620                 return 0;
 621         }
 622         catch(exception& e) {
 623                 m->errorOut(e, "SffInfoCommand", "readHeader");
 624                 exit(1);
 625         }
 626 }
 627 //**********************************************************************************************************************
 628 int SffInfoCommand::readSeqData(ifstream& in, seqRead& read, int numFlowReads, int numBases){
 629         try {
 630
 631                 if (!in.eof()) {
 632
 633                         //read flowgram
 634                         read.flowgram.resize(numFlowReads);
 635                         for (int i = 0; i < numFlowReads; i++) {
 636                                 char buffer [2];
 637                                 in.read(buffer, 2);
 638                                 read.flowgram[i] = be_int2(*(unsigned short *)(&buffer));
 639                         }
 640
 641                         //read flowIndex
 642                         read.flowIndex.resize(numBases);
 643                         for (int i = 0; i < numBases; i++) {
 644                                 char temp[1];
 645                                 in.read(temp, 1);
 646                                 read.flowIndex[i] = be_int1(*(unsigned char *)(&temp));
 647                         }
 648
 649                         //read bases
 650                         char* tempBuffer = new char[numBases];
 651                         in.read(&(*tempBuffer), numBases);
 652                         read.bases = tempBuffer;
 653                         if (read.bases.length() > numBases) { read.bases = read.bases.substr(0, numBases);  }
 654                         delete[] tempBuffer;
 655
 656                         //read qual scores
 657                         read.qualScores.resize(numBases);
 658                         for (int i = 0; i < numBases; i++) {
 659                                 char temp[1];
 660                                 in.read(temp, 1);
 661                                 read.qualScores[i] = be_int1(*(unsigned char *)(&temp));
 662                         }
 663
 664                         /* Pad to 8 chars */
 665                         unsigned long long spotInFile = in.tellg();
 666                         unsigned long long spot = (spotInFile + 7)& ~7;
 667                         in.seekg(spot);
 668
 669                 }else{
 670                         m->mothurOut("Error reading."); m->mothurOutEndLine();
 671                 }
 672
 673                 return 0;
 674         }
 675         catch(exception& e) {
 676                 m->errorOut(e, "SffInfoCommand", "readSeqData");
 677                 exit(1);
 678         }
 679 }
 680 //**********************************************************************************************************************
 681 int SffInfoCommand::decodeName(string& timestamp, string& region, string& xy, string name) {
 682         try {
 683
 684                 if (name.length() >= 6) {
 685                         string time = name.substr(0, 6);
 686                         unsigned int timeNum = m->fromBase36(time);
 687
 688                         int q1 = timeNum / 60;
 689                         int sec = timeNum - 60 * q1;
 690                         int q2 = q1 / 60;
 691                         int minute = q1 - 60 * q2;
 692                         int q3 = q2 / 24;
 693                         int hr = q2 - 24 * q3;
 694                         int q4 = q3 / 32;
 695                         int day = q3 - 32 * q4;
 696                         int q5 = q4 / 13;
 697                         int mon = q4 - 13 * q5;
 698                         int year = 2000 + q5;
 699
 700                         timestamp = toString(year) + "_" + toString(mon) + "_" + toString(day) + "_" + toString(hr) + "_" + toString(minute) + "_" + toString(sec);
 701                 }
 702
 703                 if (name.length() >= 9) {
 704                         region = name.substr(7, 2);
 705
 706                         string xyNum = name.substr(9);
 707                         unsigned int myXy = m->fromBase36(xyNum);
 708                         int x = myXy >> 12;
 709                         int y = myXy & 4095;
 710
 711                         xy = toString(x) + "_" + toString(y);
 712                 }
 713
 714                 return 0;
 715         }
 716         catch(exception& e) {
 717                 m->errorOut(e, "SffInfoCommand", "decodeName");
 718                 exit(1);
 719         }
 720 }
 721 //**********************************************************************************************************************
 722 int SffInfoCommand::printCommonHeader(ofstream& out, CommonHeader& header) {
 723         try {
 724
 725                 out << "Common Header:\nMagic Number: " << header.magicNumber << endl;
 726                 out << "Version: " << header.version << endl;
 727                 out << "Index Offset: " << header.indexOffset << endl;
 728                 out << "Index Length: " << header.indexLength << endl;
 729                 out << "Number of Reads: " << header.numReads << endl;
 730                 out << "Header Length: " << header.headerLength << endl;
 731                 out << "Key Length: " << header.keyLength << endl;
 732                 out << "Number of Flows: " << header.numFlowsPerRead << endl;
 733                 out << "Format Code: " << header.flogramFormatCode << endl;
 734                 out << "Flow Chars: " << header.flowChars << endl;
 735                 out << "Key Sequence: " << header.keySequence << endl << endl;
 736
 737                 return 0;
 738         }
 739         catch(exception& e) {
 740                 m->errorOut(e, "SffInfoCommand", "printCommonHeader");
 741                 exit(1);
 742         }
 743 }
 744 //**********************************************************************************************************************
 745 int SffInfoCommand::printHeader(ofstream& out, Header& header) {
 746         try {
 747
 748                 out << ">" << header.name << endl;
 749                 out << "Run Prefix: " << header.timestamp << endl;
 750                 out << "Region #:  " << header.region << endl;
 751                 out << "XY Location: " << header.xy << endl << endl;
 752
 753                 out << "Run Name:  " << endl;
 754                 out << "Analysis Name:  " << endl;
 755                 out << "Full Path: " << endl << endl;
 756
 757                 out << "Read Header Len: " << header.headerLength << endl;
 758                 out << "Name Length: " << header.nameLength << endl;
 759                 out << "# of Bases: " << header.numBases << endl;
 760                 out << "Clip Qual Left: " << header.clipQualLeft << endl;
 761                 out << "Clip Qual Right: " << header.clipQualRight << endl;
 762                 out << "Clip Adap Left: " << header.clipAdapterLeft << endl;
 763                 out << "Clip Adap Right: " << header.clipAdapterRight << endl << endl;
 764
 765                 return 0;
 766         }
 767         catch(exception& e) {
 768                 m->errorOut(e, "SffInfoCommand", "printHeader");
 769                 exit(1);
 770         }
 771 }
 772 //**********************************************************************************************************************
 773 bool SffInfoCommand::sanityCheck(Header& header, seqRead& read) {
 774         try {
 775         bool okay = true;
 776         string message = "[WARNING]: Your sff file may be corrupted! Sequence: " + header.name + "\n";
 777
 778         if (header.clipQualLeft > read.bases.length()) {
 779             okay = false; message += "Clip Qual Left = " + toString(header.clipQualLeft) + ", but we only read " + toString(read.bases.length()) + " bases.\n";
 780         }
 781         if (header.clipQualRight > read.bases.length()) {
 782             okay = false; message += "Clip Qual Right = " + toString(header.clipQualRight) + ", but we only read " + toString(read.bases.length()) + " bases.\n";
 783         }
 784         if (header.clipQualLeft > read.qualScores.size()) {
 785             okay = false; message += "Clip Qual Left = " + toString(header.clipQualLeft) + ", but we only read " + toString(read.qualScores.size()) + " quality scores.\n";
 786         }
 787         if (header.clipQualRight > read.qualScores.size()) {
 788             okay = false; message += "Clip Qual Right = " + toString(header.clipQualRight) + ", but we only read " + toString(read.qualScores.size()) + " quality scores.\n";
 789         }
 790
 791         if (okay == false) {
 792             m->mothurOut(message); m->mothurOutEndLine();
 793         }
 794
 795                 return okay;
 796         }
 797         catch(exception& e) {
 798                 m->errorOut(e, "SffInfoCommand", "sanityCheck");
 799                 exit(1);
 800         }
 801 }
 802 //**********************************************************************************************************************
 803 int SffInfoCommand::printSffTxtSeqData(ofstream& out, seqRead& read, Header& header) {
 804         try {
 805                 out << "Flowgram: ";
 806                 for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << (read.flowgram[i]/(float)100) << '\t';  }
 807
 808                 out << endl <<  "Flow Indexes: ";
 809                 int sum = 0;
 810                 for (int i = 0; i < read.flowIndex.size(); i++) {  sum +=  read.flowIndex[i];  out << sum << '\t'; }
 811
 812                 //make the bases you want to clip lowercase and the bases you want to keep upper case
 813                 if(header.clipQualRight == 0){  header.clipQualRight = read.bases.length();     }
 814                 for (int i = 0; i < (header.clipQualLeft-1); i++) { read.bases[i] = tolower(read.bases[i]); }
 815                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   read.bases[i] = toupper(read.bases[i]);  }
 816                 for (int i = (header.clipQualRight-1); i < read.bases.length(); i++) {   read.bases[i] = tolower(read.bases[i]);  }
 817
 818                 out << endl <<  "Bases: " << read.bases << endl << "Quality Scores: ";
 819                 for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
 820
 821
 822                 out << endl << endl;
 823
 824                 return 0;
 825         }
 826         catch(exception& e) {
 827                 m->errorOut(e, "SffInfoCommand", "printSffTxtSeqData");
 828                 exit(1);
 829         }
 830 }
 831 //**********************************************************************************************************************
 832 int SffInfoCommand::printFastaSeqData(ofstream& out, seqRead& read, Header& header) {
 833         try {
 834                 string seq = read.bases;
 835
 836         if (trim) {
 837                         if(header.clipQualRight < header.clipQualLeft){
 838                                 seq = "NNNN";
 839                         }
 840                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
 841                                 seq = seq.substr((header.clipQualLeft-1), (header.clipQualRight-header.clipQualLeft));
 842                         }
 843                         else {
 844                                 seq = seq.substr(header.clipQualLeft-1);
 845                         }
 846                 }else{
 847                         //if you wanted the sfftxt then you already converted the bases to the right case
 848                         if (!sfftxt) {
 849                                 //make the bases you want to clip lowercase and the bases you want to keep upper case
 850                                 if(header.clipQualRight == 0){  header.clipQualRight = seq.length();    }
 851                                 for (int i = 0; i < (header.clipQualLeft-1); i++) { seq[i] = tolower(seq[i]);  }
 852                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++)  {   seq[i] = toupper(seq[i]);  }
 853                                 for (int i = (header.clipQualRight-1); i < seq.length(); i++) {   seq[i] = tolower(seq[i]);  }
 854                         }
 855                 }
 856
 857                 out << ">" << header.name  << " xy=" << header.xy << endl;
 858                 out << seq << endl;
 859
 860                 return 0;
 861         }
 862         catch(exception& e) {
 863                 m->errorOut(e, "SffInfoCommand", "printFastaSeqData");
 864                 exit(1);
 865         }
 866 }
 867
 868 //**********************************************************************************************************************
 869 int SffInfoCommand::printQualSeqData(ofstream& out, seqRead& read, Header& header) {
 870         try {
 871
 872                 if (trim) {
 873                         if(header.clipQualRight < header.clipQualLeft){
 874                                 out << ">" << header.name << " xy=" << header.xy << endl;
 875                                 out << "0\t0\t0\t0";
 876                         }
 877                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
 878                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
 879                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   out << read.qualScores[i] << '\t'; }
 880                         }
 881                         else{
 882                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
 883                                 for (int i = (header.clipQualLeft-1); i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';   }
 884                         }
 885                 }else{
 886                         out << ">" << header.name << " xy=" << header.xy << " length=" << read.qualScores.size() << endl;
 887                         for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
 888                 }
 889
 890                 out << endl;
 891
 892                 return 0;
 893         }
 894         catch(exception& e) {
 895                 m->errorOut(e, "SffInfoCommand", "printQualSeqData");
 896                 exit(1);
 897         }
 898 }
 899
 900 //**********************************************************************************************************************
 901 int SffInfoCommand::printFlowSeqData(ofstream& out, seqRead& read, Header& header) {
 902         try {
 903                 if(header.clipQualRight > header.clipQualLeft){
 904
 905                         int rightIndex = 0;
 906                         for (int i = 0; i < header.clipQualRight; i++) {  rightIndex +=  read.flowIndex[i];     }
 907
 908                         out << header.name << ' ' << rightIndex;
 909                         for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << ' ' << (read.flowgram[i]/(float)100);  }
 910                         out << endl;
 911                 }
 912
 913
 914                 return 0;
 915         }
 916         catch(exception& e) {
 917                 m->errorOut(e, "SffInfoCommand", "printFlowSeqData");
 918                 exit(1);
 919         }
 920 }
 921 //**********************************************************************************************************************
 922 int SffInfoCommand::readAccnosFile(string filename) {
 923         try {
 924                 //remove old names
 925                 seqNames.clear();
 926
 927                 ifstream in;
 928                 m->openInputFile(filename, in);
 929                 string name;
 930
 931                 while(!in.eof()){
 932                         in >> name; m->gobble(in);
 933
 934                         seqNames.insert(name);
 935
 936                         if (m->control_pressed) { seqNames.clear(); break; }
 937                 }
 938                 in.close();
 939
 940                 return 0;
 941         }
 942         catch(exception& e) {
 943                 m->errorOut(e, "SffInfoCommand", "readAccnosFile");
 944                 exit(1);
 945         }
 946 }
 947 //**********************************************************************************************************************
 948 int SffInfoCommand::parseSffTxt() {
 949         try {
 950
 951                 ifstream inSFF;
 952                 m->openInputFile(sfftxtFilename, inSFF);
 953
 954                 if (outputDir == "") {  outputDir += m->hasPath(sfftxtFilename); }
 955
 956                 //output file names
 957                 ofstream outFasta, outQual, outFlow;
 958                 string outFastaFileName, outQualFileName;
 959                 string fileRoot = m->getRootName(m->getSimpleName(sfftxtFilename));
 960                 if (fileRoot.length() > 0) {
 961                         //rip off last .
 962                         fileRoot = fileRoot.substr(0, fileRoot.length()-1);
 963                         fileRoot = m->getRootName(fileRoot);
 964                 }
 965
 966                 string outFlowFileName = outputDir + fileRoot + getOutputFileNameTag("flow");
 967                 if (trim) {
 968                         outFastaFileName = outputDir + fileRoot + getOutputFileNameTag("fasta");
 969                         outQualFileName = outputDir + fileRoot + getOutputFileNameTag("qfile");
 970                 }else{
 971                         outFastaFileName = outputDir + fileRoot + "raw." + getOutputFileNameTag("fasta");
 972                         outQualFileName = outputDir + fileRoot + "raw." + getOutputFileNameTag("qfile");
 973                 }
 974
 975                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
 976                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qfile"].push_back(outQualFileName);  }
 977                 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);  }
 978
 979                 //read common header
 980                 string commonHeader = m->getline(inSFF);
 981                 string magicNumber = m->getline(inSFF);
 982                 string version = m->getline(inSFF);
 983                 string indexOffset = m->getline(inSFF);
 984                 string indexLength = m->getline(inSFF);
 985                 int numReads = parseHeaderLineToInt(inSFF);
 986                 string headerLength = m->getline(inSFF);
 987                 string keyLength = m->getline(inSFF);
 988                 int numFlows = parseHeaderLineToInt(inSFF);
 989                 string flowgramCode = m->getline(inSFF);
 990                 string flowChars = m->getline(inSFF);
 991                 string keySequence = m->getline(inSFF);
 992                 m->gobble(inSFF);
 993
 994                 string seqName;
 995
 996                 if (flow)       {       outFlow << numFlows << endl;    }
 997
 998                 for(int i=0;i<numReads;i++){
 999
1000                         //sanity check
1001                         if (inSFF.eof()) { m->mothurOut("[ERROR]: Expected " + toString(numReads) + " but reached end of file at " + toString(i+1) + "."); m->mothurOutEndLine(); break; }
1002
1003                         Header header;
1004
1005                         //parse read header
1006                         inSFF >> seqName;
1007                         seqName = seqName.substr(1);
1008                         m->gobble(inSFF);
1009                         header.name = seqName;
1010
1011                         string runPrefix = parseHeaderLineToString(inSFF);              header.timestamp = runPrefix;
1012                         string regionNumber = parseHeaderLineToString(inSFF);   header.region = regionNumber;
1013                         string xyLocation = parseHeaderLineToString(inSFF);             header.xy = xyLocation;
1014                         m->gobble(inSFF);
1015
1016                         string runName = parseHeaderLineToString(inSFF);
1017                         string analysisName = parseHeaderLineToString(inSFF);
1018                         string fullPath = parseHeaderLineToString(inSFF);
1019                         m->gobble(inSFF);
1020
1021                         string readHeaderLen = parseHeaderLineToString(inSFF);  convert(readHeaderLen, header.headerLength);
1022                         string nameLength = parseHeaderLineToString(inSFF);             convert(nameLength, header.nameLength);
1023                         int numBases = parseHeaderLineToInt(inSFF);                             header.numBases = numBases;
1024                         string clipQualLeft = parseHeaderLineToString(inSFF);   convert(clipQualLeft, header.clipQualLeft);
1025                         int clipQualRight = parseHeaderLineToInt(inSFF);                header.clipQualRight = clipQualRight;
1026                         string clipAdapLeft = parseHeaderLineToString(inSFF);   convert(clipAdapLeft, header.clipAdapterLeft);
1027                         string clipAdapRight = parseHeaderLineToString(inSFF);  convert(clipAdapRight, header.clipAdapterRight);
1028                         m->gobble(inSFF);
1029
1030                         seqRead read;
1031
1032                         //parse read
1033                         vector<unsigned short> flowVector = parseHeaderLineToFloatVector(inSFF, numFlows);      read.flowgram = flowVector;
1034                         vector<unsigned int> flowIndices = parseHeaderLineToIntVector(inSFF, numBases);
1035
1036                         //adjust for print
1037                         vector<unsigned int> flowIndicesAdjusted; flowIndicesAdjusted.push_back(flowIndices[0]);
1038                         for (int j = 1; j < flowIndices.size(); j++) {   flowIndicesAdjusted.push_back(flowIndices[j] - flowIndices[j-1]);   }
1039                         read.flowIndex = flowIndicesAdjusted;
1040
1041                         string bases = parseHeaderLineToString(inSFF);                                                                          read.bases = bases;
1042                         vector<unsigned int> qualityScores = parseHeaderLineToIntVector(inSFF, numBases);       read.qualScores = qualityScores;
1043                         m->gobble(inSFF);
1044
1045                         //if you have provided an accosfile and this seq is not in it, then dont print
1046                         bool print = true;
1047                         if (seqNames.size() != 0) {   if (seqNames.count(header.name) == 0) { print = false; }  }
1048
1049                         //print
1050                         if (print) {
1051                                 if (fasta)      {       printFastaSeqData(outFasta, read, header);      }
1052                                 if (qual)       {       printQualSeqData(outQual, read, header);        }
1053                                 if (flow)       {       printFlowSeqData(outFlow, read, header);        }
1054                         }
1055
1056                         //report progress
1057                         if((i+1) % 10000 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine();             }
1058
1059                         if (m->control_pressed) {  break;  }
1060                 }
1061
1062                 //report progress
1063                 if (!m->control_pressed) {   if((numReads) % 10000 != 0){       m->mothurOut(toString(numReads)); m->mothurOutEndLine();                }  }
1064
1065                 inSFF.close();
1066
1067                 if (fasta)      {  outFasta.close();    }
1068                 if (qual)       {  outQual.close();             }
1069                 if (flow)       {  outFlow.close();             }
1070
1071                 return 0;
1072         }
1073         catch(exception& e) {
1074                 m->errorOut(e, "SffInfoCommand", "parseSffTxt");
1075                 exit(1);
1076         }
1077 }
1078 //**********************************************************************************************************************
1079
1080 int SffInfoCommand::parseHeaderLineToInt(ifstream& file){
1081         try {
1082                 int number;
1083
1084                 while (!file.eof())     {
1085
1086                         char c = file.get();
1087                         if (c == ':'){
1088                                 file >> number;
1089                                 break;
1090                         }
1091
1092                 }
1093                 m->gobble(file);
1094                 return number;
1095         }
1096         catch(exception& e) {
1097                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToInt");
1098                 exit(1);
1099         }
1100
1101 }
1102
1103 //**********************************************************************************************************************
1104
1105 string SffInfoCommand::parseHeaderLineToString(ifstream& file){
1106         try {
1107                 string text;
1108
1109                 while (!file.eof())     {
1110                         char c = file.get();
1111
1112                         if (c == ':'){
1113                                 //m->gobble(file);
1114                                 //text = m->getline(file);
1115                                 file >> text;
1116                                 break;
1117                         }
1118                 }
1119                 m->gobble(file);
1120
1121                 return text;
1122         }
1123         catch(exception& e) {
1124                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToString");
1125                 exit(1);
1126         }
1127 }
1128
1129 //**********************************************************************************************************************
1130
1131 vector<unsigned short> SffInfoCommand::parseHeaderLineToFloatVector(ifstream& file, int length){
1132         try {
1133                 vector<unsigned short> floatVector(length);
1134
1135                 while (!file.eof())     {
1136                         char c = file.get();
1137                         if (c == ':'){
1138                                 float temp;
1139                                 for(int i=0;i<length;i++){
1140                                         file >> temp;
1141                                         floatVector[i] = temp * 100;
1142                                 }
1143                                 break;
1144                         }
1145                 }
1146                 m->gobble(file);
1147                 return floatVector;
1148         }
1149         catch(exception& e) {
1150                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToFloatVector");
1151                 exit(1);
1152         }
1153 }
1154
1155 //**********************************************************************************************************************
1156
1157 vector<unsigned int> SffInfoCommand::parseHeaderLineToIntVector(ifstream& file, int length){
1158         try {
1159                 vector<unsigned int> intVector(length);
1160
1161                 while (!file.eof())     {
1162                         char c = file.get();
1163                         if (c == ':'){
1164                                 for(int i=0;i<length;i++){
1165                                         file >> intVector[i];
1166                                 }
1167                                 break;
1168                         }
1169                 }
1170                 m->gobble(file);
1171                 return intVector;
1172         }
1173         catch(exception& e) {
1174                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToIntVector");
1175                 exit(1);
1176         }
1177 }
1178
1179 //**********************************************************************************************************************
1180
1181
1182
1183