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