parselistscommand.cpp

   1 /*
   2  *  parselistcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 2/24/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "parselistscommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> ParseListCommand::setParameters(){
  14         try {
  15                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none","list",false,true,true); parameters.push_back(plist);
  16         CommandParameter pcount("count", "InputTypes", "", "", "CountGroup", "CountGroup", "none","",false,false,true); parameters.push_back(pcount);
  17                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "CountGroup", "none","",false,false,true); parameters.push_back(pgroup);
  18                 CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
  19                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  20                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  21
  22                 vector<string> myArray;
  23                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  24                 return myArray;
  25         }
  26         catch(exception& e) {
  27                 m->errorOut(e, "ParseListCommand", "setParameters");
  28                 exit(1);
  29         }
  30 }
  31 //**********************************************************************************************************************
  32 string ParseListCommand::getHelpString(){
  33         try {
  34                 string helpString = "";
  35                 helpString += "The parse.list command reads a list and group or count file and generates a list file for each group in the group or count file. \n";
  36                 helpString += "The parse.list command parameters are list, group, count and label.\n";
  37                 helpString += "The list and group or count parameters are required.\n";
  38         helpString += "If a count file is provided, mothur assumes the list file contains only unique names.\n";
  39         helpString += "If a group file is provided, mothur assumes the list file contains all names.\n";
  40                 helpString += "The label parameter is used to read specific labels in your input you want to use.\n";
  41                 helpString += "The parse.list command should be used in the following format: parse.list(list=yourListFile, group=yourGroupFile, label=yourLabels).\n";
  42                 helpString += "Example: parse.list(list=abrecovery.fn.list, group=abrecovery.groups, label=0.03).\n";
  43                 helpString += "Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListfile).\n";
  44                 return helpString;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "ParseListCommand", "getHelpString");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 string ParseListCommand::getOutputPattern(string type) {
  53     try {
  54         string pattern = "";
  55
  56         if (type == "list") {  pattern = "[filename],[group],list"; }
  57         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  58
  59         return pattern;
  60     }
  61     catch(exception& e) {
  62         m->errorOut(e, "ParseListCommand", "getOutputPattern");
  63         exit(1);
  64     }
  65 }
  66 //**********************************************************************************************************************
  67 ParseListCommand::ParseListCommand(){
  68         try {
  69                 abort = true; calledHelp = true;
  70                 setParameters();
  71                 vector<string> tempOutNames;
  72                 outputTypes["list"] = tempOutNames;
  73         }
  74         catch(exception& e) {
  75                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
  76                 exit(1);
  77         }
  78 }
  79 //**********************************************************************************************************************
  80 ParseListCommand::ParseListCommand(string option)  {
  81         try {
  82                 abort = false; calledHelp = false;
  83                 allLines = 1;
  84
  85                 //allow user to run help
  86                 if(option == "help") { help(); abort = true; calledHelp = true; }
  87                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  88
  89                 else {
  90                         vector<string> myArray = setParameters();
  91
  92                         OptionParser parser(option);
  93                         map<string, string> parameters = parser.getParameters();
  94
  95                         ValidParameters validParameter;
  96                         map<string, string>::iterator it;
  97
  98                         //check to make sure all parameters are valid for command
  99                         for (it = parameters.begin(); it != parameters.end(); it++) {
 100                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 101                         }
 102
 103                         //initialize outputTypes
 104                         vector<string> tempOutNames;
 105                         outputTypes["list"] = tempOutNames;
 106
 107                         //if the user changes the input directory command factory will send this info to us in the output parameter
 108                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 109                         if (inputDir == "not found"){   inputDir = "";          }
 110                         else {
 111                                 string path;
 112                                 it = parameters.find("list");
 113                                 //user has given a template file
 114                                 if(it != parameters.end()){
 115                                         path = m->hasPath(it->second);
 116                                         //if the user has not given a path then, add inputdir. else leave path alone.
 117                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 118                                 }
 119
 120                                 it = parameters.find("group");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 126                                 }
 127
 128                 it = parameters.find("count");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 134                                 }
 135                         }
 136
 137
 138
 139
 140                         //check for required parameters
 141                         listfile = validParameter.validFile(parameters, "list", true);
 142                         if (listfile == "not open") { abort = true; }
 143                         else if (listfile == "not found") {
 144                                 listfile = m->getListFile();
 145                                 if (listfile != "") {  m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 146                                 else {
 147                                         m->mothurOut("No valid current list file. You must provide a list file."); m->mothurOutEndLine();
 148                                         abort = true;
 149
 150                                 }
 151                         }else { m->setListFile(listfile); }
 152
 153             //if the user changes the output directory command factory will send this info to us in the output parameter
 154                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(listfile);       }
 155
 156             groupfile = validParameter.validFile(parameters, "group", true);
 157                         if (groupfile == "not found") { groupfile =  "";   groupMap = NULL; }
 158                         else if (groupfile == "not open") { abort = true; groupfile =  ""; groupMap = NULL; }
 159                         else {
 160                 m->setGroupFile(groupfile);
 161                                 groupMap = new GroupMap(groupfile);
 162
 163                                 int error = groupMap->readMap();
 164                                 if (error == 1) { abort = true; }
 165             }
 166
 167             countfile = validParameter.validFile(parameters, "count", true);
 168                         if (countfile == "not found") { countfile =  "";   }
 169                         else if (countfile == "not open") { abort = true; countfile =  ""; }
 170                         else {
 171                 m->setCountTableFile(countfile);
 172                 ct.readTable(countfile, true);
 173                 if (!ct.hasGroupInfo()) {
 174                     abort = true;
 175                     m->mothurOut("[ERROR]: The parse.list command requires group info to be present in your countfile, quitting."); m->mothurOutEndLine();
 176                 }
 177
 178             }
 179
 180             if ((groupfile != "") && (countfile != "")) {
 181                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 182             }else if ((groupfile == "") && (countfile == "")) {
 183                 m->mothurOut("[ERROR]: you must provide one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 184             }
 185
 186                         //check for optional parameter and set defaults
 187                         // ...at some point should added some additional type checking...
 188                         label = validParameter.validFile(parameters, "label", false);
 189                         if (label == "not found") { label = "";  allLines = 1; }
 190                         else {
 191                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 192                                 else { allLines = 1;  }
 193                         }
 194                 }
 195
 196         }
 197         catch(exception& e) {
 198                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
 199                 exit(1);
 200         }
 201 }
 202 //**********************************************************************************************************************
 203 int ParseListCommand::execute(){
 204         try {
 205
 206                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 207
 208                 //set fileroot
 209                 map<string, string> variables;
 210         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(listfile));
 211
 212                 //fill filehandles with neccessary ofstreams
 213                 int i;
 214                 ofstream* temp;
 215                 vector<string> gGroups;
 216         if (groupfile != "") { gGroups = groupMap->getNamesOfGroups(); }
 217         else { gGroups = ct.getNamesOfGroups(); }
 218
 219                 for (i=0; i<gGroups.size(); i++) {
 220                         temp = new ofstream;
 221                         filehandles[gGroups[i]] = temp;
 222
 223             variables["[group]"] = gGroups[i];
 224                         string filename = getOutputFileName("list",variables);
 225                         outputNames.push_back(filename); outputTypes["list"].push_back(filename);
 226                         m->openOutputFile(filename, *temp);
 227                 }
 228
 229                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 230                 set<string> processedLabels;
 231                 set<string> userLabels = labels;
 232
 233                 InputData input(listfile, "list");
 234                 list = input.getListVector();
 235                 string lastLabel = list->getLabel();
 236
 237                 if (m->control_pressed) {
 238                         delete list; if (groupfile != "") { delete groupMap; }
 239                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 240                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 241                         return 0;
 242                 }
 243
 244                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 245
 246                         if (m->control_pressed) {
 247                                 delete list; if (groupfile != "") { delete groupMap; }
 248                                 for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 249                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 250                                 return 0;
 251                         }
 252
 253                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 254
 255                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 256                                         parse(list);
 257
 258                                         processedLabels.insert(list->getLabel());
 259                                         userLabels.erase(list->getLabel());
 260                         }
 261
 262                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 263                                         string saveLabel = list->getLabel();
 264
 265                                         list = input.getListVector(lastLabel); //get new list vector to process
 266
 267                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 268                                         parse(list);
 269
 270                                         processedLabels.insert(list->getLabel());
 271                                         userLabels.erase(list->getLabel());
 272
 273                                         //restore real lastlabel to save below
 274                                         list->setLabel(saveLabel);
 275                         }
 276
 277
 278                         lastLabel = list->getLabel();
 279
 280                         delete list;
 281                         list = input.getListVector(); //get new list vector to process
 282                 }
 283
 284                 if (m->control_pressed) {
 285                         if (groupfile != "") { delete groupMap; }
 286                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 287                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 288                         return 0;
 289                 }
 290
 291                 //output error messages about any remaining user labels
 292                 set<string>::iterator it;
 293                 bool needToRun = false;
 294                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 295                         m->mothurOut("Your file does not include the label " + *it);
 296                         if (processedLabels.count(lastLabel) != 1) {
 297                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 298                                 needToRun = true;
 299                         }else {
 300                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 301                         }
 302
 303                 }
 304
 305                 if (m->control_pressed) {
 306                         if (groupfile != "") { delete groupMap; }
 307                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 308                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 309                         return 0;
 310                 }
 311
 312                 //run last label if you need to
 313                 if (needToRun == true)  {
 314                         if (list != NULL) {     delete list;    }
 315                         list = input.getListVector(lastLabel); //get new list vector to process
 316
 317                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 318                         parse(list);
 319
 320                         delete list;
 321                 }
 322
 323                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 324                         (*(filehandles[it3->first])).close();
 325                         delete it3->second;
 326                 }
 327
 328                 if (groupfile != "") { delete groupMap; }
 329
 330                 if (m->control_pressed) {
 331                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 332                         return 0;
 333                 }
 334
 335                 //set fasta file as new current fastafile
 336                 string current = "";
 337                 itTypes = outputTypes.find("list");
 338                 if (itTypes != outputTypes.end()) {
 339                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 340                 }
 341
 342                 m->mothurOutEndLine();
 343                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 344                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 345                 m->mothurOutEndLine();
 346
 347                 return 0;
 348         }
 349         catch(exception& e) {
 350                 m->errorOut(e, "ParseListCommand", "execute");
 351                 exit(1);
 352         }
 353 }
 354 /**********************************************************************************************************************/
 355 int ParseListCommand::parse(ListVector* thisList) {
 356         try {
 357
 358                 map<string, string> groupVector;
 359                 map<string, string>::iterator itGroup;
 360                 map<string, int> groupNumBins;
 361
 362                 //print label
 363                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 364                         groupNumBins[it3->first] = 0;
 365                         groupVector[it3->first] = "";
 366                 }
 367
 368
 369                 for (int i = 0; i < thisList->getNumBins(); i++) {
 370                         if (m->control_pressed) { return 0; }
 371
 372                         map<string, string> groupBins;
 373                         string bin = list->get(i);
 374
 375                         vector<string> names;
 376                         m->splitAtComma(bin, names);  //parses bin into individual sequence names
 377
 378                         //parse bin into list of sequences in each group
 379                         for (int j = 0; j < names.size(); j++) {
 380                 if (groupfile != "") {
 381                     string group = groupMap->getGroup(names[j]);
 382
 383                     if (group == "not found") { m->mothurOut(names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); exit(1); }
 384
 385                     itGroup = groupBins.find(group);
 386                     if(itGroup == groupBins.end()) {
 387                         groupBins[group] = names[j];  //add first name
 388                         groupNumBins[group]++;
 389                     }else{ //add another name
 390                         groupBins[group] = groupBins[group] + "," + names[j];
 391                     }
 392                 }else{
 393                     vector<string> thisSeqsGroups = ct.getGroups(names[j]);
 394
 395                     for (int k = 0; k < thisSeqsGroups.size(); k++) {
 396                         string group = thisSeqsGroups[k];
 397                         itGroup = groupBins.find(group);
 398                         if(itGroup == groupBins.end()) {
 399                             groupBins[group] = names[j];  //add first name
 400                             groupNumBins[group]++;
 401                         }else{ //add another name
 402                             groupBins[group] = groupBins[group] + "," + names[j];
 403                         }
 404
 405                     }
 406                 }
 407                         }
 408
 409                         //print parsed bin info to files
 410                         for (itGroup = groupBins.begin(); itGroup != groupBins.end(); itGroup++) {
 411                                 groupVector[itGroup->first] +=  itGroup->second + '\t';
 412                         }
 413
 414                 }
 415
 416                 //end list vector
 417                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 418                         (*(filehandles[it3->first])) << thisList->getLabel() << '\t' << groupNumBins[it3->first] << '\t' << groupVector[it3->first] << endl;  // label numBins  listvector for that group
 419                 }
 420
 421                 return 0;
 422
 423         }
 424         catch(exception& e) {
 425                 m->errorOut(e, "ParseListCommand", "parse");
 426                 exit(1);
 427         }
 428 }
 429
 430 /**********************************************************************************************************************/
 431
 432