getlistcountcommand.cpp

   1 /*
   2  *  getlistcountcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/12/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "getlistcountcommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> GetListCountCommand::getValidParameters(){
  14         try {
  15                 string Array[] =  {"list","label","sort","outputdir","inputdir"};
  16                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  17                 return myArray;
  18         }
  19         catch(exception& e) {
  20                 m->errorOut(e, "GetListCountCommand", "getValidParameters");
  21                 exit(1);
  22         }
  23 }
  24 //**********************************************************************************************************************
  25 GetListCountCommand::GetListCountCommand(){
  26         try {
  27                 abort = true; calledHelp = true;
  28                 vector<string> tempOutNames;
  29                 outputTypes["otu"] = tempOutNames;
  30         }
  31         catch(exception& e) {
  32                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
  33                 exit(1);
  34         }
  35 }
  36 //**********************************************************************************************************************
  37 vector<string> GetListCountCommand::getRequiredParameters(){
  38         try {
  39                 string Array[] =  {"list"};
  40                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  41                 return myArray;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "GetListCountCommand", "getRequiredParameters");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 vector<string> GetListCountCommand::getRequiredFiles(){
  50         try {
  51                 vector<string> myArray;
  52                 return myArray;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "GetListCountCommand", "getRequiredFiles");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 GetListCountCommand::GetListCountCommand(string option)  {
  61         try {
  62                 globaldata = GlobalData::getInstance();
  63                 abort = false; calledHelp = false;
  64                 allLines = 1;
  65                 labels.clear();
  66
  67                 //allow user to run help
  68                 if(option == "help") { help(); abort = true; calledHelp = true; }
  69
  70                 else {
  71                         //valid paramters for this command
  72                         string AlignArray[] =  {"list","label","sort","outputdir","inputdir"};
  73                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  74
  75                         OptionParser parser(option);
  76                         map<string, string> parameters = parser.getParameters();
  77
  78                         ValidParameters validParameter;
  79                         map<string, string>::iterator it;
  80
  81                         //check to make sure all parameters are valid for command
  82                         for (it = parameters.begin(); it != parameters.end(); it++) {
  83                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  84                         }
  85
  86                         //initialize outputTypes
  87                         vector<string> tempOutNames;
  88                         outputTypes["otu"] = tempOutNames;
  89
  90                         string ranRead = globaldata->getListFile();
  91
  92                         //if the user changes the input directory command factory will send this info to us in the output parameter
  93                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  94                         if (inputDir == "not found"){   inputDir = "";          }
  95                         else {
  96                                 string path;
  97                                 it = parameters.find("list");
  98                                 //user has given a template file
  99                                 if(it != parameters.end()){
 100                                         path = m->hasPath(it->second);
 101                                         //if the user has not given a path then, add inputdir. else leave path alone.
 102                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 103                                 }
 104                         }
 105
 106                         //if the user changes the output directory command factory will send this info to us in the output parameter
 107                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 108
 109                         //check for required parameters
 110                         listfile = validParameter.validFile(parameters, "list", true);
 111                         if ((listfile == "not found") && (globaldata->getListFile() == ""))  { m->mothurOut("You must read a listfile before running the get.listcount command.");  m->mothurOutEndLine(); abort = true; }
 112                         else if ((listfile == "not found") && (globaldata->getListFile() != "")) { listfile = globaldata->getListFile(); }
 113                         else if (listfile == "not open") { abort = true; }
 114                         else { globaldata->setListFile(listfile); }
 115
 116                         //check for optional parameter and set defaults
 117                         // ...at some point should added some additional type checking...
 118                         sort = validParameter.validFile(parameters, "sort", false);       if (sort == "not found") { sort = "otu"; }
 119                         if ((sort != "otu") && (sort != "name")) { m->mothurOut( sort + " is not a valid sort option. Options are otu and name. I will use otu."); m->mothurOutEndLine(); sort = "otu"; }
 120
 121                         label = validParameter.validFile(parameters, "label", false);
 122                         if (label == "not found") { label = ""; }
 123                         else {
 124                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 125                                 else { allLines = 1;  }
 126                         }
 127
 128                         //if the user has not specified any labels use the ones from read.otu
 129                         if ((label == "") && (ranRead != "")) {
 130                                 allLines = globaldata->allLines;
 131                                 labels = globaldata->labels;
 132                         }
 133                 }
 134         }
 135         catch(exception& e) {
 136                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
 137                 exit(1);
 138         }
 139 }
 140 //**********************************************************************************************************************
 141
 142 void GetListCountCommand::help(){
 143         try {
 144                 m->mothurOut("The get.otulist command can only be executed after a successful read.otu command of a listfile or providing a list file using the list parameter.\n");
 145                 m->mothurOut("The get.otulist command parameters are list, sort and label.  No parameters are required.\n");
 146                 m->mothurOut("The label parameter allows you to select what distance levels you would like a output files created for, and are separated by dashes.\n");
 147                 m->mothurOut("The sort parameter allows you to select how you want the output displayed. Options are otu and name.\n");
 148                 m->mothurOut("If otu is selected the output will be otu number followed by the list of names in that otu.\n");
 149                 m->mothurOut("If name is selected the output will be a sequence name followed by its otu number.\n");
 150                 m->mothurOut("The get.otulist command should be in the following format: get.otulist(list=yourlistFile, label=yourLabels).\n");
 151                 m->mothurOut("Example get.otulist(list=amazon.fn.list, label=0.10).\n");
 152                 m->mothurOut("The default value for label is all lines in your inputfile.\n");
 153                 m->mothurOut("The get.otulist command outputs a .otu file for each distance you specify listing the bin number and the names of the sequences in that bin.\n");
 154                 m->mothurOut("Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListFile).\n\n");
 155         }
 156         catch(exception& e) {
 157                 m->errorOut(e, "GetListCountCommand", "help");
 158                 exit(1);
 159         }
 160 }
 161
 162 //**********************************************************************************************************************
 163
 164 GetListCountCommand::~GetListCountCommand(){}
 165
 166 //**********************************************************************************************************************
 167
 168 int GetListCountCommand::execute(){
 169         try {
 170                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 171
 172                 globaldata->setFormat("list");
 173
 174                 //read list file
 175                 read = new ReadOTUFile(listfile);
 176                 read->read(&*globaldata);
 177
 178                 input = globaldata->ginput;
 179                 list = globaldata->gListVector;
 180                 string lastLabel = list->getLabel();
 181
 182                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 183                 set<string> processedLabels;
 184                 set<string> userLabels = labels;
 185
 186                 if (m->control_pressed) {
 187                         delete read;
 188                         delete input;
 189                         delete list;
 190                         globaldata->gListVector = NULL;
 191                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 192                         return 0;
 193                 }
 194
 195                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 196
 197                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 198
 199                                 process(list);
 200
 201                                 if (m->control_pressed) {
 202                                         delete read;
 203                                         delete input;
 204                                         delete list;
 205                                         globaldata->gListVector = NULL;
 206                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 207                                         return 0;
 208                                 }
 209
 210                                 processedLabels.insert(list->getLabel());
 211                                 userLabels.erase(list->getLabel());
 212                         }
 213
 214                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 215                                 string saveLabel = list->getLabel();
 216
 217                                 delete list;
 218                                 list = input->getListVector(lastLabel);
 219
 220                                 process(list);
 221
 222                                 if (m->control_pressed) {
 223                                         delete read;
 224                                         delete input;
 225                                         delete list;
 226                                         globaldata->gListVector = NULL;
 227                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 228                                         return 0;
 229                                 }
 230
 231                                 processedLabels.insert(list->getLabel());
 232                                 userLabels.erase(list->getLabel());
 233
 234                                 //restore real lastlabel to save below
 235                                 list->setLabel(saveLabel);
 236                         }
 237
 238                         lastLabel = list->getLabel();
 239
 240                         delete list;
 241                         list = input->getListVector();
 242                 }
 243
 244
 245                 //output error messages about any remaining user labels
 246                 set<string>::iterator it;
 247                 bool needToRun = false;
 248                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 249                         m->mothurOut("Your file does not include the label " + *it);
 250                         if (processedLabels.count(lastLabel) != 1) {
 251                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 252                                 needToRun = true;
 253                         }else {
 254                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 255                         }
 256                 }
 257
 258                 //run last label if you need to
 259                 if (needToRun == true)  {
 260                         if (list != NULL) {             delete list;    }
 261                         list = input->getListVector(lastLabel);
 262
 263                         process(list);
 264
 265                         if (m->control_pressed) {
 266                                         delete read;
 267                                         delete input;
 268                                         delete list;
 269                                         globaldata->gListVector = NULL;
 270                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 271                                         return 0;
 272                         }
 273
 274                         delete list;
 275                 }
 276
 277                 delete read;
 278                 delete input;
 279                 globaldata->gListVector = NULL;
 280
 281                 m->mothurOutEndLine();
 282                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 283                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 284                 m->mothurOutEndLine();
 285
 286                 return 0;
 287         }
 288         catch(exception& e) {
 289                 m->errorOut(e, "GetListCountCommand", "execute");
 290                 exit(1);
 291         }
 292 }
 293
 294 //**********************************************************************************************************************
 295 //return 1 if error, 0 otherwise
 296 void GetListCountCommand::process(ListVector* list) {
 297         try {
 298                 string binnames;
 299                 if (outputDir == "") { outputDir += m->hasPath(listfile); }
 300                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(listfile)) + list->getLabel() + ".otu";
 301                 m->openOutputFile(outputFileName, out);
 302                 outputNames.push_back(outputFileName); outputTypes["otu"].push_back(outputFileName);
 303
 304                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 305
 306                 //for each bin in the list vector
 307                 for (int i = 0; i < list->getNumBins(); i++) {
 308                         if (m->control_pressed) { break; }
 309
 310                         binnames = list->get(i);
 311
 312                         if (sort == "otu") {
 313                                 out << i+1 << '\t' << binnames << endl;
 314                         }else{ //sort = name
 315                                 vector<string> names;
 316                                 m->splitAtComma(binnames, names);
 317
 318                                 for (int j = 0; j < names.size(); j++) {
 319                                         out << names[j] << '\t' << i+1 << endl;
 320                                 }
 321                         }
 322                 }
 323
 324                 out.close();
 325         }
 326         catch(exception& e) {
 327                 m->errorOut(e, "GetListCountCommand", "process");
 328                 exit(1);
 329         }
 330 }
 331 //**********************************************************************************************************************
 332
 333