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;
  28                 //initialize outputTypes
  29                 vector<string> tempOutNames;
  30                 outputTypes["otu"] = tempOutNames;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 vector<string> GetListCountCommand::getRequiredParameters(){
  39         try {
  40                 string Array[] =  {"list"};
  41                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  42                 return myArray;
  43         }
  44         catch(exception& e) {
  45                 m->errorOut(e, "GetListCountCommand", "getRequiredParameters");
  46                 exit(1);
  47         }
  48 }
  49 //**********************************************************************************************************************
  50 vector<string> GetListCountCommand::getRequiredFiles(){
  51         try {
  52                 vector<string> myArray;
  53                 return myArray;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "GetListCountCommand", "getRequiredFiles");
  57                 exit(1);
  58         }
  59 }
  60 //**********************************************************************************************************************
  61 GetListCountCommand::GetListCountCommand(string option)  {
  62         try {
  63                 globaldata = GlobalData::getInstance();
  64                 abort = false;
  65                 allLines = 1;
  66                 labels.clear();
  67
  68                 //allow user to run help
  69                 if(option == "help") { help(); abort = true; }
  70
  71                 else {
  72                         //valid paramters for this command
  73                         string AlignArray[] =  {"list","label","sort","outputdir","inputdir"};
  74                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  75
  76                         OptionParser parser(option);
  77                         map<string, string> parameters = parser.getParameters();
  78
  79                         ValidParameters validParameter;
  80                         map<string, string>::iterator it;
  81
  82                         //check to make sure all parameters are valid for command
  83                         for (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["otu"] = tempOutNames;
  90
  91                         string ranRead = globaldata->getListFile();
  92
  93                         //if the user changes the input directory command factory will send this info to us in the output parameter
  94                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  95                         if (inputDir == "not found"){   inputDir = "";          }
  96                         else {
  97                                 string path;
  98                                 it = parameters.find("list");
  99                                 //user has given a template file
 100                                 if(it != parameters.end()){
 101                                         path = m->hasPath(it->second);
 102                                         //if the user has not given a path then, add inputdir. else leave path alone.
 103                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 104                                 }
 105                         }
 106
 107                         //if the user changes the output directory command factory will send this info to us in the output parameter
 108                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 109
 110                         //check for required parameters
 111                         listfile = validParameter.validFile(parameters, "list", true);
 112                         if ((listfile == "not found") && (globaldata->getListFile() == ""))  { m->mothurOut("You must read a listfile before running the get.listcount command.");  m->mothurOutEndLine(); abort = true; }
 113                         else if ((listfile == "not found") && (globaldata->getListFile() != "")) { listfile = globaldata->getListFile(); }
 114                         else if (listfile == "not open") { abort = true; }
 115                         else { globaldata->setListFile(listfile); }
 116
 117                         //check for optional parameter and set defaults
 118                         // ...at some point should added some additional type checking...
 119                         sort = validParameter.validFile(parameters, "sort", false);       if (sort == "not found") { sort = "otu"; }
 120                         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"; }
 121
 122                         label = validParameter.validFile(parameters, "label", false);
 123                         if (label == "not found") { label = ""; }
 124                         else {
 125                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 126                                 else { allLines = 1;  }
 127                         }
 128
 129                         //if the user has not specified any labels use the ones from read.otu
 130                         if ((label == "") && (ranRead != "")) {
 131                                 allLines = globaldata->allLines;
 132                                 labels = globaldata->labels;
 133                         }
 134                 }
 135         }
 136         catch(exception& e) {
 137                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
 138                 exit(1);
 139         }
 140 }
 141 //**********************************************************************************************************************
 142
 143 void GetListCountCommand::help(){
 144         try {
 145                 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");
 146                 m->mothurOut("The get.otulist command parameters are list, sort and label.  No parameters are required.\n");
 147                 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");
 148                 m->mothurOut("The sort parameter allows you to select how you want the output displayed. Options are otu and name.\n");
 149                 m->mothurOut("If otu is selected the output will be otu number followed by the list of names in that otu.\n");
 150                 m->mothurOut("If name is selected the output will be a sequence name followed by its otu number.\n");
 151                 m->mothurOut("The get.otulist command should be in the following format: get.otulist(list=yourlistFile, label=yourLabels).\n");
 152                 m->mothurOut("Example get.otulist(list=amazon.fn.list, label=0.10).\n");
 153                 m->mothurOut("The default value for label is all lines in your inputfile.\n");
 154                 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");
 155                 m->mothurOut("Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListFile).\n\n");
 156         }
 157         catch(exception& e) {
 158                 m->errorOut(e, "GetListCountCommand", "help");
 159                 exit(1);
 160         }
 161 }
 162
 163 //**********************************************************************************************************************
 164
 165 GetListCountCommand::~GetListCountCommand(){}
 166
 167 //**********************************************************************************************************************
 168
 169 int GetListCountCommand::execute(){
 170         try {
 171                 if (abort == true) {    return 0;       }
 172
 173                 globaldata->setFormat("list");
 174
 175                 //read list file
 176                 read = new ReadOTUFile(listfile);
 177                 read->read(&*globaldata);
 178
 179                 input = globaldata->ginput;
 180                 list = globaldata->gListVector;
 181                 string lastLabel = list->getLabel();
 182
 183                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 184                 set<string> processedLabels;
 185                 set<string> userLabels = labels;
 186
 187                 if (m->control_pressed) {
 188                         delete read;
 189                         delete input;
 190                         delete list;
 191                         globaldata->gListVector = NULL;
 192                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }
 193                         return 0;
 194                 }
 195
 196                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 197
 198                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 199
 200                                 process(list);
 201
 202                                 if (m->control_pressed) {
 203                                         delete read;
 204                                         delete input;
 205                                         delete list;
 206                                         globaldata->gListVector = NULL;
 207                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 208                                         return 0;
 209                                 }
 210
 211                                 processedLabels.insert(list->getLabel());
 212                                 userLabels.erase(list->getLabel());
 213                         }
 214
 215                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 216                                 string saveLabel = list->getLabel();
 217
 218                                 delete list;
 219                                 list = input->getListVector(lastLabel);
 220
 221                                 process(list);
 222
 223                                 if (m->control_pressed) {
 224                                         delete read;
 225                                         delete input;
 226                                         delete list;
 227                                         globaldata->gListVector = NULL;
 228                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 229                                         return 0;
 230                                 }
 231
 232                                 processedLabels.insert(list->getLabel());
 233                                 userLabels.erase(list->getLabel());
 234
 235                                 //restore real lastlabel to save below
 236                                 list->setLabel(saveLabel);
 237                         }
 238
 239                         lastLabel = list->getLabel();
 240
 241                         delete list;
 242                         list = input->getListVector();
 243                 }
 244
 245
 246                 //output error messages about any remaining user labels
 247                 set<string>::iterator it;
 248                 bool needToRun = false;
 249                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 250                         m->mothurOut("Your file does not include the label " + *it);
 251                         if (processedLabels.count(lastLabel) != 1) {
 252                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 253                                 needToRun = true;
 254                         }else {
 255                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 256                         }
 257                 }
 258
 259                 //run last label if you need to
 260                 if (needToRun == true)  {
 261                         if (list != NULL) {             delete list;    }
 262                         list = input->getListVector(lastLabel);
 263
 264                         process(list);
 265
 266                         if (m->control_pressed) {
 267                                         delete read;
 268                                         delete input;
 269                                         delete list;
 270                                         globaldata->gListVector = NULL;
 271                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } outputTypes.clear();
 272                                         return 0;
 273                         }
 274
 275                         delete list;
 276                 }
 277
 278                 delete read;
 279                 delete input;
 280                 globaldata->gListVector = NULL;
 281
 282                 m->mothurOutEndLine();
 283                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 284                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 285                 m->mothurOutEndLine();
 286
 287                 return 0;
 288         }
 289         catch(exception& e) {
 290                 m->errorOut(e, "GetListCountCommand", "execute");
 291                 exit(1);
 292         }
 293 }
 294
 295 //**********************************************************************************************************************
 296 //return 1 if error, 0 otherwise
 297 void GetListCountCommand::process(ListVector* list) {
 298         try {
 299                 string binnames;
 300                 if (outputDir == "") { outputDir += m->hasPath(listfile); }
 301                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(listfile)) + list->getLabel() + ".otu";
 302                 m->openOutputFile(outputFileName, out);
 303                 outputNames.push_back(outputFileName); outputTypes["otu"].push_back(outputFileName);
 304
 305                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 306
 307                 //for each bin in the list vector
 308                 for (int i = 0; i < list->getNumBins(); i++) {
 309                         if (m->control_pressed) { break; }
 310
 311                         binnames = list->get(i);
 312
 313                         if (sort == "otu") {
 314                                 out << i+1 << '\t' << binnames << endl;
 315                         }else{ //sort = name
 316                                 vector<string> names;
 317                                 m->splitAtComma(binnames, names);
 318
 319                                 for (int j = 0; j < names.size(); j++) {
 320                                         out << names[j] << '\t' << i+1 << endl;
 321                                 }
 322                         }
 323                 }
 324
 325                 out.close();
 326         }
 327         catch(exception& e) {
 328                 m->errorOut(e, "GetListCountCommand", "process");
 329                 exit(1);
 330         }
 331 }
 332 //**********************************************************************************************************************
 333
 334