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