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::setParameters(){
  14         try {
  15                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(plist);
  16                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  17                 CommandParameter parasort("sort", "Multiple", "name-otu", "otu", "", "", "",false,false); parameters.push_back(parasort);
  18                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  19                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  20
  21                 vector<string> myArray;
  22                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  23                 return myArray;
  24         }
  25         catch(exception& e) {
  26                 m->errorOut(e, "GetListCountCommand", "setParameters");
  27                 exit(1);
  28         }
  29 }
  30 //**********************************************************************************************************************
  31 string GetListCountCommand::getHelpString(){
  32         try {
  33                 string helpString = "";
  34                 helpString += "The get.otulist command parameters are list, sort and label.  list is required, unless you have a valid current list file.\n";
  35                 helpString += "The label parameter allows you to select what distance levels you would like a output files created for, and are separated by dashes.\n";
  36                 helpString += "The sort parameter allows you to select how you want the output displayed. Options are otu and name.\n";
  37                 helpString += "If otu is selected the output will be otu number followed by the list of names in that otu.\n";
  38                 helpString += "If name is selected the output will be a sequence name followed by its otu number.\n";
  39                 helpString += "The get.otulist command should be in the following format: get.otulist(list=yourlistFile, label=yourLabels).\n";
  40                 helpString += "Example get.otulist(list=amazon.fn.list, label=0.10).\n";
  41                 helpString += "The default value for label is all lines in your inputfile.\n";
  42                 helpString += "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";
  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, "GetListCountCommand", "getHelpString");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 GetListCountCommand::GetListCountCommand(){
  53         try {
  54                 abort = true; calledHelp = true;
  55                 setParameters();
  56                 vector<string> tempOutNames;
  57                 outputTypes["otu"] = tempOutNames;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 GetListCountCommand::GetListCountCommand(string option)  {
  66         try {
  67                 abort = false; calledHelp = false;
  68                 allLines = 1;
  69
  70                 //allow user to run help
  71                 if(option == "help") { help(); abort = true; calledHelp = true; }
  72                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  73
  74                 else {
  75                         vector<string> myArray = setParameters();
  76
  77                         OptionParser parser(option);
  78                         map<string, string> parameters = parser.getParameters();
  79
  80                         ValidParameters validParameter;
  81                         map<string, string>::iterator it;
  82
  83                         //check to make sure all parameters are valid for command
  84                         for (it = parameters.begin(); it != parameters.end(); it++) {
  85                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  86                         }
  87
  88                         //initialize outputTypes
  89                         vector<string> tempOutNames;
  90                         outputTypes["otu"] = tempOutNames;
  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")  {
 112                                 listfile = m->getListFile();
 113                                 if (listfile != "") { m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 114                                 else {  m->mothurOut("You have no current list file and the list parameter is required."); m->mothurOutEndLine(); abort = true; }
 115                         }
 116                         else if (listfile == "not open") { abort = true; }
 117                         else { m->setListFile(listfile); }
 118
 119
 120                         //check for optional parameter and set defaults
 121                         // ...at some point should added some additional type checking...
 122                         sort = validParameter.validFile(parameters, "sort", false);       if (sort == "not found") { sort = "otu"; }
 123                         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"; }
 124
 125                         label = validParameter.validFile(parameters, "label", false);
 126                         if (label == "not found") { label = ""; }
 127                         else {
 128                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 129                                 else { allLines = 1;  }
 130                         }
 131                 }
 132         }
 133         catch(exception& e) {
 134                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
 135                 exit(1);
 136         }
 137 }
 138 //**********************************************************************************************************************
 139
 140 int GetListCountCommand::execute(){
 141         try {
 142                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 143
 144                 input = new InputData(listfile, "list");
 145                 list = input->getListVector();
 146                 string lastLabel = list->getLabel();
 147
 148                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 149                 set<string> processedLabels;
 150                 set<string> userLabels = labels;
 151
 152                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);        } return 0;  }
 153
 154                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 155
 156                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 157
 158                                 process(list);
 159
 160                                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);        } return 0;  }
 161
 162                                 processedLabels.insert(list->getLabel());
 163                                 userLabels.erase(list->getLabel());
 164                         }
 165
 166                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 167                                 string saveLabel = list->getLabel();
 168
 169                                 delete list;
 170                                 list = input->getListVector(lastLabel);
 171
 172                                 process(list);
 173
 174                                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);        } return 0;  }
 175
 176
 177                                 processedLabels.insert(list->getLabel());
 178                                 userLabels.erase(list->getLabel());
 179
 180                                 //restore real lastlabel to save below
 181                                 list->setLabel(saveLabel);
 182                         }
 183
 184                         lastLabel = list->getLabel();
 185
 186                         delete list;
 187                         list = input->getListVector();
 188                 }
 189
 190
 191                 //output error messages about any remaining user labels
 192                 set<string>::iterator it;
 193                 bool needToRun = false;
 194                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 195                         m->mothurOut("Your file does not include the label " + *it);
 196                         if (processedLabels.count(lastLabel) != 1) {
 197                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 198                                 needToRun = true;
 199                         }else {
 200                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 201                         }
 202                 }
 203
 204                 //run last label if you need to
 205                 if (needToRun == true)  {
 206                         if (list != NULL) {             delete list;    }
 207                         list = input->getListVector(lastLabel);
 208
 209                         process(list);
 210
 211                         if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);        } return 0;  }
 212
 213                         delete list;
 214                 }
 215
 216                 delete input;
 217
 218                 m->mothurOutEndLine();
 219                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 220                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 221                 m->mothurOutEndLine();
 222
 223                 return 0;
 224         }
 225         catch(exception& e) {
 226                 m->errorOut(e, "GetListCountCommand", "execute");
 227                 exit(1);
 228         }
 229 }
 230
 231 //**********************************************************************************************************************
 232 //return 1 if error, 0 otherwise
 233 void GetListCountCommand::process(ListVector* list) {
 234         try {
 235                 string binnames;
 236                 if (outputDir == "") { outputDir += m->hasPath(listfile); }
 237                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(listfile)) + list->getLabel() + ".otu";
 238                 m->openOutputFile(outputFileName, out);
 239                 outputNames.push_back(outputFileName); outputTypes["otu"].push_back(outputFileName);
 240
 241                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 242
 243                 //for each bin in the list vector
 244                 for (int i = 0; i < list->getNumBins(); i++) {
 245                         if (m->control_pressed) { break; }
 246
 247                         binnames = list->get(i);
 248
 249                         if (sort == "otu") {
 250                                 out << i+1 << '\t' << binnames << endl;
 251                         }else{ //sort = name
 252                                 vector<string> names;
 253                                 m->splitAtComma(binnames, names);
 254
 255                                 for (int j = 0; j < names.size(); j++) {
 256                                         out << names[j] << '\t' << i+1 << endl;
 257                                 }
 258                         }
 259                 }
 260
 261                 out.close();
 262         }
 263         catch(exception& e) {
 264                 m->errorOut(e, "GetListCountCommand", "process");
 265                 exit(1);
 266         }
 267 }
 268 //**********************************************************************************************************************
 269
 270