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
  73                 else {
  74                         vector<string> myArray = setParameters();
  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                         //if the user changes the input directory command factory will send this info to us in the output parameter
  92                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  93                         if (inputDir == "not found"){   inputDir = "";          }
  94                         else {
  95                                 string path;
  96                                 it = parameters.find("list");
  97                                 //user has given a template file
  98                                 if(it != parameters.end()){
  99                                         path = m->hasPath(it->second);
 100                                         //if the user has not given a path then, add inputdir. else leave path alone.
 101                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 102                                 }
 103                         }
 104
 105                         //if the user changes the output directory command factory will send this info to us in the output parameter
 106                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 107
 108                         //check for required parameters
 109                         listfile = validParameter.validFile(parameters, "list", true);
 110                         if (listfile == "not found")  {
 111                                 listfile = m->getListFile();
 112                                 if (listfile != "") { m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 113                                 else {  m->mothurOut("You have no current list file and the list parameter is required."); m->mothurOutEndLine(); abort = true; }
 114                         }
 115                         else if (listfile == "not open") { abort = true; }
 116
 117
 118                         //check for optional parameter and set defaults
 119                         // ...at some point should added some additional type checking...
 120                         sort = validParameter.validFile(parameters, "sort", false);       if (sort == "not found") { sort = "otu"; }
 121                         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"; }
 122
 123                         label = validParameter.validFile(parameters, "label", false);
 124                         if (label == "not found") { label = ""; }
 125                         else {
 126                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 127                                 else { allLines = 1;  }
 128                         }
 129                 }
 130         }
 131         catch(exception& e) {
 132                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
 133                 exit(1);
 134         }
 135 }
 136 //**********************************************************************************************************************
 137
 138 int GetListCountCommand::execute(){
 139         try {
 140                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 141
 142                 input = new InputData(listfile, "list");
 143                 list = input->getListVector();
 144                 string lastLabel = list->getLabel();
 145
 146                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 147                 set<string> processedLabels;
 148                 set<string> userLabels = labels;
 149
 150                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } return 0;  }
 151
 152                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 153
 154                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 155
 156                                 process(list);
 157
 158                                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } return 0;  }
 159
 160                                 processedLabels.insert(list->getLabel());
 161                                 userLabels.erase(list->getLabel());
 162                         }
 163
 164                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 165                                 string saveLabel = list->getLabel();
 166
 167                                 delete list;
 168                                 list = input->getListVector(lastLabel);
 169
 170                                 process(list);
 171
 172                                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } return 0;  }
 173
 174
 175                                 processedLabels.insert(list->getLabel());
 176                                 userLabels.erase(list->getLabel());
 177
 178                                 //restore real lastlabel to save below
 179                                 list->setLabel(saveLabel);
 180                         }
 181
 182                         lastLabel = list->getLabel();
 183
 184                         delete list;
 185                         list = input->getListVector();
 186                 }
 187
 188
 189                 //output error messages about any remaining user labels
 190                 set<string>::iterator it;
 191                 bool needToRun = false;
 192                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 193                         m->mothurOut("Your file does not include the label " + *it);
 194                         if (processedLabels.count(lastLabel) != 1) {
 195                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 196                                 needToRun = true;
 197                         }else {
 198                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 199                         }
 200                 }
 201
 202                 //run last label if you need to
 203                 if (needToRun == true)  {
 204                         if (list != NULL) {             delete list;    }
 205                         list = input->getListVector(lastLabel);
 206
 207                         process(list);
 208
 209                         if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } return 0;  }
 210
 211                         delete list;
 212                 }
 213
 214                 delete input;
 215
 216                 m->mothurOutEndLine();
 217                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 218                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 219                 m->mothurOutEndLine();
 220
 221                 return 0;
 222         }
 223         catch(exception& e) {
 224                 m->errorOut(e, "GetListCountCommand", "execute");
 225                 exit(1);
 226         }
 227 }
 228
 229 //**********************************************************************************************************************
 230 //return 1 if error, 0 otherwise
 231 void GetListCountCommand::process(ListVector* list) {
 232         try {
 233                 string binnames;
 234                 if (outputDir == "") { outputDir += m->hasPath(listfile); }
 235                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(listfile)) + list->getLabel() + ".otu";
 236                 m->openOutputFile(outputFileName, out);
 237                 outputNames.push_back(outputFileName); outputTypes["otu"].push_back(outputFileName);
 238
 239                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 240
 241                 //for each bin in the list vector
 242                 for (int i = 0; i < list->getNumBins(); i++) {
 243                         if (m->control_pressed) { break; }
 244
 245                         binnames = list->get(i);
 246
 247                         if (sort == "otu") {
 248                                 out << i+1 << '\t' << binnames << endl;
 249                         }else{ //sort = name
 250                                 vector<string> names;
 251                                 m->splitAtComma(binnames, names);
 252
 253                                 for (int j = 0; j < names.size(); j++) {
 254                                         out << names[j] << '\t' << i+1 << endl;
 255                                 }
 256                         }
 257                 }
 258
 259                 out.close();
 260         }
 261         catch(exception& e) {
 262                 m->errorOut(e, "GetListCountCommand", "process");
 263                 exit(1);
 264         }
 265 }
 266 //**********************************************************************************************************************
 267
 268