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","otu",false,true, 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 string GetListCountCommand::getOutputPattern(string type) {
  53     try {
  54         string pattern = "";
  55
  56         if (type == "otu") {  pattern = "[filename],[tag],otu"; }
  57         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  58
  59         return pattern;
  60     }
  61     catch(exception& e) {
  62         m->errorOut(e, "GetListCountCommand", "getOutputPattern");
  63         exit(1);
  64     }
  65 }
  66 //**********************************************************************************************************************
  67 GetListCountCommand::GetListCountCommand(){
  68         try {
  69                 abort = true; calledHelp = true;
  70                 setParameters();
  71                 vector<string> tempOutNames;
  72                 outputTypes["otu"] = tempOutNames;
  73         }
  74         catch(exception& e) {
  75                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
  76                 exit(1);
  77         }
  78 }
  79 //**********************************************************************************************************************
  80 GetListCountCommand::GetListCountCommand(string option)  {
  81         try {
  82                 abort = false; calledHelp = false;
  83                 allLines = 1;
  84
  85                 //allow user to run help
  86                 if(option == "help") { help(); abort = true; calledHelp = true; }
  87                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  88
  89                 else {
  90                         vector<string> myArray = setParameters();
  91
  92                         OptionParser parser(option);
  93                         map<string, string> parameters = parser.getParameters();
  94
  95                         ValidParameters validParameter;
  96                         map<string, string>::iterator it;
  97
  98                         //check to make sure all parameters are valid for command
  99                         for (it = parameters.begin(); it != parameters.end(); it++) {
 100                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 101                         }
 102
 103                         //initialize outputTypes
 104                         vector<string> tempOutNames;
 105                         outputTypes["otu"] = tempOutNames;
 106
 107                         //if the user changes the input directory command factory will send this info to us in the output parameter
 108                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 109                         if (inputDir == "not found"){   inputDir = "";          }
 110                         else {
 111                                 string path;
 112                                 it = parameters.find("list");
 113                                 //user has given a template file
 114                                 if(it != parameters.end()){
 115                                         path = m->hasPath(it->second);
 116                                         //if the user has not given a path then, add inputdir. else leave path alone.
 117                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 118                                 }
 119                         }
 120
 121                         //if the user changes the output directory command factory will send this info to us in the output parameter
 122                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 123
 124                         //check for required parameters
 125                         listfile = validParameter.validFile(parameters, "list", true);
 126                         if (listfile == "not found")  {
 127                                 listfile = m->getListFile();
 128                                 if (listfile != "") { m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 129                                 else {  m->mothurOut("You have no current list file and the list parameter is required."); m->mothurOutEndLine(); abort = true; }
 130                         }
 131                         else if (listfile == "not open") { abort = true; }
 132                         else { m->setListFile(listfile); }
 133
 134
 135                         //check for optional parameter and set defaults
 136                         // ...at some point should added some additional type checking...
 137                         sort = validParameter.validFile(parameters, "sort", false);       if (sort == "not found") { sort = "otu"; }
 138                         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"; }
 139
 140                         label = validParameter.validFile(parameters, "label", false);
 141                         if (label == "not found") { label = ""; }
 142                         else {
 143                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 144                                 else { allLines = 1;  }
 145                         }
 146                 }
 147         }
 148         catch(exception& e) {
 149                 m->errorOut(e, "GetListCountCommand", "GetListCountCommand");
 150                 exit(1);
 151         }
 152 }
 153 //**********************************************************************************************************************
 154
 155 int GetListCountCommand::execute(){
 156         try {
 157                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 158
 159                 input = new InputData(listfile, "list");
 160                 list = input->getListVector();
 161                 string lastLabel = list->getLabel();
 162
 163                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 164                 set<string> processedLabels;
 165                 set<string> userLabels = labels;
 166
 167                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);        } return 0;  }
 168
 169                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 170
 171                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 172
 173                                 process(list);
 174
 175                                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);        } return 0;  }
 176
 177                                 processedLabels.insert(list->getLabel());
 178                                 userLabels.erase(list->getLabel());
 179                         }
 180
 181                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 182                                 string saveLabel = list->getLabel();
 183
 184                                 delete list;
 185                                 list = input->getListVector(lastLabel);
 186
 187                                 process(list);
 188
 189                                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);        } return 0;  }
 190
 191
 192                                 processedLabels.insert(list->getLabel());
 193                                 userLabels.erase(list->getLabel());
 194
 195                                 //restore real lastlabel to save below
 196                                 list->setLabel(saveLabel);
 197                         }
 198
 199                         lastLabel = list->getLabel();
 200
 201                         delete list;
 202                         list = input->getListVector();
 203                 }
 204
 205
 206                 //output error messages about any remaining user labels
 207                 set<string>::iterator it;
 208                 bool needToRun = false;
 209                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 210                         m->mothurOut("Your file does not include the label " + *it);
 211                         if (processedLabels.count(lastLabel) != 1) {
 212                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 213                                 needToRun = true;
 214                         }else {
 215                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 216                         }
 217                 }
 218
 219                 //run last label if you need to
 220                 if (needToRun == true)  {
 221                         if (list != NULL) {             delete list;    }
 222                         list = input->getListVector(lastLabel);
 223
 224                         process(list);
 225
 226                         if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);        } return 0;  }
 227
 228                         delete list;
 229                 }
 230
 231                 delete input;
 232
 233                 m->mothurOutEndLine();
 234                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 235                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 236                 m->mothurOutEndLine();
 237
 238                 return 0;
 239         }
 240         catch(exception& e) {
 241                 m->errorOut(e, "GetListCountCommand", "execute");
 242                 exit(1);
 243         }
 244 }
 245
 246 //**********************************************************************************************************************
 247 //return 1 if error, 0 otherwise
 248 void GetListCountCommand::process(ListVector* list) {
 249         try {
 250                 string binnames;
 251                 if (outputDir == "") { outputDir += m->hasPath(listfile); }
 252         map<string, string> variables;
 253                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(listfile));
 254         variables["[tag]"] = list->getLabel();
 255                 string outputFileName = getOutputFileName("otu", variables);
 256
 257                 m->openOutputFile(outputFileName, out);
 258                 outputNames.push_back(outputFileName); outputTypes["otu"].push_back(outputFileName);
 259
 260                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 261
 262                 //for each bin in the list vector
 263                 for (int i = 0; i < list->getNumBins(); i++) {
 264                         if (m->control_pressed) { break; }
 265
 266                         binnames = list->get(i);
 267
 268                         if (sort == "otu") {
 269                                 out << i+1 << '\t' << binnames << endl;
 270                         }else{ //sort = name
 271                                 vector<string> names;
 272                                 m->splitAtComma(binnames, names);
 273
 274                                 for (int j = 0; j < names.size(); j++) {
 275                                         out << names[j] << '\t' << i+1 << endl;
 276                                 }
 277                         }
 278                 }
 279
 280                 out.close();
 281         }
 282         catch(exception& e) {
 283                 m->errorOut(e, "GetListCountCommand", "process");
 284                 exit(1);
 285         }
 286 }
 287 //**********************************************************************************************************************
 288
 289