parselistscommand.cpp

   1 /*
   2  *  parselistcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 2/24/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "parselistscommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> ParseListCommand::setParameters(){
  14         try {
  15                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(plist);
  16                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pgroup);
  17                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  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, "ParseListCommand", "setParameters");
  27                 exit(1);
  28         }
  29 }
  30 //**********************************************************************************************************************
  31 string ParseListCommand::getHelpString(){
  32         try {
  33                 string helpString = "";
  34                 helpString += "The parse.list command reads a list and group file and generates a list file for each group in the groupfile. \n";
  35                 helpString += "The parse.list command parameters are list, group and label.\n";
  36                 helpString += "The list and group parameters are required.\n";
  37                 helpString += "The label parameter is used to read specific labels in your input you want to use.\n";
  38                 helpString += "The parse.list command should be used in the following format: parse.list(list=yourListFile, group=yourGroupFile, label=yourLabels).\n";
  39                 helpString += "Example: parse.list(list=abrecovery.fn.list, group=abrecovery.groups, label=0.03).\n";
  40                 helpString += "Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListfile).\n";
  41                 return helpString;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "ParseListCommand", "getHelpString");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 ParseListCommand::ParseListCommand(){
  50         try {
  51                 abort = true; calledHelp = true;
  52                 setParameters();
  53                 vector<string> tempOutNames;
  54                 outputTypes["list"] = tempOutNames;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
  58                 exit(1);
  59         }
  60 }
  61 //**********************************************************************************************************************
  62 ParseListCommand::ParseListCommand(string option)  {
  63         try {
  64                 abort = false; calledHelp = false;
  65                 allLines = 1;
  66
  67                 //allow user to run help
  68                 if(option == "help") { help(); abort = true; calledHelp = true; }
  69                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  70
  71                 else {
  72                         vector<string> myArray = setParameters();
  73
  74                         OptionParser parser(option);
  75                         map<string, string> parameters = parser.getParameters();
  76
  77                         ValidParameters validParameter;
  78                         map<string, string>::iterator it;
  79
  80                         //check to make sure all parameters are valid for command
  81                         for (it = parameters.begin(); it != parameters.end(); it++) {
  82                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  83                         }
  84
  85                         //initialize outputTypes
  86                         vector<string> tempOutNames;
  87                         outputTypes["list"] = tempOutNames;
  88
  89                         //if the user changes the input directory command factory will send this info to us in the output parameter
  90                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  91                         if (inputDir == "not found"){   inputDir = "";          }
  92                         else {
  93                                 string path;
  94                                 it = parameters.find("list");
  95                                 //user has given a template file
  96                                 if(it != parameters.end()){
  97                                         path = m->hasPath(it->second);
  98                                         //if the user has not given a path then, add inputdir. else leave path alone.
  99                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 100                                 }
 101
 102                                 it = parameters.find("group");
 103                                 //user has given a template file
 104                                 if(it != parameters.end()){
 105                                         path = m->hasPath(it->second);
 106                                         //if the user has not given a path then, add inputdir. else leave path alone.
 107                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 108                                 }
 109                         }
 110
 111
 112                         //if the user changes the output directory command factory will send this info to us in the output parameter
 113                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 114
 115                         //check for required parameters
 116                         listfile = validParameter.validFile(parameters, "list", true);
 117                         if (listfile == "not open") { abort = true; }
 118                         else if (listfile == "not found") {
 119                                 listfile = m->getListFile();
 120                                 if (listfile != "") {  m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 121                                 else {
 122                                         m->mothurOut("No valid current list file. You must provide a list file."); m->mothurOutEndLine();
 123                                         abort = true;
 124
 125                                 }
 126                         }else { m->setListFile(listfile); }
 127
 128                         groupfile = validParameter.validFile(parameters, "group", true);
 129                         if (groupfile == "not open") { abort = true; }
 130                         else if (groupfile == "not found") {
 131                                 groupfile = m->getListFile();
 132                                 if (groupfile != "") {
 133                                         m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine();
 134                                         groupMap = new GroupMap(groupfile);
 135
 136                                         int error = groupMap->readMap();
 137                                         if (error == 1) { abort = true; }
 138                                 }else { m->mothurOut("No valid current group file. You must provide a group file."); m->mothurOutEndLine();  abort = true; }
 139                         }else {
 140                                 m->setGroupFile(groupfile);
 141                                 groupMap = new GroupMap(groupfile);
 142
 143                                 int error = groupMap->readMap();
 144                                 if (error == 1) { abort = true; }
 145                         }
 146
 147                         //do you have all files needed
 148                         if ((listfile == "") || (groupfile == "")) { m->mothurOut("You must enter both a listfile and groupfile for the parse.list command. "); m->mothurOutEndLine(); abort = true;  }
 149
 150                         //check for optional parameter and set defaults
 151                         // ...at some point should added some additional type checking...
 152                         label = validParameter.validFile(parameters, "label", false);
 153                         if (label == "not found") { label = "";  allLines = 1; }
 154                         else {
 155                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 156                                 else { allLines = 1;  }
 157                         }
 158                 }
 159
 160         }
 161         catch(exception& e) {
 162                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
 163                 exit(1);
 164         }
 165 }
 166 //**********************************************************************************************************************
 167 int ParseListCommand::execute(){
 168         try {
 169
 170                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 171
 172                 //set fileroot
 173                 string fileroot = outputDir + m->getRootName(m->getSimpleName(listfile));
 174
 175                 //fill filehandles with neccessary ofstreams
 176                 int i;
 177                 ofstream* temp;
 178                 vector<string> gGroups = groupMap->getNamesOfGroups();
 179                 for (i=0; i<gGroups.size(); i++) {
 180                         temp = new ofstream;
 181                         filehandles[gGroups[i]] = temp;
 182
 183                         string filename = fileroot +  gGroups[i] + ".list";
 184                         outputNames.push_back(filename); outputTypes["list"].push_back(filename);
 185                         m->openOutputFile(filename, *temp);
 186                 }
 187
 188                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 189                 set<string> processedLabels;
 190                 set<string> userLabels = labels;
 191
 192                 input = new InputData(listfile, "list");
 193                 list = input->getListVector();
 194                 string lastLabel = list->getLabel();
 195
 196                 if (m->control_pressed) {
 197                         delete input; delete list; delete groupMap;
 198                         vector<string> gGroups = groupMap->getNamesOfGroups();
 199                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 200                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 201                         return 0;
 202                 }
 203
 204                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 205
 206                         if (m->control_pressed) {
 207                                 delete input; delete list; delete groupMap;
 208                                 for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 209                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 210                                 return 0;
 211                         }
 212
 213                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 214
 215                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 216                                         parse(list);
 217
 218                                         processedLabels.insert(list->getLabel());
 219                                         userLabels.erase(list->getLabel());
 220                         }
 221
 222                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 223                                         string saveLabel = list->getLabel();
 224
 225                                         delete list;
 226                                         list = input->getListVector(lastLabel); //get new list vector to process
 227
 228                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 229                                         parse(list);
 230
 231                                         processedLabels.insert(list->getLabel());
 232                                         userLabels.erase(list->getLabel());
 233
 234                                         //restore real lastlabel to save below
 235                                         list->setLabel(saveLabel);
 236                         }
 237
 238
 239                         lastLabel = list->getLabel();
 240
 241                         delete list;
 242                         list = input->getListVector(); //get new list vector to process
 243                 }
 244
 245                 if (m->control_pressed) {
 246                         delete input; delete groupMap;
 247                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 248                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 249                         return 0;
 250                 }
 251
 252                 //output error messages about any remaining user labels
 253                 set<string>::iterator it;
 254                 bool needToRun = false;
 255                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 256                         m->mothurOut("Your file does not include the label " + *it);
 257                         if (processedLabels.count(lastLabel) != 1) {
 258                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 259                                 needToRun = true;
 260                         }else {
 261                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 262                         }
 263
 264                 }
 265
 266                 if (m->control_pressed) {
 267                         delete input; delete groupMap;
 268                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 269                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 270                         return 0;
 271                 }
 272
 273                 //run last label if you need to
 274                 if (needToRun == true)  {
 275                         if (list != NULL) {     delete list;    }
 276                         list = input->getListVector(lastLabel); //get new list vector to process
 277
 278                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 279                         parse(list);
 280
 281                         delete list;
 282                 }
 283
 284                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 285                         (*(filehandles[it3->first])).close();
 286                         delete it3->second;
 287                 }
 288
 289
 290                 delete groupMap;
 291                 delete input;
 292
 293                 if (m->control_pressed) {
 294                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 295                         return 0;
 296                 }
 297
 298                 //set fasta file as new current fastafile
 299                 string current = "";
 300                 itTypes = outputTypes.find("list");
 301                 if (itTypes != outputTypes.end()) {
 302                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 303                 }
 304
 305                 m->mothurOutEndLine();
 306                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 307                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 308                 m->mothurOutEndLine();
 309
 310                 return 0;
 311         }
 312         catch(exception& e) {
 313                 m->errorOut(e, "ParseListCommand", "execute");
 314                 exit(1);
 315         }
 316 }
 317 /**********************************************************************************************************************/
 318 int ParseListCommand::parse(ListVector* thisList) {
 319         try {
 320
 321                 map<string, string> groupVector;
 322                 map<string, string>::iterator itGroup;
 323                 map<string, int> groupNumBins;
 324
 325                 //print label
 326                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 327                         groupNumBins[it3->first] = 0;
 328                         groupVector[it3->first] = "";
 329                 }
 330
 331
 332                 for (int i = 0; i < thisList->getNumBins(); i++) {
 333                         if (m->control_pressed) { return 0; }
 334
 335                         map<string, string> groupBins;
 336                         string bin = list->get(i);
 337
 338                         vector<string> names;
 339                         m->splitAtComma(bin, names);  //parses bin into individual sequence names
 340
 341                         //parse bin into list of sequences in each group
 342                         for (int j = 0; j < names.size(); j++) {
 343                                 string group = groupMap->getGroup(names[j]);
 344
 345                                 if (group == "not found") { m->mothurOut(names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); exit(1); }
 346
 347                                 itGroup = groupBins.find(group);
 348                                 if(itGroup == groupBins.end()) {
 349                                         groupBins[group] = names[j];  //add first name
 350                                         groupNumBins[group]++;
 351                                 }else{ //add another name
 352                                         groupBins[group] = groupBins[group] + "," + names[j];
 353                                 }
 354                         }
 355
 356                         //print parsed bin info to files
 357                         for (itGroup = groupBins.begin(); itGroup != groupBins.end(); itGroup++) {
 358                                 groupVector[itGroup->first] +=  itGroup->second + '\t';
 359                         }
 360
 361                 }
 362
 363                 //end list vector
 364                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 365                         (*(filehandles[it3->first])) << thisList->getLabel() << '\t' << groupNumBins[it3->first] << '\t' << groupVector[it3->first] << endl;  // label numBins  listvector for that group
 366                 }
 367
 368                 return 0;
 369
 370         }
 371         catch(exception& e) {
 372                 m->errorOut(e, "ParseListCommand", "parse");
 373                 exit(1);
 374         }
 375 }
 376
 377 /**********************************************************************************************************************/
 378
 379