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","list",false,true,true); parameters.push_back(plist);
  16                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none","",false,true,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 string ParseListCommand::getOutputPattern(string type) {
  50     try {
  51         string pattern = "";
  52
  53         if (type == "list") {  pattern = "[filename],[group],list"; }
  54         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  55
  56         return pattern;
  57     }
  58     catch(exception& e) {
  59         m->errorOut(e, "ParseListCommand", "getOutputPattern");
  60         exit(1);
  61     }
  62 }
  63 //**********************************************************************************************************************
  64 ParseListCommand::ParseListCommand(){
  65         try {
  66                 abort = true; calledHelp = true;
  67                 setParameters();
  68                 vector<string> tempOutNames;
  69                 outputTypes["list"] = tempOutNames;
  70         }
  71         catch(exception& e) {
  72                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
  73                 exit(1);
  74         }
  75 }
  76 //**********************************************************************************************************************
  77 ParseListCommand::ParseListCommand(string option)  {
  78         try {
  79                 abort = false; calledHelp = false;
  80                 allLines = 1;
  81
  82                 //allow user to run help
  83                 if(option == "help") { help(); abort = true; calledHelp = true; }
  84                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  85
  86                 else {
  87                         vector<string> myArray = setParameters();
  88
  89                         OptionParser parser(option);
  90                         map<string, string> parameters = parser.getParameters();
  91
  92                         ValidParameters validParameter;
  93                         map<string, string>::iterator it;
  94
  95                         //check to make sure all parameters are valid for command
  96                         for (it = parameters.begin(); it != parameters.end(); it++) {
  97                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  98                         }
  99
 100                         //initialize outputTypes
 101                         vector<string> tempOutNames;
 102                         outputTypes["list"] = tempOutNames;
 103
 104                         //if the user changes the input directory command factory will send this info to us in the output parameter
 105                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 106                         if (inputDir == "not found"){   inputDir = "";          }
 107                         else {
 108                                 string path;
 109                                 it = parameters.find("list");
 110                                 //user has given a template file
 111                                 if(it != parameters.end()){
 112                                         path = m->hasPath(it->second);
 113                                         //if the user has not given a path then, add inputdir. else leave path alone.
 114                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 115                                 }
 116
 117                                 it = parameters.find("group");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 123                                 }
 124                         }
 125
 126
 127                         //if the user changes the output directory command factory will send this info to us in the output parameter
 128                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 129
 130                         //check for required parameters
 131                         listfile = validParameter.validFile(parameters, "list", true);
 132                         if (listfile == "not open") { abort = true; }
 133                         else if (listfile == "not found") {
 134                                 listfile = m->getListFile();
 135                                 if (listfile != "") {  m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 136                                 else {
 137                                         m->mothurOut("No valid current list file. You must provide a list file."); m->mothurOutEndLine();
 138                                         abort = true;
 139
 140                                 }
 141                         }else { m->setListFile(listfile); }
 142
 143                         groupfile = validParameter.validFile(parameters, "group", true);
 144                         if (groupfile == "not open") { abort = true; }
 145                         else if (groupfile == "not found") {
 146                                 groupfile = m->getListFile();
 147                                 if (groupfile != "") {
 148                                         m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine();
 149                                         groupMap = new GroupMap(groupfile);
 150
 151                                         int error = groupMap->readMap();
 152                                         if (error == 1) { abort = true; }
 153                                 }else { m->mothurOut("No valid current group file. You must provide a group file."); m->mothurOutEndLine();  abort = true; }
 154                         }else {
 155                                 m->setGroupFile(groupfile);
 156                                 groupMap = new GroupMap(groupfile);
 157
 158                                 int error = groupMap->readMap();
 159                                 if (error == 1) { abort = true; }
 160                         }
 161
 162                         //do you have all files needed
 163                         if ((listfile == "") || (groupfile == "")) { m->mothurOut("You must enter both a listfile and groupfile for the parse.list command. "); m->mothurOutEndLine(); abort = true;  }
 164
 165                         //check for optional parameter and set defaults
 166                         // ...at some point should added some additional type checking...
 167                         label = validParameter.validFile(parameters, "label", false);
 168                         if (label == "not found") { label = "";  allLines = 1; }
 169                         else {
 170                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 171                                 else { allLines = 1;  }
 172                         }
 173                 }
 174
 175         }
 176         catch(exception& e) {
 177                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
 178                 exit(1);
 179         }
 180 }
 181 //**********************************************************************************************************************
 182 int ParseListCommand::execute(){
 183         try {
 184
 185                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 186
 187                 //set fileroot
 188                 map<string, string> variables;
 189         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(listfile));
 190
 191                 //fill filehandles with neccessary ofstreams
 192                 int i;
 193                 ofstream* temp;
 194                 vector<string> gGroups = groupMap->getNamesOfGroups();
 195                 for (i=0; i<gGroups.size(); i++) {
 196                         temp = new ofstream;
 197                         filehandles[gGroups[i]] = temp;
 198
 199             variables["[group]"] = gGroups[i];
 200                         string filename = getOutputFileName("list",variables);
 201                         outputNames.push_back(filename); outputTypes["list"].push_back(filename);
 202                         m->openOutputFile(filename, *temp);
 203                 }
 204
 205                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 206                 set<string> processedLabels;
 207                 set<string> userLabels = labels;
 208
 209                 input = new InputData(listfile, "list");
 210                 list = input->getListVector();
 211                 string lastLabel = list->getLabel();
 212
 213                 if (m->control_pressed) {
 214                         delete input; delete list; delete groupMap;
 215                         vector<string> gGroups = groupMap->getNamesOfGroups();
 216                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 217                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 218                         return 0;
 219                 }
 220
 221                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 222
 223                         if (m->control_pressed) {
 224                                 delete input; delete list; delete groupMap;
 225                                 for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 226                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 227                                 return 0;
 228                         }
 229
 230                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 231
 232                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 233                                         parse(list);
 234
 235                                         processedLabels.insert(list->getLabel());
 236                                         userLabels.erase(list->getLabel());
 237                         }
 238
 239                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 240                                         string saveLabel = list->getLabel();
 241
 242                                         delete list;
 243                                         list = input->getListVector(lastLabel); //get new list vector to process
 244
 245                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 246                                         parse(list);
 247
 248                                         processedLabels.insert(list->getLabel());
 249                                         userLabels.erase(list->getLabel());
 250
 251                                         //restore real lastlabel to save below
 252                                         list->setLabel(saveLabel);
 253                         }
 254
 255
 256                         lastLabel = list->getLabel();
 257
 258                         delete list;
 259                         list = input->getListVector(); //get new list vector to process
 260                 }
 261
 262                 if (m->control_pressed) {
 263                         delete input; delete groupMap;
 264                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 265                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 266                         return 0;
 267                 }
 268
 269                 //output error messages about any remaining user labels
 270                 set<string>::iterator it;
 271                 bool needToRun = false;
 272                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 273                         m->mothurOut("Your file does not include the label " + *it);
 274                         if (processedLabels.count(lastLabel) != 1) {
 275                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 276                                 needToRun = true;
 277                         }else {
 278                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 279                         }
 280
 281                 }
 282
 283                 if (m->control_pressed) {
 284                         delete input; delete groupMap;
 285                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 286                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 287                         return 0;
 288                 }
 289
 290                 //run last label if you need to
 291                 if (needToRun == true)  {
 292                         if (list != NULL) {     delete list;    }
 293                         list = input->getListVector(lastLabel); //get new list vector to process
 294
 295                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 296                         parse(list);
 297
 298                         delete list;
 299                 }
 300
 301                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 302                         (*(filehandles[it3->first])).close();
 303                         delete it3->second;
 304                 }
 305
 306
 307                 delete groupMap;
 308                 delete input;
 309
 310                 if (m->control_pressed) {
 311                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 312                         return 0;
 313                 }
 314
 315                 //set fasta file as new current fastafile
 316                 string current = "";
 317                 itTypes = outputTypes.find("list");
 318                 if (itTypes != outputTypes.end()) {
 319                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 320                 }
 321
 322                 m->mothurOutEndLine();
 323                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 324                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 325                 m->mothurOutEndLine();
 326
 327                 return 0;
 328         }
 329         catch(exception& e) {
 330                 m->errorOut(e, "ParseListCommand", "execute");
 331                 exit(1);
 332         }
 333 }
 334 /**********************************************************************************************************************/
 335 int ParseListCommand::parse(ListVector* thisList) {
 336         try {
 337
 338                 map<string, string> groupVector;
 339                 map<string, string>::iterator itGroup;
 340                 map<string, int> groupNumBins;
 341
 342                 //print label
 343                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 344                         groupNumBins[it3->first] = 0;
 345                         groupVector[it3->first] = "";
 346                 }
 347
 348
 349                 for (int i = 0; i < thisList->getNumBins(); i++) {
 350                         if (m->control_pressed) { return 0; }
 351
 352                         map<string, string> groupBins;
 353                         string bin = list->get(i);
 354
 355                         vector<string> names;
 356                         m->splitAtComma(bin, names);  //parses bin into individual sequence names
 357
 358                         //parse bin into list of sequences in each group
 359                         for (int j = 0; j < names.size(); j++) {
 360                                 string group = groupMap->getGroup(names[j]);
 361
 362                                 if (group == "not found") { m->mothurOut(names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); exit(1); }
 363
 364                                 itGroup = groupBins.find(group);
 365                                 if(itGroup == groupBins.end()) {
 366                                         groupBins[group] = names[j];  //add first name
 367                                         groupNumBins[group]++;
 368                                 }else{ //add another name
 369                                         groupBins[group] = groupBins[group] + "," + names[j];
 370                                 }
 371                         }
 372
 373                         //print parsed bin info to files
 374                         for (itGroup = groupBins.begin(); itGroup != groupBins.end(); itGroup++) {
 375                                 groupVector[itGroup->first] +=  itGroup->second + '\t';
 376                         }
 377
 378                 }
 379
 380                 //end list vector
 381                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 382                         (*(filehandles[it3->first])) << thisList->getLabel() << '\t' << groupNumBins[it3->first] << '\t' << groupVector[it3->first] << endl;  // label numBins  listvector for that group
 383                 }
 384
 385                 return 0;
 386
 387         }
 388         catch(exception& e) {
 389                 m->errorOut(e, "ParseListCommand", "parse");
 390                 exit(1);
 391         }
 392 }
 393
 394 /**********************************************************************************************************************/
 395
 396