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 string ParseListCommand::getOutputFileNameTag(string type, string inputName=""){
  50         try {
  51         string outputFileName = "";
  52                 map<string, vector<string> >::iterator it;
  53
  54         //is this a type this command creates
  55         it = outputTypes.find(type);
  56         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  57         else {
  58             if (type == "list") {  outputFileName =  "list"; }
  59             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  60         }
  61         return outputFileName;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "ParseListCommand", "getOutputFileNameTag");
  65                 exit(1);
  66         }
  67 }
  68
  69 //**********************************************************************************************************************
  70 ParseListCommand::ParseListCommand(){
  71         try {
  72                 abort = true; calledHelp = true;
  73                 setParameters();
  74                 vector<string> tempOutNames;
  75                 outputTypes["list"] = tempOutNames;
  76         }
  77         catch(exception& e) {
  78                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
  79                 exit(1);
  80         }
  81 }
  82 //**********************************************************************************************************************
  83 ParseListCommand::ParseListCommand(string option)  {
  84         try {
  85                 abort = false; calledHelp = false;
  86                 allLines = 1;
  87
  88                 //allow user to run help
  89                 if(option == "help") { help(); abort = true; calledHelp = true; }
  90                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  91
  92                 else {
  93                         vector<string> myArray = setParameters();
  94
  95                         OptionParser parser(option);
  96                         map<string, string> parameters = parser.getParameters();
  97
  98                         ValidParameters validParameter;
  99                         map<string, string>::iterator it;
 100
 101                         //check to make sure all parameters are valid for command
 102                         for (it = parameters.begin(); it != parameters.end(); it++) {
 103                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 104                         }
 105
 106                         //initialize outputTypes
 107                         vector<string> tempOutNames;
 108                         outputTypes["list"] = tempOutNames;
 109
 110                         //if the user changes the input directory command factory will send this info to us in the output parameter
 111                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 112                         if (inputDir == "not found"){   inputDir = "";          }
 113                         else {
 114                                 string path;
 115                                 it = parameters.find("list");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 121                                 }
 122
 123                                 it = parameters.find("group");
 124                                 //user has given a template file
 125                                 if(it != parameters.end()){
 126                                         path = m->hasPath(it->second);
 127                                         //if the user has not given a path then, add inputdir. else leave path alone.
 128                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 129                                 }
 130                         }
 131
 132
 133                         //if the user changes the output directory command factory will send this info to us in the output parameter
 134                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 135
 136                         //check for required parameters
 137                         listfile = validParameter.validFile(parameters, "list", true);
 138                         if (listfile == "not open") { abort = true; }
 139                         else if (listfile == "not found") {
 140                                 listfile = m->getListFile();
 141                                 if (listfile != "") {  m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 142                                 else {
 143                                         m->mothurOut("No valid current list file. You must provide a list file."); m->mothurOutEndLine();
 144                                         abort = true;
 145
 146                                 }
 147                         }else { m->setListFile(listfile); }
 148
 149                         groupfile = validParameter.validFile(parameters, "group", true);
 150                         if (groupfile == "not open") { abort = true; }
 151                         else if (groupfile == "not found") {
 152                                 groupfile = m->getListFile();
 153                                 if (groupfile != "") {
 154                                         m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine();
 155                                         groupMap = new GroupMap(groupfile);
 156
 157                                         int error = groupMap->readMap();
 158                                         if (error == 1) { abort = true; }
 159                                 }else { m->mothurOut("No valid current group file. You must provide a group file."); m->mothurOutEndLine();  abort = true; }
 160                         }else {
 161                                 m->setGroupFile(groupfile);
 162                                 groupMap = new GroupMap(groupfile);
 163
 164                                 int error = groupMap->readMap();
 165                                 if (error == 1) { abort = true; }
 166                         }
 167
 168                         //do you have all files needed
 169                         if ((listfile == "") || (groupfile == "")) { m->mothurOut("You must enter both a listfile and groupfile for the parse.list command. "); m->mothurOutEndLine(); abort = true;  }
 170
 171                         //check for optional parameter and set defaults
 172                         // ...at some point should added some additional type checking...
 173                         label = validParameter.validFile(parameters, "label", false);
 174                         if (label == "not found") { label = "";  allLines = 1; }
 175                         else {
 176                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 177                                 else { allLines = 1;  }
 178                         }
 179                 }
 180
 181         }
 182         catch(exception& e) {
 183                 m->errorOut(e, "ParseListCommand", "ParseListCommand");
 184                 exit(1);
 185         }
 186 }
 187 //**********************************************************************************************************************
 188 int ParseListCommand::execute(){
 189         try {
 190
 191                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 192
 193                 //set fileroot
 194                 string fileroot = outputDir + m->getRootName(m->getSimpleName(listfile));
 195
 196                 //fill filehandles with neccessary ofstreams
 197                 int i;
 198                 ofstream* temp;
 199                 vector<string> gGroups = groupMap->getNamesOfGroups();
 200                 for (i=0; i<gGroups.size(); i++) {
 201                         temp = new ofstream;
 202                         filehandles[gGroups[i]] = temp;
 203
 204                         string filename = fileroot +  gGroups[i] + "." + getOutputFileNameTag("list");
 205                         outputNames.push_back(filename); outputTypes["list"].push_back(filename);
 206                         m->openOutputFile(filename, *temp);
 207                 }
 208
 209                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 210                 set<string> processedLabels;
 211                 set<string> userLabels = labels;
 212
 213                 input = new InputData(listfile, "list");
 214                 list = input->getListVector();
 215                 string lastLabel = list->getLabel();
 216
 217                 if (m->control_pressed) {
 218                         delete input; delete list; delete groupMap;
 219                         vector<string> gGroups = groupMap->getNamesOfGroups();
 220                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 221                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 222                         return 0;
 223                 }
 224
 225                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 226
 227                         if (m->control_pressed) {
 228                                 delete input; delete list; delete groupMap;
 229                                 for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 230                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 231                                 return 0;
 232                         }
 233
 234                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 235
 236                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 237                                         parse(list);
 238
 239                                         processedLabels.insert(list->getLabel());
 240                                         userLabels.erase(list->getLabel());
 241                         }
 242
 243                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 244                                         string saveLabel = list->getLabel();
 245
 246                                         delete list;
 247                                         list = input->getListVector(lastLabel); //get new list vector to process
 248
 249                                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 250                                         parse(list);
 251
 252                                         processedLabels.insert(list->getLabel());
 253                                         userLabels.erase(list->getLabel());
 254
 255                                         //restore real lastlabel to save below
 256                                         list->setLabel(saveLabel);
 257                         }
 258
 259
 260                         lastLabel = list->getLabel();
 261
 262                         delete list;
 263                         list = input->getListVector(); //get new list vector to process
 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                 //output error messages about any remaining user labels
 274                 set<string>::iterator it;
 275                 bool needToRun = false;
 276                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 277                         m->mothurOut("Your file does not include the label " + *it);
 278                         if (processedLabels.count(lastLabel) != 1) {
 279                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 280                                 needToRun = true;
 281                         }else {
 282                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 283                         }
 284
 285                 }
 286
 287                 if (m->control_pressed) {
 288                         delete input; delete groupMap;
 289                         for (i=0; i<gGroups.size(); i++) {  (*(filehandles[gGroups[i]])).close();  delete filehandles[gGroups[i]]; }
 290                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 291                         return 0;
 292                 }
 293
 294                 //run last label if you need to
 295                 if (needToRun == true)  {
 296                         if (list != NULL) {     delete list;    }
 297                         list = input->getListVector(lastLabel); //get new list vector to process
 298
 299                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 300                         parse(list);
 301
 302                         delete list;
 303                 }
 304
 305                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 306                         (*(filehandles[it3->first])).close();
 307                         delete it3->second;
 308                 }
 309
 310
 311                 delete groupMap;
 312                 delete input;
 313
 314                 if (m->control_pressed) {
 315                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]); } outputTypes.clear();
 316                         return 0;
 317                 }
 318
 319                 //set fasta file as new current fastafile
 320                 string current = "";
 321                 itTypes = outputTypes.find("list");
 322                 if (itTypes != outputTypes.end()) {
 323                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 324                 }
 325
 326                 m->mothurOutEndLine();
 327                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 328                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 329                 m->mothurOutEndLine();
 330
 331                 return 0;
 332         }
 333         catch(exception& e) {
 334                 m->errorOut(e, "ParseListCommand", "execute");
 335                 exit(1);
 336         }
 337 }
 338 /**********************************************************************************************************************/
 339 int ParseListCommand::parse(ListVector* thisList) {
 340         try {
 341
 342                 map<string, string> groupVector;
 343                 map<string, string>::iterator itGroup;
 344                 map<string, int> groupNumBins;
 345
 346                 //print label
 347                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 348                         groupNumBins[it3->first] = 0;
 349                         groupVector[it3->first] = "";
 350                 }
 351
 352
 353                 for (int i = 0; i < thisList->getNumBins(); i++) {
 354                         if (m->control_pressed) { return 0; }
 355
 356                         map<string, string> groupBins;
 357                         string bin = list->get(i);
 358
 359                         vector<string> names;
 360                         m->splitAtComma(bin, names);  //parses bin into individual sequence names
 361
 362                         //parse bin into list of sequences in each group
 363                         for (int j = 0; j < names.size(); j++) {
 364                                 string group = groupMap->getGroup(names[j]);
 365
 366                                 if (group == "not found") { m->mothurOut(names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); exit(1); }
 367
 368                                 itGroup = groupBins.find(group);
 369                                 if(itGroup == groupBins.end()) {
 370                                         groupBins[group] = names[j];  //add first name
 371                                         groupNumBins[group]++;
 372                                 }else{ //add another name
 373                                         groupBins[group] = groupBins[group] + "," + names[j];
 374                                 }
 375                         }
 376
 377                         //print parsed bin info to files
 378                         for (itGroup = groupBins.begin(); itGroup != groupBins.end(); itGroup++) {
 379                                 groupVector[itGroup->first] +=  itGroup->second + '\t';
 380                         }
 381
 382                 }
 383
 384                 //end list vector
 385                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 386                         (*(filehandles[it3->first])) << thisList->getLabel() << '\t' << groupNumBins[it3->first] << '\t' << groupVector[it3->first] << endl;  // label numBins  listvector for that group
 387                 }
 388
 389                 return 0;
 390
 391         }
 392         catch(exception& e) {
 393                 m->errorOut(e, "ParseListCommand", "parse");
 394                 exit(1);
 395         }
 396 }
 397
 398 /**********************************************************************************************************************/
 399
 400