--- /dev/null
+/*
+ * getgroupscommand.cpp
+ * Mothur
+ *
+ * Created by westcott on 11/10/10.
+ * Copyright 2010 Schloss Lab. All rights reserved.
+ *
+ */
+
+#include "getgroupscommand.h"
+#include "sequence.hpp"
+#include "listvector.hpp"
+#include "sharedutilities.h"
+#include "inputdata.h"
+
+//**********************************************************************************************************************
+vector<string> GetGroupsCommand::setParameters(){
+ try {
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(pfasta);
+ CommandParameter pshared("shared", "InputTypes", "", "", "none", "sharedGroup", "none",false,false); parameters.push_back(pshared);
+ CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
+ CommandParameter pgroup("group", "InputTypes", "", "", "none", "sharedGroup", "FNGLT",false,false); parameters.push_back(pgroup);
+ CommandParameter pdesign("design", "InputTypes", "", "", "none", "sharedGroup", "FNGLT",false,false); parameters.push_back(pdesign);
+ CommandParameter plist("list", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(plist);
+ CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "none", "FNGLT",false,false); parameters.push_back(ptaxonomy);
+ CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
+ CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "setParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+string GetGroupsCommand::getHelpString(){
+ try {
+ string helpString = "";
+ helpString += "The get.groups command selects sequences from a specfic group or set of groups from the following file types: fasta, name, group, list, taxonomy, design or shared file.\n";
+ helpString += "It outputs a file containing the sequences in the those specified groups, or a sharedfile containing only those groups.\n";
+ helpString += "The get.groups command parameters are accnos, fasta, name, group, list, taxonomy, shared, design and groups. The group parameter is required, unless you have a current group file, or are using a shared file.\n";
+ helpString += "You must also provide an accnos containing the list of groups to get or set the groups parameter to the groups you wish to select.\n";
+ helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like. You can separate group names with dashes.\n";
+ helpString += "The get.groups command should be in the following format: get.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n";
+ helpString += "Example get.groups(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n";
+ helpString += "or get.groups(groups=pasture, fasta=amazon.fasta, group=amazon.groups).\n";
+ helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
+ return helpString;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "getHelpString");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+GetGroupsCommand::GetGroupsCommand(){
+ try {
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["taxonomy"] = tempOutNames;
+ outputTypes["name"] = tempOutNames;
+ outputTypes["group"] = tempOutNames;
+ outputTypes["list"] = tempOutNames;
+ outputTypes["shared"] = tempOutNames;
+ outputTypes["design"] = tempOutNames;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+GetGroupsCommand::GetGroupsCommand(string option) {
+ try {
+ abort = false; calledHelp = false;
+
+ //allow user to run help
+ if(option == "help") { help(); abort = true; calledHelp = true; }
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
+
+ else {
+ vector<string> myArray = setParameters();
+
+ OptionParser parser(option);
+ map<string,string> parameters = parser.getParameters();
+
+ ValidParameters validParameter;
+ map<string,string>::iterator it;
+
+ //check to make sure all parameters are valid for command
+ for (it = parameters.begin(); it != parameters.end(); it++) {
+ if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
+ }
+
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["taxonomy"] = tempOutNames;
+ outputTypes["name"] = tempOutNames;
+ outputTypes["group"] = tempOutNames;
+ outputTypes["list"] = tempOutNames;
+ outputTypes["shared"] = tempOutNames;
+ outputTypes["design"] = tempOutNames;
+
+
+ //if the user changes the output directory command factory will send this info to us in the output parameter
+ outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ outputDir = ""; }
+
+ //if the user changes the input directory command factory will send this info to us in the output parameter
+ string inputDir = validParameter.validFile(parameters, "inputdir", false);
+ if (inputDir == "not found"){ inputDir = ""; }
+ else {
+ string path;
+ it = parameters.find("fasta");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["fasta"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("accnos");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["accnos"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("list");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["list"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("name");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["name"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("group");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["group"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("taxonomy");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["taxonomy"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("shared");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["shared"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("design");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["design"] = inputDir + it->second; }
+ }
+ }
+
+
+ //check for required parameters
+ accnosfile = validParameter.validFile(parameters, "accnos", true);
+ if (accnosfile == "not open") { abort = true; }
+ else if (accnosfile == "not found") { accnosfile = ""; }
+ else { m->setAccnosFile(accnosfile); }
+
+ fastafile = validParameter.validFile(parameters, "fasta", true);
+ if (fastafile == "not open") { fastafile = ""; abort = true; }
+ else if (fastafile == "not found") { fastafile = ""; }
+ else { m->setFastaFile(fastafile); }
+
+ namefile = validParameter.validFile(parameters, "name", true);
+ if (namefile == "not open") { namefile = ""; abort = true; }
+ else if (namefile == "not found") { namefile = ""; }
+ else { m->setNameFile(namefile); }
+
+ groupfile = validParameter.validFile(parameters, "group", true);
+ if (groupfile == "not open") { groupfile = ""; abort = true; }
+ else if (groupfile == "not found") { groupfile = ""; }
+ else { m->setGroupFile(groupfile); }
+
+ listfile = validParameter.validFile(parameters, "list", true);
+ if (listfile == "not open") { abort = true; }
+ else if (listfile == "not found") { listfile = ""; }
+ else { m->setListFile(listfile); }
+
+ taxfile = validParameter.validFile(parameters, "taxonomy", true);
+ if (taxfile == "not open") { taxfile = ""; abort = true; }
+ else if (taxfile == "not found") { taxfile = ""; }
+ else { m->setTaxonomyFile(taxfile); }
+
+ groups = validParameter.validFile(parameters, "groups", false);
+ if (groups == "not found") { groups = ""; }
+ else {
+ m->splitAtDash(groups, Groups);
+ m->setGroups(Groups);
+ }
+
+ sharedfile = validParameter.validFile(parameters, "shared", true);
+ if (sharedfile == "not open") { sharedfile = ""; abort = true; }
+ else if (sharedfile == "not found") { sharedfile = ""; }
+ else { m->setSharedFile(sharedfile); }
+
+ groupfile = validParameter.validFile(parameters, "group", true);
+ if (groupfile == "not open") { groupfile = ""; abort = true; }
+ else if (groupfile == "not found") { groupfile = ""; }
+ else { m->setGroupFile(groupfile); }
+
+ designfile = validParameter.validFile(parameters, "design", true);
+ if (designfile == "not open") { designfile = ""; abort = true; }
+ else if (designfile == "not found") { designfile = ""; }
+ else { m->setDesignFile(designfile); }
+
+ if ((sharedfile == "") && (groupfile == "") && (designfile == "")) {
+ //is there are current file available for any of these?
+ if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
+ //give priority to group, then shared
+ groupfile = m->getGroupFile();
+ if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
+ else {
+ sharedfile = m->getSharedFile();
+ if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
+ else {
+ m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
+ }
+ }
+ }else {
+ //give priority to shared, then group
+ sharedfile = m->getSharedFile();
+ if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
+ else {
+ groupfile = m->getGroupFile();
+ if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
+ else {
+ designfile = m->getDesignFile();
+ if (designfile != "") { m->mothurOut("Using " + designfile + " as input file for the design parameter."); m->mothurOutEndLine(); }
+ else {
+ m->mothurOut("You have no current groupfile or sharedfile or designfile and one is required."); m->mothurOutEndLine(); abort = true;
+ }
+ }
+ }
+ }
+ }
+
+ if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
+
+ if ((fastafile == "") && (namefile == "") && (groupfile == "") && (designfile == "") && (sharedfile == "") && (listfile == "") && (taxfile == "")) { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group, shared, design or list."); m->mothurOutEndLine(); abort = true; }
+ if ((groupfile == "") && ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != ""))) { m->mothurOut("If using a fasta, name, taxonomy, group or list, then you must provide a group file."); m->mothurOutEndLine(); abort = true; }
+
+ if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
+ vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
+ parser.getNameFile(files);
+ }
+ }
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+
+int GetGroupsCommand::execute(){
+ try {
+
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
+
+ //get groups you want to remove
+ if (accnosfile != "") { readAccnos(); }
+
+ if (groupfile != "") {
+ groupMap = new GroupMap(groupfile);
+ groupMap->readMap();
+
+ //make sure groups are valid
+ //takes care of user setting groupNames that are invalid or setting groups=all
+ SharedUtil* util = new SharedUtil();
+ vector<string> gNamesOfGroups = groupMap->getNamesOfGroups();
+ util->setGroups(Groups, gNamesOfGroups);
+ groupMap->setNamesOfGroups(gNamesOfGroups);
+ delete util;
+
+ //fill names with names of sequences that are from the groups we want to remove
+ fillNames();
+
+ delete groupMap;
+ }
+
+ if (m->control_pressed) { return 0; }
+
+ //read through the correct file and output lines you want to keep
+ if (namefile != "") { readName(); }
+ if (fastafile != "") { readFasta(); }
+ if (groupfile != "") { readGroup(); }
+ if (listfile != "") { readList(); }
+ if (taxfile != "") { readTax(); }
+ if (sharedfile != "") { readShared(); }
+ if (designfile != "") { readDesign(); }
+
+ if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
+
+
+ if (outputNames.size() != 0) {
+ m->mothurOutEndLine();
+ m->mothurOut("Output File names: "); m->mothurOutEndLine();
+ for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
+ m->mothurOutEndLine();
+
+ //set fasta file as new current fastafile
+ string current = "";
+ itTypes = outputTypes.find("fasta");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
+ }
+
+ itTypes = outputTypes.find("name");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
+ }
+
+ itTypes = outputTypes.find("group");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
+ }
+
+ itTypes = outputTypes.find("list");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
+ }
+
+ itTypes = outputTypes.find("taxonomy");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
+ }
+
+ itTypes = outputTypes.find("shared");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
+ }
+
+ itTypes = outputTypes.find("design");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setDesignFile(current); }
+ }
+ }
+
+ return 0;
+ }
+
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "execute");
+ exit(1);
+ }
+}
+
+//**********************************************************************************************************************
+int GetGroupsCommand::readFasta(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(fastafile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
+ ifstream in;
+ m->openInputFile(fastafile, in);
+ string name;
+
+ bool wroteSomething = false;
+ int selectedCount = 0;
+
+ while(!in.eof()){
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
+
+ Sequence currSeq(in);
+ name = currSeq.getName();
+
+ if (name != "") {
+ //if this name is in the accnos file
+ if (names.count(name) != 0) {
+ wroteSomething = true;
+
+ currSeq.printSequence(out);
+ selectedCount++;
+ }else{
+ //if you are not in the accnos file check if you are a name that needs to be changed
+ map<string, string>::iterator it = uniqueToRedundant.find(name);
+ if (it != uniqueToRedundant.end()) {
+ wroteSomething = true;
+ currSeq.setName(it->second);
+ currSeq.printSequence(out);
+ selectedCount++;
+ }
+ }
+ }
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) { m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine(); }
+ outputTypes["fasta"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
+
+
+ return 0;
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "readFasta");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+int GetGroupsCommand::readShared(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(sharedfile); }
+
+ InputData input(sharedfile, "sharedfile");
+ vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
+
+ bool wroteSomething = false;
+
+ while(lookup[0] != NULL) {
+
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + ".pick" + m->getExtension(sharedfile);
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+ outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ if (m->control_pressed) { out.close(); m->mothurRemove(outputFileName); for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
+
+ lookup[0]->printHeaders(out);
+
+ for (int i = 0; i < lookup.size(); i++) {
+ out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
+ lookup[i]->print(out);
+ wroteSomething = true;
+
+ }
+
+ //get next line to process
+ //prevent memory leak
+ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
+ lookup = input.getSharedRAbundVectors();
+
+ out.close();
+ }
+
+ if (wroteSomething == false) { m->mothurOut("Your file contains only the groups you wish to remove."); m->mothurOutEndLine(); }
+
+ string groupsString = "";
+ for (int i = 0; i < Groups.size()-1; i++) { groupsString += Groups[i] + ", "; }
+ groupsString += Groups[Groups.size()-1];
+
+ m->mothurOut("Selected groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
+
+ return 0;
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "readShared");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+int GetGroupsCommand::readList(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(listfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" + m->getExtension(listfile);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
+ ifstream in;
+ m->openInputFile(listfile, in);
+
+ bool wroteSomething = false;
+ int selectedCount = 0;
+
+ while(!in.eof()){
+
+ selectedCount = 0;
+
+ //read in list vector
+ ListVector list(in);
+
+ //make a new list vector
+ ListVector newList;
+ newList.setLabel(list.getLabel());
+
+ //for each bin
+ for (int i = 0; i < list.getNumBins(); i++) {
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
+
+ //parse out names that are in accnos file
+ string binnames = list.get(i);
+
+ string newNames = "";
+ while (binnames.find_first_of(',') != -1) {
+ string name = binnames.substr(0,binnames.find_first_of(','));
+ binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
+
+ //if that name is in the .accnos file, add it
+ if (names.count(name) != 0) { newNames += name + ","; selectedCount++; }
+ else{
+ //if you are not in the accnos file check if you are a name that needs to be changed
+ map<string, string>::iterator it = uniqueToRedundant.find(name);
+ if (it != uniqueToRedundant.end()) {
+ newNames += it->second + ",";
+ selectedCount++;
+ }
+ }
+ }
+
+ //get last name
+ if (names.count(binnames) != 0) { newNames += binnames + ","; selectedCount++; }
+ else{
+ //if you are not in the accnos file check if you are a name that needs to be changed
+ map<string, string>::iterator it = uniqueToRedundant.find(binnames);
+ if (it != uniqueToRedundant.end()) {
+ newNames += it->second + ",";
+ selectedCount++;
+ }
+ }
+
+ //if there are names in this bin add to new list
+ if (newNames != "") {
+ newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
+ newList.push_back(newNames);
+ }
+ }
+
+ //print new listvector
+ if (newList.getNumBins() != 0) {
+ wroteSomething = true;
+ newList.print(out);
+ }
+
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) { m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine(); }
+ outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
+
+ return 0;
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "readList");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+int GetGroupsCommand::readName(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(namefile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
+ ifstream in;
+ m->openInputFile(namefile, in);
+ string name, firstCol, secondCol;
+
+ bool wroteSomething = false;
+ int selectedCount = 0;
+
+ while(!in.eof()){
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
+
+ in >> firstCol; m->gobble(in);
+ in >> secondCol;
+
+ vector<string> parsedNames;
+ m->splitAtComma(secondCol, parsedNames);
+
+ vector<string> validSecond; validSecond.clear();
+ for (int i = 0; i < parsedNames.size(); i++) {
+ if (names.count(parsedNames[i]) != 0) {
+ validSecond.push_back(parsedNames[i]);
+ }
+ }
+
+ selectedCount += validSecond.size();
+
+ //if the name in the first column is in the set then print it and any other names in second column also in set
+ if (names.count(firstCol) != 0) {
+
+ wroteSomething = true;
+
+ out << firstCol << '\t';
+
+ //you know you have at least one valid second since first column is valid
+ for (int i = 0; i < validSecond.size()-1; i++) { out << validSecond[i] << ','; }
+ out << validSecond[validSecond.size()-1] << endl;
+
+ //make first name in set you come to first column and then add the remaining names to second column
+ }else {
+
+ //you want part of this row
+ if (validSecond.size() != 0) {
+
+ wroteSomething = true;
+
+ out << validSecond[0] << '\t';
+
+ //you know you have at least one valid second since first column is valid
+ for (int i = 0; i < validSecond.size()-1; i++) { out << validSecond[i] << ','; }
+ out << validSecond[validSecond.size()-1] << endl;
+ uniqueToRedundant[firstCol] = validSecond[0];
+ }
+ }
+
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) { m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine(); }
+ outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "readName");
+ exit(1);
+ }
+}
+
+//**********************************************************************************************************************
+int GetGroupsCommand::readGroup(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(groupfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
+ ifstream in;
+ m->openInputFile(groupfile, in);
+ string name, group;
+
+ bool wroteSomething = false;
+ int selectedCount = 0;
+
+ while(!in.eof()){
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
+
+ in >> name; //read from first column
+ in >> group; //read from second column
+
+ //if this name is in the accnos file
+ if (names.count(name) != 0) {
+ wroteSomething = true;
+ out << name << '\t' << group << endl;
+ selectedCount++;
+ }
+
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) { m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine(); }
+ outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "readGroup");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+int GetGroupsCommand::readDesign(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(designfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(designfile)) + "pick" + m->getExtension(designfile);
+
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
+ ifstream in;
+ m->openInputFile(designfile, in);
+ string name, group;
+
+ bool wroteSomething = false;
+ int selectedCount = 0;
+
+ while(!in.eof()){
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
+
+ in >> name; //read from first column
+ in >> group; //read from second column
+
+ //if this name is in the accnos file
+ if (m->inUsersGroups(name, Groups)) {
+ wroteSomething = true;
+ out << name << '\t' << group << endl;
+ selectedCount++;
+ }
+
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) { m->mothurOut("Your file does NOT contain groups from the groups you wish to get."); m->mothurOutEndLine(); }
+ outputTypes["design"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ m->mothurOut("Selected " + toString(selectedCount) + " groups from your design file."); m->mothurOutEndLine();
+
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "readDesign");
+ exit(1);
+ }
+}
+
+
+//**********************************************************************************************************************
+int GetGroupsCommand::readTax(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(taxfile); }
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+
+ ifstream in;
+ m->openInputFile(taxfile, in);
+ string name, tax;
+
+ bool wroteSomething = false;
+
+ while(!in.eof()){
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
+
+ in >> name; //read from first column
+ in >> tax; //read from second column
+
+ //if this name is in the accnos file
+ if (names.count(name) != 0) {
+ wroteSomething = true;
+ out << name << '\t' << tax << endl;
+ }else{
+ //if you are not in the accnos file check if you are a name that needs to be changed
+ map<string, string>::iterator it = uniqueToRedundant.find(name);
+ if (it != uniqueToRedundant.end()) {
+ wroteSomething = true;
+ out << it->second << '\t' << tax << endl;
+ }
+ }
+
+ m->gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) { m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine(); }
+ outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "readTax");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+void GetGroupsCommand::readAccnos(){
+ try {
+ Groups.clear();
+
+ ifstream in;
+ m->openInputFile(accnosfile, in);
+ string name;
+
+ while(!in.eof()){
+ in >> name;
+
+ Groups.push_back(name);
+
+ m->gobble(in);
+ }
+ in.close();
+
+ m->setGroups(Groups);
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "readAccnos");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+int GetGroupsCommand::fillNames(){
+ try {
+ vector<string> seqs = groupMap->getNamesSeqs();
+
+ for (int i = 0; i < seqs.size(); i++) {
+
+ if (m->control_pressed) { return 0; }
+
+ string group = groupMap->getGroup(seqs[i]);
+
+ if (m->inUsersGroups(group, Groups)) {
+ names.insert(seqs[i]);
+ }
+ }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "fillNames");
+ exit(1);
+ }
+}
+
+//**********************************************************************************************************************
+
+
projects / mothur.git / blobdiff