else {
//valid paramters for this command
- string Array[] = {"fasta","name", "group", "alignreport", "accnos", "list","outputdir","inputdir", "dups" };
+ string Array[] = {"fasta","name", "group", "alignreport", "accnos", "list","taxonomy","outputdir","inputdir", "dups" };
vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
OptionParser parser(option);
//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 = 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; }
+ }
}
if (listfile == "not open") { abort = true; }
else if (listfile == "not found") { listfile = ""; }
+ taxfile = validParameter.validFile(parameters, "taxonomy", true);
+ if (taxfile == "not open") { abort = true; }
+ else if (taxfile == "not found") { taxfile = ""; }
+
+
string usedDups = "true";
string temp = validParameter.validFile(parameters, "dups", false); if (temp == "not found") { temp = "false"; usedDups = ""; }
dups = isTrue(temp);
- if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "")) { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport or list."); m->mothurOutEndLine(); abort = true; }
-
- int okay = 2;
- if (outputDir != "") { okay++; }
- if (usedDups != "") { okay++; }
-
- if ((usedDups != "") && (namefile == "")) { m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine(); abort = true; }
+ if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "")) { m->mothurOut("You must provide at least one of the following: fasta, name, group, taxonomy, alignreport or list."); m->mothurOutEndLine(); abort = true; }
- if (parameters.size() > okay) { m->mothurOut("You may only enter one of the following: fasta, name, group, alignreport, or list."); m->mothurOutEndLine(); abort = true; }
+ if ((usedDups != "") && (namefile == "")) { m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine(); abort = true; }
}
}
void RemoveSeqsCommand::help(){
try {
- m->mothurOut("The remove.seqs command reads an .accnos file and one of the following file types: fasta, name, group, list or alignreport file.\n");
+ m->mothurOut("The remove.seqs command reads an .accnos file and at least one of the following file types: fasta, name, group, list, taxonomy or alignreport file.\n");
m->mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
- m->mothurOut("The remove.seqs command parameters are accnos, fasta, name, group, list, alignreport and dups. You must provide accnos and one of the file parameters.\n");
- m->mothurOut("The dups parameter allows you to remove the entire line from a name file if you remove any name from the line. default=false. If dups=true, then remove.seqs outputs a new .accnos file containing all the sequences removed. \n");
+ m->mothurOut("The remove.seqs command parameters are accnos, fasta, name, group, list, taxonomy, alignreport and dups. You must provide accnos and at least one of the file parameters.\n");
+ m->mothurOut("The dups parameter allows you to remove the entire line from a name file if you remove any name from the line. default=false. \n");
m->mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
m->mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
if (m->control_pressed) { return 0; }
//read through the correct file and output lines you want to keep
+ if (namefile != "") { readName(); }
if (fastafile != "") { readFasta(); }
- else if (namefile != "") { readName(); }
- else if (groupfile != "") { readGroup(); }
- else if (alignfile != "") { readAlign(); }
- else if (listfile != "") { readList(); }
+ if (groupfile != "") { readGroup(); }
+ if (alignfile != "") { readAlign(); }
+ if (listfile != "") { readList(); }
+ if (taxfile != "") { readTax(); }
if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
int RemoveSeqsCommand::readFasta(){
try {
if (outputDir == "") { outputDir += hasPath(fastafile); }
- string outputFileName = getRootName(fastafile) + "pick" + getExtension(fastafile);
+ string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "pick" + getExtension(fastafile);
ofstream out;
openOutputFile(outputFileName, out);
wroteSomething = true;
currSeq.printSequence(out);
- }else { names.erase(name); }
+ }
}
gobble(in);
}
in.close();
out.close();
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputNames.push_back(outputFileName);
return 0;
int RemoveSeqsCommand::readList(){
try {
if (outputDir == "") { outputDir += hasPath(listfile); }
- string outputFileName = getRootName(listfile) + "pick" + getExtension(listfile);
+ string outputFileName = outputDir + getRootName(getSimpleName(listfile)) + "pick" + getExtension(listfile);
ofstream out;
openOutputFile(outputFileName, out);
}
//get last name
- if (names.count(binnames) == 0) { newNames += binnames; }
+ if (names.count(binnames) == 0) { newNames += binnames + ","; }
//if there are names in this bin add to new list
- if (newNames != "") { newList.push_back(newNames); }
+ if (newNames != "") {
+ newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
+ newList.push_back(newNames);
+ }
}
//print new listvector
in.close();
out.close();
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
-
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputNames.push_back(outputFileName);
+
return 0;
}
int RemoveSeqsCommand::readName(){
try {
if (outputDir == "") { outputDir += hasPath(namefile); }
- string outputFileName = getRootName(namefile) + "pick" + getExtension(namefile);
- string outputFileName2 = getRootName(namefile) + "dups.accnos";
+ string outputFileName = outputDir + getRootName(getSimpleName(namefile)) + "pick" + getExtension(namefile);
- ofstream out2;
- if (dups) { openOutputFile(outputFileName2, out2); }
- bool wroteDups = false;
-
ofstream out;
openOutputFile(outputFileName, out);
bool wroteSomething = false;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); if (dups) { out2.close(); remove(outputFileName2.c_str()); } return 0; }
+ if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
in >> firstCol;
in >> secondCol;
}
}
- if ((dups) && (validSecond.size() != parsedNames.size())) {
- wroteDups = true;
- for (int i = 0; i < parsedNames.size(); i++) { out2 << parsedNames[i] << endl; }
+ if ((dups) && (validSecond.size() != parsedNames.size())) { //if dups is true and we want to get rid of anyone, get rid of everyone
+ for (int i = 0; i < parsedNames.size(); i++) { names.insert(parsedNames[i]); }
}else {
//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) {
}
in.close();
out.close();
-
- if (dups) { out2.close(); }
- if (wroteDups == false) {
- remove(outputFileName2.c_str());
- }else { outputNames.push_back(outputFileName2); }
-
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
-
+
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputNames.push_back(outputFileName);
+
return 0;
}
catch(exception& e) {
int RemoveSeqsCommand::readGroup(){
try {
if (outputDir == "") { outputDir += hasPath(groupfile); }
- string outputFileName = getRootName(groupfile) + "pick" + getExtension(groupfile);
+ string outputFileName = outputDir + getRootName(getSimpleName(groupfile)) + "pick" + getExtension(groupfile);
ofstream out;
openOutputFile(outputFileName, out);
if (names.count(name) == 0) {
wroteSomething = true;
out << name << '\t' << group << endl;
- }else { names.erase(name); }
+ }
gobble(in);
}
in.close();
out.close();
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputNames.push_back(outputFileName);
return 0;
}
exit(1);
}
}
+//**********************************************************************************************************************
+int RemoveSeqsCommand::readTax(){
+ try {
+ if (outputDir == "") { outputDir += hasPath(taxfile); }
+ string outputFileName = outputDir + getRootName(getSimpleName(taxfile)) + "pick" + getExtension(taxfile);
+ ofstream out;
+ openOutputFile(outputFileName, out);
+ ifstream in;
+ openInputFile(taxfile, in);
+ string name, tax;
+
+ bool wroteSomething = false;
+
+ while(!in.eof()){
+ if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); 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;
+ }
+
+ gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputNames.push_back(outputFileName);
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "RemoveSeqsCommand", "readTax");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
//alignreport file has a column header line then all other lines contain 16 columns. we just want the first column since that contains the name
int RemoveSeqsCommand::readAlign(){
try {
if (outputDir == "") { outputDir += hasPath(alignfile); }
- string outputFileName = getRootName(getRootName(alignfile)) + "pick.align.report";
+ string outputFileName = outputDir + getRootName(getSimpleName(alignfile)) + "pick.align.report";
ofstream out;
openOutputFile(outputFileName, out);
out << endl;
}else {//still read just don't do anything with it
- names.erase(name);
//read rest
for (int i = 0; i < 15; i++) {
in.close();
out.close();
- if (wroteSomething == false) {
- m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
- remove(outputFileName.c_str());
- }else { outputNames.push_back(outputFileName); }
+ if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine(); }
+ outputNames.push_back(outputFileName);
return 0;
projects / mothur.git / blobdiff