Adds error message for missing headers

Prevents addition of dummy group or removal of valid group #779 #769
mothur · Jul 23, 2021 · c55e885 · c55e885
1 parent f1b4624
commit c55e885
Show file tree

Hide file tree

Showing 13 changed files with 32 additions and 69 deletions.
diff --git a/source/commands/amovacommand.cpp b/source/commands/amovacommand.cpp
@@ -154,6 +154,8 @@ int AmovaCommand::execute(){
 
 		//read design file
 		designMap = new DesignMap(designFileName);
+
+        if (m->getControl_pressed()) { delete designMap; return 0; }
 
 		if (outputdir == "") { outputdir = util.hasPath(phylipFileName); }
 

diff --git a/source/commands/anosimcommand.cpp b/source/commands/anosimcommand.cpp
@@ -127,7 +127,7 @@ int AnosimCommand::execute(){
 		if (abort) { if (calledHelp) { return 0; }  return 2;	}
 
 		//read design file
-		designMap = new DesignMap(designFileName);
+        designMap = new DesignMap(designFileName); if (m->getControl_pressed()) { delete designMap; return 0; }
 
 		if (outputdir == "") { outputdir = util.hasPath(phylipFileName); }
 

diff --git a/source/commands/classifysvmsharedcommand.cpp b/source/commands/classifysvmsharedcommand.cpp
@@ -367,7 +367,7 @@ void ClassifySvmSharedCommand::readSharedAndDesignFiles(const string& sharedFile
     Groups = lookup->getNamesGroups();
 
     DesignMap designMap;
-    designMap.read(designFilePath);
+    designMap.read(designFilePath); if (m->getControl_pressed()) { return ; }
 
     while ( lookup != NULL ) {
         vector<SharedRAbundVector*> data = lookup->getSharedRAbundVectors();

diff --git a/source/commands/getgroupscommand.cpp b/source/commands/getgroupscommand.cpp
@@ -720,7 +720,7 @@ void GetGroupsCommand::readDesign(){
         variables["[extension]"] = util.getExtension(designfile);
 		string outputFileName = getOutputFileName("design", variables);
 
-        DesignMap designMap(designfile);
+        DesignMap designMap(designfile); if (m->getControl_pressed()) { return ; }
 
         bool wroteSomething = false;
 
@@ -1025,7 +1025,7 @@ void GetGroupsCommand::fillNames(){
 //**********************************************************************************************************************
 void GetGroupsCommand::fillGroupsFromDesign(){
     try {
-        DesignMap designMap(designfile);
+        DesignMap designMap(designfile); if (m->getControl_pressed()) { return ; }
         Groups = designMap.getNamesGroups(Sets);
     }
     catch(exception& e) {

diff --git a/source/commands/homovacommand.cpp b/source/commands/homovacommand.cpp
@@ -138,7 +138,7 @@ int HomovaCommand::execute(){
 		if (abort) { if (calledHelp) { return 0; }  return 2;	}
 
 		//read design file
-		DesignMap* designMap = new DesignMap(designFileName);
+        DesignMap* designMap = new DesignMap(designFileName); if (m->getControl_pressed()) { delete designMap; return 0; }
 
 		//read in distance matrix and square it
 		ReadPhylipVector readMatrix(phylipFileName);

diff --git a/source/commands/indicatorcommand.cpp b/source/commands/indicatorcommand.cpp
@@ -180,7 +180,7 @@ int IndicatorCommand::execute(){
 		//read designfile if given and set up groups for read of sharedfiles
         vector<string> allGroups;
 		if (designfile != "") {
-			designMap = new DesignMap(designfile);
+            designMap = new DesignMap(designfile); if (m->getControl_pressed()) { delete designMap; return 0; }
 
             if (Groups.size() == 0) { Groups = designMap->getCategory(); }
             allGroups = designMap->getCategory();

diff --git a/source/commands/kruskalwalliscommand.cpp b/source/commands/kruskalwalliscommand.cpp
@@ -127,7 +127,7 @@ int KruskalWallisCommand::execute(){
 
 		if (abort) { if (calledHelp) { return 0; }  return 2;	}
 
-        DesignMap designMap(designfile);
+        DesignMap designMap(designfile); if (m->getControl_pressed()) { return 0; }
 
         //if user did not select class use first column
         if (mclass == "") {  mclass = designMap.getDefaultClass(); m->mothurOut("\nYou did not provide a class, using " + mclass +".\n\n"); }

diff --git a/source/commands/lefsecommand.cpp b/source/commands/lefsecommand.cpp
@@ -230,7 +230,7 @@ int LefseCommand::execute(){
         m->setRandomSeed(1982);
 		if (abort) { if (calledHelp) { return 0; }  return 2;	}
 
-        DesignMap designMap(designfile);
+        DesignMap designMap(designfile); if (m->getControl_pressed()) { return 0; }
 
         //if user did not select class use first column
         if (mclass == "") {  mclass = designMap.getDefaultClass(); m->mothurOut("\nYou did not provide a class, using " + mclass +".\n\n"); if (subclass == "") { subclass = mclass; } }

diff --git a/source/commands/makelefsecommand.cpp b/source/commands/makelefsecommand.cpp
@@ -194,7 +194,7 @@ int MakeLefseCommand::runRelabund(map<int, consTax2>& consTax, SharedRAbundFloat
         DesignMap* designMap = NULL;
         vector<string> namesOfGroups = lookup->getNamesGroups();
         if (designfile != "") {
-            designMap = new DesignMap(designfile);
+            designMap = new DesignMap(designfile);  if (m->getControl_pressed()) { out.close(); delete designMap; return 0; }
             vector<string> categories = designMap->getNamesOfCategories();
 
             if (categories.size() > 3) {  m->mothurOut("\n[NOTE]: LEfSe input files allow for a class, subclass and subject.  More than 3 categories can cause formatting errors.\n\n"); }

diff --git a/source/commands/mergegroupscommand.cpp b/source/commands/mergegroupscommand.cpp
@@ -195,7 +195,7 @@ int MergeGroupsCommand::execute(){
 
 		if (abort) { if (calledHelp) { return 0; }  return 2;	}
 
-		designMap = new DesignMap(designfile);
+        designMap = new DesignMap(designfile);  if (m->getControl_pressed()) { delete designMap; return 0; }
 
         if (method != "sum") {
             string defaultClass = designMap->getDefaultClass();

diff --git a/source/commands/metastatscommand.cpp b/source/commands/metastatscommand.cpp
@@ -183,7 +183,7 @@ int MetaStatsCommand::execute(){
 
         if (abort) { if (calledHelp) { return 0; }  return 2;	}
 
-		DesignMap* designMap = new DesignMap(designfile);
+        DesignMap* designMap = new DesignMap(designfile);  if (m->getControl_pressed()) { delete designMap; return 0; }
 
 		InputData input(inputfile, format, Groups);
 		set<string> processedLabels;

diff --git a/source/commands/removegroupscommand.cpp b/source/commands/removegroupscommand.cpp
@@ -769,7 +769,7 @@ void RemoveGroupsCommand::readDesign(){
         variables["[extension]"] = util.getExtension(designfile);
 		string outputFileName = getOutputFileName("design", variables);
 
-        DesignMap designMap(designfile);
+        DesignMap designMap(designfile);  if (m->getControl_pressed()) {  return ; }
 
         vector<string> groupsToKeep;
         vector<string> allGroups = designMap.getNamesGroups();
@@ -1091,7 +1091,7 @@ void RemoveGroupsCommand::fillNames(){
 //**********************************************************************************************************************
 void RemoveGroupsCommand::fillGroupsFromDesign(){
     try {
-        DesignMap designMap(designfile);
+        DesignMap designMap(designfile);  if (m->getControl_pressed()) {  return ; }
         Groups = designMap.getNamesGroups(Sets);
     }
     catch(exception& e) {

diff --git a/source/datastructures/designmap.cpp b/source/datastructures/designmap.cpp
@@ -23,36 +23,33 @@ DesignMap::DesignMap(string file) {
 /************************************************************/
 int DesignMap::read(string file) {
     try {
-        ifstream in;
-        util.openInputFile(file, in);
+        namesOfCategories.clear();
+        indexCategoryMap.clear();
+        indexGroupNameMap.clear();
+        designMap.clear();
+
+        ifstream in; util.openInputFile(file, in);
 
         string temp = "";
-        in >> temp; util.gobble(in);
+        string headers = util.getline(in); util.gobble(in);
+        vector<string> tempColumnHeaders = util.splitWhiteSpace(headers);
+        if (tempColumnHeaders.size() != 0) { temp = tempColumnHeaders[0]; }
+        else { m->setControl_pressed(true); return 0; }
 
         vector<string> columnHeaders;
-        vector<string> tempColumnHeaders;
         if ((temp == "group") || (temp == "Group")) {
-            string headers = util.getline(in); util.gobble(in);
-            columnHeaders = util.splitWhiteSpace(headers);
-            columnHeaders.insert(columnHeaders.begin(), "group");
-        }else {
-            string headers = util.getline(in); util.gobble(in);
-            tempColumnHeaders = util.splitWhiteSpace(headers);
-            int num = tempColumnHeaders.size();
             columnHeaders.push_back("group");
-            for (int i = 0; i < num; i++) { columnHeaders.push_back("value" + toString(i)); }
+            for (int i = 1; i < tempColumnHeaders.size(); i++) { columnHeaders.push_back(tempColumnHeaders[i]); }
+        }else {
+            m->mothurOut("\n[ERROR]: Expected 'group' and found '" + temp + "'. Mothur expects the design file to have column headers. The first column header should be 'group'.\n\nSomething like: group\ttreatment\tmetadata.\n\nWithout the headers, mothur is unable to determine if the first row is group information or a header of the wrong name. This can result in a dummy group or missing group which will results in errors. Quitting, please correct.\n\n");
+            m->setControl_pressed(true); in.close(); return 0;
         }
 
-        namesOfCategories.clear();
-        indexCategoryMap.clear();
-        indexGroupNameMap.clear();
-        designMap.clear();
+
         map<int, string> originalGroupIndexes;
         for (int i = 1; i < columnHeaders.size(); i++) {  namesOfCategories.push_back(columnHeaders[i]);  originalGroupIndexes[i-1] = columnHeaders[i]; }
         if (columnHeaders.size() > 1) { defaultClass = columnHeaders[1]; }
-        else {
-            m->mothurOut("[ERROR]: Your design file contains only one column. Please correct.\n");  m->setControl_pressed(true);
-        }
+        else { m->mothurOut("[ERROR]: Your design file contains only one column. Please correct.\n");  m->setControl_pressed(true); }
 
         //sort groups to keep consistent with how we store the groups in groupmap
         sort(namesOfCategories.begin(), namesOfCategories.end());
@@ -64,48 +61,12 @@ int DesignMap::read(string file) {
         totalCategories.resize(numCategories);
         int count = 0;
 
-        //file without headers, fix it
-        if ((temp != "group") && (temp != "Group")){
-            group = temp;
-            util.checkGroupName(group);
-            if (m->getDebug()) { m->mothurOut("[DEBUG]: group = " + group + "\n"); }
-
-            //if group info, then read it
-            vector<string> categoryValues; categoryValues.resize(numCategories, "not found");
-            for (int i = 0; i < numCategories; i++) {
-                int thisIndex = indexCategoryMap[originalGroupIndexes[i]]; //find index of this category because we sort the values.
-                string temp = tempColumnHeaders[i];
-                util.checkGroupName(temp);
-                categoryValues[thisIndex] = temp;
-
-                if (m->getDebug()) { m->mothurOut("[DEBUG]: value = " + temp + "\n"); }
-
-                //do we have this value for this category already
-                map<string, int>::iterator it = totalCategories[thisIndex].find(temp);
-                if (it == totalCategories[thisIndex].end()) { totalCategories[thisIndex][temp] = 1; }
-                else {  totalCategories[thisIndex][temp]++; }
-            }
-
-
-            map<string, int>::iterator it = indexGroupNameMap.find(group);
-            if (it == indexGroupNameMap.end()) {
-                groups.push_back(group);
-                indexGroupNameMap[group] = count;
-                designMap.push_back(categoryValues);
-                count++;
-            }else {
-                error = true;
-                m->mothurOut("[ERROR]: Your design file contains more than 1 group named " + group + ", group names must be unique. Please correct.\n"); 
-            }
-        }
-
         while (!in.eof()) {
 
             if (m->getControl_pressed()) { break; }
 
             in >> group; util.gobble(in);
-            util.checkGroupName(group);
-            if (m->getDebug()) { m->mothurOut("[DEBUG]: group = " + group + "\n"); }
+            util.checkGroupName(group);  if (m->getDebug()) { m->mothurOut("[DEBUG]: group = " + group + "\n"); }
 
             //if group info, then read it
             vector<string> categoryValues; categoryValues.resize(numCategories, "not found");