fix: tnt format

main.cpp

@@ -3,6 +3,7 @@
 #include <iostream>
 #include <sstream>
 #include <string>
+#include <vector>
 
 using namespace std;
 
@@ -24,7 +25,7 @@ class Sample {
   };
 };
 
-Basic_arg procargs(int nargs, char** arg, char* itn, char* otn);
+Basic_arg procargs(int nargs, char** arg);
 Sample read_input(char* itn, int intype);
 void show_help(int help_num);
 Sample readFas(char* itn);
@@ -40,6 +41,8 @@ bool isNum(string strnum);
 bool checkalign(class Sample sam);
 string to_lower(string stri);
 string add_space(char x, string str_old);
+string rep_space(string str_old);
+string del_space(string str_old);
 string checktype(string str);
 int countfre(string str, char c);
 int checkextension(string str);
@@ -59,7 +62,7 @@ Sample readPhy(char* itn) {
   nchar = stoi(snseq);  // string to int
   Sample sam(ntax, nchar);
   // read sequence
-  int lennum;
+  unsigned int lennum;
   for (lennum = 0; lennum < sam.ntax; lennum++) {
     getline(matrixfile, snall);
     istringstream istr(snall);
@@ -75,55 +78,53 @@ Sample readFas(char* itn) {
   int ntax, nchar, lnum;
   ifstream matrixfile;
   matrixfile.open(itn);
+  // use vector to read once use serveral times
+  vector<string> file_content;
   // check line number and taxa number
   ntax = 0;
   string temln;
   for (lnum = 0; getline(matrixfile, temln); lnum++) {
+    file_content.push_back(temln);
     if (temln[0] == '>') {
       ntax++;
     }
   }
-  matrixfile.clear();
-  matrixfile.seekg(0);
+  matrixfile.close();
   // check the nchar
-  string* str_a = new string;
-  string* str_b = new string;
-  int r = lnum / ntax;
+  string str_i;
+  int r;
+  if (file_content.size() != 0) {
+    r = file_content.size() / ntax;
+  } else {
+    cout << "MiMi:\tInput file contains 0 line" << endl;
+    exit(0);
+  }
   for (int i = 0; i < r; i++) {
-    getline(matrixfile, *str_a);
     if (i > 0) {
-      *str_b = *str_b + *str_a;
+      file_content[i] = del_space(file_content[i]);
+      str_i = str_i + file_content[i];
     }
   }
-  nchar = (*str_b).length();
-  delete str_a;
-  str_a = nullptr;
-  delete str_b;
-  str_b = nullptr;
-  matrixfile.clear();
-  matrixfile.seekg(0);
+  nchar = str_i.length();
   // create class
   Sample sam(ntax, nchar);
   // get class
-  string* str_c = new string;
-  for (int a = 1, b = 0; a <= lnum; a++) {
-    if (a % r == 1) {
-      getline(matrixfile, sam.taxas[b]);
+  for (int a = 0, b = 0; a < (int)file_content.size(); a++) {
+    if ((a + 1) % r == 1) {
+      sam.taxas[b] = file_content[a];
       sam.taxas[b].erase(0, 1);
+      sam.taxas[b] = rep_space(sam.taxas[b]);
     }
-    if (a % r > 1) {
-      getline(matrixfile, *str_c);
-      sam.chars[b] = sam.chars[b] + *str_c;
+    if ((a + 1) % r > 1) {
+      file_content[a] = del_space(file_content[a]);
+      sam.chars[b] = sam.chars[b] + file_content[a];
     }
-    if (a % r == 0) {
-      getline(matrixfile, *str_c);
-      sam.chars[b] = sam.chars[b] + *str_c;
+    if ((a + 1) % r == 0) {
+      file_content[a] = del_space(file_content[a]);
+      sam.chars[b] = sam.chars[b] + file_content[a];
       b++;
     }
   }
-  delete str_c;
-  str_c = nullptr;
-  matrixfile.close();
   return sam;
 }
 
@@ -147,7 +148,7 @@ Sample readTnt(char* itn) {
   // create class
   Sample sam(ntax, nchar);
   // get class
-  int lennum;
+  unsigned int lennum;
   for (lennum = 0; lennum < sam.ntax; lennum++) {
     getline(matrixfile, stri);
     istringstream istr(stri);
@@ -169,22 +170,26 @@ Sample readNex(char* itn) {
   // open file
   ifstream matrixfile;
   matrixfile.open(itn);
-  // some tem
+  // some var
   string snall, stri, str_a, str_b;
-  bool found = false, found_ntax = false, found_nchar = false,
-       found_equal = false;
+  bool found_ntax = false, found_nchar = false, found_equal = false;
   char x = '=';
-  int lnum, eulnum;
+  // get line number and read line to vector
+  int lnum;
+  unsigned int eulnum;
+  vector<string> file_content;
+  while (getline(matrixfile, snall)) {
+    file_content.push_back(snall);
+  }
+  matrixfile.close();
   // getline line by line
-  for (lnum = 0; getline(matrixfile, snall); lnum++) {
-    str_a = to_lower(snall);
+  for (lnum = 0; lnum < (int)file_content.size(); lnum++) {
+    str_a = to_lower(file_content[lnum]);
     str_b = add_space(x, str_a);
     istringstream istr(str_b);
     // convert to words
     while (istr >> stri) {
-      if (stri == "dimensions") {
-        found = true;
-      } else if (stri == "ntax") {
+      if (stri == "ntax") {
         found_ntax = true;
       } else if (stri == "nchar") {
         found_nchar = true;
@@ -193,7 +198,6 @@ Sample readNex(char* itn) {
       } else if (found_ntax && found_equal) {
         if (stri.back() == ';') {
           stri.pop_back();
-          found = false;
         }
         ntax = stoi(stri);
         found_equal = false;
@@ -201,7 +205,6 @@ Sample readNex(char* itn) {
       } else if (found_nchar && found_equal) {
         if (stri.back() == ';') {
           stri.pop_back();
-          found = false;
         }
         nchar = stoi(stri);
         found_equal = false;
@@ -211,32 +214,24 @@ Sample readNex(char* itn) {
       }
     }
   }
-  // go back
-  matrixfile.clear();
-  matrixfile.seekg(0);
   // create class
   Sample sam(ntax, nchar);
-  // some temp, z is line number, l is the string arrary number
-  int z = 0;
+  // read line by line, limit line number
   int l = 0;
-  // read line by line
-  while (getline(matrixfile, snall)) {
+  for (unsigned int z = eulnum; z < (eulnum + sam.ntax); z++) {
     // convert to word
-    istringstream istr(snall);
-    // limit the read line number
-    if (z > (eulnum - 1) && z < (eulnum + sam.ntax)) {
-      istr >> sam.taxas[l];
-      istr >> sam.chars[l];
-      l++;
-    }
-    z++;
+    istringstream istr(file_content[z]);
+    istr >> sam.taxas[l];
+    istr >> sam.chars[l];
+    l++;
   }
   return sam;
 }
+
 string add_space(char x, string str_old) {
   int i;
   string str_new;
-  for (i = 0; i < str_old.length(); i++) {
+  for (i = 0; i < (int)str_old.length(); i++) {
     if (str_old[i] != x) {
       str_new = str_new + str_old[i];
     } else {
@@ -246,6 +241,32 @@ string add_space(char x, string str_old) {
   return str_new;
 }
 
+string del_space(string str_old) {
+  int i;
+  string str_new;
+  char x = ' ';
+  for (i = 0; i < (int)str_old.length(); i++) {
+    if (str_old[i] != x) {
+      str_new = str_new + str_old[i];
+    }
+  }
+  return str_new;
+}
+
+string rep_space(string str_old) {
+  int i;
+  string str_new;
+  char x = ' ', c = '_';
+  for (i = 0; i < (int)str_old.length(); i++) {
+    if (str_old[i] != x) {
+      str_new = str_new + str_old[i];
+    } else {
+      str_new = str_new + c;
+    }
+  }
+  return str_new;
+}
+
 string to_lower(string stri) {
   transform(stri.begin(), stri.end(), stri.begin(), ::tolower);
   return stri;
@@ -253,7 +274,7 @@ string to_lower(string stri) {
 
 void writeFas(class Sample sam, char* otn) {
   ofstream matrixfile(otn);
-  for (int i = 0; i < sam.ntax; i++) {
+  for (unsigned int i = 0; i < sam.ntax; i++) {
     matrixfile << ">" << sam.taxas[i] << endl;
     matrixfile << sam.chars[i] << endl;
   }
@@ -263,7 +284,7 @@ void writeFas(class Sample sam, char* otn) {
 void writePhy(class Sample sam, char* otn) {
   ofstream matrixfile(otn);
   matrixfile << sam.ntax << " " << sam.nchar << endl;
-  for (int i = 0; i < sam.ntax; i++) {
+  for (unsigned int i = 0; i < sam.ntax; i++) {
     matrixfile << sam.taxas[i] << "\t" << sam.chars[i] << endl;
   }
   matrixfile.close();
@@ -273,16 +294,14 @@ void writeNex(class Sample sam, char* otn) {
   ofstream matrixfile(otn);
   string datatype;
   datatype = checktype(sam.chars[0]);
-  matrixfile << "#NEXUS" << endl;
-  matrixfile << "Begin data;" << endl
-             << "\tDimensions nchar=" << sam.nchar << " ntax=" << sam.ntax
-             << ";" << endl
-             << "\tFormat datatype=" << datatype << " missing=? gap=-;" << endl
-             << "\tMatrix" << endl;
-  for (int i2 = 0; i2 < sam.ntax; i2++) {
+  matrixfile << "#NEXUS\nBegin data;\n\tDimensions nchar=" << sam.nchar
+             << " ntax=" << sam.ntax << ";\n\tFormat datatype=" << datatype
+             << " missing=? gap=-;\n\tMatrix" << endl;
+
+  for (unsigned int i2 = 0; i2 < sam.ntax; i2++) {
     matrixfile << "\t\t" << sam.taxas[i2] << "\t" << sam.chars[i2] << endl;
   }
-  matrixfile << "\t;" << endl << "End;" << endl;
+  matrixfile << "\t;\nEnd;" << endl;
   matrixfile.close();
 }
 
@@ -323,18 +342,18 @@ int countfre(string str, char c) {
 
 void writeTnt(class Sample sam, char* otn) {
   ofstream matrixfile(otn);
-  matrixfile << "xread" << endl << "\' \'" << endl;
-  matrixfile << sam.nchar << " " << sam.ntax << endl;
-  for (int i = 0; i < sam.ntax; i++) {
+  matrixfile << "xread\n\' \'\n" << sam.nchar << " " << sam.ntax << endl;
+  for (unsigned int i = 0; i < sam.ntax; i++) {
     matrixfile << sam.taxas[i] << "\t" << sam.chars[i] << endl;
   }
-  matrixfile << "proc / ;" << endl;
+  matrixfile << "\n;\nproc / ;" << endl;
   matrixfile.close();
 }
 
-Basic_arg procargs(int nargs, char** arg, char* itn, char* otn) {
+Basic_arg procargs(int nargs, char** arg) {
   int i, sta = 0, intype = 0, outype = 0;
   string para, inputfile, outputfile;
+  char *itn, *otn;
   // no arg, show help
   if (nargs == 1) {
     show_help(0);
@@ -345,28 +364,36 @@ Basic_arg procargs(int nargs, char** arg, char* itn, char* otn) {
   for (i = 1; i < nargs; i++) {
     // to string
     string para(arg[i]);
-    if (para == "-h" | para == "--help") {
+    if ((para == "-h") || (para == "--help")) {
       show_help(1);
       sta = 2;
-    } else if (para == "-i" | para == "--input") {
-      i++;
-      itn = arg[i];
-      string inputfile(arg[i]);
-      intype = checkextension(inputfile);
-      sta++;
-    } else if (para == "-o" | para == "--output") {
-      i++;
-      otn = arg[i];
-      string outputfile(arg[i]);
-      outype = checkextension(outputfile);
-      sta++;
+    } else if ((para == "-i") || (para == "--input")) {
+      if ((i + 1) < nargs) {
+        i++;
+        itn = arg[i];
+        string inputfile(arg[i]);
+        intype = checkextension(inputfile);
+        sta++;
+      } else {
+        cout << "MiMi:\tOInput file name must be defined" << endl;
+      }
+    } else if (((para == "-o") || (para == "--output"))) {
+      if ((i + 1) < nargs) {
+        i++;
+        otn = arg[i];
+        string outputfile(arg[i]);
+        outype = checkextension(outputfile);
+        sta++;
+      } else {
+        cout << "MiMi:\tOutput file name must be defined" << endl;
+      }
     } else {
-      cout << "MiMi\tUnknown arguments, please use -h to check" << endl;
+      cout << "MiMi:\tUnknown arguments, please use -h to check" << endl;
       exit(0);
     }
   }
   if (sta != 2) {
-    cout << "MiMi\tInput and Output can't be empty" << endl;
+    cout << "MiMi:\tInput and Output can't be empty" << endl;
     exit(0);
   }
   Basic_arg arguvar(intype, outype, itn, otn);
@@ -380,20 +407,20 @@ int checkextension(string str) {
   if (loc) {
     extension = str.substr(loc + 1);
   } else {
-    cout << "MiMi\tPlease sepecifc the extension name" << endl;
+    cout << "MiMi:\tPlease sepecifc the extension name" << endl;
     exit(0);
   }
   extension = to_lower(extension);
-  if (extension == "fas" | extension == "fasta") {
+  if ((extension == "fas") || (extension == "fasta")) {
     type = 1;
-  } else if (extension == "nex" | extension == "nexus") {
+  } else if ((extension == "nex") || (extension == "nexus")) {
     type = 2;
-  } else if (extension == "phy" | extension == "phylip") {
+  } else if ((extension == "phy") || (extension == "phylip")) {
     type = 3;
-  } else if (extension == "tnt" | extension == "ss") {
+  } else if ((extension == "tnt") || (extension == "ss")) {
     type = 4;
   } else {
-    cout << "MiMi\tUnknown format" << endl;
+    cout << "MiMi:\tUnknown format" << endl;
     exit(0);
   }
   return type;
@@ -405,22 +432,17 @@ void show_help(int help_num) {
          << endl;
   } else {
     cout << "\n  ／l、    \t.  . .  .\n（ﾟ､ ｡ ７    \t|\\/|*|\\/|*\n  l  ~ヽ   "
-            " \t|  |||  ||\n  じしf_,)ノ\t|  |||  ||\n"
+            " \t|  |||  ||\n  じしf_,)ノ\t|  |||  ||\n\nMorphology into "
+            "Molecules into\nGPL;\tGuoyi "
+            "Zhang;\t2023\n\nArguments:\n-h\t--help;\n-i\t--input\t\t${"
+            "filename};\n-o\t--output\t${filename};\n\nAccepted "
+            "formats:\nfas\tfasta;\nnex\tnexus;\nphy\tphylip;\ntnt\tss;"
          << endl;
-    cout << "Morphology into Molecules into\n"
-         << "GPL;\tGuoyi Zhang;\t2023\n"
-         << endl;
-    cout << "-h\t--help;\n-i\t--input\t\t${filename};\n-o\t--output\t${"
-            "filename};\n"
-         << endl;
-    cout
-        << "Accepted formats:\nfas\tfasta;\nnex\tnexus;\nphy\tphylip;\ntnt\tss;"
-        << endl;
   }
 }
 
 Sample read_input(char* itn, int intype) {
-  int ntax, nchar;
+  int ntax = 0, nchar = 0;
   Sample sam(ntax, nchar);
   ifstream matrixfile;
   matrixfile.open(itn);
@@ -430,7 +452,7 @@ Sample read_input(char* itn, int intype) {
     if (intype == 3) sam = readPhy(itn);
     if (intype == 4) sam = readTnt(itn);
   } else {
-    cout << "MiMi\tInput file can't be open" << endl;
+    cout << "MiMi:\tInput file can't be open" << endl;
     exit(0);
   }
   return sam;
@@ -444,16 +466,17 @@ void write_output(class Sample sam, char* otn, int outype) {
     if (outype == 3) writePhy(sam, otn);
     if (outype == 4) writeTnt(sam, otn);
   } else {
-    cout << "MiMi\tOutput file can't be open" << endl;
+    cout << "MiMi:\tOutput file can't be open" << endl;
     exit(0);
   }
 }
 
 bool checkalign(class Sample sam) {
-  int a = 0, b = 0, x = 0;
+  int a = 0, b = 0;
+  unsigned int x = 0;
   a = sam.nchar;
   bool aligned = true;
-  for (int i = 0; i < sam.ntax; i++) {
+  for (unsigned int i = 0; i < sam.ntax; i++) {
     b = sam.chars[i].length();
     if (a == b) {
       x++;
@@ -466,8 +489,7 @@ bool checkalign(class Sample sam) {
 }
 
 int main(int argc, char** argv) {
-  char *itn, *otn;
-  Basic_arg arguvar = procargs(argc, argv, itn, otn);
+  Basic_arg arguvar = procargs(argc, argv);
   if (arguvar.intype != 0 && arguvar.outype != 0) {
     Sample sam = read_input(arguvar.itn, arguvar.intype);
     cout << "MiMi:\tInput\tfinished" << endl;

mimi.tcl

@@ -0,0 +1,108 @@
+package require Tk
+
+proc ? L {lindex $L [expr {int (rand ()* [llength $L])}]}
+
+proc validateFile name {
+  set ext [file extension $name]
+  if {$ext ni {.fas .fasta .nex .nexus .tnt .ss .phy .phylip}} {
+    tk_messageBox -icon error -type ok -message "extension of file must be fas fasta nex nexus tnt ss phy phylip"
+    return 0
+  } else {
+    return 1
+  }
+}
+
+proc runMimi {} {
+  global input output mimi
+  if {[validateFile $input] && [validateFile $output]} {
+    if {[file exists $mimi]} {
+      exec $mimi -i $input -o $output
+      tk_messageBox -icon info -type ok -message "Success"
+    } else {
+      tk_messageBox -icon error -type ok -message "Can't find the path of MiMi"
+    }
+  }
+}
+
+proc clearAll {} {
+  global input output mimi
+  set input ""
+  set output ""
+}
+
+proc chooseMimi {} {
+  global mimi
+  set file [tk_getOpenFile]
+  if {$file ne ""} {
+    set mimi $file
+  }
+}
+
+proc chooseInput {} {
+  global input
+  set file [tk_getOpenFile]
+  if {$file ne ""} {
+    set input $file
+  }
+}
+
+proc chooseOutput {} {
+  global output
+  set file [tk_getSaveFile]
+  if {$file ne ""} {
+    set output $file
+  }
+}
+
+proc quitApp {} {
+  exit
+}
+
+proc helpApp {} { 
+  tk_messageBox -icon info -type ok -message "Thanks to my beloved daughter MiMi (Ragdoll cat) for accompanying me in my life, mom loves you forever.\n\nGuoyi"
+}
+
+#bind . <Escape> {quitApp}
+wm protocol . WM_DELETE_WINDOW {quitApp}
+
+wm iconphoto . [image create photo -file logo.gif]
+
+set input ""
+set output ""
+set mimi ""
+
+wm title . "Mimi GUI"
+
+# 创建菜单栏
+menu .menubar
+.menubar add cascade -label "Settings" -menu .menubar.file
+menu .menubar.file
+.menubar.file add command -label "Select MiMi Path" -command chooseMimi
+.menubar.file add separator
+.menubar.file add command -label "Quit" -command quitApp
+
+.menubar add cascade -label "Help" -menu .menubar.help
+menu .menubar.help
+.menubar.help add command -label "About" -command helpApp
+
+# 配置菜单栏到顶层窗口
+. configure -menu .menubar
+
+# 创建输入输出框和按钮
+grid [ttk::frame .c -padding "3 3 12 12"] -column 0 -row 0 -sticky nwes
+grid columnconfigure . 0 -weight 1; grid rowconfigure . 0 -weight 1
+
+grid [ttk::label .c.ilbl -text "Input"] -column 1 -row 1 -sticky e
+grid [ttk::entry .c.input -width 20 -textvariable input] -column 2 -row 1 -sticky we
+grid [ttk::button .c.ibtn -text "Select" -command chooseInput] -column 3 -row 1
+
+grid [ttk::label .c.olbl -text "Output"] -column 1 -row 2 -sticky e
+grid [ttk::entry .c.output -width 20 -textvariable output] -column 2 -row 2 -sticky we
+grid [ttk::button .c.obtn -text "Select" -command chooseOutput] -column 3 -row 2
+
+grid [ttk::button .c.run -text "Run" -command runMimi] -column 2 -row 3
+grid [ttk::button .c.clear -text "Clean" -command clearAll] -column 3 -row 3
+
+foreach w [winfo children .c] {grid configure $w -padx 5 -pady 5 }
+
+vwait forever