binaries/src/muscle/physeq.cpp

   1 #include "muscle.h"\r
   2 #include "msa.h"\r
   3 #include "textfile.h"\r
   4 \r
   5 const int BLOCKSIZE = 60;\r
   6 \r
   7 static char FixChar(char c)\r
   8         {\r
   9         switch (c)\r
  10                 {\r
  11         case '(':\r
  12         case ')':\r
  13         case '[':\r
  14         case ']':\r
  15         case ':':\r
  16         case ';':\r
  17         case ',':\r
  18                 return '_';\r
  19                 }\r
  20         if (!isprint(c))\r
  21                 return '_';\r
  22         return c;\r
  23         }\r
  24 \r
  25 static void FixName(char Name[])\r
  26         {\r
  27         while (char c = *Name)\r
  28                 *Name++ = FixChar(c);\r
  29         }\r
  30 \r
  31 void MSA::ToPhySequentialFile(TextFile &File) const\r
  32         {\r
  33         const unsigned SeqCount = GetSeqCount();\r
  34         const unsigned ColCount = GetColCount();\r
  35 \r
  36         File.PutFormat("%d %d\n", SeqCount, ColCount);\r
  37 \r
  38         if (0 == ColCount)\r
  39                 return;\r
  40 \r
  41         for (unsigned Seq = 0; Seq < SeqCount; ++Seq)\r
  42                 {\r
  43                 char Name[11];\r
  44                 const char *ptrName = GetSeqName(Seq);\r
  45                 size_t n = strlen(ptrName);\r
  46                 if (n > 10)\r
  47                         n = 10;\r
  48                 memcpy(Name, ptrName, n);\r
  49                 Name[n] = 0;\r
  50                 FixName(Name);\r
  51                 File.PutFormat("%-10.10s", Name);\r
  52 \r
  53                 int BlockIndex = 0;\r
  54                 int Col = 0;\r
  55                 for (;;)\r
  56                         {\r
  57                         const unsigned MaxCols = (BlockIndex == 0) ? (BLOCKSIZE - 10) : BLOCKSIZE;\r
  58                         for (unsigned ColsThisBlock = 0; ColsThisBlock < MaxCols; ++ColsThisBlock)\r
  59                                 {\r
  60                                 if (Col == ColCount)\r
  61                                         break;\r
  62                                 if (ColsThisBlock%10 == 0 && (BlockIndex == 0 || ColsThisBlock > 0))\r
  63                                         File.PutChar(' ');\r
  64                                 char c = GetChar(Seq, Col);\r
  65                                 if (isalpha(c))\r
  66                                         c = toupper(c);\r
  67                                 File.PutChar(c);\r
  68                                 ++Col;\r
  69                                 }\r
  70                         File.PutChar('\n');\r
  71                         if (Col == ColCount)\r
  72                                 break;\r
  73                         ++BlockIndex;\r
  74                         }\r
  75                 }\r
  76         }\r
  77 \r
  78 void MSA::ToPhyInterleavedFile(TextFile &File) const\r
  79         {\r
  80         const unsigned SeqCount = GetSeqCount();\r
  81         const unsigned ColCount = GetColCount();\r
  82 \r
  83         File.PutFormat("%d %d\n", SeqCount, ColCount);\r
  84 \r
  85         if (0 == ColCount)\r
  86                 return;\r
  87 \r
  88         int Col = 0;\r
  89         for (;;)\r
  90                 {\r
  91                 const unsigned ColBlockStart = Col;\r
  92                 const unsigned MaxCols = (ColBlockStart == 0) ? (BLOCKSIZE - 10) : BLOCKSIZE;\r
  93 \r
  94                 for (unsigned Seq = 0; Seq < SeqCount; ++Seq)\r
  95                         {\r
  96                         if (0 == ColBlockStart)\r
  97                                 {\r
  98                                 char Name[11];\r
  99                                 const char *ptrName = GetSeqName(Seq);\r
 100                                 size_t n = strlen(ptrName);\r
 101                                 if (n > 10)\r
 102                                         n = 10;\r
 103                                 memcpy(Name, ptrName, n);\r
 104                                 Name[n] = 0;\r
 105                                 FixName(Name);\r
 106                                 File.PutFormat("%-10.10s", Name);\r
 107                                 }\r
 108 \r
 109                         Col = ColBlockStart;\r
 110                         for (unsigned ColsThisBlock = 0; ColsThisBlock < MaxCols; ++ColsThisBlock)\r
 111                                 {\r
 112                                 if (Col == ColCount)\r
 113                                         break;\r
 114                                 if (ColsThisBlock%10 == 0 && (0 == ColBlockStart || ColsThisBlock > 0))\r
 115                                         File.PutChar(' ');\r
 116                                 char c = GetChar(Seq, Col);\r
 117                                 if (isalpha(c))\r
 118                                         c = toupper(c);\r
 119                                 File.PutChar(c);\r
 120                                 ++Col;\r
 121                                 }\r
 122                         File.PutChar('\n');\r
 123                         }\r
 124                 if (Col == ColCount)\r
 125                         break;\r
 126                 File.PutChar('\n');\r
 127                 }\r
 128         }\r