BGI/SOAPsnp/FileListManager.h

   1 /*
   2  ******************************************************************************
   3  *Copyright 2010
   4  *      BGI-SHENZHEN
   5  *All Rights Reserved
   6  *ATHUOR : Bill Tang
   7  *CREATE DATE : 2010-8-9
   8  *CLASS NAME: FileListManager
   9  *FUNCTION : a class that can control file list. support open, readline, close etc
  10  *                      file functions.
  11  *FILE NAME : FileListManager.h
  12  *UPDATE DATE : 2010-8-19
  13  *UPDATE BY : Bill Tang
  14  *UPDATA DATA : 2010-8.30
  15  *UPDATA BY :ZHOU GUYUE
  16  *UPDATE: 2010-11-2 change the m_soap_file_vec's type, to vector<*igzstream>.
  17                                         change the interface which is related to soap file.
  18  *UPDATA BY :BIll Tang
  19  *******************************************************************************
  20  */
  21
  22 #pragma once
  23
  24 /*define some error number*/
  25 #define CNSFILE_ERROR           -2              //can not open cnsFile
  26 #define BASEFILE_ERROR          -1              //can not open listFile
  27 #define SOAPFILE_ERROR          -2              //can not open soapFile
  28 #define SAMFILE_ERROR           -2              //can not open sam/bamFile
  29 #define SETLIMIT_ERROR          -3              //can not set limit
  30
  31 #define SETLIMIT_OK          3
  32 #define INDEX_OVER_FLOW         NULL    // index is over flow
  33 #define COUNT_ERROR                     -3              // the readwin_vec size not equal to the out files number.
  34 #define INPUT_ERROR                     -4              // the input files has something wrong.
  35 #define FILE_END                        2               // file is end.
  36
  37 #include "SamCtrl.h"
  38 #include "Readwin.h"
  39 #include "soap_snp.h"
  40 #include "CThreadPool.h"
  41 #include "Call_winManager.h"
  42
  43 using namespace std;
  44
  45 class FileListManager
  46 {
  47         int m_file_number;     // file number in file list.
  48         my_ifstream m_base_file;        // the handle of the file that contain file list.
  49         vector<igzstream*>  m_soap_file_vec;    // the handle vector of the files which are soap format.
  50         vector<SamCtrl*>  m_sam_file_vec;       // the handle vector of the files which are sam/bam format.
  51         vector<gzoutstream*>  m_cns_file_vec;   // the handle vector of the output files.
  52         vector<fstream*> m_matrix_file_vec;     // a vector stored the file handle which are matrix files
  53         int m_stop_pos;         // record the position once one of the samples reach a break point that the chromosome changed.
  54         int m_endfile_count;                            // record the file's count that reach the end.
  55 public:
  56         FileListManager();
  57         virtual ~FileListManager();
  58         // open output file function
  59         virtual int openCnsFile(const string& path, const string& outdir, const char* mode, Parameter * para);
  60         // open all the file in the file list
  61         virtual int openAli(const string& listfilename, const char* mode);
  62         // open sam/bam file
  63         virtual int openSamFile(const string& listfilename, const char* mode);
  64         // open soap format file
  65         virtual int openSoapFile(const string& listfilename);
  66         // read a line from the *vec[index]
  67         virtual int readLine(string& line, const int index);
  68         // close alignment files
  69         virtual void closeAliFiles(void);
  70         // close consensus files
  71         virtual void closeCnsFiles(void);
  72         // read lines to the readwin_vec. The lines' start position must inside a win size. When the lines' end position exceed the win,put it to the next win.
  73         virtual int readWin(vector<Readwin>& readwin_vec, Parameter * para);
  74         // get the soap file handle pointer m_soap_file_vec[index]
  75         virtual igzstream * getSoapFile(const int index);
  76         // get the soap file handle pointer m_soap_file_vec[index]
  77         virtual SamCtrl* getSamFile(const int index);
  78         // read record from soapfile and send to the readwin.
  79         virtual int readWinFromSoap(Readwin& readwin, igzstream& soapfile);
  80         // read record from samCtrl and send to the readwin.
  81         virtual int readWinFromSam(Readwin& readwin, SamCtrl& samCtrl);
  82         // get the file number
  83         virtual int getFileNum(void);
  84         // get handle of consensus file.
  85         virtual gzoutstream* getCnsFile(const int index);
  86         // open matrix files with open_mode mode
  87         virtual int openMatrixFile(const std::string matrix_list, std::ios_base::open_mode mode, const std::string alignment_list = "");
  88         // get the matrix file handle
  89         virtual fstream* getMatrixFile(const int index);
  90         //set limit of file number;
  91         //virtual int setlimit(int filenumber); //update by zhukai on 2010-12-28
  92
  93 };
  94
  95 //2010-11-08
  96
  97 typedef struct _read_win_args
  98 {
  99         //FileListManager * fileListManager_p;
 100         Readwin * readwin_p;
 101         igzstream * m_soap_file_p;
 102         SamCtrl * m_sam_file_p;
 103         int * ret_p;
 104         int index;
 105
 106         inline _read_win_args(Readwin * b, igzstream *c, SamCtrl * f,int * d, int e)
 107         {
 108                 //fileListManager_p = a;
 109                 readwin_p = b;
 110                 m_soap_file_p = c;
 111                 m_sam_file_p = f;
 112                 //cerr <<"111 ret_p = d "<<endl;
 113                 ret_p = d;
 114                 index = e;
 115         };
 116         /*inline _read_win_args(Readwin * b,SamCtrl *c, int * d, int e)
 117         {
 118                 //fileListManager_p = a;
 119                 readwin_p = b;
 120                 m_sam_file_p = c;
 121                 m_soap_file_p = NULL;
 122                 ret_p = d;
 123                 index = e;
 124         };*/
 125         inline _read_win_args()
 126         {
 127                 //fileListManager_p = NULL;
 128                 readwin_p = NULL;
 129                 m_soap_file_p = NULL;
 130                 m_sam_file_p = NULL;
 131                 ret_p = NULL;
 132                 index = 0;
 133         };
 134 }READ_WIN_ARGS;
 135
 136 class Read_win_Task : public CTask
 137 {
 138 public:
 139         FileListManager fileListManager;
 140         // overload the function Run().
 141         int Run();
 142         /*{
 143                 _call_winManager_soap2cns_norm(this->m_ptrData);
 144                 return 1;
 145         }*/
 146 };
 147
 148 // 2010-11-12  the function readWin's parameter structure.
 149 typedef struct _big_read_win_args
 150 {
 151         FileListManager * fileListManager;
 152         vector<Readwin> * readwin_vec;
 153         Parameter * para;
 154
 155         inline _big_read_win_args(FileListManager * a, vector<Readwin> * b, Parameter * c)
 156         {
 157                 fileListManager = a;
 158                 readwin_vec = b;
 159                 para = c;
 160         };
 161
 162         inline _big_read_win_args()
 163         {
 164                 fileListManager = NULL;
 165                 readwin_vec = NULL;
 166                 para = NULL;
 167         };
 168 } BIG_READ_WIN_ARGS;
 169
 170 void *_flieListManager_readWin(void * args);