source: GSEXmlParser/GSEXmlParser.cpp @ 195

#include "GSEXmlParser.h"

void   Data::Dump()
{
    LOG("Sample iid: " << iid);
    LOG("Source: " << source);
    LOG("Supplementary-Data Type: " << supp_type);
    LOG("Supplementary-Data: " << supp_data);
    LOG("\n-------- Description --------");
    for(int i = 0; i < desc.GetCount(); i++)
        LOG("   " << desc.GetKey(i) << " ..... " << desc[i]);
    LOG("-------- Description --------\n");
}

String   Data::ToXML()
{
    String descs; //descriptions
    for(int i = 0; i < desc.GetCount(); i++)
        descs << XmlTag("Description")("name", desc.GetKey(i)).Text(desc[i]);

    return XmlTag("Sample")("iid", iid)(
                     XmlTag("Source").Text(source) + 
                     XmlTag("Supplementary-Data")("type", supp_type).Text(supp_data) +
                     XmlTag("Descriptions") (descs) 
                 );
}

String Data::ToTSV()
{
  String tsv, tab, nl;
  tab = "\t";
  nl = "\n";
  tsv = source + tab + "Sample-iid" + tab + iid + nl +
        source + tab + "Supplementary-Data:" + supp_type + tab + supp_data + nl;
  for(int i = 0; i < desc.GetCount(); i++)      
      tsv += source + tab + "Description:" + desc.GetKey(i) + tab + desc[i] + nl; 
        
  return tsv;
}

void   GSEParser::Dump()
{
  LOG("\n======== GSE XML ========\n");
  for (int i=0; i < data.GetCount(); i++) {
    data[i].Dump();
  }
}

void GSEParser::LoadXML(String filename) {
  //Load XML File
    //LOG("\n---- GSE 2109 XML ----");
    
    String gse;
    /*
    Cout() << filename << "\n";
    String fn;
    fn = GetDataFile(filename);
    Cout() << fn << "\n";
    */
    gse = LoadFile(filename);
    //Cout() << gse << "\n";
    XmlNode rnode = ParseXML(gse);
    
    //Clear Data Array
    data.Clear();
    
    //Get Root Node    
    const XmlNode &xnode = rnode["MINiML"];
    
    //Save Data Array, one by one
    Vector<String> lines, des;
    String str, lstr, rstr;
    for (int i=0; i<xnode.GetCount(); i++) {
       if (xnode[i].GetTag()=="Sample") {  
          //LOG("Sample(iid):" << xnode[i].Attr("iid"));
          //LOG("Source:" << xnode[i]["Channel"]["Source"].GatherText());
          //LOG("\n-------- Description --------");
          Data d;
          d.iid = xnode[i].Attr("iid");
          d.source = xnode[i]["Channel"]["Source"].GatherText();
          d.supp_type = xnode[i]["Supplementary-Data"].Attr("type");
          d.supp_data = xnode[i]["Supplementary-Data"].GatherText();
          lines = Split(xnode[i]["Description"].GatherText(),'\n');
          for (int j=0; j<lines.GetCount(); j++) {
              if (lines[j].Find("=") > -1) {
                  des = Split(lines[j], '=');
              } else if (lines[j].Find(":") > -1) {
                  des = Split(lines[j], ':');
              }
              if (des.GetCount()>1) {
                  lstr = TrimSpace(des[0]);
                  rstr = TrimSpace(des[1]);                  
                  //LOG(lstr << ":" << rstr);
                  String& v = d.desc.GetAdd(lstr);
                  v = rstr;
              }
          }
          data.Add(d);
          //LOG("-------- Description --------\n");
       }
    }
    //    
}

String   GSEParser::ToXML()
{
  String datas; // Datas
    for (int i = 0; i < data.GetCount(); i++) {
        datas << data[i].ToXML();
    }
    return XmlHeader() + XmlTag("GSEFAMILY")(datas);
}

String   GSEParser::ToTSV()
{
  String tsv, tab; 
  tab = "\t";
  tsv = "# rowkey" + tab + "columnkey" + tab + "value\n";
  
  for (int i = 0; i < data.GetCount(); i++) {
    tsv << data[i].ToTSV();
  }
  return tsv;
}

void   GSEParser::ToMySQL(String username, String password, String db, String db_ip)
{
  //
  MySqlSession session;
  if(!session.Connect(username, password, db, db_ip)) {
    printf("Can't connect with MySql\n");
    return;
  }
  Sql sql(session);
  sql.Execute("use gse"); 
  String cmd;
  //
  Cout() << "Total Record Count:" << data.GetCount() << "\n";
  int sid;
    int did;
    String name;  
  for (int i = 0; i < data.GetCount(); i++) {   
    // check gsefamily record
    cmd = "select * from gsefamily where iid = \"" + data[i].iid + "\"";
    sql.Execute(cmd);   
    if (sql.GetRowsProcessed()>0) {
      Cout() << "Warnning: IID ["+data[i].iid+"] is exist." << "\n";      
    } else {
      // add a new gsefamily record
      cmd = "insert into gsefamily (iid,source) values (\""+data[i].iid+"\",\""+data[i].source+"\");";
      if (!sql.Execute(cmd)) {
        Cout() << "Execution Error: " << sql.GetErrorCodeString() << "  Cmd: " << sql.GetErrorStatement() << "\n";      
      }
    }
    //Cout() << "IID ["+data[i].iid+"]" << "\n";
    
    sid = 0;
    // check supplementary    
        cmd = "select sid from supplementary where type = \"" + data[i].supp_type + "\";";        
        sql.Execute(cmd);        
        if (sql.GetRowsProcessed()==0) {  
            //new a supplementary record
            cmd = "insert into supplementary (type) values (\"" + data[i].supp_type + "\");";            
            if (!sql.Execute(cmd)) {
                Cout() << "Execution Error: " << sql.GetErrorCode() << "  Cmd: " << sql.GetErrorStatement() << "\n";      
            }       
            sid = sql.GetInsertedId();     
        } else {
            if (sql.Fetch())
               sid = sql[0];
            //Cout() << "sid: " << sid << "\n";
        }
        
        //new a supplementary-map record        
        cmd = "insert into `supplementary-map` (iid,sid,value) values (\""+data[i].iid+"\",\""+AsString(sid)+"\",\""+data[i].supp_data+"\")";        
        if (!sql.Execute(cmd)) {            
            Cout() << "\nExecution Error: " << sql.GetErrorCode() << "  Cmd: " << sql.GetErrorStatement() << "\n";      
        }
                       
        //Cout() << "data[" << i << "].desc.GetCount() = " << data[i].desc.GetCount() << "\n";
        VectorMap<String, String> desc = data[i].desc;
        for (int j=0; j<desc.GetCount(); j++) {
            
      // check description
      name = desc.GetKey(j);      
          cmd = "select did from description where name = \"" + name + "\"";
          sql.Execute(cmd);
          
          did = 0;
          if (sql.GetRowsProcessed()==0) {  
              //new a description record
              cmd = "insert into description (name) values (\"" + name + "\");";
              if (!sql.Execute(cmd)) {
                  Cout() << "Execution Error: " << sql.GetErrorCodeString() << "  Cmd: " << sql.GetErrorStatement() << "\n";      
              }   
              did = sql.GetInsertedId();         
          } else {
              if (sql.Fetch())
                did = sql[0];
          }
          
          cmd = "insert into `description-map` (iid,did,value) values (\"" + data[i].iid + "\",\"" + AsString(did) + "\",\"" + desc.Get(name) + "\")";
          if (!sql.Execute(cmd)) {
              Cout() << "Execution Error: " << sql.GetErrorCodeString() << "  Cmd: " << sql.GetErrorStatement() << "\n";      
          }
                    
        }
        
  } 
}

String TrimSpace(String str) {
  while (str.StartsWith(" ") && str.GetLength()) {
    str = str.Mid(1);
  }
  while (str.EndsWith(" ") && str.GetLength()) {
    str = str.Left(str.GetLength()-1);
  }
  return str;
}

CONSOLE_APP_MAIN
{
  String in, out, fmt, user, pwd, db, ip, arg_flag;
  /*
  in = "GSE2109_family.xml";
  out = "GSEFamily.tsv";
  fmt = "tsv";
  */
  const Vector<String>& cmdline = CommandLine();
    
  for(int i = 0; i < cmdline.GetCount(); i++) {
    if (cmdline[i].StartsWith("-")){
      arg_flag = cmdline[i];      
    } else if (arg_flag.IsEqual("-i")) {
      in = cmdline[i];    
    } else if (arg_flag.IsEqual("-o")) {
      out = cmdline[i];     
    } else if (arg_flag.IsEqual("-f")) {
      fmt = cmdline[i];
    } else if (arg_flag.IsEqual("-u")) {
      user = cmdline[i];      
    } else if (arg_flag.IsEqual("-p")) {
      pwd = cmdline[i];     
    } else if (arg_flag.IsEqual("-db")) {
      db = cmdline[i];      
    } else if (arg_flag.IsEqual("-ip")) {
      ip = cmdline[i];      
    } else if (arg_flag.StartsWith("-")) {
      //Error argument
      Cout() << "Usage Error: invalid argument "<< arg_flag << "\n";
      exit(0);
    }
  }
  
  if (cmdline.GetCount()==0 || 
      ( (fmt.IsEqual("xml") || fmt.IsEqual("tsv")) && (in.IsEmpty() || out.IsEmpty() || fmt.IsEmpty()) ) ||
      ( fmt.IsEqual("mysql") && (in.IsEmpty() || user.IsEmpty() || db.IsEmpty() || ip.IsEmpty()) )
     ) {
    Cout() << "# Usage #\n";
    Cout() << "For linux:   GSEXmlParser -f (xml|tsv) -i (input file) -o (output file)\n";
    Cout() << "For windows: GSEXmlParser.exe -f (xml|tsv) -i (input file) -o (output file)\n";
    Cout() << "             GSEXmlParser.exe -f (mysql) -i (input file) -u (username) -p (password) -db (database) -ip (db host)\n";
    exit(0);  
  } 
  GSEParser parser;
  parser.LoadXML(in);
  if (fmt.IsEqual("xml"))
    SaveFile(out,parser.ToXML());
  else if (fmt.IsEqual("tsv"))  {
    SaveFile(out,parser.ToTSV());
  } else if (fmt.IsEqual("mysql")) {
    parser.ToMySQL(user,pwd,db,ip);
    Cout() << "Completed.\n";
  }
  
  
  
}