source: GSEXmlParser/GSEXmlParser.cpp @ 29

#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;
    gse = LoadFile(GetDataFile(filename));
    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;
}

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, 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.StartsWith("-")) {
      //Error argument
      Cout() << "Usage Error: invalid argument "<< arg_flag << "\n";
      exit(0);
    }
  }
  
  if (cmdline.GetCount()==0 || in.IsEmpty() || out.IsEmpty() || fmt.IsEmpty()) {
    Cout() << "Usage: GSEXmlParser.exe -f (xml|tsv) -i (input file) -o (output file)\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());
  }
  
  
  
}