datadb/compbio/cassandra/JpredParserLocalFile.java

   1 package compbio.cassandra;
   2
   3 import java.io.BufferedReader;
   4 import java.io.File;
   5 import java.io.IOException;
   6 import java.io.InputStreamReader;
   7 import java.net.HttpURLConnection;
   8 import java.net.MalformedURLException;
   9 import java.io.FileInputStream;
  10 import java.text.ParseException;
  11 import java.text.SimpleDateFormat;
  12 import java.util.ArrayList;
  13 import java.util.Calendar;
  14 import java.util.Date;
  15 import java.util.List;
  16
  17 public class JpredParserLocalFile {
  18         private CassandraCreate cc = new CassandraCreate();
  19         private String dirprefix;
  20
  21         public void setSource (String newsourceprefix) {
  22                 this.dirprefix = newsourceprefix;
  23         }
  24
  25         JpredParserLocalFile() {
  26                 this.dirprefix = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/data.dat";
  27         }
  28
  29         JpredParserLocalFile(String sourceurl) {
  30                 this.dirprefix = sourceurl;
  31         }
  32
  33         public void Parsing(String source, int nDays) {
  34                 Calendar cal = Calendar.getInstance();
  35                 cal.add(Calendar.DATE, -nDays);
  36                 for (int i = 0; i < nDays; ++i) {
  37                         cal.add(Calendar.DATE, 1);
  38                         int month = cal.get(Calendar.MONTH) + 1;
  39                         int year = cal.get(Calendar.YEAR);
  40                         int day = cal.get(Calendar.DATE);
  41                         String date = year + "/" + month + "/" + day;
  42                         ParsingForDate(source, date);
  43                 }
  44         }
  45
  46         private void ParsingForDate(String input, String date) {
  47                 int totalcount = 0;
  48                 int countNoData = 0;
  49                 int countUnclearFASTAid = 0;
  50                 int countinsertions = 0;
  51                 int countinserted = 0;
  52                 int counAlignments = 0;
  53                 int countStrange = 0;
  54
  55                 System.out.println("Inserting jobs for " + date);
  56                 try {
  57                         File file = new File(input);
  58                         BufferedReader alljobs = new BufferedReader(new InputStreamReader(new FileInputStream(file)));
  59                         String line;
  60
  61                         while ((line = alljobs.readLine()) != null) {
  62                                 if (line.matches(date + "(.*)jp_[^\\s]+")) {
  63                                         String[] table = line.split("\\s+");
  64                                         String id = table[table.length - 1];
  65                                         totalcount++;
  66                                         if (!cc.CheckID(id)) {
  67                                                 String confilename = dirprefix + "/" + id + "/" + id + ".concise";
  68                                                 File confile = new File(confilename);
  69                                                 if (confile.exists()) {
  70                                                         try {
  71                                                                 final FastaReader fr = new FastaReader(confilename);
  72                                                                 final List<FastaSequence> seqs = new ArrayList<FastaSequence>();
  73                                                                 String newprotein = "";
  74                                                                 while (fr.hasNext()) {
  75                                                                         final FastaSequence fs = fr.next();
  76                                                                         if (fs.getId().equals("QUERY") || fs.getId().equals(id))
  77                                                                                 newprotein = fs.getSequence().replaceAll("\n", "");
  78                                                                         else
  79                                                                                 seqs.add(fs);
  80                                                                 }
  81                                                                 if (newprotein.equals("")) {
  82                                                                         countUnclearFASTAid++;
  83                                                                 } else {
  84                                                                         SimpleDateFormat formatter = new SimpleDateFormat("yyyy/MM/dd");
  85                                                                         String dateInString1 = table[0].substring(0, table[0].indexOf(":"));
  86                                                                         long dateWork1 = 0;
  87                                                                         try {
  88                                                                                 Date dat1 = formatter.parse(dateInString1);
  89                                                                                 dateWork1 = dat1.getTime();
  90                                                                         } catch (ParseException e) {
  91                                                                                 e.printStackTrace();
  92                                                                         }
  93                                                                         cc.InsertData(dateWork1, table[0], table[1], table[2], id, "OK", "OK", newprotein, seqs);
  94                                                                         ++countinsertions;
  95                                                                         // flush every 100 insertions
  96                                                                         if (0 == countinsertions % 100) {
  97                                                                                 cc.flushData();
  98                                                                         }
  99                                                                 }
 100                                                         } catch (IOException e) {
 101                                                                 e.printStackTrace();
 102                                                         }
 103                                                 } else {
 104                                                         countNoData++;
 105                                                 }
 106                                         } else {
 107                                                 ++countinserted;
 108                                         }
 109                                 } else {
 110                                         if (line.matches(date + "(.*)Sequence0/(.*)")) {
 111                                                 ++counAlignments;
 112                                         } else {
 113                                                 ++countStrange;
 114                                         }
 115                                 }
 116                         }
 117                         alljobs.close();
 118                         System.out.println("Total number of jobs = " + totalcount);
 119                         System.out.println("   " + countinserted + " jobs inserted already");
 120                         System.out.println("   " + counAlignments + " jalview jobs");
 121                         System.out.println("   " + countStrange + " not analysed jobs");
 122                         System.out.println("   " + countNoData + " jobs without *.concise.fasta file");
 123                         System.out.println("   " + countUnclearFASTAid + " jobs with unclear FASTA protein id in *.concise.fasta");
 124                         System.out.println("   " + countinsertions + " new job insertions\n");
 125                 } catch (MalformedURLException e) {
 126                         e.printStackTrace();
 127                 } catch (IOException e) {
 128                         e.printStackTrace();
 129                 }
 130         }
 131 }