datadb/compbio/cassandra/DataParsing.java

   1 package compbio.cassandra;
   2
   3 import java.io.BufferedReader;
   4 import java.io.IOException;
   5 import java.io.InputStreamReader;
   6 import java.net.HttpURLConnection;
   7 import java.net.MalformedURLException;
   8 import java.net.URL;
   9 import java.net.URLConnection;
  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 DataParsing {
  18         private CassandraCreate cc = new CassandraCreate();
  19         private String dirprefix = "http://www.compbio.dundee.ac.uk/www-jpred/results";
  20
  21         public void setDirPrefix (String newprefix) {
  22                 this.dirprefix = newprefix;
  23         }
  24
  25         public void Parsing(String input, int nDays) {
  26                 Calendar cal = Calendar.getInstance();
  27                 cal.add(Calendar.DATE, -nDays);
  28                 for (int i = 0; i < nDays; ++i) {
  29                         cal.add(Calendar.DATE, 1);
  30                         int month = cal.get(Calendar.MONTH) + 1;
  31                         int year = cal.get(Calendar.YEAR);
  32                         int day = cal.get(Calendar.DATE);
  33                         String date = year + "/" + month + "/" + day;
  34                         ParsingForDate(input, date);
  35                 }
  36         }
  37
  38         private void ParsingForDate(String input, String date) {
  39                 int totalcount = 0;
  40                 int countNoData = 0;
  41                 int countUnclearFASTAid = 0;
  42                 int countinsertions = 0;
  43                 int countinserted = 0;
  44                 int counAlignments = 0;
  45                 int countStrange = 0;
  46
  47                 System.out.println("Inserting jobs for " + date);
  48                 try {
  49                         URL url = new URL(input);
  50                         URLConnection conn = url.openConnection();
  51                         BufferedReader alljobs = new BufferedReader(new InputStreamReader(conn.getInputStream()));
  52                         String line;
  53
  54                         while ((line = alljobs.readLine()) != null) {
  55                                 if (line.matches(date + "(.*)jp_[^\\s]+")) {
  56                                         String[] table = line.split("\\s+");
  57                                         String id = table[table.length - 1];
  58                                         totalcount++;
  59                                         if (!cc.CheckID(id)) {
  60                                                 URL urltable = new URL(dirprefix + "/" + id + "/" + id + ".concise.fasta");
  61                                                 HttpURLConnection httpConnection = (HttpURLConnection) urltable.openConnection();
  62                                                 int responsecode = httpConnection.getResponseCode();
  63                                                 if (199 < responsecode && responsecode < 300) {
  64                                                         try {
  65                                                                 final FastaReader fr = new FastaReader(urltable.openStream());
  66                                                                 final List<FastaSequence> seqs = new ArrayList<FastaSequence>();
  67                                                                 String newprotein = "";
  68                                                                 while (fr.hasNext()) {
  69                                                                         final FastaSequence fs = fr.next();
  70                                                                         if (fs.getId().equals("QUERY") || fs.getId().equals(id))
  71                                                                                 newprotein = fs.getSequence().replaceAll("\n", "");
  72                                                                         else
  73                                                                                 seqs.add(fs);
  74                                                                 }
  75                                                                 if (newprotein.equals("")) {
  76                                                                         countUnclearFASTAid++;
  77                                                                 } else {
  78                                                                         SimpleDateFormat formatter = new SimpleDateFormat("yyyy/MM/dd");
  79                                                                         String dateInString1 = table[0].substring(0, table[0].indexOf(":"));
  80                                                                         long dateWork1 = 0;
  81                                                                         try {
  82                                                                                 Date dat1 = formatter.parse(dateInString1);
  83                                                                                 dateWork1 = dat1.getTime();
  84                                                                         } catch (ParseException e) {
  85                                                                                 e.printStackTrace();
  86                                                                         }
  87                                                                         cc.InsertData(dateWork1, table[0], table[1], table[2], id, "OK", "OK", newprotein, seqs);
  88                                                                         ++countinsertions;
  89                                                                         // flush every 100 insertions
  90                                                                         if (0 == countinsertions % 100) {
  91                                                                                 cc.flushData();
  92                                                                         }
  93                                                                 }
  94                                                         } catch (IOException e) {
  95                                                                 e.printStackTrace();
  96                                                         }
  97                                                 } else {
  98                                                         countNoData++;
  99                                                 }
 100                                         } else {
 101                                                 ++countinserted;
 102                                         }
 103                                 } else {
 104                                         if (line.matches(date + "(.*)Sequence0/(.*)")) {
 105                                                 ++counAlignments;
 106                                         } else {
 107                                                 ++countStrange;
 108                                         }
 109                                 }
 110                         }
 111                         alljobs.close();
 112                         System.out.println("Total number of jobs = " + totalcount);
 113                         System.out.println("   " + countinserted + " jobs inserted already");
 114                         System.out.println("   " + counAlignments + " jalview jobs");
 115                         System.out.println("   " + countStrange + " not analysed jobs");
 116                         System.out.println("   " + countNoData + " jobs without *.concise.fasta file");
 117                         System.out.println("   " + countUnclearFASTAid + " jobs with unclear FASTA protein id in *.concise.fasta");
 118                         System.out.println("   " + countinsertions + " new job insertions\n");
 119                 } catch (MalformedURLException e) {
 120                         e.printStackTrace();
 121                 } catch (IOException e) {
 122                         e.printStackTrace();
 123                 }
 124         }
 125 }