datadb/compbio/cassandra/JpredParserHTTP.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 import compbio.cassandra.JpredParser;
  18
  19 public class JpredParserHTTP implements JpredParser {
  20         private CassandraCreate cc = new CassandraCreate();
  21         private String dirprefix;
  22
  23         JpredParserHTTP() {
  24                 dirprefix = "http://www.compbio.dundee.ac.uk/www-jpred/results";
  25         }
  26
  27         JpredParserHTTP(String sourceurl) {
  28                 dirprefix = sourceurl;
  29         }
  30
  31         public void setSource(String newsourceprefix) {
  32                 dirprefix = newsourceprefix;
  33         }
  34
  35         public void Parsing(String source, int nDays) {
  36                 Calendar cal = Calendar.getInstance();
  37                 cal.add(Calendar.DATE, -nDays);
  38                 for (int i = 0; i < nDays; ++i) {
  39                         cal.add(Calendar.DATE, 1);
  40                         int month = cal.get(Calendar.MONTH) + 1;
  41                         int year = cal.get(Calendar.YEAR);
  42                         int day = cal.get(Calendar.DATE);
  43                         String date = year + "/" + month + "/" + day;
  44                         if (0 < ParsingForDate(source, date)) {
  45                                 cc.flushData();
  46                         }
  47                 }
  48         }
  49
  50         private int ParsingForDate(String input, String date) {
  51                 int totalcount = 0;
  52                 int countNoData = 0;
  53                 int countUnclearFASTAid = 0;
  54                 int countinsertions = 0;
  55                 int countinserted = 0;
  56                 int counAlignments = 0;
  57                 int countStrange = 0;
  58                 int njobs = 0;
  59
  60                 System.out.println("Inserting jobs for " + date);
  61                 try {
  62                         URL url = new URL(input);
  63                         URLConnection conn = url.openConnection();
  64                         BufferedReader alljobs = new BufferedReader(new InputStreamReader(conn.getInputStream()));
  65                         String line;
  66
  67                         while ((line = alljobs.readLine()) != null) {
  68                                 if (line.matches(date + "(.*)jp_[^\\s]+")) {
  69                                         String[] table = line.split("\\s+");
  70                                         // Format of a record:
  71                                         // starttime endtime ip email jobid (directory)
  72                                         // 013/10/25:21:55:7 2013/10/25:21:59:13 201.239.98.172 unknown_email jp_J9HBCBT
  73                                         String id = table[table.length - 1];
  74                                         totalcount++;
  75                                         if (!cc.CheckID(id)) {
  76                                                 String datalink = dirprefix + "/" + id + "/" + id + ".concise.fasta";
  77                                                 URL urltable = new URL(datalink);
  78                                                 HttpURLConnection httpConnection = (HttpURLConnection) urltable.openConnection();
  79                                                 int responsecode = httpConnection.getResponseCode();
  80                                                 if (199 < responsecode && responsecode < 300) {
  81                                                         try {
  82                                                                 final FastaReader fr = new FastaReader(urltable.openStream());
  83                                                                 final List<FastaSequence> seqs = new ArrayList<FastaSequence>();
  84                                                                 String newprotein = "";
  85                                                                 while (fr.hasNext()) {
  86                                                                         final FastaSequence fs = fr.next();
  87                                                                         if (fs.getId().equals("QUERY") || fs.getId().equals(id))
  88                                                                                 newprotein = fs.getSequence().replaceAll("\n", "");
  89                                                                         else
  90                                                                                 seqs.add(fs);
  91                                                                 }
  92                                                                 if (newprotein.equals("")) {
  93                                                                         countUnclearFASTAid++;
  94                                                                 } else {
  95                                                                         SimpleDateFormat formatter = new SimpleDateFormat("yyyy/MM/dd");
  96                                                                         String dateInString1 = table[0].substring(0, table[0].indexOf(":"));
  97                                                                         long dateWork1 = 0;
  98                                                                         try {
  99                                                                                 Date dat1 = formatter.parse(dateInString1);
 100                                                                                 dateWork1 = dat1.getTime();
 101                                                                         } catch (ParseException e) {
 102                                                                                 e.printStackTrace();
 103                                                                         }
 104                                                                         cc.InsertData(dateWork1, table[0], table[1], table[2], id, "OK", "OK", newprotein, seqs);
 105                                                                         ++countinsertions;
 106                                                                         ++njobs;
 107                                                                         // flush every 50 insertions
 108                                                                         if (0 == countinsertions % 50) {
 109                                                                                 cc.flushData();
 110                                                                                 njobs -= 50;
 111                                                                         }
 112                                                                 }
 113                                                         } catch (IOException e) {
 114                                                                 e.printStackTrace();
 115                                                         }
 116                                                 } else {
 117                                                         countNoData++;
 118                                                 }
 119                                         } else {
 120                                                 ++countinserted;
 121                                         }
 122                                 } else {
 123                                         if (line.matches(date + "(.*)Sequence0/(.*)")) {
 124                                                 ++counAlignments;
 125                                         } else {
 126                                                 ++countStrange;
 127                                         }
 128                                 }
 129                         }
 130                         alljobs.close();
 131                         System.out.println("Total number of jobs = " + totalcount);
 132                         System.out.println("   " + countinserted + " jobs inserted already");
 133                         System.out.println("   " + counAlignments + " jalview jobs");
 134                         System.out.println("   " + countStrange + " not analysed jobs");
 135                         System.out.println("   " + countNoData + " jobs without *.concise.fasta file");
 136                         System.out.println("   " + countUnclearFASTAid + " jobs with unclear FASTA protein id in *.concise.fasta");
 137                         System.out.println("   " + countinsertions + " new job insertions\n");
 138                 } catch (MalformedURLException e) {
 139                         e.printStackTrace();
 140                 } catch (IOException e) {
 141                         e.printStackTrace();
 142                 }
 143                 return njobs;
 144         }
 145 }