datadb/compbio/cassandra/JpredParserLocalFile.java

   1 package compbio.cassandra;
   2
   3 import java.io.BufferedReader;
   4 import java.io.File;
   5 import java.io.FileNotFoundException;
   6 import java.io.IOException;
   7 import java.io.InputStreamReader;
   8 import java.net.HttpURLConnection;
   9 import java.net.MalformedURLException;
  10 import java.io.FileInputStream;
  11 import java.text.ParseException;
  12 import java.text.SimpleDateFormat;
  13 import java.util.ArrayList;
  14 import java.util.Calendar;
  15 import java.util.Date;
  16 import java.util.List;
  17
  18 import compbio.data.sequence.FastaReader;
  19 import compbio.data.sequence.FastaSequence;
  20
  21 public class JpredParserLocalFile implements JpredParser {
  22         private CassandraWriter cw = new CassandraWriter();
  23         private String dirprefix;
  24
  25         public void setSource(String newsourceprefix) {
  26                 this.dirprefix = newsourceprefix;
  27         }
  28
  29         public JpredParserLocalFile() {
  30                 this.dirprefix = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/data.dat";
  31         }
  32
  33         public JpredParserLocalFile(String sourceurl) {
  34                 this.dirprefix = sourceurl;
  35         }
  36
  37         public void Parsing(String source, int nDays) throws IOException {
  38                 Calendar cal = Calendar.getInstance();
  39                 cal.add(Calendar.DATE, -nDays);
  40                 List<String> alljobs = new ArrayList<String>();
  41                 File file = new File(source);
  42                 BufferedReader alljobsfile = new BufferedReader(new InputStreamReader(new FileInputStream(file)));
  43                 String line;
  44
  45                 while ((line = alljobsfile.readLine()) != null) {
  46                         alljobs.add(line);
  47                 }
  48                 alljobsfile.close();
  49
  50                 System.out.println("Inserting jobs for " + nDays + " days, " + alljobs.size() + " jobs in total");
  51                 final long startTime = System.currentTimeMillis();
  52                 for (int i = 0; i < nDays; ++i) {
  53                         cal.add(Calendar.DATE, 1);
  54                         int month = cal.get(Calendar.MONTH) + 1;
  55                         int year = cal.get(Calendar.YEAR);
  56                         int day = cal.get(Calendar.DATE);
  57                         String date = year + "/" + month + "/" + day;
  58                         ParsingForDate(alljobs, date);
  59                 }
  60                 final long execTime = System.currentTimeMillis() - startTime;
  61                 System.out.println("Execution Time = " + execTime + " ms");
  62         }
  63
  64         private void ParsingForDate(List<String> input, String date) {
  65                 int totalcount = 0;
  66                 int countNoData = 0;
  67                 int countUnclearFASTAid = 0;
  68                 int countinsertions = 0;
  69                 int countinserted = 0;
  70                 int counAlignments = 0;
  71                 int countStrange = 0;
  72
  73                 System.out.println("Inserting jobs for " + date);
  74                 for (String in : input) {
  75                         if (in.matches(date + ":(.*)jp_[^\\s]+")) {
  76                                 String[] table = in.split("\\s+");
  77                                 String starttime = table[0];
  78                                 String finishtime = table[1];
  79                                 String ip = table[2];
  80                                 String id = table[table.length - 1];
  81                                 totalcount++;
  82                                 String confilename = dirprefix + "/" + id + "/" + id + ".concise";
  83                                 File confile = new File(confilename);
  84                                 if (confile.exists()) {
  85                                         try {
  86                                                 final FastaReader fr = new FastaReader(confilename);
  87                                                 final List<FastaSequence> seqs = new ArrayList<FastaSequence>();
  88                                                 String newprotein = "";
  89                                                 while (fr.hasNext()) {
  90                                                         final FastaSequence fs = fr.next();
  91                                                         if (fs.getId().equals("QUERY") || fs.getId().equals(id))
  92                                                                 newprotein = fs.getSequence().replaceAll("\n", "");
  93                                                         else if (fs.getId().equals("jnetpred") || fs.getId().equals("JNETPRED")) {
  94                                                                 seqs.add(fs);
  95                                                         }
  96                                                 }
  97                                                 if (newprotein.equals("")) {
  98                                                         countUnclearFASTAid++;
  99                                                 } else {
 100                                                         SimpleDateFormat formatter = new SimpleDateFormat("yyyy/MM/dd");
 101                                                         String dateInString1 = starttime.substring(0, starttime.indexOf(":"));
 102                                                         long insertdate = 0;
 103                                                         try {
 104                                                                 Date dat = formatter.parse(dateInString1);
 105                                                                 insertdate = dat.getTime();
 106                                                         } catch (ParseException e) {
 107                                                                 e.printStackTrace();
 108                                                         }
 109                                                         // countinsertions += cw.FormQueryTables(insertdate,
 110                                                         // starttime, finishtime, ip, id, "OK", "OK",
 111                                                         // newprotein, seqs);
 112                                                 }
 113                                                 fr.close();
 114                                         } catch (IOException e) {
 115                                                 e.printStackTrace();
 116                                         }
 117                                 } else {
 118                                         countNoData++;
 119                                 }
 120                         } else {
 121                                 if (in.matches(date + "(.*)Sequence0/(.*)")) {
 122                                         ++counAlignments;
 123                                 } else {
 124                                         ++countStrange;
 125                                 }
 126                         }
 127                 }
 128                 if (true) {
 129                         System.out.println("Total number of jobs = " + totalcount);
 130                         System.out.println("   " + countinserted + " jobs inserted already");
 131                         System.out.println("   " + counAlignments + " jalview jobs");
 132                         System.out.println("   " + countStrange + " not analysed jobs");
 133                         System.out.println("   " + countNoData + " jobs without *.concise.fasta file");
 134                         System.out.println("   " + countUnclearFASTAid + " jobs with unclear FASTA protein id in *.concise.fasta");
 135                         System.out.println("   " + countinsertions + " new job insertions\n");
 136                 }
 137         }
 138
 139 }