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