datadb/compbio/cassandra/CassandraReader.java

   1 package compbio.cassandra;
   2
   3 import java.util.Date;
   4 import java.util.HashMap;
   5 import java.util.List;
   6 import java.util.ArrayList;
   7 import java.util.Map;
   8
   9 import org.apache.log4j.Logger;
  10
  11 import com.datastax.driver.core.Row;
  12 import com.datastax.driver.core.Session;
  13 import com.datastax.driver.core.ResultSet;
  14
  15 public class CassandraReader {
  16         private Session session;
  17         private static Logger log = Logger.getLogger(CassandraNativeConnector.class);
  18
  19         public CassandraReader() {
  20                 Session inis = CassandraNativeConnector.getSession();
  21                 setSession (inis);
  22         }
  23
  24         public void setSession(Session s) {
  25                 assert s != null;
  26                 session = s;
  27         }
  28
  29         /*
  30          * getting data from the db
  31          */
  32         public List<Pair<String, String>> ReadProteinDataTable() {
  33                 final long startTime = System.currentTimeMillis();
  34                 String com = "SELECT DataBegin,DataEnd FROM ProteinLog;";
  35                 System.out.println("Command: " + com);
  36                 ResultSet results = session.execute(com);
  37                 final long queryTime = System.currentTimeMillis();
  38                 List<Row> rows = results.all();
  39                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  40
  41                 List<Pair<String, String>> res = new ArrayList<Pair<String, String>>();
  42                 int c = 0;
  43                 for (Row r : rows) {
  44                         Pair<String, String> pair = new Pair<String, String>(r.getString("DataBegin"), r.getString("DataEnd"));
  45                         res.add(pair);
  46                         ++c;
  47                 }
  48                 final long endTime = System.currentTimeMillis();
  49                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  50                 return res;
  51         }
  52
  53         /*
  54          * getting data from the db JobDateInfo
  55          */
  56         public List<Long> ReadDateTable(long queryDate) {
  57                 ResultSet results = session.execute("SELECT * FROM JobDateInfo WHERE jobday = " + queryDate + ";");
  58                 if (results.isExhausted())
  59                         return null;
  60                 Row therow = results.one();
  61                 List<Long> res = new ArrayList<Long>();
  62                 res.add(therow.getLong("Total"));
  63                 res.add(therow.getLong("TotalOK"));
  64                 res.add(therow.getLong("TotalStopped"));
  65                 res.add(therow.getLong("TotalError"));
  66                 res.add(therow.getLong("TotalTimeOut"));
  67                 if (!results.isExhausted()) {
  68                         Date date = new Date (queryDate);
  69                         log.warn("CassandraReader.ReadDateTable: date row for " + date.toString () + " ("+ queryDate + ") duplicated ");
  70                 }
  71                 return res;
  72         }
  73         /*
  74          * getting whole protein sequence from the db ProteinRow
  75          */
  76         public List<AnnotatedProteinSequenceBean> ReadWholeSequence(String queryProtein) {
  77                 final long startTime = System.currentTimeMillis();
  78                 String com = "SELECT JobID, Predictions FROM ProteinRow WHERE Protein = '" + queryProtein + "';";
  79                 System.out.println("Command: " + com);
  80                 ResultSet results = session.execute(com);
  81                 if (results.isExhausted())
  82                         return null;
  83                 final long queryTime = System.currentTimeMillis();
  84                 List<Row> rows = results.all();
  85                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  86                 System.out.println(" rows analysed,  " + rows.size());
  87                 List<AnnotatedProteinSequenceBean> res = new ArrayList<AnnotatedProteinSequenceBean>();
  88                 int c = 0;
  89                 for (Row r : rows) {
  90                         AnnotatedProteinSequenceBean structure = new AnnotatedProteinSequenceBean(queryProtein, r.getString("JobID"), r.getMap(
  91                                         "Predictions", String.class, String.class));
  92                         res.add(structure);
  93                         ++c;
  94                 }
  95                 final long endTime = System.currentTimeMillis();
  96                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  97                 return res;
  98         }
  99
 100         /*
 101          * getting part of protein sequence from the db ProteinRow
 102          */
 103         public List<AnnotatedProteinSequenceBean> ReadPartOfSequence(String queryProtein) {
 104                 final long startTime = System.currentTimeMillis();
 105                 String com = "SELECT * FROM ProteinRow;";
 106                 System.out.println("Command: " + com);
 107                 ResultSet results = session.execute(com);
 108                 if (results.isExhausted())
 109                         return null;
 110                 final long queryTime = System.currentTimeMillis();
 111                 List<Row> rows = results.all();
 112                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 113                 System.out.println(" rows analysed,  " + rows.size());
 114                 List<AnnotatedProteinSequenceBean> res = new ArrayList<AnnotatedProteinSequenceBean>();
 115                 int c = 0;
 116                 for (Row r : rows) {
 117                         String prot = r.getString("Protein");
 118                         if (prot.matches("(.*)" + queryProtein + "(.*)")) {
 119                                 AnnotatedProteinSequenceBean structure = new AnnotatedProteinSequenceBean(prot, r.getString("JobID"), r.getMap("Predictions",
 120                                                 String.class, String.class));
 121                                 res.add(structure);
 122                                 ++c;
 123                         }
 124                 }
 125                 final long endTime = System.currentTimeMillis();
 126                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 127                 return res;
 128         }
 129
 130         /*
 131          * getting protein sequences by counter
 132          */
 133         public Map<String, Integer> ReadProteinSequenceByCounter() {
 134                 final long startTime = System.currentTimeMillis();
 135                 String com = "SELECT Protein, JobID FROM ProteinRow;";
 136                 System.out.println("Command: " + com);
 137                 ResultSet results = session.execute(com);
 138                 if (results.isExhausted())
 139                         return null;
 140                 final long queryTime = System.currentTimeMillis();
 141                 List<Row> rows = results.all();
 142                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 143                 System.out.println(" rows analysed,  " + rows.size());
 144                 Map<String, Integer> res = new HashMap<String, Integer>();
 145                 int c = 0;
 146                 for (Row r : rows) {
 147                         String protein = r.getString("Protein");
 148                         String id = r.getString("JobID");
 149                         if (res.containsKey(protein))
 150                                 res.put(protein, res.get(protein) + 1);
 151                         else
 152                                 res.put(protein, 1);
 153                 }
 154                 final long endTime = System.currentTimeMillis();
 155                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 156                 return res;
 157         }
 158
 159         /*
 160          * getting protein sequences by counter
 161          */
 162         public StructureJobLog ReadJobLog(String jobid) {
 163                 final long startTime = System.currentTimeMillis();
 164                 String com = "SELECT * FROM ProteinLog WHERE JobID = '" + jobid + "';";
 165                 System.out.println("Command: " + com);
 166                 ResultSet results = session.execute(com);
 167                 if (results.isExhausted())
 168                         return null;
 169                 final long queryTime = System.currentTimeMillis();
 170                 Row row = results.one();
 171                 String com1 = "SELECT * FROM ProteinRow WHERE JobID = '" + jobid + "' ALLOW FILTERING;";
 172                 System.out.println("Command: " + com1);
 173                 ResultSet results1 = session.execute(com1);
 174                 if (results1.isExhausted())
 175                         return null;
 176                 Row row1 = results1.one();
 177                 StructureJobLog res = new StructureJobLog(row.getString("Protein"), row.getString("JobID"), row.getString("DataBegin"),
 178                                 row.getString("DataEnd"), row.getString("ip"), row1.getMap("Predictions", String.class, String.class));
 179                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 180                 final long endTime = System.currentTimeMillis();
 181                 System.out.println(" rows analysed, execution time is " + (endTime - startTime) + " msec");
 182                 return res;
 183         }
 184 }