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 long ReadDateTable(long queryDate) {
  57                 ResultSet results = session.execute("SELECT Total FROM JobDateInfo WHERE jobday = " + queryDate + ";");
  58                 if (results.isExhausted())
  59                         return 0;
  60                 Row therow = results.one();
  61                 long res = therow.getLong("Total");
  62                 if (!results.isExhausted()) {
  63                         Date date = new Date (queryDate);
  64                         log.warn("CassandraReader.ReadDateTable: date row for " + date.toString () + " ("+ queryDate + ") duplicated ");
  65                 }
  66                 return res;
  67         }
  68         /*
  69          * getting whole protein sequence from the db ProteinRow
  70          */
  71         public List<StructureProteinPrediction> ReadWholeSequence(String queryProtein) {
  72                 final long startTime = System.currentTimeMillis();
  73                 String com = "SELECT JobID, Predictions FROM ProteinRow WHERE Protein = '" + queryProtein + "';";
  74                 System.out.println("Command: " + com);
  75                 ResultSet results = session.execute(com);
  76                 if (results.isExhausted())
  77                         return null;
  78                 final long queryTime = System.currentTimeMillis();
  79                 List<Row> rows = results.all();
  80                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  81                 System.out.println(" rows analysed,  " + rows.size());
  82                 List<StructureProteinPrediction> res = new ArrayList<StructureProteinPrediction>();
  83                 int c = 0;
  84                 for (Row r : rows) {
  85                         StructureProteinPrediction structure = new StructureProteinPrediction(queryProtein, r.getString("JobID"), r.getMap(
  86                                         "Predictions", String.class, String.class));
  87                         res.add(structure);
  88                         ++c;
  89                 }
  90                 final long endTime = System.currentTimeMillis();
  91                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  92                 return res;
  93         }
  94
  95         /*
  96          * getting part of protein sequence from the db ProteinRow
  97          */
  98         public List<StructureProteinPrediction> ReadPartOfSequence(String queryProtein) {
  99                 final long startTime = System.currentTimeMillis();
 100                 String com = "SELECT * FROM ProteinRow;";
 101                 System.out.println("Command: " + com);
 102                 ResultSet results = session.execute(com);
 103                 if (results.isExhausted())
 104                         return null;
 105                 final long queryTime = System.currentTimeMillis();
 106                 List<Row> rows = results.all();
 107                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 108                 System.out.println(" rows analysed,  " + rows.size());
 109                 List<StructureProteinPrediction> res = new ArrayList<StructureProteinPrediction>();
 110                 int c = 0;
 111                 for (Row r : rows) {
 112                         String prot = r.getString("Protein");
 113                         if (prot.matches("(.*)" + queryProtein + "(.*)")) {
 114                                 StructureProteinPrediction structure = new StructureProteinPrediction(prot, r.getString("JobID"), r.getMap("Predictions",
 115                                                 String.class, String.class));
 116                                 res.add(structure);
 117                                 ++c;
 118                         }
 119                 }
 120                 final long endTime = System.currentTimeMillis();
 121                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 122                 return res;
 123         }
 124
 125         /*
 126          * getting protein sequences by counter
 127          */
 128         public Map<String, Integer> ReadProteinSequenceByCounter() {
 129                 final long startTime = System.currentTimeMillis();
 130                 String com = "SELECT Protein, JobID FROM ProteinRow;";
 131                 System.out.println("Command: " + com);
 132                 ResultSet results = session.execute(com);
 133                 if (results.isExhausted())
 134                         return null;
 135                 final long queryTime = System.currentTimeMillis();
 136                 List<Row> rows = results.all();
 137                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 138                 System.out.println(" rows analysed,  " + rows.size());
 139                 Map<String, Integer> res = new HashMap<String, Integer>();
 140                 int c = 0;
 141                 for (Row r : rows) {
 142                         String protein = r.getString("Protein");
 143                         String id = r.getString("JobID");
 144                         System.out.println(id + ", " + protein);
 145                         if (res.containsKey(protein))
 146                                 res.put(protein, res.get(protein) + 1);
 147                         else
 148                                 res.put(protein, 1);
 149                 }
 150                 final long endTime = System.currentTimeMillis();
 151                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 152                 return res;
 153         }
 154
 155         /*
 156          * getting protein sequences by counter
 157          */
 158         public StructureJobLog ReadJobLog(String jobid) {
 159                 final long startTime = System.currentTimeMillis();
 160                 String com = "SELECT * FROM ProteinLog WHERE JobID = '" + jobid + "';";
 161                 System.out.println("Command: " + com);
 162                 ResultSet results = session.execute(com);
 163                 if (results.isExhausted())
 164                         return null;
 165                 final long queryTime = System.currentTimeMillis();
 166                 Row row = results.one();
 167                 String com1 = "SELECT * FROM ProteinRow WHERE JobID = '" + jobid + "' ALLOW FILTERING;";
 168                 System.out.println("Command: " + com1);
 169                 ResultSet results1 = session.execute(com1);
 170                 if (results1.isExhausted())
 171                         return null;
 172                 Row row1 = results1.one();
 173                 StructureJobLog res = new StructureJobLog(row.getString("Protein"), row.getString("JobID"), row.getString("DataBegin"),
 174                                 row.getString("DataEnd"), row.getString("ip"), row1.getMap("Predictions", String.class, String.class));
 175                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 176                 final long endTime = System.currentTimeMillis();
 177                 System.out.println(" rows analysed, execution time is " + (endTime - startTime) + " msec");
 178                 return res;
 179         }
 180 }