datadb/compbio/cassandra/CassandraReader.java

   1 package compbio.cassandra;
   2
   3 import java.util.Calendar;
   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 import compbio.engine.ProteoCachePropertyHelperManager;
  16 import compbio.util.PropertyHelper;
  17
  18 public class CassandraReader {
  19         private Session session;
  20         private static Logger log = Logger.getLogger(CassandraNativeConnector.class);
  21
  22         public CassandraReader() {
  23                 Session inis = CassandraNativeConnector.getSession();
  24                 setSession (inis);
  25         }
  26
  27         public void setSession(Session s) {
  28                 assert s != null;
  29                 session = s;
  30         }
  31
  32         /*
  33          * getting data from the db
  34          */
  35         public List<Pair<String, String>> ReadProteinDataTable() {
  36                 final long startTime = System.currentTimeMillis();
  37                 String com = "SELECT DataBegin,DataEnd FROM ProteinLog;";
  38                 System.out.println("Command: " + com);
  39                 ResultSet results = session.execute(com);
  40                 final long queryTime = System.currentTimeMillis();
  41                 List<Row> rows = results.all();
  42                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  43
  44                 List<Pair<String, String>> res = new ArrayList<Pair<String, String>>();
  45                 int c = 0;
  46                 for (Row r : rows) {
  47                         Pair<String, String> pair = new Pair<String, String>(r.getString("DataBegin"), r.getString("DataEnd"));
  48                         res.add(pair);
  49                         ++c;
  50                 }
  51                 final long endTime = System.currentTimeMillis();
  52                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  53                 return res;
  54         }
  55
  56         /*
  57          * getting data from the db ProteinData
  58          */
  59         public Integer ReadDateTable(long queryDate) {
  60                 final long startTime = System.currentTimeMillis();
  61                 String com = "SELECT jobtime, JobID FROM ProteinData WHERE jobtime = " + queryDate + ";";
  62                 System.out.println("Command: " + com);
  63                 ResultSet results = session.execute(com);
  64                 final long queryTime = System.currentTimeMillis();
  65                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  66                 if (results.isExhausted())
  67                         return 0;
  68                 List<Row> rows = results.all();
  69                 final long endTime = System.currentTimeMillis();
  70                 System.out.println("Processing time is " + (endTime - queryTime) + " msec");
  71                 return rows.size();
  72         }
  73
  74         /*
  75          * getting whole protein sequence from the db ProteinRow
  76          */
  77         public List<StructureProteinPrediction> ReadWholeSequence(String queryProtein) {
  78                 final long startTime = System.currentTimeMillis();
  79                 String com = "SELECT JobID, Predictions FROM ProteinRow WHERE Protein = '" + queryProtein + "';";
  80                 System.out.println("Command: " + com);
  81                 ResultSet results = session.execute(com);
  82                 if (results.isExhausted())
  83                         return null;
  84                 final long queryTime = System.currentTimeMillis();
  85                 List<Row> rows = results.all();
  86                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  87                 System.out.println(" rows analysed,  " + rows.size());
  88                 List<StructureProteinPrediction> res = new ArrayList<StructureProteinPrediction>();
  89                 int c = 0;
  90                 for (Row r : rows) {
  91                         StructureProteinPrediction structure = new StructureProteinPrediction(queryProtein, r.getString("JobID"), r.getMap(
  92                                         "Predictions", String.class, String.class));
  93                         res.add(structure);
  94                         ++c;
  95                 }
  96                 final long endTime = System.currentTimeMillis();
  97                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  98                 return res;
  99         }
 100
 101         /*
 102          * getting part of protein sequence from the db ProteinRow
 103          */
 104         public List<StructureProteinPrediction> ReadPartOfSequence(String queryProtein) {
 105                 final long startTime = System.currentTimeMillis();
 106                 String com = "SELECT * FROM ProteinRow;";
 107                 System.out.println("Command: " + com);
 108                 ResultSet results = session.execute(com);
 109                 if (results.isExhausted())
 110                         return null;
 111                 final long queryTime = System.currentTimeMillis();
 112                 List<Row> rows = results.all();
 113                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 114                 System.out.println(" rows analysed,  " + rows.size());
 115                 List<StructureProteinPrediction> res = new ArrayList<StructureProteinPrediction>();
 116                 int c = 0;
 117                 for (Row r : rows) {
 118                         String prot = r.getString("Protein");
 119                         if (prot.matches("(.*)" + queryProtein + "(.*)")) {
 120                                 StructureProteinPrediction structure = new StructureProteinPrediction(prot, r.getString("JobID"), r.getMap("Predictions",
 121                                                 String.class, String.class));
 122                                 res.add(structure);
 123                                 ++c;
 124                         }
 125                 }
 126                 final long endTime = System.currentTimeMillis();
 127                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 128                 return res;
 129         }
 130
 131         /*
 132          * getting protein sequences by counter
 133          */
 134         public Map<String, Integer> ReadProteinDataByCounter() {
 135                 final long startTime = System.currentTimeMillis();
 136                 String com = "SELECT Protein FROM ProteinRow;";
 137                 System.out.println("Command: " + com);
 138                 ResultSet results = session.execute(com);
 139                 if (results.isExhausted())
 140                         return null;
 141                 final long queryTime = System.currentTimeMillis();
 142                 List<Row> rows = results.all();
 143                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 144                 System.out.println(" rows analysed,  " + rows.size());
 145                 Map<String, Integer> res = new HashMap<String, Integer>();
 146                 int c = 0;
 147                 for (Row r : rows) {
 148                         String protein = r.getString("Protein");
 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 }