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 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 JobDateInfo
  58          */
  59         public long ReadDateTable(long queryDate) {
  60                 ResultSet results = session.execute("SELECT Total FROM JobDateInfo WHERE jobday = " + queryDate + ";");
  61                 if (results.isExhausted())
  62                         return 0;
  63                 Row therow = results.one();
  64                 long res = therow.getLong("Total");
  65                 if (!results.isExhausted()) {
  66                         Date date = new Date (queryDate);
  67                         log.warn("CassandraReader.ReadDateTable: date row for " + date.toString () + " ("+ queryDate + ") duplicated ");
  68                 }
  69                 return res;
  70         }
  71         /*
  72          * getting whole protein sequence from the db ProteinRow
  73          */
  74         public List<StructureProteinPrediction> ReadWholeSequence(String queryProtein) {
  75                 final long startTime = System.currentTimeMillis();
  76                 String com = "SELECT JobID, Predictions FROM ProteinRow WHERE Protein = '" + queryProtein + "';";
  77                 System.out.println("Command: " + com);
  78                 ResultSet results = session.execute(com);
  79                 if (results.isExhausted())
  80                         return null;
  81                 final long queryTime = System.currentTimeMillis();
  82                 List<Row> rows = results.all();
  83                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  84                 System.out.println(" rows analysed,  " + rows.size());
  85                 List<StructureProteinPrediction> res = new ArrayList<StructureProteinPrediction>();
  86                 int c = 0;
  87                 for (Row r : rows) {
  88                         StructureProteinPrediction structure = new StructureProteinPrediction(queryProtein, r.getString("JobID"), r.getMap(
  89                                         "Predictions", String.class, String.class));
  90                         res.add(structure);
  91                         ++c;
  92                 }
  93                 final long endTime = System.currentTimeMillis();
  94                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  95                 return res;
  96         }
  97
  98         /*
  99          * getting part of protein sequence from the db ProteinRow
 100          */
 101         public List<StructureProteinPrediction> ReadPartOfSequence(String queryProtein) {
 102                 final long startTime = System.currentTimeMillis();
 103                 String com = "SELECT * FROM ProteinRow;";
 104                 System.out.println("Command: " + com);
 105                 ResultSet results = session.execute(com);
 106                 if (results.isExhausted())
 107                         return null;
 108                 final long queryTime = System.currentTimeMillis();
 109                 List<Row> rows = results.all();
 110                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 111                 System.out.println(" rows analysed,  " + rows.size());
 112                 List<StructureProteinPrediction> res = new ArrayList<StructureProteinPrediction>();
 113                 int c = 0;
 114                 for (Row r : rows) {
 115                         String prot = r.getString("Protein");
 116                         if (prot.matches("(.*)" + queryProtein + "(.*)")) {
 117                                 StructureProteinPrediction structure = new StructureProteinPrediction(prot, r.getString("JobID"), r.getMap("Predictions",
 118                                                 String.class, String.class));
 119                                 res.add(structure);
 120                                 ++c;
 121                         }
 122                 }
 123                 final long endTime = System.currentTimeMillis();
 124                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 125                 return res;
 126         }
 127
 128         /*
 129          * getting protein sequences by counter
 130          */
 131         public Map<String, Integer> ReadProteinSequenceByCounter() {
 132                 final long startTime = System.currentTimeMillis();
 133                 String com = "SELECT Protein FROM ProteinRow;";
 134                 System.out.println("Command: " + com);
 135                 ResultSet results = session.execute(com);
 136                 if (results.isExhausted())
 137                         return null;
 138                 final long queryTime = System.currentTimeMillis();
 139                 List<Row> rows = results.all();
 140                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 141                 System.out.println(" rows analysed,  " + rows.size());
 142                 Map<String, Integer> res = new HashMap<String, Integer>();
 143                 int c = 0;
 144                 for (Row r : rows) {
 145                         String protein = r.getString("Protein");
 146                         if (res.containsKey(protein))
 147                                 res.put(protein, res.get(protein) + 1);
 148                         else
 149                                 res.put(protein, 1);
 150                 }
 151                 final long endTime = System.currentTimeMillis();
 152                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 153                 return res;
 154         }
 155
 156         /*
 157          * getting protein sequences by counter
 158          */
 159         public StructureJobLog ReadJobLog(String jobid) {
 160                 final long startTime = System.currentTimeMillis();
 161                 String com = "SELECT * FROM ProteinLog WHERE JobID = '" + jobid + "';";
 162                 System.out.println("Command: " + com);
 163                 ResultSet results = session.execute(com);
 164                 if (results.isExhausted())
 165                         return null;
 166                 final long queryTime = System.currentTimeMillis();
 167                 Row row = results.one();
 168                 String com1 = "SELECT * FROM ProteinRow WHERE JobID = '" + jobid + "' ALLOW FILTERING;";
 169                 System.out.println("Command: " + com1);
 170                 ResultSet results1 = session.execute(com1);
 171                 if (results1.isExhausted())
 172                         return null;
 173                 Row row1 = results1.one();
 174                 StructureJobLog res = new StructureJobLog(row.getString("Protein"), row.getString("JobID"), row.getString("DataBegin"),
 175                                 row.getString("DataEnd"), row.getString("ip"), row1.getMap("Predictions", String.class, String.class));
 176                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 177                 final long endTime = System.currentTimeMillis();
 178                 System.out.println(" rows analysed, execution time is " + (endTime - startTime) + " msec");
 179                 return res;
 180         }
 181 }