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
  55          */
  56         public List<Pair<String, String>> ReadProteinData(long day) {
  57                 final long startTime = System.currentTimeMillis();
  58                 String com = "SELECT JobID, Protein FROM ProteinData WHERE jobtime = " + day + ";";
  59                 System.out.println("Command: " + com);
  60                 ResultSet results = session.execute(com);
  61                 if (results.isExhausted())
  62                         return null;
  63                 final long queryTime = System.currentTimeMillis();
  64                 List<Row> rows = results.all();
  65                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  66                 List<Pair<String, String>> res = new ArrayList<Pair<String, String>>();
  67                 int c = 0;
  68                 for (Row r : rows) {
  69                         Pair<String, String> pair = new Pair<String, String>(r.getString("JobID"), r.getString("Protein"));
  70                         res.add(pair);
  71                         ++c;
  72                 }
  73                 final long endTime = System.currentTimeMillis();
  74                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  75                 return res;
  76         }
  77         /*
  78          * getting data from the db JobDateInfo
  79          */
  80         public List<Long> ReadDateTable(long queryDate) {
  81                 ResultSet results = session.execute("SELECT * FROM JobDateInfo WHERE jobday = " + queryDate + ";");
  82                 if (results.isExhausted())
  83                         return null;
  84                 Row therow = results.one();
  85                 List<Long> res = new ArrayList<Long>();
  86                 res.add(therow.getLong("Total"));
  87                 res.add(therow.getLong("TotalOK"));
  88                 res.add(therow.getLong("TotalStopped"));
  89                 res.add(therow.getLong("TotalError"));
  90                 res.add(therow.getLong("TotalTimeOut"));
  91                 if (!results.isExhausted()) {
  92                         Date date = new Date (queryDate);
  93                         log.warn("CassandraReader.ReadDateTable: date row for " + date.toString () + " ("+ queryDate + ") duplicated ");
  94                 }
  95                 return res;
  96         }
  97         /*
  98          * getting whole protein sequence from the db ProteinRow
  99          */
 100         public List<AnnotatedProteinSequenceBean> ReadWholeSequence(String queryProtein) {
 101                 final long startTime = System.currentTimeMillis();
 102                 String com = "SELECT JobID, Predictions FROM ProteinRow WHERE Protein = '" + queryProtein + "';";
 103                 System.out.println("Command: " + com);
 104                 ResultSet results = session.execute(com);
 105                 if (results.isExhausted())
 106                         return null;
 107                 final long queryTime = System.currentTimeMillis();
 108                 List<Row> rows = results.all();
 109                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 110                 System.out.println(" rows analysed,  " + rows.size());
 111                 List<AnnotatedProteinSequenceBean> res = new ArrayList<AnnotatedProteinSequenceBean>();
 112                 int c = 0;
 113                 for (Row r : rows) {
 114                         AnnotatedProteinSequenceBean structure = new AnnotatedProteinSequenceBean(queryProtein, r.getString("JobID"), r.getMap(
 115                                         "Predictions", String.class, String.class));
 116                         if (structure.getPrediction().containsKey("jnetpred"))
 117                                 structure.setJnetpred(structure.getPrediction().get("jnetpred"));
 118                         else
 119                                 structure.setJnetpred("");
 120                         res.add(structure);
 121                         ++c;
 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 part of protein sequence from the db ProteinRow
 130          */
 131         public List<AnnotatedProteinSequenceBean> ReadPartOfSequence(String queryProtein) {
 132                 final long startTime = System.currentTimeMillis();
 133                 String com = "SELECT * 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                 List<AnnotatedProteinSequenceBean> res = new ArrayList<AnnotatedProteinSequenceBean>();
 143                 int c = 0;
 144                 for (Row r : rows) {
 145                         String prot = r.getString("Protein");
 146                         if (prot.matches("(.*)" + queryProtein + "(.*)")) {
 147                                 AnnotatedProteinSequenceBean structure = new AnnotatedProteinSequenceBean(prot, r.getString("JobID"), r.getMap("Predictions",
 148                                                 String.class, String.class));
 149                                 res.add(structure);
 150                                 ++c;
 151                         }
 152                 }
 153                 final long endTime = System.currentTimeMillis();
 154                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 155                 return res;
 156         }
 157
 158         /*
 159          * getting protein sequences by counter
 160          */
 161         public Map<String, Integer> ReadProteinSequenceByCounter() {
 162                 final long startTime = System.currentTimeMillis();
 163                 String com = "SELECT Protein, JobID FROM ProteinRow;";
 164                 System.out.println("Command: " + com);
 165                 ResultSet results = session.execute(com);
 166                 if (results.isExhausted())
 167                         return null;
 168                 final long queryTime = System.currentTimeMillis();
 169                 List<Row> rows = results.all();
 170                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 171                 System.out.println(" rows analysed,  " + rows.size());
 172                 Map<String, Integer> res = new HashMap<String, Integer>();
 173                 int c = 0;
 174                 for (Row r : rows) {
 175                         String protein = r.getString("Protein");
 176                         String id = r.getString("JobID");
 177                         if (res.containsKey(protein))
 178                                 res.put(protein, res.get(protein) + 1);
 179                         else
 180                                 res.put(protein, 1);
 181                 }
 182                 final long endTime = System.currentTimeMillis();
 183                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 184                 return res;
 185         }
 186
 187         /*
 188          * getting log info for jobid
 189          */
 190         public StructureJobLog ReadJobLog(String jobid) {
 191                 final long startTime = System.currentTimeMillis();
 192                 String com = "SELECT * FROM ProteinLog WHERE JobID = '" + jobid + "';";
 193                 System.out.println("Command: " + com);
 194                 ResultSet results = session.execute(com);
 195                 if (results.isExhausted())
 196                         return null;
 197                 final long queryTime = System.currentTimeMillis();
 198                 Row row = results.one();
 199                 String com1 = "SELECT * FROM ProteinRow WHERE JobID = '" + jobid + "' ALLOW FILTERING;";
 200                 System.out.println("Command: " + com1);
 201                 ResultSet results1 = session.execute(com1);
 202                 if (results1.isExhausted())
 203                         return null;
 204                 Row row1 = results1.one();
 205                 StructureJobLog res = new StructureJobLog(row.getString("Protein"), row.getString("JobID"), row.getString("DataBegin"),
 206                                 row.getString("DataEnd"), row.getString("ip"), row1.getMap("Predictions", String.class, String.class));
 207                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 208                 final long endTime = System.currentTimeMillis();
 209                 System.out.println(" rows analysed, execution time is " + (endTime - startTime) + " msec");
 210                 return res;
 211         }
 212 }