webservices/compbio/nosql/cassandra/CassandraCreate.java

   1 package combio.cassandra;
   2
   3 import java.util.Arrays;
   4
   5 import me.prettyprint.cassandra.serializers.LongSerializer;
   6 import me.prettyprint.cassandra.serializers.StringSerializer;
   7 import me.prettyprint.cassandra.service.ThriftKsDef;
   8 import me.prettyprint.hector.api.Cluster;
   9 import me.prettyprint.hector.api.Keyspace;
  10 import me.prettyprint.hector.api.beans.ColumnSlice;
  11 import me.prettyprint.hector.api.ddl.ColumnFamilyDefinition;
  12 import me.prettyprint.hector.api.ddl.ComparatorType;
  13 import me.prettyprint.hector.api.ddl.KeyspaceDefinition;
  14 import me.prettyprint.hector.api.factory.HFactory;
  15 import me.prettyprint.hector.api.mutation.Mutator;
  16 import me.prettyprint.hector.api.query.QueryResult;
  17 import me.prettyprint.hector.api.query.SliceQuery;
  18
  19 public class CassandraCreate {
  20         private static Keyspace ksp;
  21         private static Cluster cluster;
  22         private static Mutator<Long> mutatorLong;
  23         private static Mutator<String> mutatorString;
  24         private static Mutator<String> mutatorLog;
  25         StringSerializer ss = StringSerializer.get();
  26         LongSerializer ls = LongSerializer.get();
  27
  28
  29         // connect to cluster
  30         public void Connection() {
  31                 cluster = HFactory.getOrCreateCluster("Protein Cluster", "127.0.0.1:9160");
  32                 KeyspaceDefinition keyspaceDef = cluster.describeKeyspace("ProteinKeyspace");
  33                 // If keyspace does not exist, the CFs don't exist either. => create them.
  34 /*              if (keyspaceDef != null) {
  35                         cluster.dropColumnFamily("ProteinKeyspace", "ProteinRow", true);
  36                         cluster.dropColumnFamily("ProteinKeyspace", "ProteinData", true);
  37                         cluster.dropKeyspace("ProteinKeyspace", true);
  38                         System.out.println("ProteinKeyspace has been dropped");
  39         } else*/ if (keyspaceDef == null) {     // create column family
  40                         System.out.println("ProteinKeyspace has been null");
  41                         ColumnFamilyDefinition cfProtein = HFactory.createColumnFamilyDefinition("ProteinKeyspace", "ProteinRow", ComparatorType.ASCIITYPE);
  42                         ColumnFamilyDefinition cfLog = HFactory.createColumnFamilyDefinition("ProteinKeyspace", "ProteinLog", ComparatorType.ASCIITYPE);
  43                         ColumnFamilyDefinition cfData = HFactory.createColumnFamilyDefinition("ProteinKeyspace", "ProteinData", ComparatorType.ASCIITYPE);
  44
  45                         KeyspaceDefinition newKeyspace = HFactory.createKeyspaceDefinition("ProteinKeyspace",
  46                                         ThriftKsDef.DEF_STRATEGY_CLASS, 1, Arrays.asList(cfProtein, cfLog, cfData));
  47                         //Add the schema to the cluster.
  48                         //"true" as the second param means that Hector will block until all nodes see the change.
  49                         cluster.addKeyspace(newKeyspace, true);
  50                         cluster.addColumnFamily(cfProtein, true);
  51                         cluster.addColumnFamily(cfLog, true);
  52                         cluster.addColumnFamily(cfData, true);
  53                 }
  54                 ksp = HFactory.createKeyspace("ProteinKeyspace", cluster);
  55                 System.out.println("Cassantra has been connected");
  56         }
  57
  58         // parsing data from http://www.compbio.dundee.ac.uk/www-jpred/results/usage-new/alljobs.dat
  59         public void Parsing() {
  60                 mutatorString = HFactory.createMutator(ksp, ss); // CF ProteinRow store protein and prediction
  61                 mutatorLog = HFactory.createMutator(ksp, ss); // CF ProteinLog store log informations (ip, id, dates start and dates of end)
  62                 mutatorLong = HFactory.createMutator(ksp, ls); // CF ProteinData store id and protein per data
  63                 System.out.println("Parsing......");
  64                 String in = "http://www.compbio.dundee.ac.uk/www-jpred/results/usage-new/alljobs.dat";
  65                 DataParsing datParsing = new DataParsing();
  66                 datParsing.ParsingTest(in);
  67                 mutatorString.execute();
  68                 mutatorLong.execute();
  69                 mutatorLog.execute();
  70                 System.out.println("Data Inserted");
  71         }
  72
  73         public void Closing() {
  74                 cluster.getConnectionManager().shutdown();
  75                 System.out.println("Cassantra has been closed");
  76         }
  77
  78         // check whether this id exists in the cassandra DB
  79         public boolean CheckIP(String ip) {
  80                 SliceQuery<String, String, String> sliceQuery = HFactory.createSliceQuery(ksp, ss, ss, ss);
  81                 sliceQuery.setColumnFamily("ProteinLog").setKey(ip).setRange("", "", false, 100);
  82                 QueryResult<ColumnSlice<String, String>> result = sliceQuery.execute();
  83                 if (result.get().getColumns().size() > 0)
  84                         return true;
  85                 else
  86                         return false;
  87         }
  88
  89         public void InsertData(long dataWork, String dataBegin, String dataEnd, String ip, String id, String statusEx, String statusFinal, String protein, String jnetpred) {
  90                 mutatorLog.addInsertion(id, "ProteinLog", HFactory.createColumn("ip", ip, ss, ss))
  91                 .addInsertion(id, "ProteinLog", HFactory.createColumn("DataBegin", dataBegin, ss, ss))
  92                 .addInsertion(id, "ProteinLog", HFactory.createColumn("DataEnd", dataEnd, ss, ss))
  93                 .addInsertion(id, "ProteinLog", HFactory.createColumn("Status ex", statusEx, ss, ss))
  94                 .addInsertion(id, "ProteinLog", HFactory.createColumn("Status final", statusFinal, ss, ss))
  95                 .addInsertion(id, "ProteinLog", HFactory.createColumn("Protein", protein, ss, ss));
  96                 mutatorString.addInsertion(protein, "ProteinRow", HFactory.createColumn(id, jnetpred, ss, ss));
  97                 mutatorLong.addInsertion(dataWork, "ProteinData", HFactory.createColumn(id, protein, ss, ss));
  98           }
  99
 100         public Keyspace GetKeyspace() {
 101                 return ksp;
 102         }
 103 }