datadb/compbio/cassandra/CassandraNativeConnector.java

   1 package compbio.cassandra;
   2
   3 import java.io.IOException;
   4 import java.util.List;
   5
   6 import com.datastax.driver.core.Cluster;
   7 import com.datastax.driver.core.Host;
   8 import com.datastax.driver.core.Metadata;
   9 import com.datastax.driver.core.Session;
  10 import com.datastax.driver.core.ResultSet;
  11
  12 public class CassandraNativeConnector {
  13         private static Cluster cluster;
  14         private static Session session;
  15
  16         /*
  17          * private static Keyspace ksp; private static Mutator<Long> mutatorLong;
  18          * private static Mutator<String> mutatorString; private static
  19          * Mutator<String> mutatorLog; StringSerializer ss = StringSerializer.get();
  20          * LongSerializer ls = LongSerializer.get();
  21          */
  22
  23         /*
  24          * connect to the cluster and look weather the dababase has any data inside
  25          */
  26         public void Connect() {
  27                 // local cassandra cluster
  28                 cluster = Cluster.builder().addContactPoint("localhost").build();
  29                 // distributed cassandra cluster
  30                 /* cluster = Cluster.builder().addContactPoint("10.0.115.190").build(); */
  31                 Metadata metadata = cluster.getMetadata();
  32                 System.out.printf("Connected to cluster: %s\n", metadata.getClusterName());
  33                 for (Host host : metadata.getAllHosts()) {
  34                         System.out.printf("Datatacenter: %s; Host: %s; Rack: %s\n", host.getDatacenter(), host.getAddress(), host.getRack());
  35                 }
  36
  37                 session = cluster.connect();
  38                 session.execute("CREATE KEYSPACE IF NOT EXISTS ProteinKeyspace WITH replication = {'class':'SimpleStrategy', 'replication_factor':3};");
  39                 session.execute("CREATE COLUMNFAMILY IF NOT EXISTS ProteinKeyspace.ProteinRow (Protein ascii, JobID ascii, Predictions map<ascii,ascii>, PRIMARY KEY(JobID));");
  40                 session.execute("CREATE COLUMNFAMILY IF NOT EXISTS ProteinKeyspace.ProteinLog "
  41                                 + "(JobID ascii, DataBegin ascii, DataEnd ascii, ip ascii, FinalStatus ascii, ExecutionStatus ascii, Protein ascii, PRIMARY KEY(JobID));");
  42                 session.execute("CREATE COLUMNFAMILY IF NOT EXISTS ProteinKeyspace.ProteinData (jobtime bigint, JobID ascii, Protein ascii, PRIMARY KEY(JobID));");
  43
  44                 session.execute("CREATE INDEX ProteinSeq ON ProteinKeyspace.ProteinRow (protein);");
  45                 session.execute("CREATE INDEX JobDateStamp ON ProteinKeyspace.ProteinData (jobtime);");
  46
  47                 System.out.println("Cassandra connected");
  48         }
  49
  50         /*
  51          * parsing data source and filling the database
  52          */
  53         public void Parsing() throws IOException {
  54                 if (false) {
  55                         // if (source.equals("http")) {
  56                         // get data from real Jpred production server
  57                         System.out.println("Parsing web data source......");
  58                         String datasrc = "http://www.compbio.dundee.ac.uk/www-jpred/results/usage-new/alljobs.dat";
  59                         String prefix = "http://www.compbio.dundee.ac.uk/www-jpred/results";
  60                         JpredParserHTTP parser = new JpredParserHTTP(prefix);
  61                         parser.Parsing(datasrc, 4);
  62                 }
  63                 if (true) {
  64                         // if (source.equals("file")) {
  65                         // get irtifical data generated for the DB stress tests
  66                         System.out.println("Parsing local file data source......");
  67                         String datasrc = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/data.dat";
  68                         String prefix = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/Jpreddata";
  69                         JpredParserLocalFile parser = new JpredParserLocalFile(prefix);
  70                         parser.Parsing(datasrc, 190);
  71                 }
  72         }
  73
  74         public void Closing() {
  75                 cluster.shutdown();
  76                 System.out.println("Cassandra has been shut down");
  77         }
  78
  79         /*
  80          * prepare data for insertion into the db
  81          */
  82         public void InsertData(long jobtime, String startdate, String enddate, String ip, String jobid, String statusEx, String statusFinal,
  83                         String protein, List<FastaSequence> predictions) {
  84
  85                 String check1 = "SELECT * FROM ProteinKeyspace.ProteinLog WHERE JobID = '" + jobid + "';";
  86                 //System.out.println(check1);
  87                 ResultSet results1 = session.execute(check1);
  88                 if (results1.isExhausted()) {
  89                         String com1 = "INSERT INTO ProteinKeyspace.ProteinLog "
  90                                         + "(JobID, IP, DataBegin, DataEnd, FinalStatus, ExecutionStatus, Protein)" + " VALUES ('" + jobid + "','" + ip + "','"
  91                                         + startdate + "','" + enddate + "','" + statusFinal + "','" + statusEx + "','" + protein + "');";
  92                         //System.out.println(com1);
  93                         session.execute(com1);
  94
  95                         String com2 = "INSERT INTO ProteinKeyspace.ProteinData " + "(jobtime, JobID, Protein)" + " VALUES (" + jobtime + ",'" + jobid
  96                                         + "','" + protein + "');";
  97                         //System.out.println(com2);
  98                         session.execute(com2);
  99
 100                         String allpredictions = "";
 101                         for (FastaSequence pred : predictions) {
 102                                 String predictionname = pred.getId();
 103                                 String prediction = pred.getSequence().replaceAll("\n", "");
 104                                 allpredictions += "'" + predictionname + "':'" + prediction + "',";
 105                         }
 106                         String final_prediction = "";
 107                         if (null != allpredictions) {
 108                                 final_prediction = allpredictions.substring(0, allpredictions.length() - 1);
 109                         }
 110
 111                         String check2 = "SELECT * FROM ProteinKeyspace.ProteinRow WHERE JobID = '" + jobid + "';";
 112                         //System.out.println(check2);
 113                         ResultSet results2 = session.execute(check2);
 114                         if (results2.isExhausted()) {
 115                                 String com3 = "INSERT INTO ProteinKeyspace.ProteinRow " + "(Protein, JobID, Predictions)" + " VALUES ('"
 116                         + protein + "','" + jobid + "',{" + final_prediction + "});";
 117                                 //System.out.println(com3);
 118                                 session.execute(com3);
 119                         }
 120                 }
 121         }
 122
 123 }