datadb/compbio/cassandra/CassandraNativeConnector.java

   1 package compbio.cassandra;
   2
   3 import java.io.IOException;
   4 import java.util.Calendar;
   5 import java.util.HashMap;
   6 import java.util.List;
   7 import java.util.ArrayList;
   8
   9 import com.datastax.driver.core.Cluster;
  10 import com.datastax.driver.core.Host;
  11 import com.datastax.driver.core.Metadata;
  12 import com.datastax.driver.core.Row;
  13 import com.datastax.driver.core.Session;
  14 import com.datastax.driver.core.ResultSet;
  15
  16 public class CassandraNativeConnector {
  17         private static Cluster cluster;
  18         private static Session session;
  19         /*
  20          * connect to the cluster and look weather the dababase has any data inside
  21          */
  22         public void Connect() {
  23                 // local cassandra cluster
  24                 cluster = Cluster.builder().addContactPoint("localhost").build();
  25                 // distributed cassandra cluster
  26                 /* cluster = Cluster.builder().addContactPoint("10.0.115.190").build(); */
  27                 Metadata metadata = cluster.getMetadata();
  28                 System.out.printf("Connected to cluster: %s\n", metadata.getClusterName());
  29                 for (Host host : metadata.getAllHosts()) {
  30                         System.out.printf("Datatacenter: %s; Host: %s; Rack: %s\n", host.getDatacenter(), host.getAddress(), host.getRack());
  31                 }
  32
  33                 session = cluster.connect();
  34                 session.execute("CREATE KEYSPACE IF NOT EXISTS ProteinKeyspace WITH replication = {'class':'SimpleStrategy', 'replication_factor':3};");
  35                 session.execute("CREATE COLUMNFAMILY IF NOT EXISTS ProteinKeyspace.ProteinRow (Protein ascii, JobID ascii, Predictions map<ascii,ascii>, PRIMARY KEY(JobID));");
  36                 session.execute("CREATE COLUMNFAMILY IF NOT EXISTS ProteinKeyspace.ProteinLog "
  37                                 + "(JobID ascii, DataBegin ascii, DataEnd ascii, ip ascii, FinalStatus ascii, ExecutionStatus ascii, Protein ascii, PRIMARY KEY(JobID));");
  38                 session.execute("CREATE COLUMNFAMILY IF NOT EXISTS ProteinKeyspace.ProteinData (jobtime bigint, JobID ascii, Protein ascii, PRIMARY KEY(JobID));");
  39
  40                 session.execute("CREATE INDEX IF NOT EXISTS ProteinSeq ON ProteinKeyspace.ProteinRow (protein);");
  41                 session.execute("CREATE INDEX IF NOT EXISTS JobDateStamp ON ProteinKeyspace.ProteinData (jobtime);");
  42
  43                 System.out.println("Cassandra connected");
  44         }
  45
  46         /*
  47          * parsing data source and filling the database
  48          */
  49         public void Parsing() throws IOException {
  50                 if (true) {
  51                         // if (source.equals("http")) {
  52                         // get data from real Jpred production server
  53                         System.out.println("Parsing web data source......");
  54                         String datasrc = "http://www.compbio.dundee.ac.uk/www-jpred/results/usage-new/alljobs.dat";
  55                         String prefix = "http://www.compbio.dundee.ac.uk/www-jpred/results";
  56                         JpredParserHTTP parser = new JpredParserHTTP(prefix);
  57                         parser.Parsing(datasrc, 4);
  58                 }
  59                 if (false) {
  60                         // if (source.equals("file")) {
  61                         // get irtifical data generated for the DB stress tests
  62                         System.out.println("Parsing local file data source......");
  63                         String datasrc = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/data.dat";
  64                         String prefix = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/Jpreddata";
  65                         JpredParserLocalFile parser = new JpredParserLocalFile(prefix);
  66                         parser.Parsing(datasrc, 190);
  67                 }
  68         }
  69
  70         public void Closing() {
  71                 session.shutdown();
  72                 cluster.shutdown();
  73                 System.out.println("Cassandra has been shut down");
  74         }
  75
  76         /*
  77          * inserting data into the db
  78          */
  79         public void InsertData(long jobtime, String startdate, String enddate, String ip, String jobid, String statusEx, String statusFinal,
  80                         String protein, List<FastaSequence> predictions) {
  81
  82                 String check1 = "SELECT * FROM ProteinKeyspace.ProteinLog WHERE JobID = '" + jobid + "';";
  83                 ResultSet results1 = session.execute(check1);
  84                 if (results1.isExhausted()) {
  85                         String com1 = "INSERT INTO ProteinKeyspace.ProteinLog "
  86                                         + "(JobID, IP, DataBegin, DataEnd, FinalStatus, ExecutionStatus, Protein)" + " VALUES ('" + jobid + "','" + ip + "','"
  87                                         + startdate + "','" + enddate + "','" + statusFinal + "','" + statusEx + "','" + protein + "');";
  88                         session.execute(com1);
  89
  90                         String com2 = "INSERT INTO ProteinKeyspace.ProteinData " + "(jobtime, JobID, Protein)" + " VALUES (" + jobtime + ",'" + jobid
  91                                         + "','" + protein + "');";
  92                         session.execute(com2);
  93
  94                         String allpredictions = "";
  95                         for (FastaSequence pred : predictions) {
  96                                 String predictionname = pred.getId();
  97                                 String prediction = pred.getSequence().replaceAll("\n", "");
  98                                 allpredictions += "'" + predictionname + "':'" + prediction + "',";
  99                         }
 100                         String final_prediction = "";
 101                         if (null != allpredictions) {
 102                                 final_prediction = allpredictions.substring(0, allpredictions.length() - 1);
 103                         }
 104
 105                         String check2 = "SELECT * FROM ProteinKeyspace.ProteinRow WHERE JobID = '" + jobid + "';";
 106                         ResultSet results2 = session.execute(check2);
 107                         if (results2.isExhausted()) {
 108                                 String com3 = "INSERT INTO ProteinKeyspace.ProteinRow " + "(Protein, JobID, Predictions)" + " VALUES ('"
 109                         + protein + "','" + jobid + "',{" + final_prediction + "});";
 110                                 session.execute(com3);
 111                         }
 112                 }
 113         }
 114
 115         /*
 116          * getting data from the db
 117          */
 118         public List<Pair<String, String>> ReadProteinDataTable() {
 119                 final long startTime = System.currentTimeMillis();
 120                 String com = "SELECT DataBegin,DataEnd FROM ProteinKeyspace.ProteinLog;";
 121                 System.out.println("Command: " + com);
 122                 ResultSet results = session.execute(com);
 123                 final long queryTime = System.currentTimeMillis();
 124                 List<Row> rows = results.all();
 125                 System.out.println ("Query time is " + (queryTime - startTime) + " msec");
 126
 127                 List<Pair<String, String>> res = new ArrayList<Pair<String, String>>();
 128                 int c = 0;
 129                 for (Row r : rows) {
 130                         Pair<String, String> pair = new Pair<String, String>(r.getString("DataBegin"),r.getString("DataEnd"));
 131                         res.add(pair);
 132                         ++c;
 133                 }
 134                 final long endTime = System.currentTimeMillis();
 135                 System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 136                 return res;
 137         }
 138
 139         /*
 140          * getting earlest date of jobs from the db
 141          */
 142         public long getEarliestDateInDB() {
 143                 final long startTime = System.currentTimeMillis();
 144                 String com = "SELECT jobtime FROM ProteinKeyspace.ProteinData;";
 145                 System.out.println("Command: " + com);
 146                 ResultSet results = session.execute(com);
 147                 final long queryTime = System.currentTimeMillis();
 148                 System.out.println ("Query time is " + (queryTime - startTime) + " msec");
 149
 150                 Calendar cal = Calendar.getInstance();
 151                 long res = cal.getTimeInMillis();
 152                 int c = 0;
 153                 while (!results.isExhausted()) {
 154                         Row r = results.one();
 155                         long d1 = r.getLong("jobtime");
 156                         if (res > d1) {
 157                                 res = d1;
 158                         }
 159                         ++c;
 160                 }
 161                 final long endTime = System.currentTimeMillis();
 162                 System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 163                 return res;
 164         }
 165
 166 }