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 import java.util.Map;
   9
  10 import com.datastax.driver.core.Cluster;
  11 import com.datastax.driver.core.Host;
  12 import com.datastax.driver.core.Metadata;
  13 import com.datastax.driver.core.Row;
  14 import com.datastax.driver.core.Session;
  15 import com.datastax.driver.core.ResultSet;
  16
  17 public class CassandraNativeConnector {
  18         private static Cluster cluster;
  19         private static Session session;
  20         /*
  21          * connect to the cluster and look weather the dababase has any data inside
  22          */
  23         public void Connect() {
  24                 // local cassandra cluster
  25                 cluster = Cluster.builder().addContactPoint("localhost").build();
  26                 // distributed cassandra cluster
  27                 /* cluster = Cluster.builder().addContactPoint("10.0.115.190").build(); */
  28                 Metadata metadata = cluster.getMetadata();
  29                 System.out.printf("Connected to cluster: %s\n", metadata.getClusterName());
  30                 for (Host host : metadata.getAllHosts()) {
  31                         System.out.printf("Datatacenter: %s; Host: %s; Rack: %s\n", host.getDatacenter(), host.getAddress(), host.getRack());
  32                 }
  33
  34                 session = cluster.connect();
  35                 session.execute("CREATE KEYSPACE IF NOT EXISTS ProteinKeyspace WITH replication = {'class':'SimpleStrategy', 'replication_factor':3};");
  36                 session.execute("CREATE COLUMNFAMILY IF NOT EXISTS ProteinKeyspace.ProteinRow (Protein ascii, JobID ascii, Predictions map<ascii,ascii>, PRIMARY KEY(JobID));");
  37                 session.execute("CREATE COLUMNFAMILY IF NOT EXISTS ProteinKeyspace.ProteinLog "
  38                                 + "(JobID ascii, DataBegin ascii, DataEnd ascii, ip ascii, FinalStatus ascii, ExecutionStatus ascii, Protein ascii, PRIMARY KEY(JobID));");
  39                 session.execute("CREATE COLUMNFAMILY IF NOT EXISTS ProteinKeyspace.ProteinData (jobtime bigint, JobID ascii, Protein ascii, PRIMARY KEY(JobID));");
  40
  41                 session.execute("CREATE INDEX IF NOT EXISTS ProteinSeq ON ProteinKeyspace.ProteinRow (protein);");
  42                 session.execute("CREATE INDEX IF NOT EXISTS JobDateStamp ON ProteinKeyspace.ProteinData (jobtime);");
  43
  44                 System.out.println("Cassandra connected");
  45         }
  46
  47         /*
  48          * parsing data source and filling the database
  49          */
  50         public void Parsing() throws IOException {
  51                 if (true) {
  52                         // if (source.equals("http")) {
  53                         // get data from real Jpred production server
  54                         System.out.println("Parsing web data source......");
  55                         String datasrc = "http://www.compbio.dundee.ac.uk/www-jpred/results/usage-new/alljobs.dat";
  56                         String prefix = "http://www.compbio.dundee.ac.uk/www-jpred/results";
  57                         JpredParserHTTP parser = new JpredParserHTTP(prefix);
  58                         parser.Parsing(datasrc, 4);
  59                 }
  60                 if (false) {
  61                         // if (source.equals("file")) {
  62                         // get irtifical data generated for the DB stress tests
  63                         System.out.println("Parsing local file data source......");
  64                         String datasrc = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/data.dat";
  65                         String prefix = "/home/asherstnev/Projects/Java.projects/proteocache/data_stress_test/Jpreddata";
  66                         JpredParserLocalFile parser = new JpredParserLocalFile(prefix);
  67                         parser.Parsing(datasrc, 190);
  68                 }
  69         }
  70
  71         public void Closing() {
  72                 session.shutdown();
  73                 cluster.shutdown();
  74                 System.out.println("Cassandra has been shut down");
  75         }
  76
  77         /*
  78          * inserting data into the db
  79          */
  80         public void InsertData(long jobtime, String startdate, String enddate, String ip, String jobid, String statusEx, String statusFinal,
  81                         String protein, List<FastaSequence> predictions) {
  82
  83                 String check1 = "SELECT * FROM ProteinKeyspace.ProteinLog WHERE JobID = '" + jobid + "';";
  84                 ResultSet results1 = session.execute(check1);
  85                 if (results1.isExhausted()) {
  86                         String com1 = "INSERT INTO ProteinKeyspace.ProteinLog "
  87                                         + "(JobID, IP, DataBegin, DataEnd, FinalStatus, ExecutionStatus, Protein)" + " VALUES ('" + jobid + "','" + ip + "','"
  88                                         + startdate + "','" + enddate + "','" + statusFinal + "','" + statusEx + "','" + protein + "');";
  89                         session.execute(com1);
  90
  91                         String com2 = "INSERT INTO ProteinKeyspace.ProteinData " + "(jobtime, JobID, Protein)" + " VALUES (" + jobtime + ",'" + jobid
  92                                         + "','" + protein + "');";
  93                         session.execute(com2);
  94
  95                         String allpredictions = "";
  96                         for (FastaSequence pred : predictions) {
  97                                 String predictionname = pred.getId();
  98                                 String prediction = pred.getSequence().replaceAll("\n", "");
  99                                 allpredictions += "'" + predictionname + "':'" + prediction + "',";
 100                         }
 101                         String final_prediction = "";
 102                         if (null != allpredictions) {
 103                                 final_prediction = allpredictions.substring(0, allpredictions.length() - 1);
 104                         }
 105
 106                         String check2 = "SELECT * FROM ProteinKeyspace.ProteinRow WHERE JobID = '" + jobid + "';";
 107                         ResultSet results2 = session.execute(check2);
 108                         if (results2.isExhausted()) {
 109                                 String com3 = "INSERT INTO ProteinKeyspace.ProteinRow " + "(Protein, JobID, Predictions)" + " VALUES ('"
 110                         + protein + "','" + jobid + "',{" + final_prediction + "});";
 111                                 session.execute(com3);
 112                         }
 113                 }
 114         }
 115
 116         /*
 117          * getting data from the db
 118          */
 119         public List<Pair<String, String>> ReadProteinDataTable() {
 120                 final long startTime = System.currentTimeMillis();
 121                 String com = "SELECT DataBegin,DataEnd FROM ProteinKeyspace.ProteinLog;";
 122                 System.out.println("Command: " + com);
 123                 ResultSet results = session.execute(com);
 124                 final long queryTime = System.currentTimeMillis();
 125                 List<Row> rows = results.all();
 126                 System.out.println ("Query time is " + (queryTime - startTime) + " msec");
 127
 128                 List<Pair<String, String>> res = new ArrayList<Pair<String, String>>();
 129                 int c = 0;
 130                 for (Row r : rows) {
 131                         Pair<String, String> pair = new Pair<String, String>(r.getString("DataBegin"),r.getString("DataEnd"));
 132                         res.add(pair);
 133                         ++c;
 134                 }
 135                 final long endTime = System.currentTimeMillis();
 136                 System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 137                 return res;
 138         }
 139
 140         /*
 141          * getting data from the db ProteinData
 142          */
 143         public Integer ReadDateTable(long queryDate) {
 144                 final long startTime = System.currentTimeMillis();
 145                 String com = "SELECT jobtime, JobID FROM ProteinKeyspace.ProteinData WHERE jobtime = " + queryDate + ";";
 146                 System.out.println("Command: " + com);
 147                 ResultSet results = session.execute(com);
 148                 if (results.isExhausted())
 149                         return null;
 150                 final long queryTime = System.currentTimeMillis();
 151                 List<Row> rows = results.all();
 152                 System.out.println ("Query time is " + (queryTime - startTime) + " msec");
 153                 return rows.size();
 154         }
 155
 156         /*
 157          * getting whole protein sequence from the db ProteinRow
 158          */
 159         public List<StructureProteinPrediction> ReadWholeSequence(String queryProtein) {
 160                 final long startTime = System.currentTimeMillis();
 161                 String com = "SELECT JobID, Predictions FROM ProteinKeyspace.ProteinRow WHERE Protein = '" + queryProtein + "';";
 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                 List<Row> rows = results.all();
 168                 System.out.println ("Query time is " + (queryTime - startTime) + " msec");
 169                 System.out.println (" rows analysed,  " + rows.size());
 170                 List<StructureProteinPrediction> res = new ArrayList<StructureProteinPrediction>();
 171                 int c = 0;
 172                 for (Row r : rows) {
 173                         StructureProteinPrediction structure = new StructureProteinPrediction(queryProtein, r.getString("JobID"), r.getMap("Predictions", String.class, String.class));
 174                         res.add(structure);
 175                         ++c;
 176                 }
 177                 final long endTime = System.currentTimeMillis();
 178                 System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 179                 return res;
 180         }
 181
 182         /*
 183          * getting part of protein sequence from the db ProteinRow
 184          */
 185         public List<StructureProteinPrediction>  ReadPartOfSequence(String queryProtein) {
 186                 final long startTime = System.currentTimeMillis();
 187                 String com = "SELECT * FROM ProteinKeyspace.ProteinRow;";
 188                 System.out.println("Command: " + com);
 189                 ResultSet results = session.execute(com);
 190                 if (results.isExhausted())
 191                         return null;
 192                 final long queryTime = System.currentTimeMillis();
 193                 List<Row> rows = results.all();
 194                 System.out.println ("Query time is " + (queryTime - startTime) + " msec");
 195                 System.out.println (" rows analysed,  " + rows.size());
 196                 List<StructureProteinPrediction>  res = new ArrayList<StructureProteinPrediction>();
 197                 int c = 0;
 198                 for (Row r : rows) {
 199                         String prot = r.getString("Protein");
 200                         if (prot.matches("(.*)" + queryProtein + "(.*)")) {
 201                         //      System.out.println(prot);
 202                                 StructureProteinPrediction structure = new StructureProteinPrediction(prot, r.getString("JobID"), r.getMap("Predictions", String.class, String.class));
 203                                 res.add(structure);
 204                                 ++c;
 205                         }
 206                 }
 207                 final long endTime = System.currentTimeMillis();
 208                 System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 209                 return res;
 210         }
 211
 212         /*
 213          * getting protein sequences by counter
 214          */
 215         public List<Pair<String, Integer>>  ReadProteinDataByCounter(int counter) {
 216                 final long startTime = System.currentTimeMillis();
 217                 String com = "SELECT DISTINCT Protein FROM ProteinKeyspace.ProteinRow;";
 218                 System.out.println("Command: " + com);
 219                 ResultSet results = session.execute(com);
 220                 if (results.isExhausted())
 221                         return null;
 222                 final long queryTime = System.currentTimeMillis();
 223                 List<Row> rows = results.all();
 224                 System.out.println ("Query time is " + (queryTime - startTime) + " msec");
 225                 System.out.println (" rows analysed,  " + rows.size());
 226                 List<Pair<String, Integer>>  res = new ArrayList<Pair<String, Integer>>();
 227                 int c = 0;
 228                 for (Row r : rows) {
 229                         String prot = r.getString("Protein");
 230
 231                 }
 232                 final long endTime = System.currentTimeMillis();
 233                 System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 234                 return res;
 235         }
 236
 237         /*
 238          * getting earlest date of jobs from the db
 239          */
 240         public long getEarliestDateInDB() {
 241                 final long startTime = System.currentTimeMillis();
 242                 String com = "SELECT jobtime,JobID FROM ProteinKeyspace.ProteinData;";
 243                 System.out.println("Command: " + com);
 244                 ResultSet results = session.execute(com);
 245                 final long queryTime = System.currentTimeMillis();
 246                 System.out.println ("Query time is  " + (queryTime - startTime) + " msec");
 247
 248                 Calendar cal = Calendar.getInstance();
 249                 long res = cal.getTimeInMillis();
 250                 int c = 0;
 251                 while (!results.isExhausted()) {
 252                         Row r = results.one();
 253                         long d1 = r.getLong("jobtime");
 254                         if (res > d1) {
 255                                 res = d1;
 256                         }
 257                         ++c;
 258                 }
 259                 final long endTime = System.currentTimeMillis();
 260                 System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 261                 return res;
 262         }
 263
 264 }