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 import compbio.engine.JobStatus;
  15
  16 public class CassandraReader {
  17         private Session session;
  18         private static Logger log = Logger.getLogger(CassandraNativeConnector.class);
  19
  20         public CassandraReader() {
  21                 Session inis = CassandraNativeConnector.getSession();
  22                 setSession(inis);
  23         }
  24
  25         public void setSession(Session s) {
  26                 assert s != null;
  27                 session = s;
  28         }
  29
  30         /*
  31          * getting data from the db
  32          */
  33         public List<Pair<String, String>> ReadProteinDataTable() {
  34                 final long startTime = System.currentTimeMillis();
  35                 String com = "SELECT DataBegin,DataEnd FROM ProteinLog;";
  36                 System.out.println("Command: " + com);
  37                 ResultSet results = session.execute(com);
  38                 final long queryTime = System.currentTimeMillis();
  39                 List<Row> rows = results.all();
  40                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  41
  42                 List<Pair<String, String>> res = new ArrayList<Pair<String, String>>();
  43                 int c = 0;
  44                 for (Row r : rows) {
  45                         Pair<String, String> pair = new Pair<String, String>(r.getString("DataBegin"), r.getString("DataEnd"));
  46                         res.add(pair);
  47                         ++c;
  48                 }
  49                 final long endTime = System.currentTimeMillis();
  50                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  51                 return res;
  52         }
  53
  54         /*
  55          * getting data from the db
  56          */
  57         public DateBean ReadProteinData(long day, String date) {
  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                 List<Row> rows = results.all();
  64                 DateBean res = new DateBean(date);
  65                 for (Row r : rows) {
  66                         res.setJobidAndSeq(r.getString("JobID"), r.getString("Protein"));
  67                 }
  68                 return res;
  69         }
  70
  71         /**
  72          * getting data from the db
  73          */
  74         public DateBean ReadFailedJobs(long day, String date, JobStatus status) {
  75                 // FailLog (jobtime, JobID, ExecTime, ip, FinalStatus)
  76                 String com = "SELECT JobID FROM FailLog WHERE jobtime = " + day + " and FinalStatus = '" + status.name() + "';";
  77                 ResultSet results = session.execute(com);
  78                 if (results.isExhausted())
  79                         return null;
  80                 List<Row> rows = results.all();
  81                 DateBean res = new DateBean(date);
  82                 for (Row r : rows) {
  83                         String jobid = r.getString("JobID");
  84                         String com1 = "SELECT Protein FROM ProteinLog WHERE JobID = '" + jobid + "';";
  85                         System.out.println("Command: " + com1);
  86                         ResultSet results2 = session.execute(com1);
  87                         List<Row> jrows = results2.all();
  88                         if (1 == jrows.size()) {
  89                                 String protein = jrows.get(0).getString("Protein");
  90                                 res.setJobidAndSeq(jobid, protein);
  91                         }
  92                 }
  93                 return res;
  94         }
  95
  96         /*
  97          * getting data from the db JobDateInfo
  98          */
  99         public Total ReadDateTable(long queryDate) {
 100                 ResultSet results = session.execute("SELECT * FROM JobDateInfo WHERE jobday = " + queryDate + ";");
 101                 if (results.isExhausted())
 102                         return null;
 103                 Row therow = results.one();
 104                 Total res = new Total(therow.getLong("Total"), therow.getLong("TotalOK"), therow.getLong("TotalStopped"),
 105                                 therow.getLong("TotalError"), therow.getLong("TotalTimeOut"));
 106                 if (!results.isExhausted()) {
 107                         Date date = new Date(queryDate);
 108                         log.warn("CassandraReader.ReadDateTable: date row for " + date.toString() + " (" + queryDate + ") duplicated ");
 109                 }
 110                 return res;
 111         }
 112
 113         /*
 114          * getting whole protein sequence from the db ProteinRow
 115          */
 116         public List<ProteinBean> ReadWholeSequence(String queryProtein) {
 117                 final long startTime = System.currentTimeMillis();
 118                 String com = "SELECT JobID, Predictions FROM ProteinRow WHERE Protein = '" + queryProtein + "';";
 119                 System.out.println("Command: " + com);
 120                 ResultSet results = session.execute(com);
 121                 if (results.isExhausted())
 122                         return null;
 123                 final long queryTime = System.currentTimeMillis();
 124                 List<Row> rows = results.all();
 125                 System.out.println("first size : " + rows.size());
 126                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 127                 System.out.println(" rows analysed,  " + rows.size());
 128                 List<ProteinBean> res = new ArrayList<ProteinBean>();
 129                 ProteinBean structure = new ProteinBean(queryProtein, rows.get(0).getMap("Predictions", String.class, String.class));
 130                 System.out.println("second size : " + rows.size());
 131                 int c = 0;
 132                 for (Row r : rows) {
 133                         structure.setJobid(r.getString("JobID"));
 134                         ++c;
 135                 }
 136                 res.add(structure);
 137                 final long endTime = System.currentTimeMillis();
 138                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 139                 return res;
 140         }
 141
 142         /*
 143          * getting jobs by ip
 144          */
 145         public Map<String, String[]> ReadIpWithJobs(String ip) {
 146                 final long startTime = System.currentTimeMillis();
 147                 String com = "SELECT JobID, Protein, FinalStatus, DataBegin FROM ProteinLog WHERE ip = '" + ip + "';";
 148                 System.out.println("Command: " + com);
 149                 ResultSet results = session.execute(com);
 150                 if (results.isExhausted())
 151                         return null;
 152                 final long queryTime = System.currentTimeMillis();
 153                 List<Row> rows = results.all();
 154                 Map<String, String[]> res = new HashMap<String, String[]>();
 155                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 156                 System.out.println(" rows analysed,  " + rows.size());
 157                 int c = 0;
 158                 for (Row r : rows) {
 159                         if (r.getString("FinalStatus").equals("OK")) {
 160                                 String date = r.getString("DataBegin");
 161                                 res.put(r.getString("JobID"), new String[] { date.substring(0, date.indexOf(":")), r.getString("Protein") });
 162                                 ++c;
 163                         }
 164                 }
 165                 final long endTime = System.currentTimeMillis();
 166                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 167                 return res;
 168         }
 169
 170         /*
 171          * getting part of protein sequence from the db ProteinRow
 172          */
 173         public List<ProteinBean> ReadPartOfSequence(String queryProtein) {
 174                 final long startTime = System.currentTimeMillis();
 175                 String com = "SELECT * FROM ProteinRow;";
 176                 System.out.println("Command: " + com);
 177                 ResultSet results = session.execute(com);
 178                 if (results.isExhausted())
 179                         return null;
 180                 final long queryTime = System.currentTimeMillis();
 181                 List<Row> rows = results.all();
 182                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 183                 System.out.println(" rows analysed,  " + rows.size());
 184                 List<ProteinBean> res = new ArrayList<ProteinBean>();
 185                 int c = 0;
 186                 for (Row r : rows) {
 187                         String prot = r.getString("Protein");
 188                         if (prot.matches("(.*)" + queryProtein + "(.*)")) {
 189                                 ProteinBean structure = new ProteinBean(prot, r.getMap("Predictions", String.class, String.class));
 190                                 structure.setJobid(r.getString("JobID"));
 191                                 res.add(structure);
 192                                 ++c;
 193                         }
 194                 }
 195                 final long endTime = System.currentTimeMillis();
 196                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 197                 return res;
 198         }
 199
 200         /*
 201          * getting protein sequences by counter
 202          */
 203         public Map<String, Integer> ReadProteinSequenceByCounter() {
 204                 final long startTime = System.currentTimeMillis();
 205                 String com = "SELECT Protein, JobID FROM ProteinRow;";
 206                 System.out.println("Command: " + com);
 207                 ResultSet results = session.execute(com);
 208                 if (results.isExhausted())
 209                         return null;
 210                 final long queryTime = System.currentTimeMillis();
 211                 List<Row> rows = results.all();
 212                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 213                 System.out.println(" rows analysed,  " + rows.size());
 214                 Map<String, Integer> res = new HashMap<String, Integer>();
 215                 int c = 0;
 216                 for (Row r : rows) {
 217                         String protein = r.getString("Protein");
 218                         String id = r.getString("JobID");
 219                         if (res.containsKey(protein))
 220                                 res.put(protein, res.get(protein) + 1);
 221                         else
 222                                 res.put(protein, 1);
 223                 }
 224                 final long endTime = System.currentTimeMillis();
 225                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 226                 return res;
 227         }
 228
 229         /*
 230          * getting ip by counter
 231          */
 232         public Map<String, Integer> ReadIpByCounter() {
 233                 final long startTime = System.currentTimeMillis();
 234                 String com = "SELECT JobID, ip FROM ProteinLog;";
 235                 System.out.println("Command: " + com);
 236                 ResultSet results = session.execute(com);
 237                 if (results.isExhausted())
 238                         return null;
 239                 final long queryTime = System.currentTimeMillis();
 240                 List<Row> rows = results.all();
 241                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 242                 System.out.println(" rows analysed,  " + rows.size());
 243                 Map<String, Integer> res = new HashMap<String, Integer>();
 244                 int c = 0;
 245                 for (Row r : rows) {
 246                         String ip = r.getString("ip");
 247                         String id = r.getString("JobID");
 248                         if (res.containsKey(ip))
 249                                 res.put(ip, res.get(ip) + 1);
 250                         else
 251                                 res.put(ip, 1);
 252                 }
 253                 final long endTime = System.currentTimeMillis();
 254                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 255                 return res;
 256         }
 257
 258         /*
 259          * getting log info for jobid
 260          */
 261         public JobBean ReadJobLog(String jobid) {
 262                 final long startTime = System.currentTimeMillis();
 263                 String com = "SELECT * FROM ProteinLog WHERE JobID = '" + jobid + "';";
 264                 System.out.println("Command: " + com);
 265                 ResultSet results = session.execute(com);
 266                 if (results.isExhausted())
 267                         return null;
 268                 final long queryTime = System.currentTimeMillis();
 269                 Row row = results.one();
 270                 String com1 = "SELECT * FROM ProteinRow WHERE JobID = '" + jobid + "' ALLOW FILTERING;";
 271                 System.out.println("Command: " + com1);
 272                 ResultSet results1 = session.execute(com1);
 273                 if (results1.isExhausted())
 274                         return null;
 275                 Row row1 = results1.one();
 276                 JobBean res = new JobBean(row.getString("Protein"), row.getString("JobID"), row.getString("DataBegin"), row.getString("DataEnd"),
 277                                 row.getString("ip"), row1.getMap("Predictions", String.class, String.class));
 278                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 279                 final long endTime = System.currentTimeMillis();
 280                 System.out.println(" rows analysed, execution time is " + (endTime - startTime) + " msec");
 281                 return res;
 282         }
 283 }