datadb/compbio/cassandra/CassandraReaderOld.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
  15 import compbio.beans.DateBean;
  16 import compbio.beans.JobBean;
  17 import compbio.beans.ProteinBean;
  18 import compbio.beans.Total;
  19 import compbio.engine.JobStatus;
  20
  21 public class CassandraReaderOld {
  22         private Session session;
  23         private static Logger log = Logger.getLogger(CassandraNativeConnector.class);
  24
  25         public CassandraReaderOld() {
  26                 Session inis = CassandraNativeConnector.getSession();
  27                 setSession(inis);
  28         }
  29
  30         public void setSession(Session s) {
  31                 assert s != null;
  32                 session = s;
  33         }
  34
  35         /*
  36          * getting data from the db
  37          */
  38         public List<Pair<String, String>> ReadProteinDataTable() {
  39                 final long startTime = System.currentTimeMillis();
  40                 String com = "SELECT DataBegin,DataEnd FROM ProteinLog;";
  41                 System.out.println("Command: " + com);
  42                 ResultSet results = session.execute(com);
  43                 final long queryTime = System.currentTimeMillis();
  44                 List<Row> rows = results.all();
  45                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  46
  47                 List<Pair<String, String>> res = new ArrayList<Pair<String, String>>();
  48                 int c = 0;
  49                 for (Row r : rows) {
  50                         Pair<String, String> pair = new Pair<String, String>(r.getString("DataBegin"), r.getString("DataEnd"));
  51                         res.add(pair);
  52                         ++c;
  53                 }
  54                 final long endTime = System.currentTimeMillis();
  55                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  56                 return res;
  57         }
  58
  59         /*
  60          * getting data from the db
  61          */
  62         public DateBean ReadProteinData(long day, String date) {
  63                 String com = "SELECT JobID, Protein FROM ProteinData WHERE jobtime = " + day + ";";
  64                 System.out.println("Command: " + com);
  65                 ResultSet results = session.execute(com);
  66                 if (results.isExhausted())
  67                         return null;
  68                 List<Row> rows = results.all();
  69                 DateBean res = new DateBean(date);
  70                 for (Row r : rows) {
  71                         res.setJobidAndSeq(r.getString("JobID"), r.getString("Protein"));
  72                 }
  73                 return res;
  74         }
  75
  76         /**
  77          * getting data from the db
  78          */
  79         public DateBean ReadFailedJobs(long day, String date, JobStatus status) {
  80                 // FailLog (jobtime, JobID, ExecTime, ip, FinalStatus)
  81                 String com = "SELECT JobID FROM FailLog WHERE jobtime = " + day + " and FinalStatus = '" + status.name() + "';";
  82                 ResultSet results = session.execute(com);
  83                 if (results.isExhausted())
  84                         return null;
  85                 List<Row> rows = results.all();
  86                 DateBean res = new DateBean(date);
  87                 for (Row r : rows) {
  88                         String jobid = r.getString("JobID");
  89                         String com1 = "SELECT Protein FROM ProteinLog WHERE JobID = '" + jobid + "';";
  90                         System.out.println("Command: " + com1);
  91                         ResultSet results2 = session.execute(com1);
  92                         List<Row> jrows = results2.all();
  93                         if (1 == jrows.size()) {
  94                                 String protein = jrows.get(0).getString("Protein");
  95                                 res.setJobidAndSeq(jobid, protein);
  96                         }
  97                 }
  98                 return res;
  99         }
 100
 101         /*
 102          * getting data from the db JobDateInfo
 103          */
 104         public Total ReadDateTable(long queryDate) {
 105                 ResultSet results = session.execute("SELECT * FROM JobDateInfo WHERE jobday = " + queryDate + ";");
 106                 if (results.isExhausted())
 107                         return null;
 108                 Row therow = results.one();
 109                 Total res = new Total(therow.getLong("Total"), therow.getLong("TotalOK"), therow.getLong("TotalStopped"),
 110                                 therow.getLong("TotalError"), therow.getLong("TotalTimeOut"));
 111                 if (!results.isExhausted()) {
 112                         Date date = new Date(queryDate);
 113                         log.warn("CassandraReader.ReadDateTable: date row for " + date.toString() + " (" + queryDate + ") duplicated ");
 114                 }
 115                 return res;
 116         }
 117
 118         /*
 119          * getting whole protein sequence from the db ProteinRow
 120          */
 121         public List<ProteinBean> ReadWholeSequence(String queryProtein) {
 122                 final long startTime = System.currentTimeMillis();
 123                 String com = "SELECT JobID, Predictions FROM ProteinRow WHERE Protein = '" + queryProtein + "';";
 124                 System.out.println("Command: " + com);
 125                 ResultSet results = session.execute(com);
 126                 if (results.isExhausted())
 127                         return null;
 128                 final long queryTime = System.currentTimeMillis();
 129                 List<Row> rows = results.all();
 130                 System.out.println("first size : " + rows.size());
 131                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 132                 System.out.println(" rows analysed,  " + rows.size());
 133                 List<ProteinBean> res = new ArrayList<ProteinBean>();
 134                 ProteinBean structure = new ProteinBean(queryProtein, rows.get(0).getMap("Predictions", String.class, String.class));
 135                 System.out.println("second size : " + rows.size());
 136                 int c = 0;
 137                 for (Row r : rows) {
 138                         structure.setJobid(r.getString("JobID"));
 139                         ++c;
 140                 }
 141                 res.add(structure);
 142                 final long endTime = System.currentTimeMillis();
 143                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 144                 return res;
 145         }
 146
 147         /*
 148          * getting jobs by ip
 149          */
 150         public Map<String, String[]> ReadIpWithJobs(String ip) {
 151                 final long startTime = System.currentTimeMillis();
 152                 String com = "SELECT JobID, Protein, FinalStatus, DataBegin FROM ProteinLog WHERE ip = '" + ip + "';";
 153                 System.out.println("Command: " + com);
 154                 ResultSet results = session.execute(com);
 155                 if (results.isExhausted())
 156                         return null;
 157                 final long queryTime = System.currentTimeMillis();
 158                 List<Row> rows = results.all();
 159                 Map<String, String[]> res = new HashMap<String, String[]>();
 160                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 161                 System.out.println(" rows analysed,  " + rows.size());
 162                 int c = 0;
 163                 for (Row r : rows) {
 164                         if (r.getString("FinalStatus").equals("OK")) {
 165                                 String date = r.getString("DataBegin");
 166                                 res.put(r.getString("JobID"), new String[] { date.substring(0, date.indexOf(":")), r.getString("Protein") });
 167                                 ++c;
 168                         }
 169                 }
 170                 final long endTime = System.currentTimeMillis();
 171                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 172                 return res;
 173         }
 174
 175         /*
 176          * getting part of protein sequence from the db ProteinRow
 177          */
 178         public List<ProteinBean> ReadPartOfSequence(String queryProtein) {
 179                 final long startTime = System.currentTimeMillis();
 180                 String com = "SELECT * FROM ProteinRow;";
 181                 System.out.println("Command: " + com);
 182                 ResultSet results = session.execute(com);
 183                 if (results.isExhausted())
 184                         return null;
 185                 final long queryTime = System.currentTimeMillis();
 186                 List<Row> rows = results.all();
 187                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 188                 System.out.println(" rows analysed,  " + rows.size());
 189                 List<ProteinBean> res = new ArrayList<ProteinBean>();
 190                 int c = 0;
 191                 for (Row r : rows) {
 192                         String prot = r.getString("Protein");
 193                         if (prot.matches("(.*)" + queryProtein + "(.*)")) {
 194                                 ProteinBean structure = new ProteinBean(prot, r.getMap("Predictions", String.class, String.class));
 195                                 structure.setJobid(r.getString("JobID"));
 196                                 res.add(structure);
 197                                 ++c;
 198                         }
 199                 }
 200                 final long endTime = System.currentTimeMillis();
 201                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 202                 return res;
 203         }
 204
 205         /*
 206          * getting protein sequence from the db ProteinRow
 207          */
 208         public Map<String, String> ReadProtein() {
 209                 final long startTime = System.currentTimeMillis();
 210                 String com = "SELECT * FROM ProteinRow;";
 211                 System.out.println("Command: " + com);
 212                 ResultSet results = session.execute(com);
 213                 if (results.isExhausted())
 214                         return null;
 215                 final long queryTime = System.currentTimeMillis();
 216                 List<Row> rows = results.all();
 217                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 218                 System.out.println(" rows analysed,  " + rows.size());
 219                 Map<String, String> res = new HashMap<String, String>();
 220                 int c = 0;
 221                 for (Row r : rows) {
 222                         String prot = r.getString("Protein");
 223                         String prediction = findJnetpred(r.getMap("Predictions", String.class, String.class));
 224                         if (prot != null || prediction != null)
 225                                 res.put(prot, prediction);
 226                 }
 227                 final long endTime = System.currentTimeMillis();
 228                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 229                 return res;
 230         }
 231
 232         private String findJnetpred(Map<String, String> pred) {
 233                 if (pred != null) {
 234                         if (pred.containsKey("jnetpred"))
 235                                 return pred.get("jnetpred");
 236                 }
 237                 return null;
 238         }
 239
 240         /*
 241          * getting protein sequences by counter
 242          */
 243         public Map<String, Integer> ReadProteinSequenceByCounter() {
 244                 final long startTime = System.currentTimeMillis();
 245                 String com = "SELECT Protein, JobID FROM ProteinRow;";
 246                 System.out.println("Command: " + com);
 247                 ResultSet results = session.execute(com);
 248                 if (results.isExhausted())
 249                         return null;
 250                 final long queryTime = System.currentTimeMillis();
 251                 List<Row> rows = results.all();
 252                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 253                 System.out.println(" rows analysed,  " + rows.size());
 254                 Map<String, Integer> res = new HashMap<String, Integer>();
 255                 int c = 0;
 256                 for (Row r : rows) {
 257                         String protein = r.getString("Protein");
 258                         String id = r.getString("JobID");
 259                         if (res.containsKey(protein))
 260                                 res.put(protein, res.get(protein) + 1);
 261                         else
 262                                 res.put(protein, 1);
 263                 }
 264                 final long endTime = System.currentTimeMillis();
 265                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 266                 return res;
 267         }
 268
 269         /*
 270          * getting ip by counter
 271          */
 272         public Map<String, Integer> ReadIpByCounter() {
 273                 final long startTime = System.currentTimeMillis();
 274                 String com = "SELECT JobID, ip FROM ProteinLog;";
 275                 System.out.println("Command: " + com);
 276                 ResultSet results = session.execute(com);
 277                 if (results.isExhausted())
 278                         return null;
 279                 final long queryTime = System.currentTimeMillis();
 280                 List<Row> rows = results.all();
 281                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 282                 System.out.println(" rows analysed,  " + rows.size());
 283                 Map<String, Integer> res = new HashMap<String, Integer>();
 284                 int c = 0;
 285                 for (Row r : rows) {
 286                         String ip = r.getString("ip");
 287                         String id = r.getString("JobID");
 288                         if (res.containsKey(ip))
 289                                 res.put(ip, res.get(ip) + 1);
 290                         else
 291                                 res.put(ip, 1);
 292                 }
 293                 final long endTime = System.currentTimeMillis();
 294                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 295                 return res;
 296         }
 297
 298         /*
 299          * getting log info for jobid
 300          */
 301         public JobBean ReadJobLog(String jobid) {
 302                 final long startTime = System.currentTimeMillis();
 303                 String com = "SELECT * FROM ProteinLog WHERE JobID = '" + jobid + "';";
 304                 System.out.println("Command: " + com);
 305                 ResultSet results = session.execute(com);
 306                 if (results.isExhausted())
 307                         return null;
 308                 final long queryTime = System.currentTimeMillis();
 309                 Row row = results.one();
 310                 String com1 = "SELECT * FROM ProteinRow WHERE JobID = '" + jobid + "' ALLOW FILTERING;";
 311                 System.out.println("Command: " + com1);
 312                 ResultSet results1 = session.execute(com1);
 313                 if (results1.isExhausted())
 314                         return null;
 315                 Row row1 = results1.one();
 316                 JobBean res = new JobBean(row.getString("Protein"), row.getString("JobID"), row.getString("DataBegin"), row.getString("DataEnd"),
 317                                 row.getString("ip"), row1.getMap("Predictions", String.class, String.class), row.getString("ProgrammeName"), row.getString("Version"));
 318                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 319                 final long endTime = System.currentTimeMillis();
 320                 System.out.println(" rows analysed, execution time is " + (endTime - startTime) + " msec");
 321                 return res;
 322         }
 323 }