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 import compbio.engine.Pair;
  21
  22 public class CassandraReaderOld {
  23         private Session session;
  24         private static Logger log = Logger.getLogger(CassandraNativeConnector.class);
  25
  26         public CassandraReaderOld() {
  27                 Session inis = CassandraNativeConnector.getSession();
  28                 setSession(inis);
  29         }
  30
  31         public void setSession(Session s) {
  32                 assert s != null;
  33                 session = s;
  34         }
  35
  36         /*
  37          * getting data from the db
  38          */
  39         public List<Pair<String, String>> ReadProteinDataTable() {
  40                 final long startTime = System.currentTimeMillis();
  41                 String com = "SELECT DataBegin,DataEnd FROM ProteinLog;";
  42                 System.out.println("Command: " + com);
  43                 ResultSet results = session.execute(com);
  44                 final long queryTime = System.currentTimeMillis();
  45                 List<Row> rows = results.all();
  46                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
  47
  48                 List<Pair<String, String>> res = new ArrayList<Pair<String, String>>();
  49                 int c = 0;
  50                 for (Row r : rows) {
  51                         Pair<String, String> pair = new Pair<String, String>(r.getString("DataBegin"), r.getString("DataEnd"));
  52                         res.add(pair);
  53                         ++c;
  54                 }
  55                 final long endTime = System.currentTimeMillis();
  56                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
  57                 return res;
  58         }
  59
  60         /*
  61          * getting data from the db
  62          */
  63         public DateBean ReadProteinData(long day, String date) {
  64                 String com = "SELECT JobID, Protein FROM ProteinData WHERE jobtime = " + day + ";";
  65                 System.out.println("Command: " + com);
  66                 ResultSet results = session.execute(com);
  67                 if (results.isExhausted())
  68                         return null;
  69                 List<Row> rows = results.all();
  70                 DateBean res = new DateBean(date);
  71                 for (Row r : rows) {
  72                         res.setJobidAndSeq(r.getString("JobID"), r.getString("Protein"));
  73                 }
  74                 return res;
  75         }
  76
  77         /**
  78          * getting data from the db
  79          */
  80         public DateBean ReadFailedJobs(long day, String date, JobStatus status) {
  81                 // FailLog (jobtime, JobID, ExecTime, ip, FinalStatus)
  82                 String com = "SELECT JobID FROM FailLog WHERE jobtime = " + day + " and FinalStatus = '" + status.name() + "';";
  83                 ResultSet results = session.execute(com);
  84                 if (results.isExhausted())
  85                         return null;
  86                 List<Row> rows = results.all();
  87                 DateBean res = new DateBean(date);
  88                 for (Row r : rows) {
  89                         String jobid = r.getString("JobID");
  90                         String com1 = "SELECT Protein FROM ProteinLog WHERE JobID = '" + jobid + "';";
  91                         System.out.println("Command: " + com1);
  92                         ResultSet results2 = session.execute(com1);
  93                         List<Row> jrows = results2.all();
  94                         if (1 == jrows.size()) {
  95                                 String protein = jrows.get(0).getString("Protein");
  96                                 res.setJobidAndSeq(jobid, protein);
  97                         }
  98                 }
  99                 return res;
 100         }
 101
 102         /*
 103          * getting data from the db JobDateInfo
 104          */
 105         public Total ReadDateTable(long queryDate) {
 106                 ResultSet results = session.execute("SELECT * FROM JobDateInfo WHERE jobday = " + queryDate + ";");
 107                 if (results.isExhausted())
 108                         return null;
 109                 Row therow = results.one();
 110                 Total res = new Total(therow.getLong("Total"), therow.getLong("TotalOK"), therow.getLong("TotalStopped"),
 111                                 therow.getLong("TotalError"), therow.getLong("TotalTimeOut"));
 112                 if (!results.isExhausted()) {
 113                         Date date = new Date(queryDate);
 114                         log.warn("CassandraReader.ReadDateTable: date row for " + date.toString() + " (" + queryDate + ") duplicated ");
 115                 }
 116                 return res;
 117         }
 118
 119         /*
 120          * getting whole protein sequence from the db ProteinRow
 121          */
 122         public List<ProteinBean> ReadWholeSequence(String queryProtein) {
 123                 final long startTime = System.currentTimeMillis();
 124                 String com = "SELECT JobID, Predictions FROM ProteinRow WHERE Protein = '" + queryProtein + "';";
 125                 System.out.println("Command: " + com);
 126                 ResultSet results = session.execute(com);
 127                 if (results.isExhausted())
 128                         return null;
 129                 final long queryTime = System.currentTimeMillis();
 130                 List<Row> rows = results.all();
 131                 System.out.println("first size : " + rows.size());
 132                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 133                 System.out.println(" rows analysed,  " + rows.size());
 134                 List<ProteinBean> res = new ArrayList<ProteinBean>();
 135                 ProteinBean structure = new ProteinBean(queryProtein, rows.get(0).getMap("Predictions", String.class, String.class));
 136                 System.out.println("second size : " + rows.size());
 137                 int c = 0;
 138                 for (Row r : rows) {
 139                         structure.setJobid(r.getString("JobID"));
 140                         ++c;
 141                 }
 142                 res.add(structure);
 143                 final long endTime = System.currentTimeMillis();
 144                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 145                 return res;
 146         }
 147
 148         /*
 149          * getting jobs by ip
 150          */
 151         public Map<String, String[]> ReadIpWithJobs(String ip) {
 152                 final long startTime = System.currentTimeMillis();
 153                 String com = "SELECT JobID, Protein, FinalStatus, DataBegin FROM ProteinLog WHERE ip = '" + ip + "';";
 154                 System.out.println("Command: " + com);
 155                 ResultSet results = session.execute(com);
 156                 if (results.isExhausted())
 157                         return null;
 158                 final long queryTime = System.currentTimeMillis();
 159                 List<Row> rows = results.all();
 160                 Map<String, String[]> res = new HashMap<String, String[]>();
 161                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 162                 System.out.println(" rows analysed,  " + rows.size());
 163                 int c = 0;
 164                 for (Row r : rows) {
 165                         if (r.getString("FinalStatus").equals("OK")) {
 166                                 String date = r.getString("DataBegin");
 167                                 res.put(r.getString("JobID"), new String[] { date.substring(0, date.indexOf(":")), r.getString("Protein") });
 168                                 ++c;
 169                         }
 170                 }
 171                 final long endTime = System.currentTimeMillis();
 172                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 173                 return res;
 174         }
 175
 176         /*
 177          * getting part of protein sequence from the db ProteinRow
 178          */
 179         public List<ProteinBean> ReadPartOfSequence(String queryProtein) {
 180                 final long startTime = System.currentTimeMillis();
 181                 String com = "SELECT * FROM ProteinRow;";
 182                 System.out.println("Command: " + com);
 183                 ResultSet results = session.execute(com);
 184                 if (results.isExhausted())
 185                         return null;
 186                 final long queryTime = System.currentTimeMillis();
 187                 List<Row> rows = results.all();
 188                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 189                 System.out.println(" rows analysed,  " + rows.size());
 190                 List<ProteinBean> res = new ArrayList<ProteinBean>();
 191                 int c = 0;
 192                 for (Row r : rows) {
 193                         String prot = r.getString("Protein");
 194                         if (prot.matches("(.*)" + queryProtein + "(.*)")) {
 195                                 ProteinBean structure = new ProteinBean(prot, r.getMap("Predictions", String.class, String.class));
 196                                 structure.setJobid(r.getString("JobID"));
 197                                 res.add(structure);
 198                                 ++c;
 199                         }
 200                 }
 201                 final long endTime = System.currentTimeMillis();
 202                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 203                 return res;
 204         }
 205
 206         /*
 207          * getting protein sequence from the db ProteinRow
 208          */
 209         public Map<String, String> ReadProtein() {
 210                 //final long startTime = System.currentTimeMillis();
 211                 String com = "SELECT * FROM ProteinRow;";
 212                 System.out.println("Command: " + com);
 213                 ResultSet results = session.execute(com);
 214                 if (results.isExhausted())
 215                         return null;
 216                 //final long queryTime = System.currentTimeMillis();
 217                 List<Row> rows = results.all();
 218                 Map<String, String> output = new HashMap<String, String>();
 219                 for (Row r : rows) {
 220                         String protein = r.getString("Protein");
 221                         String prediction = findJnetpred(r.getMap("Predictions", String.class, String.class));
 222                         if (protein != null && prediction != null) {
 223                                 output.put(protein, prediction);
 224                         }
 225                 }
 226                 //final long endTime = System.currentTimeMillis();
 227                 return output;
 228         }
 229
 230         private String findJnetpred(Map<String, String> pred) {
 231                 if (pred != null) {
 232                         if (pred.containsKey("jnetpred"))
 233                                 return pred.get("jnetpred");
 234                 }
 235                 return null;
 236         }
 237
 238         /*
 239          * getting protein sequences by counter
 240          */
 241         public Map<String, Integer> ReadProteinSequenceByCounter() {
 242                 final long startTime = System.currentTimeMillis();
 243                 String com = "SELECT Protein, JobID FROM ProteinRow;";
 244                 System.out.println("Command: " + com);
 245                 ResultSet results = session.execute(com);
 246                 if (results.isExhausted())
 247                         return null;
 248                 final long queryTime = System.currentTimeMillis();
 249                 List<Row> rows = results.all();
 250                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 251                 System.out.println(" rows analysed,  " + rows.size());
 252                 Map<String, Integer> res = new HashMap<String, Integer>();
 253                 int c = 0;
 254                 for (Row r : rows) {
 255                         String protein = r.getString("Protein");
 256                         if (res.containsKey(protein))
 257                                 res.put(protein, res.get(protein) + 1);
 258                         else
 259                                 res.put(protein, 1);
 260                 }
 261                 final long endTime = System.currentTimeMillis();
 262                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 263                 return res;
 264         }
 265
 266         /*
 267          * getting ip by counter
 268          */
 269         public Map<String, Integer> ReadIpByCounter() {
 270                 final long startTime = System.currentTimeMillis();
 271                 String com = "SELECT JobID, ip FROM ProteinLog;";
 272                 System.out.println("Command: " + com);
 273                 ResultSet results = session.execute(com);
 274                 if (results.isExhausted())
 275                         return null;
 276                 final long queryTime = System.currentTimeMillis();
 277                 List<Row> rows = results.all();
 278                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 279                 System.out.println(" rows analysed,  " + rows.size());
 280                 Map<String, Integer> res = new HashMap<String, Integer>();
 281                 int c = 0;
 282                 for (Row r : rows) {
 283                         String ip = r.getString("ip");
 284                         if (res.containsKey(ip))
 285                                 res.put(ip, res.get(ip) + 1);
 286                         else
 287                                 res.put(ip, 1);
 288                 }
 289                 final long endTime = System.currentTimeMillis();
 290                 System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 291                 return res;
 292         }
 293
 294         /*
 295          * getting log info for jobid
 296          */
 297         public JobBean ReadJobLog(String jobid) {
 298                 final long startTime = System.currentTimeMillis();
 299                 String com = "SELECT * FROM ProteinLog WHERE JobID = '" + jobid + "';";
 300                 System.out.println("Command: " + com);
 301                 ResultSet results = session.execute(com);
 302                 if (results.isExhausted())
 303                         return null;
 304                 final long queryTime = System.currentTimeMillis();
 305                 Row row = results.one();
 306                 String com1 = "SELECT * FROM ProteinRow WHERE JobID = '" + jobid + "' ALLOW FILTERING;";
 307                 System.out.println("Command: " + com1);
 308                 ResultSet results1 = session.execute(com1);
 309                 if (results1.isExhausted())
 310                         return null;
 311                 Row row1 = results1.one();
 312                 JobBean res = new JobBean(row.getString("Protein"), row.getString("JobID"), row.getString("DataBegin"), row.getString("DataEnd"),
 313                                 row.getString("ip"), row1.getMap("Predictions", String.class, String.class));
 314                 String program = row.getString("ProgramName");
 315                 String version = row.getString("ProgramVersion");
 316                 if (null != program && null != version) {
 317                         res.setProgramName(program);
 318                         res.setProgramVersion(version);
 319                 }
 320                 System.out.println("Query time is " + (queryTime - startTime) + " msec");
 321                 final long endTime = System.currentTimeMillis();
 322                 System.out.println(" rows analysed, execution time is " + (endTime - startTime) + " msec");
 323                 return res;
 324         }
 325 }