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