datadb/compbio/cassandra/CassandraWriter.java

   1 package compbio.cassandra;
   2
   3 import java.util.List;
   4
   5 import org.apache.log4j.Logger;
   6 import org.springframework.dao.DataIntegrityViolationException;
   7
   8 import com.datastax.driver.core.Row;
   9 import com.datastax.driver.core.Session;
  10 import com.datastax.driver.core.ResultSet;
  11 import com.datastax.driver.core.exceptions.QueryExecutionException;
  12 import com.datastax.driver.core.exceptions.QueryValidationException;
  13
  14 import compbio.data.sequence.FastaSequence;
  15 import compbio.engine.JpredJob;
  16 import compbio.engine.ProteoCachePropertyHelperManager;
  17 import compbio.util.PropertyHelper;
  18
  19 public class CassandraWriter {
  20         private Session session;
  21         private static final PropertyHelper ph = ProteoCachePropertyHelperManager.getPropertyHelper();
  22         private static Logger log = Logger.getLogger(CassandraNativeConnector.class);
  23
  24         CassandraWriter() {
  25                 Session inis = CassandraNativeConnector.getSession();
  26                 setSession(inis);
  27         }
  28
  29         public void setSession(Session s) {
  30                 assert s != null;
  31                 session = s;
  32         }
  33
  34         private ResultSet execute(String command) {
  35                 try {
  36                         ResultSet results = session.execute(command);
  37                         return results;
  38                 } catch (QueryExecutionException e) {
  39                         String mess = "CassandraWriter: query execution exception...";
  40                         System.out.println(mess);
  41                         log.error(mess);
  42                         log.error(e.getLocalizedMessage(), e.getCause());
  43                         return null;
  44                 } catch (QueryValidationException e) {
  45                         String mess = "CassandraWriter: query validation exception... Command: " + command;
  46                         System.out.println(mess);
  47                         log.error(mess);
  48                         log.error(e.getLocalizedMessage(), e.getCause());
  49                         return null;
  50                 }
  51         }
  52
  53         public boolean JobisNotInsterted(String jobid) {
  54                 ResultSet results = execute("SELECT * FROM ProteinLog WHERE JobID = '" + jobid + "';");
  55                 if (null != results && results.isExhausted()) {
  56                         return true;
  57                 }
  58                 return false;
  59         }
  60
  61         public boolean JobisNotArchived(String jobid) {
  62                 ResultSet results = execute("SELECT * FROM JpredArchive WHERE JobID = '" + jobid + "';");
  63                 if (null != results && results.isExhausted()) {
  64                         return true;
  65                 }
  66                 return false;
  67         }
  68
  69         /*
  70          * inserting data into the tables for queries
  71          */
  72         public int FormQueryTables(JpredJob job) {
  73                 if (JobisNotInsterted(job.getJobID())) {
  74                         String id = job.getJobID();
  75                         String protein = job.getProtein();
  76                         String finalstatus = job.getFinalStatus();
  77                         String execstatus = job.getExecutionStatus();
  78
  79                         String com1 = "INSERT INTO ProteinLog (JobID, IP, DataBegin, DataEnd, FinalStatus, ExecutionStatus, Protein) VALUES ('" + id
  80                                         + "','" + job.getIP() + "','" + job.getStartingTimeStr() + "','" + job.getEndTimeStr() + "','" + finalstatus + "','"
  81                                         + execstatus + "','" + protein + "');";
  82                         ResultSet insert = execute(com1);
  83                         if (null == insert) {
  84                                 System.out.println("CassandraWriter.FormQueryTables: couldn't insert into ProteinLog");
  85                                 // return 0;
  86                         }
  87                         if (finalstatus.equals("OK")) {
  88                                 String com2 = "INSERT INTO ProteinData " + "(jobtime, JobID, ExecTime, Protein)" + " VALUES (" + job.getStartingDate()
  89                                                 + ",'" + id + "'," + job.getExecutionTime() + ",'" + protein + "');";
  90                                 if (null == execute(com2)) {
  91                                         System.out.println("CassandraWriter.FormQueryTables: couldn't insert into ProteinData");
  92                                         // return 0;
  93                                 }
  94
  95                                 String allpredictions = "";
  96                                 List<FastaSequence> pr = job.getPredictions();
  97                                 for (FastaSequence pred : pr) {
  98                                         String predictionname = pred.getId();
  99                                         String prediction = pred.getSequence().replaceAll("\n", "");
 100                                         allpredictions += "'" + predictionname + "':'" + prediction + "',";
 101                                 }
 102                                 String final_prediction = "";
 103                                 if (!allpredictions.equals("")) {
 104                                         final_prediction = allpredictions.substring(0, allpredictions.length() - 1);
 105                                 }
 106
 107                                 ResultSet results2 = execute("SELECT * FROM ProteinRow WHERE JobID = '" + job.getJobID() + "';");
 108                                 if (null != results2 && results2.isExhausted()) {
 109                                         String com3 = "INSERT INTO ProteinRow (Protein, JobID, Predictions) VALUES ('" + protein + "','" + id + "',{"
 110                                                         + final_prediction + "});";
 111                                         if (null == execute(com3)) {
 112                                                 System.out.println("CassandraWriter.FormQueryTables: couldn't insert into ProteinRow");
 113                                                 return 0;
 114                                         }
 115                                 }
 116                         } else {
 117                                 String com5 = "INSERT INTO FailLog (jobtime, JobID, ExecTime, ip, FinalStatus) VALUES (" + job.getStartingDate() + ",'"
 118                                                 + id + "'," + job.getExecutionTime() + ",'" + job.getIP() + "', '" + finalstatus + "');";
 119                                 if (null == execute(com5)) {
 120                                         System.out.println("CassandraWriter.FormQueryTables: couldn't insert into FailLog");
 121                                         return 0;
 122                                 }
 123                         }
 124
 125                         // update Main parameters if the job is the earliest job so far
 126                         ResultSet results3 = execute("SELECT * FROM MainParameters WHERE Name = 'EarliestJobDate';");
 127                         if (null == results3) {
 128                                 System.out.println("CassandraWriter.FormQueryTables: couldn't get results from MainParameters");
 129                                 // return 0;
 130                         }
 131                         boolean updateparameter = true;
 132                         if (!results3.isExhausted()) {
 133                                 Row r = results3.one();
 134                                 if (job.getStartingDate() >= Long.parseLong(r.getString("Value")))
 135                                         updateparameter = false;
 136                         }
 137                         if (updateparameter) {
 138                                 String com6 = "INSERT INTO MainParameters (Name, Value) VALUES ('EarliestJobDate','" + job.getStartingDate() + "');";
 139                                 if (null == execute(com6)) {
 140                                         System.out.println("CassandraWriter.FormQueryTables: couldn't insert into MainParameters");
 141                                         return 0;
 142                                 }
 143                         }
 144
 145                         // update internal job counts (used by the Daily Statistics
 146                         // requests)
 147                         // TODO I don't like the bit of code. There should not be so many
 148                         // counters...
 149                         int njobsTotal = 1;
 150                         int njobsOk = 0;
 151                         int njobsStop = 0;
 152                         int njobsError = 0;
 153                         int njobsTimeOut = 0;
 154                         if (finalstatus.equals("OK"))
 155                                 njobsOk = 1;
 156                         else if (finalstatus.equals("TIMEDOUT"))
 157                                 njobsTimeOut = 1;
 158                         else if (finalstatus.equals("JPREDERROR"))
 159                                 njobsError = 1;
 160                         else if (finalstatus.equals("STOPPED"))
 161                                 njobsStop = 1;
 162                         ResultSet results4 = execute("SELECT * FROM JobDateInfo WHERE jobday = " + job.getStartingDate() + ";");
 163                         if (null == results4) {
 164                                 System.out.println("CassandraWriter.FormQueryTables: couldn't get data from JobDateInfo");
 165                                 // return 0;
 166                         }
 167                         if (!results4.isExhausted()) {
 168                                 Row r = results4.one();
 169                                 njobsTotal += r.getLong("Total");
 170                                 njobsOk += r.getLong("TotalOK");
 171                                 njobsError += r.getLong("TotalError");
 172                                 njobsStop += r.getLong("TotalStopped");
 173                                 njobsTimeOut += r.getLong("TotalTimeOut");
 174                         }
 175                         String com = "INSERT INTO JobDateInfo " + "(jobday, Total, TotalOK, TotalStopped, TotalError, TotalTimeOut)" + " VALUES ("
 176                                         + job.getStartingDate() + "," + njobsTotal + "," + njobsOk + "," + njobsStop + "," + njobsError + "," + njobsTimeOut
 177                                         + ");";
 178                         if (null == execute(com)) {
 179                                 System.out.println("CassandraWriter.FormQueryTables: couldn't insert into JobDateInfo");
 180                                 // return 0;
 181                         }
 182                         return 1;
 183                 }
 184                 return 0;
 185         }
 186
 187         /*
 188          * insert data from a real Jpred job: timing+IP, Execution Status, Final
 189          * status, protein sequence, predictions, alignment, LOG and tar.gz files
 190          */
 191         public int ArchiveData(JpredJob job, String archivepath) {
 192                 if (JobisNotArchived(job.getJobID())) {
 193                         String id = job.getJobID();
 194                         String log = job.getLog().replaceAll("'", "");
 195                         String com = "INSERT INTO JpredArchive (JobID, Protein, IP, StartTime, ExecTime, FinalStatus, ExecutionStatus, LOG, ArchiveLink) VALUES ('"
 196                                         + id
 197                                         + "','"
 198                                         + job.getProtein()
 199                                         + "','"
 200                                         + job.getIP()
 201                                         + "',"
 202                                         + job.getStartingTime()
 203                                         + ","
 204                                         + job.getExecutionTime()
 205                                         + ",'" + job.getFinalStatus() + "','" + job.getExecutionStatus() + "','" + log + "','" + archivepath + "');";
 206                         if (null == execute(com)) {
 207                                 System.out.println("CassandraWriter.ArchiveData: couldn't insert into JpredArchive");
 208                         }
 209                         List<FastaSequence> predictions = job.getPredictions();
 210                         for (FastaSequence p : predictions) {
 211                                 if (null == execute("UPDATE JpredArchive SET predictions = predictions + {'" + p.getId() + "':'"
 212                                                 + p.getSequence().replaceAll("\n", "") + "'} WHERE JobID = '" + id + "';")) {
 213                                         System.out.println("CassandraWriter.ArchiveData: couldn't update data in JpredArchive");
 214                                 }
 215                         }
 216
 217                         List<FastaSequence> seqs = job.getAlignment();
 218
 219                         for (FastaSequence s : seqs) {
 220                                 String com2 = "UPDATE JpredArchive SET alignment = alignment + {'" + s.getId() + "':'"
 221                                                 + s.getSequence().replaceAll("\n", "") + "'} WHERE JobID = '" + id + "';";
 222                                 if (null == execute(com2)) {
 223                                         System.out.println("CassandraWriter.ArchiveData: couldn't update data in JpredArchive");
 224                                 }
 225                         }
 226                         return 1;
 227                 }
 228                 return 0;
 229         }
 230
 231 }