server/compbio/statistic/CassandraRequester.java

   1 package compbio.statistic;
   2
   3 import java.text.ParseException;
   4 import java.text.SimpleDateFormat;
   5 import java.util.ArrayList;
   6 import java.util.Calendar;
   7 import java.util.Date;
   8 import java.util.HashMap;
   9 import java.util.List;
  10 import java.util.Map;
  11
  12 import compbio.cassandra.DateBean;
  13 import compbio.cassandra.ProteinBean;
  14 import compbio.cassandra.CassandraNativeConnector;
  15 import compbio.cassandra.CassandraReaderOld;
  16 import compbio.cassandra.DataBase;
  17 import compbio.cassandra.Pair;
  18 import compbio.cassandra.JobBean;
  19 import compbio.cassandra.Total;
  20 import compbio.cassandra.TotalByCounterBean;
  21 import compbio.cassandra.TotalJobsStatisticBean;
  22 import compbio.cassandra.UserBean;
  23 import compbio.engine.JobStatus;
  24
  25 public class CassandraRequester {
  26         private CassandraReaderOld db = new CassandraReaderOld();
  27         private ArrayList<DataBase> query;
  28         private static long currentDate = 0;
  29         private static long earlestDate = 0;
  30         private final static SimpleDateFormat formatYYMMDD = new SimpleDateFormat("yyyy/MM/dd");
  31         private final static SimpleDateFormat formatDDMMYY = new SimpleDateFormat("dd/MM/yyyy");
  32
  33         /*
  34          * query: execution time for the period from date1 till date2
  35          */
  36         public List<DataBase> extractExecutionTime(String date1, String date2) {
  37                 if (null == date1) {
  38                         date1 = "1970/1/1";
  39                 }
  40                 if (null == date2) {
  41                         date1 = "2100/1/1";
  42                 }
  43                 if (!isThisDateValid(date1, formatYYMMDD) || !isThisDateValid(date2, formatYYMMDD)) {
  44                         System.out.println("CassandraRequester.extractExecutionTime: wrong format for date1 " + date1 + "or date2 " + date2);
  45                         return null;
  46                 }
  47                 SetDateRange();
  48                 int nbins = 5;
  49                 long dateStart = DateParsing(date1, formatYYMMDD);
  50                 long dateEnd = DateParsing(date2, formatYYMMDD);
  51                 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
  52                         return null;
  53                 if (dateStart < earlestDate)
  54                         dateStart = earlestDate;
  55                 if (dateEnd > currentDate)
  56                         dateStart = currentDate;
  57
  58                 Calendar start = Calendar.getInstance();
  59                 start.setTime(new Date(dateStart));
  60                 Calendar end = Calendar.getInstance();
  61                 end.setTime(new Date(dateEnd));
  62                 query = new ArrayList<DataBase>();
  63                 List<Integer> totalTime = new ArrayList<Integer>();
  64                 for (int i = 0; i < nbins; i++)
  65                         totalTime.add(i, 0);
  66                 List<Pair<String, String>> res = db.ReadProteinDataTable();
  67                 List<Pair<Date, Long>> numres = new ArrayList<Pair<Date, Long>>();
  68
  69                 for (Pair<String, String> entry : res) {
  70                         SimpleDateFormat dateformatter = new SimpleDateFormat("yyyy/MM/dd");
  71                         try {
  72                                 Date jobstartdate = dateformatter.parse(entry.getElement0());
  73                                 long date = jobstartdate.getTime();
  74                                 if (dateStart <= date && date <= dateEnd) {
  75                                         SimpleDateFormat datetimeformatter = new SimpleDateFormat("yyyy/MM/dd:H:m:s");
  76                                         Date jobstarttime = datetimeformatter.parse(entry.getElement0());
  77                                         Date jobendtime = datetimeformatter.parse(entry.getElement1());
  78                                         long diff = (jobendtime.getTime() - jobstarttime.getTime()) / 1000;
  79                                         Pair<Date, Long> pair = new Pair<Date, Long>(jobstartdate, Long.valueOf(diff));
  80                                         numres.add(pair);
  81                                 }
  82                         } catch (ParseException e) {
  83                                 e.printStackTrace();
  84                         }
  85                 }
  86
  87                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
  88                         List<Integer> timeResult = new ArrayList<Integer>();
  89                         for (int i = 0; i < nbins; i++)
  90                                 timeResult.add(i, 0);
  91                         for (Pair<Date, Long> p : numres) {
  92                                 if (date.equals(p.getElement0())) {
  93                                         long lenResult = p.getElement1().longValue();
  94                                         if (lenResult <= 30)
  95                                                 timeResult.set(0, timeResult.get(0) + 1);
  96                                         else if (lenResult > 30 && lenResult <= 60)
  97                                                 timeResult.set(1, timeResult.get(1) + 1);
  98                                         else if (lenResult > 60 && lenResult <= 120)
  99                                                 timeResult.set(2, timeResult.get(2) + 1);
 100                                         else if (lenResult > 120 && lenResult <= 600)
 101                                                 timeResult.set(3, timeResult.get(3) + 1);
 102                                         else {
 103                                                 timeResult.set(4, timeResult.get(4) + 1);
 104                                         }
 105                                 }
 106                         }
 107                         for (int i = 0; i < nbins; i++)
 108                                 totalTime.set(i, totalTime.get(i) + timeResult.get(i));
 109                         DataBase db = new DataBase();
 110                         db.setTimeRez(timeResult);
 111                         db.setDate(DateFormat(date.getTime()));
 112                         query.add(db);
 113                 }
 114
 115                 DataBase db = new DataBase();
 116                 db.setTimeTotalExec(totalTime);
 117                 query.add(db);
 118                 return query;
 119         }
 120
 121         /*
 122          * query: total number of jobs for the period from date1 till date2
 123          */
 124         public TotalJobsStatisticBean countJobs(String date1, String date2) {
 125         /*      if (null == date1) {
 126                         date1 = "1970/1/1";
 127                 }
 128                 if (null == date2) {
 129                         date1 = "2100/1/1";
 130                 }
 131                 if (!isThisDateValid(date1, formatYYMMDD) || !isThisDateValid(date2, formatYYMMDD)) {
 132                         System.out.println("CassandraRequester.countJobs: wrong format for date1 " + date1 + "or date2 " + date2);
 133                         return null;
 134                 }*/
 135                 SetDateRange();
 136                 long dateStart = DateParsing(date1, formatYYMMDD);
 137                 long dateEnd = DateParsing(date2, formatYYMMDD);
 138 /*              if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
 139                         return null;
 140                 if (dateStart < earlestDate)
 141                         dateStart = earlestDate;
 142                 if (dateEnd > currentDate)
 143                         dateStart = currentDate;*/
 144                 Calendar start = Calendar.getInstance();
 145                 start.setTime(new Date(dateStart));
 146                 Calendar end = Calendar.getInstance();
 147                 end.setTime(new Date(dateEnd));
 148                 TotalJobsStatisticBean query = new TotalJobsStatisticBean();
 149                 Total wholeTotal = new Total(0, 0, 0, 0, 0);
 150                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
 151                         Total res = db.ReadDateTable(date.getTime());
 152                         if (res == null)
 153                                 continue;
 154                         query.setDateTotal(DateFormat(date.getTime()), res);
 155                         wholeTotal.setTotal(res.getTotal() + wholeTotal.getTotal());
 156                         wholeTotal.setTotalOK(res.getTotalOK() + wholeTotal.getTotalOK());
 157                         wholeTotal.setTotalStopped(res.getTotalStopped() + wholeTotal.getTotalStopped());
 158                         wholeTotal.setTotalError(res.getTotalError() + wholeTotal.getTotalError());
 159                         wholeTotal.setTotalTimeOut(res.getTotalTimeOut() + wholeTotal.getTotalTimeOut());
 160                 }
 161                 query.setWholeTotal(wholeTotal);
 162                 return query;
 163         }
 164
 165         /*
 166          * query: jobs and sequence at date
 167          */
 168         public DateBean readJobByDay(String date, JobStatus status) {
 169                 if (!isThisDateValid(date, formatDDMMYY)) {
 170                         System.out.println("CassandraRequester.readJobByDay: Wrong date format for " + date);
 171                         return null;
 172                 }
 173                 SetDateRange();
 174                 long day = DateParsing(date, formatDDMMYY);
 175                 if (day < earlestDate || day > currentDate)
 176                         return null;
 177
 178                 if (status == JobStatus.OK) {
 179                         return db.ReadProteinData(day, date);
 180                 }
 181                 return db.ReadFailedJobs(day, date, status);
 182         }
 183
 184         /*
 185          * query: protein sequence
 186          */
 187         public List<ProteinBean> readProteins(String protIn, String searchtype) {
 188                 List<ProteinBean> result;
 189                 if (searchtype.equals("whole"))
 190                         result = db.ReadWholeSequence(protIn);
 191                 else
 192                         result = db.ReadPartOfSequence(protIn);
 193                 if (result == null)
 194                         return null;
 195
 196                 if (searchtype.equals("partial")) {
 197                         for (ProteinBean entry : result) {
 198                                 entry.setSubProt(CreateSubprot(entry.getSequence(), protIn));
 199                         }
 200                 }
 201                 return result;
 202         }
 203
 204         /*
 205          * query: protein feature
 206          */
 207         public Map<String, String> readProteinsPrediction(String feature, int percent) {
 208                 Map<String, String> result = db.ReadProtein();;
 209                 if (result == null)
 210                         return null;
 211                 Map<String, String> query = new HashMap<String, String>();
 212                 for (Map.Entry<String, String> entry : result.entrySet()) {
 213                         String pred = entry.getValue();
 214                         if (pred.replaceAll("[^"+feature+"]", "").length() > pred.length() * percent / 100 && (!entry.getKey().equals(""))) {
 215                         //      if (!entry.getKey().equals(""))
 216                                         query.put(entry.getKey(), pred);
 217                         }
 218                 }
 219                 return query;
 220         }
 221
 222         /*
 223          * query protein sequences with number of jobs
 224          */
 225         public List<TotalByCounterBean> readProteinByCounter(int minimalcounter) {
 226                 List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
 227                 Map<String, Integer> map = db.ReadProteinSequenceByCounter();
 228                 if (map == null)
 229                         return null;
 230                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
 231                         if (entry.getValue() > minimalcounter) {
 232                                 TotalByCounterBean bean = new TotalByCounterBean();
 233                                 bean.setTotaljobs(entry.getValue());
 234                                 bean.setName(entry.getKey());
 235                                 query.add(bean);
 236                         }
 237                 }
 238                 return query;
 239         }
 240
 241         /*
 242          * query ip with number of jobs
 243          */
 244         public List<TotalByCounterBean> readIpByCounter(Integer minimalcounter) {
 245                 List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
 246                 Map<String, Integer> map = db.ReadIpByCounter();
 247                 if (minimalcounter == null)
 248                         minimalcounter = 0;
 249                 if (map == null)
 250                         return null;
 251                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
 252                         if (entry.getValue() > minimalcounter) {
 253                                 TotalByCounterBean bean = new TotalByCounterBean();
 254                                 bean.setTotaljobs(entry.getValue());
 255                                 bean.setName(entry.getKey());
 256                                 query.add(bean);
 257                         }
 258                 }
 259                 return query;
 260         }
 261
 262         /*
 263          * query jobs log info
 264          */
 265         public JobBean readJobLog(String jobid) {
 266                 if (jobid == null)
 267                         return null;
 268                 return db.ReadJobLog(jobid);
 269         }
 270
 271         /*
 272          * query jobs by ipStructureJobLog
 273          */
 274         public UserBean readIp(String ip) {
 275                 if (ip == null)
 276                         return null;
 277                 Map<String, String[]> res = db.ReadIpWithJobs(ip);
 278                 if (res == null)
 279                         return null;
 280                 UserBean query = new UserBean(ip);
 281                 query.setMainInfo(res);
 282                 return query;
 283         }
 284
 285         /*
 286          * create list of parts of protein sequence;
 287          */
 288         private static List<String> CreateSubprot(String protein, String subprot) {
 289                 List<String> sub = new ArrayList<String>();
 290                 String subStr = protein;
 291                 while (subStr.length() > 0 && subStr.contains(subprot)) {
 292                         String first = subStr.substring(0, subStr.indexOf(subprot));
 293                         if (first.length() > 0)
 294                                 sub.add(first);
 295                         sub.add(subprot);
 296                         subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
 297                 }
 298                 if (subStr.length() > 0)
 299                         sub.add(subStr);
 300                 return sub;
 301         }
 302
 303         /*
 304          * convert String date into long date (miliseconds since the epoch start)
 305          */
 306         private static long DateParsing(String datInput, SimpleDateFormat formatter) {
 307                 if (datInput == null) {
 308                         return 0;
 309                 }
 310                 long dateWorkSt = 0;
 311
 312                 try {
 313                         dateWorkSt = formatter.parse(datInput).getTime();
 314                 } catch (ParseException e) {
 315                         e.printStackTrace();
 316                 }
 317                 return dateWorkSt;
 318         }
 319
 320         // convert long to date in string format
 321         private static String DateFormat(long inDate) {
 322                 SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
 323                 return datformat.format(new Date(inDate));
 324         }
 325
 326         /*
 327          * set earlest date and current dates. earlestDate is static and should be
 328          * set at the 1st call currentDate should be re-calculated every time
 329          */
 330         private static void SetDateRange() {
 331                 Calendar cal = Calendar.getInstance();
 332                 currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH),
 333                                 formatYYMMDD);
 334                 if (0 == earlestDate) {
 335                         CassandraRequester cr = new CassandraRequester();
 336                         earlestDate = cr.earliestDate();
 337                 }
 338         }
 339
 340         public boolean isThisDateValid(String dateToValidate, SimpleDateFormat sdf) {
 341                 if (dateToValidate == null || dateToValidate.equals("")) {
 342                         return false;
 343                 }
 344                 try {
 345                         // if not valid, this will throw ParseException
 346                         sdf.setLenient(false);
 347                         Date date = sdf.parse(dateToValidate);
 348                 } catch (ParseException e) {
 349                         e.printStackTrace();
 350                         return false;
 351                 }
 352                 return true;
 353         }
 354
 355         /*
 356          * find the earliest date in the database
 357          */
 358         public long earliestDate() {
 359                 earlestDate = CassandraNativeConnector.getEarliestDateInDB();
 360                 return earlestDate;
 361         }
 362
 363 }