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.beans.DateBean;
  13 import compbio.beans.JobBean;
  14 import compbio.beans.ProteinBean;
  15 import compbio.beans.Total;
  16 import compbio.beans.TotalByCounterBean;
  17 import compbio.beans.TotalJobsStatisticBean;
  18 import compbio.beans.UserBean;
  19 import compbio.cassandra.CassandraNativeConnector;
  20 import compbio.cassandra.CassandraReaderOld;
  21 import compbio.cassandra.DataBase;
  22 import compbio.engine.JobStatus;
  23 import compbio.engine.Pair;
  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                 /*
 116                  * ???? Very strange code... DataBase db = new DataBase();
 117                  * db.setTimeTotalExec(totalTime); query.add(db);
 118                  */
 119                 return query;
 120         }
 121
 122         /*
 123          * query: total number of jobs for the period from date1 till date2
 124          */
 125         public TotalJobsStatisticBean countJobs(String date1, String date2) {
 126                 /*
 127                  * if (null == date1) { date1 = "1970/1/1"; } if (null == date2) { date1
 128                  * = "2100/1/1"; } if (!isThisDateValid(date1, formatYYMMDD) ||
 129                  * !isThisDateValid(date2, formatYYMMDD)) { System.out.println(
 130                  * "CassandraRequester.countJobs: wrong format for date1 " + date1 +
 131                  * "or date2 " + date2); return null; }
 132                  */
 133                 SetDateRange();
 134                 long dateStart = DateParsing(date1, formatYYMMDD);
 135                 long dateEnd = DateParsing(date2, formatYYMMDD);
 136                 /*
 137                  * if (dateEnd < earlestDate || dateStart > currentDate || dateStart >
 138                  * dateEnd) return null; if (dateStart < earlestDate) dateStart =
 139                  * earlestDate; if (dateEnd > currentDate) dateStart = currentDate;
 140                  */
 141                 Calendar start = Calendar.getInstance();
 142                 start.setTime(new Date(dateStart));
 143                 Calendar end = Calendar.getInstance();
 144                 end.setTime(new Date(dateEnd));
 145                 TotalJobsStatisticBean query = new TotalJobsStatisticBean();
 146                 Total wholeTotal = new Total(0, 0, 0, 0, 0);
 147                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
 148                         Total res = db.ReadDateTable(date.getTime());
 149                         if (res == null)
 150                                 continue;
 151                         query.setDateTotal(DateFormat(date.getTime()), res);
 152                         wholeTotal.setTotal(res.getTotal() + wholeTotal.getTotal());
 153                         wholeTotal.setTotalOK(res.getTotalOK() + wholeTotal.getTotalOK());
 154                         wholeTotal.setTotalStopped(res.getTotalStopped() + wholeTotal.getTotalStopped());
 155                         wholeTotal.setTotalError(res.getTotalError() + wholeTotal.getTotalError());
 156                         wholeTotal.setTotalTimeOut(res.getTotalTimeOut() + wholeTotal.getTotalTimeOut());
 157                 }
 158                 query.setWholeTotal(wholeTotal);
 159                 return query;
 160         }
 161
 162         /*
 163          * query: jobs and sequence at date
 164          */
 165         public DateBean readJobByDay(String date, JobStatus status) {
 166                 if (!isThisDateValid(date, formatDDMMYY)) {
 167                         System.out.println("CassandraRequester.readJobByDay: Wrong date format for " + date);
 168                         return null;
 169                 }
 170                 SetDateRange();
 171                 long day = DateParsing(date, formatDDMMYY);
 172                 if (day < earlestDate || day > currentDate)
 173                         return null;
 174
 175                 if (status == JobStatus.OK) {
 176                         return db.ReadProteinData(day, date);
 177                 }
 178                 return db.ReadFailedJobs(day, date, status);
 179         }
 180
 181         /*
 182          * query: protein sequence
 183          */
 184         public List<ProteinBean> readProteins(String protIn, String searchtype) {
 185                 List<ProteinBean> result;
 186                 if (searchtype.equals("whole"))
 187                         result = db.ReadWholeSequence(protIn);
 188                 else
 189                         result = db.ReadPartOfSequence(protIn);
 190                 if (result == null)
 191                         return null;
 192
 193                 if (searchtype.equals("partial")) {
 194                         for (ProteinBean entry : result) {
 195                                 entry.setSubProt(CreateSubprot(entry.getSequence(), protIn));
 196                         }
 197                 }
 198                 return result;
 199         }
 200
 201         /*
 202          * query: protein feature
 203          */
 204
 205         public Map<String, String> readProteinsPrediction(String feature, int percent) {
 206                 Map<String, String> results = db.ReadProtein();
 207                 if (results == null)
 208                         return null;
 209                 Map<String, String> query = new HashMap<String, String>();
 210                 for (Map.Entry<String, String> entry : results.entrySet()) {
 211                         String prediction = entry.getValue();
 212                         String protein = entry.getKey();
 213                         if (prediction.replaceAll("[^" + feature + "]", "").length() > prediction.length() * percent / 100 && !protein.equals("")) {
 214                                 query.put(protein, prediction);
 215                         }
 216                 }
 217                 return query;
 218         }
 219
 220         /*
 221          * query protein sequences with number of jobs
 222          */
 223         public List<TotalByCounterBean> readProteinByCounter(int minimalcounter) {
 224                 List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
 225                 Map<String, Integer> map = db.ReadProteinSequenceByCounter();
 226                 if (map == null)
 227                         return null;
 228                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
 229                         if (entry.getValue() > minimalcounter) {
 230                                 TotalByCounterBean bean = new TotalByCounterBean();
 231                                 bean.setTotaljobs(entry.getValue());
 232                                 bean.setName(entry.getKey());
 233                                 query.add(bean);
 234                         }
 235                 }
 236                 return query;
 237         }
 238
 239         /*
 240          * query ip with number of jobs
 241          */
 242         public List<TotalByCounterBean> readIpByCounter(Integer minimalcounter) {
 243                 List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
 244                 Map<String, Integer> map = db.ReadIpByCounter();
 245                 if (minimalcounter == null)
 246                         minimalcounter = 0;
 247                 if (map == null)
 248                         return null;
 249                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
 250                         if (entry.getValue() > minimalcounter) {
 251                                 TotalByCounterBean bean = new TotalByCounterBean();
 252                                 bean.setTotaljobs(entry.getValue());
 253                                 bean.setName(entry.getKey());
 254                                 query.add(bean);
 255                         }
 256                 }
 257                 return query;
 258         }
 259
 260         /*
 261          * query jobs log info
 262          */
 263         public JobBean readJobLog(String jobid) {
 264                 if (jobid == null)
 265                         return null;
 266                 return db.ReadJobLog(jobid);
 267         }
 268
 269         /*
 270          * query jobs by ipStructureJobLog
 271          */
 272         public UserBean readIp(String ip) {
 273                 if (ip == null)
 274                         return null;
 275                 Map<String, String[]> res = db.ReadIpWithJobs(ip);
 276                 if (res == null)
 277                         return null;
 278                 UserBean query = new UserBean(ip);
 279                 query.setMainInfo(res);
 280                 return query;
 281         }
 282
 283         /*
 284          * create list of parts of protein sequence;
 285          */
 286         private static List<String> CreateSubprot(String protein, String subprot) {
 287                 List<String> sub = new ArrayList<String>();
 288                 String subStr = protein;
 289                 while (subStr.length() > 0 && subStr.contains(subprot)) {
 290                         String first = subStr.substring(0, subStr.indexOf(subprot));
 291                         if (first.length() > 0)
 292                                 sub.add(first);
 293                         sub.add(subprot);
 294                         subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
 295                 }
 296                 if (subStr.length() > 0)
 297                         sub.add(subStr);
 298                 return sub;
 299         }
 300
 301         /*
 302          * convert String date into long date (miliseconds since the epoch start)
 303          */
 304         private static long DateParsing(String datInput, SimpleDateFormat formatter) {
 305                 if (datInput == null) {
 306                         return 0;
 307                 }
 308                 long dateWorkSt = 0;
 309
 310                 try {
 311                         dateWorkSt = formatter.parse(datInput).getTime();
 312                 } catch (ParseException e) {
 313                         e.printStackTrace();
 314                 }
 315                 return dateWorkSt;
 316         }
 317
 318         // convert long to date in string format
 319         private static String DateFormat(long inDate) {
 320                 SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
 321                 return datformat.format(new Date(inDate));
 322         }
 323
 324         /*
 325          * set earlest date and current dates. earlestDate is static and should be
 326          * set at the 1st call currentDate should be re-calculated every time
 327          */
 328         private static void SetDateRange() {
 329                 Calendar cal = Calendar.getInstance();
 330                 currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH),
 331                                 formatYYMMDD);
 332                 if (0 == earlestDate) {
 333                         CassandraRequester cr = new CassandraRequester();
 334                         earlestDate = cr.earliestDate();
 335                 }
 336         }
 337
 338         public boolean isThisDateValid(String dateToValidate, SimpleDateFormat sdf) {
 339                 if (dateToValidate == null || dateToValidate.equals("")) {
 340                         return false;
 341                 }
 342                 try {
 343                         // if not valid, this will throw ParseException
 344                         sdf.setLenient(false);
 345                         Date date = sdf.parse(dateToValidate);
 346                 } catch (ParseException e) {
 347                         e.printStackTrace();
 348                         return false;
 349                 }
 350                 return true;
 351         }
 352
 353         /*
 354          * find the earliest date in the database
 355          */
 356         public long earliestDate() {
 357                 earlestDate = CassandraNativeConnector.getEarliestDateInDB();
 358                 return earlestDate;
 359         }
 360
 361         /**
 362          * prepares an example of either job id or IP for the DB
 363          *
 364          * @param exampletype
 365          *            defines which example you need (an existing job from the DB -
 366          *            jobid, an IP - "ip")
 367          * @return a string representation of the requested example, if the example
 368          *         type is not known empty string is returned
 369          */
 370         public String getExample(String exampletype) {
 371                 if (exampletype.equals("jobid")) {
 372                         return "jp_NzBOJKo";
 373                 } else if (exampletype.equals("ip")) {
 374                         return "127.0.0.1";
 375                 }
 376                 return "";
 377         }
 378
 379 }