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.List;
   9 import java.util.Map;
  10
  11 import compbio.cassandra.DateBean;
  12 import compbio.cassandra.ProteinBean;
  13 import compbio.cassandra.CassandraNativeConnector;
  14 import compbio.cassandra.CassandraReader;
  15 import compbio.cassandra.DataBase;
  16 import compbio.cassandra.Pair;
  17 import compbio.cassandra.JobBean;
  18 import compbio.cassandra.Total;
  19 import compbio.cassandra.TotalByCounterBean;
  20 import compbio.cassandra.TotalJobsStatisticBean;
  21 import compbio.cassandra.UserBean;
  22 import compbio.engine.JobStatus;
  23
  24 public class CassandraRequester {
  25         private CassandraReader db = new CassandraReader();
  26         private ArrayList<DataBase> query;
  27         private static long currentDate = 0;
  28         private static long earlestDate = 0;
  29         private final static SimpleDateFormat formatYYMMDD = new SimpleDateFormat("yyyy/MM/dd");
  30         private final static SimpleDateFormat formatDDMMYY = new SimpleDateFormat("dd/MM/yyyy");
  31
  32         /*
  33          * query: execution time for the period from date1 till date2
  34          */
  35         public List<DataBase> extractExecutionTime(String date1, String date2) {
  36                 if (null == date1) {
  37                         date1 = "1970/1/1";
  38                 }
  39                 if (null == date2) {
  40                         date1 = "2100/1/1";
  41                 }
  42                 if (!isThisDateValid(date1, formatYYMMDD) || !isThisDateValid(date2, formatYYMMDD)) {
  43                         System.out.println("CassandraRequester.extractExecutionTime: wrong format for date1 " + date1 + "or date2 " + date2);
  44                         return null;
  45                 }
  46                 SetDateRange();
  47                 int nbins = 5;
  48                 long dateStart = DateParsing(date1, formatYYMMDD);
  49                 long dateEnd = DateParsing(date2, formatYYMMDD);
  50                 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
  51                         return null;
  52                 if (dateStart < earlestDate)
  53                         dateStart = earlestDate;
  54                 if (dateEnd > currentDate)
  55                         dateStart = currentDate;
  56
  57                 Calendar start = Calendar.getInstance();
  58                 start.setTime(new Date(dateStart));
  59                 Calendar end = Calendar.getInstance();
  60                 end.setTime(new Date(dateEnd));
  61                 query = new ArrayList<DataBase>();
  62                 List<Integer> totalTime = new ArrayList<Integer>();
  63                 for (int i = 0; i < nbins; i++)
  64                         totalTime.add(i, 0);
  65                 List<Pair<String, String>> res = db.ReadProteinDataTable();
  66                 List<Pair<Date, Long>> numres = new ArrayList<Pair<Date, Long>>();
  67
  68                 for (Pair<String, String> entry : res) {
  69                         SimpleDateFormat dateformatter = new SimpleDateFormat("yyyy/MM/dd");
  70                         try {
  71                                 Date jobstartdate = dateformatter.parse(entry.getElement0());
  72                                 long date = jobstartdate.getTime();
  73                                 if (dateStart <= date && date <= dateEnd) {
  74                                         SimpleDateFormat datetimeformatter = new SimpleDateFormat("yyyy/MM/dd:H:m:s");
  75                                         Date jobstarttime = datetimeformatter.parse(entry.getElement0());
  76                                         Date jobendtime = datetimeformatter.parse(entry.getElement1());
  77                                         long diff = (jobendtime.getTime() - jobstarttime.getTime()) / 1000;
  78                                         Pair<Date, Long> pair = new Pair<Date, Long>(jobstartdate, Long.valueOf(diff));
  79                                         numres.add(pair);
  80                                 }
  81                         } catch (ParseException e) {
  82                                 e.printStackTrace();
  83                         }
  84                 }
  85
  86                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
  87                         List<Integer> timeResult = new ArrayList<Integer>();
  88                         for (int i = 0; i < nbins; i++)
  89                                 timeResult.add(i, 0);
  90                         for (Pair<Date, Long> p : numres) {
  91                                 if (date.equals(p.getElement0())) {
  92                                         long lenResult = p.getElement1().longValue();
  93                                         if (lenResult <= 30)
  94                                                 timeResult.set(0, timeResult.get(0) + 1);
  95                                         else if (lenResult > 30 && lenResult <= 60)
  96                                                 timeResult.set(1, timeResult.get(1) + 1);
  97                                         else if (lenResult > 60 && lenResult <= 120)
  98                                                 timeResult.set(2, timeResult.get(2) + 1);
  99                                         else if (lenResult > 120 && lenResult <= 600)
 100                                                 timeResult.set(3, timeResult.get(3) + 1);
 101                                         else {
 102                                                 timeResult.set(4, timeResult.get(4) + 1);
 103                                         }
 104                                 }
 105                         }
 106                         for (int i = 0; i < nbins; i++)
 107                                 totalTime.set(i, totalTime.get(i) + timeResult.get(i));
 108                         DataBase db = new DataBase();
 109                         db.setTimeRez(timeResult);
 110                         db.setDate(DateFormat(date.getTime()));
 111                         query.add(db);
 112                 }
 113
 114                 DataBase db = new DataBase();
 115                 db.setTimeTotalExec(totalTime);
 116                 query.add(db);
 117                 return query;
 118         }
 119
 120         /*
 121          * query: total number of jobs for the period from date1 till date2
 122          */
 123         public TotalJobsStatisticBean countJobs(String date1, String date2) {
 124                 if (null == date1) {
 125                         date1 = "1970/1/1";
 126                 }
 127                 if (null == date2) {
 128                         date1 = "2100/1/1";
 129                 }
 130                 if (!isThisDateValid(date1, formatYYMMDD) || !isThisDateValid(date2, formatYYMMDD)) {
 131                         System.out.println("CassandraRequester.countJobs: wrong format for date1 " + date1 + "or date2 " + date2);
 132                         return null;
 133                 }
 134                 SetDateRange();
 135                 long dateStart = DateParsing(date1, formatYYMMDD);
 136                 long dateEnd = DateParsing(date2, formatYYMMDD);
 137                 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
 138                         return null;
 139                 if (dateStart < earlestDate)
 140                         dateStart = earlestDate;
 141                 if (dateEnd > currentDate)
 142                         dateStart = currentDate;
 143
 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 sequences with number of jobs
 206          */
 207         public List<TotalByCounterBean> readProteinByCounter(int minimalcounter) {
 208                 List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
 209                 Map<String, Integer> map = db.ReadProteinSequenceByCounter();
 210                 if (map == null)
 211                         return null;
 212                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
 213                         if (entry.getValue() > minimalcounter) {
 214                                 TotalByCounterBean bean = new TotalByCounterBean();
 215                                 bean.setTotaljobs(entry.getValue());
 216                                 bean.setName(entry.getKey());
 217                                 query.add(bean);
 218                         }
 219                 }
 220                 return query;
 221         }
 222
 223         /*
 224          * query ip with number of jobs
 225          */
 226         public List<TotalByCounterBean> readIpByCounter(Integer minimalcounter) {
 227                 List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
 228                 Map<String, Integer> map = db.ReadIpByCounter();
 229                 if (minimalcounter == null)
 230                         minimalcounter = 0;
 231                 if (map == null)
 232                         return null;
 233                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
 234                         if (entry.getValue() > minimalcounter) {
 235                                 TotalByCounterBean bean = new TotalByCounterBean();
 236                                 bean.setTotaljobs(entry.getValue());
 237                                 bean.setName(entry.getKey());
 238                                 query.add(bean);
 239                         }
 240                 }
 241                 return query;
 242         }
 243
 244         /*
 245          * query jobs log info
 246          */
 247         public JobBean readJobLog(String jobid) {
 248                 if (jobid == null)
 249                         return null;
 250                 return db.ReadJobLog(jobid);
 251         }
 252
 253         /*
 254          * query jobs by ipStructureJobLog
 255          */
 256         public UserBean readIp(String ip) {
 257                 if (ip == null)
 258                         return null;
 259                 Map<String, String[]> res = db.ReadIpWithJobs(ip);
 260                 if (res == null)
 261                         return null;
 262                 UserBean query = new UserBean(ip);
 263                 query.setMainInfo(res);
 264                 return query;
 265         }
 266
 267         /*
 268          * create list of parts of protein sequence;
 269          */
 270         private static List<String> CreateSubprot(String protein, String subprot) {
 271                 List<String> sub = new ArrayList<String>();
 272                 String subStr = protein;
 273                 while (subStr.length() > 0 && subStr.contains(subprot)) {
 274                         String first = subStr.substring(0, subStr.indexOf(subprot));
 275                         if (first.length() > 0)
 276                                 sub.add(first);
 277                         sub.add(subprot);
 278                         subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
 279                 }
 280                 if (subStr.length() > 0)
 281                         sub.add(subStr);
 282                 return sub;
 283         }
 284
 285         /*
 286          * convert String date into long date (miliseconds since the epoch start)
 287          */
 288         private static long DateParsing(String datInput, SimpleDateFormat formatter) {
 289                 if (datInput == null) {
 290                         return 0;
 291                 }
 292                 long dateWorkSt = 0;
 293
 294                 try {
 295                         dateWorkSt = formatter.parse(datInput).getTime();
 296                 } catch (ParseException e) {
 297                         e.printStackTrace();
 298                 }
 299                 return dateWorkSt;
 300         }
 301
 302         // convert long to date in string format
 303         private static String DateFormat(long inDate) {
 304                 SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
 305                 return datformat.format(new Date(inDate));
 306         }
 307
 308         /*
 309          * set earlest date and current dates. earlestDate is static and should be
 310          * set at the 1st call currentDate should be re-calculated every time
 311          */
 312         private static void SetDateRange() {
 313                 Calendar cal = Calendar.getInstance();
 314                 currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH),
 315                                 formatYYMMDD);
 316                 if (0 == earlestDate) {
 317                         CassandraRequester cr = new CassandraRequester();
 318                         earlestDate = cr.earliestDate();
 319                 }
 320         }
 321
 322         public boolean isThisDateValid(String dateToValidate, SimpleDateFormat sdf) {
 323                 if (dateToValidate == null || dateToValidate.equals("")) {
 324                         return false;
 325                 }
 326                 try {
 327                         // if not valid, this will throw ParseException
 328                         sdf.setLenient(false);
 329                         Date date = sdf.parse(dateToValidate);
 330                 } catch (ParseException e) {
 331                         e.printStackTrace();
 332                         return false;
 333                 }
 334                 return true;
 335         }
 336
 337         /*
 338          * find the earliest date in the database
 339          */
 340         public long earliestDate() {
 341                 earlestDate = CassandraNativeConnector.getEarliestDateInDB();
 342                 return earlestDate;
 343         }
 344
 345 }