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