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.cassandra.Pair;
  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                 /* ???? Very strange code...
 116                 DataBase db = new DataBase();
 117                 db.setTimeTotalExec(totalTime);
 118                 query.add(db);
 119                 */
 120                 return query;
 121         }
 122
 123         /*
 124          * query: total number of jobs for the period from date1 till date2
 125         */
 126         public TotalJobsStatisticBean countJobs(String date1, String date2) {
 127                 /*
 128                  * if (null == date1) { date1 = "1970/1/1"; } if (null == date2) { date1
 129                  * = "2100/1/1"; } if (!isThisDateValid(date1, formatYYMMDD) ||
 130                  * !isThisDateValid(date2, formatYYMMDD)) { System.out.println(
 131                  * "CassandraRequester.countJobs: wrong format for date1 " + date1 +
 132                  * "or date2 " + date2); return null; }
 133                  */
 134                 SetDateRange();
 135                 long dateStart = DateParsing(date1, formatYYMMDD);
 136                 long dateEnd = DateParsing(date2, formatYYMMDD);
 137                 /*
 138                  * if (dateEnd < earlestDate || dateStart > currentDate || dateStart >
 139                  * dateEnd) return null; if (dateStart < earlestDate) dateStart =
 140                  * earlestDate; if (dateEnd > currentDate) 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, JobStatus status) {
 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
 176                 if (status == JobStatus.OK) {
 177                         return db.ReadProteinData(day, date);
 178                 }
 179                 return db.ReadFailedJobs(day, date, status);
 180         }
 181
 182         /*
 183          * query: protein sequence
 184          */
 185         public List<ProteinBean> readProteins(String protIn, String searchtype) {
 186                 List<ProteinBean> result;
 187                 if (searchtype.equals("whole"))
 188                         result = db.ReadWholeSequence(protIn);
 189                 else
 190                         result = db.ReadPartOfSequence(protIn);
 191                 if (result == null)
 192                         return null;
 193
 194                 if (searchtype.equals("partial")) {
 195                         for (ProteinBean entry : result) {
 196                                 entry.setSubProt(CreateSubprot(entry.getSequence(), protIn));
 197                         }
 198                 }
 199                 return result;
 200         }
 201
 202         /*
 203          * query: protein feature
 204          */
 205         public Map<String, String> readProteinsPrediction(String feature, int percent) {
 206                 Map<String, String> result = db.ReadProtein();
 207                 ;
 208                 if (result == null)
 209                         return null;
 210                 Map<String, String> query = new HashMap<String, String>();
 211                 for (Map.Entry<String, String> entry : result.entrySet()) {
 212                         String pred = entry.getValue();
 213                         if (pred.replaceAll("[^" + feature + "]", "").length() > pred.length() * percent / 100 && (!entry.getKey().equals(""))) {
 214                                 // if (!entry.getKey().equals(""))
 215                                 query.put(entry.getKey(), pred);
 216                         }
 217                 }
 218                 return query;
 219         }
 220
 221         /*
 222          * query protein sequences with number of jobs
 223          */
 224         public List<TotalByCounterBean> readProteinByCounter(int minimalcounter) {
 225                 List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
 226                 Map<String, Integer> map = db.ReadProteinSequenceByCounter();
 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         /*
 241          * query ip with number of jobs
 242          */
 243         public List<TotalByCounterBean> readIpByCounter(Integer minimalcounter) {
 244                 List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
 245                 Map<String, Integer> map = db.ReadIpByCounter();
 246                 if (minimalcounter == null)
 247                         minimalcounter = 0;
 248                 if (map == null)
 249                         return null;
 250                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
 251                         if (entry.getValue() > minimalcounter) {
 252                                 TotalByCounterBean bean = new TotalByCounterBean();
 253                                 bean.setTotaljobs(entry.getValue());
 254                                 bean.setName(entry.getKey());
 255                                 query.add(bean);
 256                         }
 257                 }
 258                 return query;
 259         }
 260
 261         /*
 262          * query jobs log info
 263          */
 264         public JobBean readJobLog(String jobid) {
 265                 if (jobid == null)
 266                         return null;
 267                 return db.ReadJobLog(jobid);
 268         }
 269
 270         /*
 271          * query jobs by ipStructureJobLog
 272          */
 273         public UserBean readIp(String ip) {
 274                 if (ip == null)
 275                         return null;
 276                 Map<String, String[]> res = db.ReadIpWithJobs(ip);
 277                 if (res == null)
 278                         return null;
 279                 UserBean query = new UserBean(ip);
 280                 query.setMainInfo(res);
 281                 return query;
 282         }
 283
 284         /*
 285          * create list of parts of protein sequence;
 286          */
 287         private static List<String> CreateSubprot(String protein, String subprot) {
 288                 List<String> sub = new ArrayList<String>();
 289                 String subStr = protein;
 290                 while (subStr.length() > 0 && subStr.contains(subprot)) {
 291                         String first = subStr.substring(0, subStr.indexOf(subprot));
 292                         if (first.length() > 0)
 293                                 sub.add(first);
 294                         sub.add(subprot);
 295                         subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
 296                 }
 297                 if (subStr.length() > 0)
 298                         sub.add(subStr);
 299                 return sub;
 300         }
 301
 302         /*
 303          * convert String date into long date (miliseconds since the epoch start)
 304          */
 305         private static long DateParsing(String datInput, SimpleDateFormat formatter) {
 306                 if (datInput == null) {
 307                         return 0;
 308                 }
 309                 long dateWorkSt = 0;
 310
 311                 try {
 312                         dateWorkSt = formatter.parse(datInput).getTime();
 313                 } catch (ParseException e) {
 314                         e.printStackTrace();
 315                 }
 316                 return dateWorkSt;
 317         }
 318
 319         // convert long to date in string format
 320         private static String DateFormat(long inDate) {
 321                 SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
 322                 return datformat.format(new Date(inDate));
 323         }
 324
 325         /*
 326          * set earlest date and current dates. earlestDate is static and should be
 327          * set at the 1st call currentDate should be re-calculated every time
 328          */
 329         private static void SetDateRange() {
 330                 Calendar cal = Calendar.getInstance();
 331                 currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH),
 332                                 formatYYMMDD);
 333                 if (0 == earlestDate) {
 334                         CassandraRequester cr = new CassandraRequester();
 335                         earlestDate = cr.earliestDate();
 336                 }
 337         }
 338
 339         public boolean isThisDateValid(String dateToValidate, SimpleDateFormat sdf) {
 340                 if (dateToValidate == null || dateToValidate.equals("")) {
 341                         return false;
 342                 }
 343                 try {
 344                         // if not valid, this will throw ParseException
 345                         sdf.setLenient(false);
 346                         Date date = sdf.parse(dateToValidate);
 347                 } catch (ParseException e) {
 348                         e.printStackTrace();
 349                         return false;
 350                 }
 351                 return true;
 352         }
 353
 354         /*
 355          * find the earliest date in the database
 356          */
 357         public long earliestDate() {
 358                 earlestDate = CassandraNativeConnector.getEarliestDateInDB();
 359                 return earlestDate;
 360         }
 361
 362         /**
 363          * prepares an example of either job id or IP for the DB
 364          *
 365          * @param exampletype
 366          *            defines which example you need (an existing job from the DB -
 367          *            jobid, an IP - "ip")
 368          * @return a string representation of the requested example, if the example
 369          *         type is not known empty string is returned
 370          */
 371         public String getExample(String exampletype) {
 372                 if (exampletype.equals("jobid")) {
 373                         return "jp_NzBOJKo";
 374                 } else if (exampletype.equals("ip")) {
 375                         return "127.0.0.1";
 376                 }
 377                 return "";
 378         }
 379
 380 }