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