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.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.StructureJobLog;
  18
  19 public class CassandraRequester {
  20         private CassandraReader db = new CassandraReader();
  21         private ArrayList<DataBase> query;
  22         private static long currentDate = 0;
  23         private static long earlestDate = 0;
  24         private final static SimpleDateFormat formatYYMMDD = new SimpleDateFormat("yyyy/MM/dd");
  25         private final static SimpleDateFormat formatDDMMYY = new SimpleDateFormat("dd/MM/yyyy");
  26
  27         /*
  28          * query: execution time for the period from date1 till date2
  29          */
  30         public List<DataBase> extractExecutionTime(String date1, String date2) {
  31                 if (null == date1) {
  32                         date1 = "1970/1/1";
  33                 }
  34                 if (null == date2) {
  35                         date1 = "2100/1/1";
  36                 }
  37                 if (!isThisDateValid(date1,formatYYMMDD) || !isThisDateValid(date2,formatYYMMDD)) {
  38                         System.out.println("Wrong date: point 3");
  39                         return null;
  40                 }
  41                 SetDateRange();
  42                 int nbins = 5;
  43                 long dateStart = DateParsing(date1, formatYYMMDD);
  44                 long dateEnd = DateParsing(date2, formatYYMMDD);
  45                 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
  46                         return null;
  47                 if (dateStart < earlestDate)
  48                         dateStart = earlestDate;
  49                 if (dateEnd > currentDate)
  50                         dateStart = currentDate;
  51
  52                 Calendar start = Calendar.getInstance();
  53                 start.setTime(new Date(dateStart));
  54                 Calendar end = Calendar.getInstance();
  55                 end.setTime(new Date(dateEnd));
  56                 query = new ArrayList<DataBase>();
  57                 List<Integer> totalTime = new ArrayList<Integer>();
  58                 for (int i = 0; i < nbins; i++)
  59                         totalTime.add(i, 0);
  60                 List<Pair<String, String>> res = db.ReadProteinDataTable();
  61                 List<Pair<Date, Long>> numres = new ArrayList<Pair<Date, Long>>();
  62
  63                 for (Pair<String, String> entry : res) {
  64                         SimpleDateFormat dateformatter = new SimpleDateFormat("yyyy/MM/dd");
  65                         try {
  66                                 Date jobstartdate = dateformatter.parse(entry.getElement0());
  67                                 long date = jobstartdate.getTime();
  68                                 if (dateStart <= date && date <= dateEnd) {
  69                                         SimpleDateFormat datetimeformatter = new SimpleDateFormat("yyyy/MM/dd:H:m:s");
  70                                         Date jobstarttime = datetimeformatter.parse(entry.getElement0());
  71                                         Date jobendtime = datetimeformatter.parse(entry.getElement1());
  72                                         long diff = (jobendtime.getTime() - jobstarttime.getTime()) / 1000;
  73                                         Pair<Date, Long> pair = new Pair<Date, Long>(jobstartdate, Long.valueOf(diff));
  74                                         numres.add(pair);
  75                                 }
  76                         } catch (ParseException e) {
  77                                 e.printStackTrace();
  78                         }
  79                 }
  80
  81                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
  82                         List<Integer> timeResult = new ArrayList<Integer>();
  83                         for (int i = 0; i < nbins; i++)
  84                                 timeResult.add(i, 0);
  85                         for (Pair<Date, Long> p : numres) {
  86                                 if (date.equals(p.getElement0())) {
  87                                         long lenResult = p.getElement1().longValue();
  88                                         if (lenResult <= 30)
  89                                                 timeResult.set(0, timeResult.get(0) + 1);
  90                                         else if (lenResult > 30 && lenResult <= 60)
  91                                                 timeResult.set(1, timeResult.get(1) + 1);
  92                                         else if (lenResult > 60 && lenResult <= 120)
  93                                                 timeResult.set(2, timeResult.get(2) + 1);
  94                                         else if (lenResult > 120 && lenResult <= 600)
  95                                                 timeResult.set(3, timeResult.get(3) + 1);
  96                                         else {
  97                                                 timeResult.set(4, timeResult.get(4) + 1);
  98                                         }
  99                                 }
 100                         }
 101                         for (int i = 0; i < nbins; i++)
 102                                 totalTime.set(i, totalTime.get(i) + timeResult.get(i));
 103                         DataBase db = new DataBase();
 104                         db.setTimeRez(timeResult);
 105                         db.setDate(DateFormat(date.getTime()));
 106                         query.add(db);
 107                 }
 108
 109                 DataBase db = new DataBase();
 110                 db.setTimeTotalExec(totalTime);
 111                 query.add(db);
 112                 System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
 113                 return query;
 114         }
 115
 116         /*
 117          * query: total number of jobs  for the period from date1 till date2
 118          */
 119         public List<DataBase> countJobs(String date1, String date2) {
 120                 if (null == date1) {
 121                         date1 = "1970/1/1";
 122                 }
 123                 if (null == date2) {
 124                         date1 = "2100/1/1";
 125                 }
 126                 if (!isThisDateValid(date1, formatYYMMDD) || !isThisDateValid(date2, formatYYMMDD)) {
 127                         System.out.println("Wrong date: point 3");
 128                         return null;
 129                 }
 130                 SetDateRange();
 131                 long dateStart = DateParsing(date1, formatYYMMDD);
 132                 long dateEnd = DateParsing(date2, formatYYMMDD);
 133                 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
 134                         return null;
 135                 if (dateStart < earlestDate)
 136                         dateStart = earlestDate;
 137                 if (dateEnd > currentDate)
 138                         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                 query = new ArrayList<DataBase>();
 145                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
 146                         List<Long> res = db.ReadDateTable(date.getTime());
 147                         if (res == null)
 148                                 continue;
 149                         DataBase bean = new DataBase();
 150                         bean.setTotal((int)(long)res.get(0));
 151                         bean.setTotalOK((int)(long)res.get(1));
 152                         bean.setTotalStopped((int)(long)res.get(2));
 153                         bean.setTotalError((int)(long)res.get(3));
 154                         bean.setTotalTimeOut((int)(long)res.get(4));
 155                         bean.setDate(DateFormat(date.getTime()));
 156                         query.add(bean);
 157                 }
 158                 System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
 159                 return query;
 160         }
 161
 162         /*
 163          * query: jobs and sequence at date
 164          */
 165         public List<DataBase> readJobByDay (String date) {
 166                 System.out.println(date);
 167                 if (null == date) {
 168                         return null;
 169                 }
 170                 if (!isThisDateValid(date, formatDDMMYY)) {
 171                         System.out.println("Wrong date: point 3");
 172                         return null;
 173                 }
 174                 SetDateRange();
 175                 long day = DateParsing(date, formatDDMMYY);
 176                 System.out.println(day);
 177                 if (day < earlestDate || day > currentDate)
 178                         return null;
 179                 List<Pair<String, String>>      res = db.ReadProteinData(day);
 180                 if (res == null)
 181                         return null;
 182                 query = new ArrayList<DataBase>();
 183                 for (Pair<String, String> entry : res) {
 184                         DataBase bean = new DataBase();
 185                         bean.setDate(date);
 186                         bean.setId(entry.getElement0());
 187                         bean.setProt(entry.getElement1());
 188                         query.add(bean);
 189                 }
 190                 System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
 191                 return query;
 192         }
 193
 194
 195         /*
 196          * query: protein sequence
 197          * */
 198         public List<ProteinBean> readProteins(String protIn, String flag) {
 199                 List<ProteinBean> result;
 200                 if (flag.equals("whole"))
 201                         result = db.ReadWholeSequence(protIn);
 202                  else
 203                          result = db.ReadPartOfSequence(protIn);
 204                 if (result == null)
 205                         return null;
 206
 207                 if (flag.equals("part")) {
 208                         for (ProteinBean entry : result) {
 209                                 entry.setSubProt(CreateSubprot(entry.getSequence(), protIn));
 210                         }
 211                 }
 212                 return result;
 213         }
 214
 215
 216         /*
 217          * query protein sequences with number of jobs
 218          */
 219         public List<DataBase> readProteinByCounter(int minimalcounter) {
 220                 query = new ArrayList<DataBase>();
 221                 Map<String, Integer> map = db.ReadProteinSequenceByCounter();
 222                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
 223                         if (entry.getValue() > minimalcounter && entry.getKey().length() > 0) {
 224                                 DataBase bean = new DataBase();
 225                                 bean.setTotalId(entry.getValue());
 226                                 bean.setProt(entry.getKey());
 227                                 query.add(bean);
 228                         }
 229                 }
 230                 return query;
 231         }
 232
 233         /*
 234          * query ip with number of jobs
 235          */
 236         public List<DataBase> readIpByCounter(Integer minimalcounter) {
 237                 query = new ArrayList<DataBase>();
 238                 Map<String, Integer> map = db.ReadIpByCounter();
 239                 if (minimalcounter == null)
 240                         minimalcounter = 0;
 241                 if (map == null)
 242                         return null;
 243                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
 244                         if (entry.getValue() > minimalcounter) {
 245                                 DataBase bean = new DataBase();
 246                                 bean.setTotalId(entry.getValue());
 247                                 bean.setIp(entry.getKey());
 248                                 query.add(bean);
 249                         }
 250                 }
 251                 return query;
 252         }
 253         /*
 254          * query jobs log info
 255          */
 256         public DataBase readJobLog(String jobid) {
 257                 if (jobid == null)
 258                         return null;
 259                 StructureJobLog res = db.ReadJobLog(jobid);
 260                 if (res == null)
 261                         return null;
 262                 DataBase query = new DataBase();
 263                 query.setLogInfo(res);
 264                 return query;
 265         }
 266
 267
 268         /*
 269          * query jobs by ipStructureJobLog
 270          */
 271         public List<DataBase> readIp(String ip) {
 272                 if (ip == null)
 273                         return null;
 274                 List<Pair<String, String>> res = db.ReadIpWithJobs(ip);
 275         //      System.out.println(res.size());
 276                 if (res == null)
 277                         return null;
 278                 query = new ArrayList<DataBase>();
 279                 for (Pair<String, String> entry : res) {
 280                         System.out.println("ip " + ip);
 281                         System.out.println("id " + entry.getElement0());
 282                         DataBase bean = new DataBase();
 283                         bean.setIp(ip);
 284                         bean.setId(entry.getElement0());
 285                         bean.setProt(entry.getElement1());
 286                         query.add(bean);
 287                 }
 288                 System.out.println(query.size());
 289                 return query;
 290         }
 291
 292         /*
 293          * create list of parts of protein sequence;
 294          */
 295         private static List<String> CreateSubprot (String protein, String subprot) {
 296                 List<String> sub = new ArrayList<String>();
 297                 String subStr = protein;
 298                 while (subStr.length() > 0 && subStr.contains(subprot)) {
 299                         String first = subStr.substring(0, subStr.indexOf(subprot));
 300                         if (first.length() > 0)
 301                                 sub.add(first);
 302                         sub.add(subprot);
 303                         subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
 304                 }
 305                 if (subStr.length() > 0)
 306                         sub.add(subStr);
 307                 return sub;
 308         }
 309         /*
 310          * convert String date into long date (miliseconds since the epoch start)
 311          */
 312         private static long DateParsing(String datInput, SimpleDateFormat formatter) {
 313                 if (datInput == null) {
 314                         return 0;
 315                 }
 316                 long dateWorkSt = 0;
 317
 318                 try {
 319                         dateWorkSt = formatter.parse(datInput).getTime();
 320                 } catch (ParseException e) {
 321                         e.printStackTrace();
 322                 }
 323                 return dateWorkSt;
 324         }
 325
 326         // convert long to date in string format
 327         private static String DateFormat(long inDate) {
 328                 SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
 329                 return datformat.format(new Date(inDate));
 330         }
 331
 332         /*
 333          * set earlest date and current dates. earlestDate is static and should be
 334          * set at the 1st call currentDate should be re-calculated every time
 335          */
 336         private static void SetDateRange() {
 337                 Calendar cal = Calendar.getInstance();
 338                 currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH), formatYYMMDD);
 339                 if (0 == earlestDate) {
 340                         CassandraRequester cr = new CassandraRequester();
 341                         earlestDate = cr.earliestDate();
 342                         System.out.println("Set earlest Date = " + earlestDate);
 343                 }
 344         }
 345
 346         public boolean isThisDateValid(String dateToValidate, SimpleDateFormat sdf) {
 347                 if (dateToValidate == null || dateToValidate.equals("")) {
 348                         System.out.println("Undefined date");
 349                         return false;
 350                 }
 351                 try {
 352                         // if not valid, this will throw ParseException
 353                         sdf.setLenient(false);
 354                         Date date = sdf.parse(dateToValidate);
 355                 } catch (ParseException e) {
 356                         e.printStackTrace();
 357                         return false;
 358                 }
 359                 return true;
 360         }
 361
 362         /*
 363          * find the earliest date in the database
 364          */
 365         public long earliestDate() {
 366                 earlestDate = CassandraNativeConnector.getEarliestDateInDB();
 367                 return earlestDate;
 368         }
 369
 370 }