new type date request
[proteocache.git] / 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.CassandraNativeConnector;
13 import compbio.cassandra.DataBase;
14 import compbio.cassandra.Pair;
15 import compbio.cassandra.StructureJobLog;
16 import compbio.cassandra.StructureProteinPrediction;
17
18 public class CassandraRequester {
19         private CassandraNativeConnector DBInstance = new CassandraNativeConnector();
20         private ArrayList<DataBase> query;
21         private static long currentDate = 0;
22         private static long earlestDate = 0;
23
24         /*
25          * query: execution time for the period from date1 till date2
26          */
27         public List<DataBase> extractExecutionTime(String date1, String date2) {
28                 if (null == date1) {
29                         date1 = "1970/1/1";
30                 }
31                 if (null == date2) {
32                         date1 = "2100/1/1";
33                 }
34                 if (!isThisDateValid(date1) || !isThisDateValid(date2)) {
35                         System.out.println("Wrong date: point 3");
36                         return null;
37                 }
38                 SetDateRange();
39                 int nbins = 5;
40                 long dateStart = DateParsing(date1);
41                 long dateEnd = DateParsing(date2);
42                 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
43                         return null;
44                 if (dateStart < earlestDate)
45                         dateStart = earlestDate;
46                 if (dateEnd > currentDate)
47                         dateStart = currentDate;
48
49                 Calendar start = Calendar.getInstance();
50                 start.setTime(new Date(dateStart));
51                 Calendar end = Calendar.getInstance();
52                 end.setTime(new Date(dateEnd));
53                 query = new ArrayList<DataBase>();
54                 List<Integer> totalTime = new ArrayList<Integer>();
55                 for (int i = 0; i < nbins; i++)
56                         totalTime.add(i, 0);
57                 List<Pair<String, String>> res = DBInstance.ReadProteinDataTable();
58                 List<Pair<Date, Long>> numres = new ArrayList<Pair<Date, Long>>();
59
60                 for (Pair<String, String> entry : res) {
61                         SimpleDateFormat dateformatter = new SimpleDateFormat("yyyy/MM/dd");
62                         try {
63                                 Date jobstartdate = dateformatter.parse(entry.getElement0());
64                                 long date = jobstartdate.getTime();
65                                 if (dateStart <= date && date <= dateEnd) {
66                                         SimpleDateFormat datetimeformatter = new SimpleDateFormat("yyyy/MM/dd:H:m:s");
67                                         Date jobstarttime = datetimeformatter.parse(entry.getElement0());
68                                         Date jobendtime = datetimeformatter.parse(entry.getElement1());
69                                         long diff = (jobendtime.getTime() - jobstarttime.getTime()) / 1000;
70                                         Pair<Date, Long> pair = new Pair<Date, Long>(jobstartdate, Long.valueOf(diff));
71                                         numres.add(pair);
72                                 }
73                         } catch (ParseException e) {
74                                 e.printStackTrace();
75                         }
76                 }
77
78                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
79                         List<Integer> timeResult = new ArrayList<Integer>();
80                         for (int i = 0; i < nbins; i++)
81                                 timeResult.add(i, 0);
82                         for (Pair<Date, Long> p : numres) {
83                                 if (date.equals(p.getElement0())) {
84                                         long lenResult = p.getElement1().longValue();
85                                         if (lenResult <= 30)
86                                                 timeResult.set(0, timeResult.get(0) + 1);
87                                         else if (lenResult > 30 && lenResult <= 60)
88                                                 timeResult.set(1, timeResult.get(1) + 1);
89                                         else if (lenResult > 60 && lenResult <= 120)
90                                                 timeResult.set(2, timeResult.get(2) + 1);
91                                         else if (lenResult > 120 && lenResult <= 600)
92                                                 timeResult.set(3, timeResult.get(3) + 1);
93                                         else {
94                                                 timeResult.set(4, timeResult.get(4) + 1);
95                                         }
96                                 }
97                         }
98                         for (int i = 0; i < nbins; i++)
99                                 totalTime.set(i, totalTime.get(i) + timeResult.get(i));
100                         DataBase db = new DataBase();
101                         db.setTimeRez(timeResult);
102                         db.setDate(DateFormat(date.getTime()));
103                         query.add(db);
104                 }
105
106                 DataBase db = new DataBase();
107                 db.setTimeTotalExec(totalTime);
108                 query.add(db);
109                 System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
110                 return query;
111         }
112         
113         /*
114          * query: total number of jobs  for the period from date1 till date2
115          */
116         public List<DataBase> countJobs(String date1, String date2) {
117                 if (null == date1) {
118                         date1 = "1970/1/1";
119                 }
120                 if (null == date2) {
121                         date1 = "2100/1/1";
122                 }
123                 if (!isThisDateValid(date1) || !isThisDateValid(date2)) {
124                         System.out.println("Wrong date: point 3");
125                         return null;
126                 }
127                 SetDateRange();
128                 long dateStart = DateParsing(date1);
129                 long dateEnd = DateParsing(date2);
130                 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
131                         return null;
132                 if (dateStart < earlestDate)
133                         dateStart = earlestDate;
134                 if (dateEnd > currentDate)
135                         dateStart = currentDate;
136
137                 Calendar start = Calendar.getInstance();
138                 start.setTime(new Date(dateStart));
139                 Calendar end = Calendar.getInstance();
140                 end.setTime(new Date(dateEnd));
141                 query = new ArrayList<DataBase>();
142                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
143                         Integer res = DBInstance.ReadDateTable(date.getTime());
144                         System.out.println("Command: " + res);
145                         if (res == null)
146                                 continue;
147                         DataBase db = new DataBase();
148                         db.setTotal(res);
149                         db.setDate(DateFormat(date.getTime()));
150                         query.add(db);
151                 }                       
152                 System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
153                 return query;
154         }
155         /* 
156          * query: protein sequence
157          * */
158         public List<DataBase> readProteins(String protIn, String flag) {
159                 query = new ArrayList<DataBase>();
160                 List<StructureProteinPrediction> res;
161                 if (flag.equals("whole")) 
162                         res = DBInstance.ReadWholeSequence(protIn);
163                  else 
164                         res = DBInstance.ReadPartOfSequence(protIn);
165                 for (StructureProteinPrediction entry : res) {
166                         Map<String,String> pred = entry.getPrediction();
167                         Iterator it = pred.entrySet().iterator();
168                         while (it.hasNext()) {
169                                 DataBase db = new DataBase();
170                                 db.setProt(entry.getSequence());
171                                 Map.Entry pairs = (Map.Entry)it.next();
172                                 db.setId(entry.getJobid());
173                                 db.setJpred(pairs.getValue().toString());
174                                 if (flag.equals("part"))
175                                         db.setSubProt(CreateSubprot (entry.getSequence(), protIn));                             
176                                 query.add(db);
177                         }
178                 }
179                 return query;
180         }
181         
182         /* 
183          * query  protein sequences with number of jobs
184          * */
185         public List<DataBase> readProteinByCounter(int counter) {
186                 query = new ArrayList<DataBase>();
187                 Map<String, Integer> map = DBInstance.ReadProteinDataByCounter();
188                 for (Map.Entry<String, Integer> entry : map.entrySet()) {
189                         if (entry.getValue() > counter) {
190                                 DataBase db = new DataBase();
191                                 db.setTotalId(entry.getValue());
192                                 db.setProt(entry.getKey());
193                                 query.add(db);
194                         }
195                 }
196                 return query;
197         }
198         
199         /*
200          * query jobs log info
201          */
202         public DataBase readJobLog(String jobid) {
203         //      query = new ArrayList<DataBase>();
204                 StructureJobLog res = DBInstance.ReadJobLog(jobid);
205                 DataBase query = new DataBase();
206                 query.setLogInfo(res);
207         //      query.setres);
208                 return query;
209         }
210         /*
211          * create list of parts of protein sequence;
212          */
213         private static List<String> CreateSubprot (String protein, String subprot) {
214                 List<String> sub = new ArrayList<String>();
215                 String subStr = protein;
216                 while (subStr.length() > 0 && subStr.contains(subprot)) {
217                         String first = subStr.substring(0, subStr.indexOf(subprot));
218                         if (first.length() > 0)
219                                 sub.add(first);
220                         sub.add(subprot);
221                         subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
222                 }
223                 if (subStr.length() > 0)
224                         sub.add(subStr);
225                 return sub;
226         }
227         /*
228          * convert String date into long date (miliseconds since the epoch start)
229          */
230         private static long DateParsing(String datInput) {
231                 if (datInput == null) {
232                         return 0;
233                 }
234                 long dateWorkSt = 0;
235                 SimpleDateFormat formatter = new SimpleDateFormat("yyyy/MM/dd");
236                 try {
237                         dateWorkSt = formatter.parse(datInput).getTime();
238                 } catch (ParseException e) {
239                         e.printStackTrace();
240                 }
241                 return dateWorkSt;
242         }
243
244         // convert long to date in string format
245         private static String DateFormat(long inDate) {
246                 SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
247                 return datformat.format(new Date(inDate));
248         }
249
250         /*
251          * set earlest date and current dates. earlestDate is static and should be
252          * set at the 1st call currentDate should be re-calculated every time
253          */
254         private static void SetDateRange() {
255                 Calendar cal = Calendar.getInstance();
256                 currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH));
257                 if (0 == earlestDate) {
258                         CassandraRequester cr = new CassandraRequester();
259                         earlestDate = cr.earliestDate();
260                         System.out.println("Set earlest Date = " + earlestDate);
261                 }
262         }
263
264         public boolean isThisDateValid(String dateToValidate) {
265                 if (dateToValidate == null || dateToValidate.equals("")) {
266                         System.out.println("Undefined date");
267                         return false;
268                 }
269                 SimpleDateFormat sdf = new SimpleDateFormat("yyyy/MM/dd");
270                 try {
271                         // if not valid, this will throw ParseException
272                         sdf.setLenient(false);
273                         Date date = sdf.parse(dateToValidate);
274                 } catch (ParseException e) {
275                         e.printStackTrace();
276                         return false;
277                 }
278                 return true;
279         }
280
281         /*
282          * find the earliest date in the database
283          */
284         public long earliestDate() {
285                 earlestDate =  DBInstance.getEarliestDateInDB();
286                 return earlestDate;
287         }
288         
289 }