53350237a8b91c43b5b1b2a50101d07ba1a16f64
[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.StructureProteinPrediction;
16
17 public class CassandraRequester {
18         private CassandraNativeConnector DBInstance = new CassandraNativeConnector();
19         private ArrayList<DataBase> query;
20         private static long currentDate = 0;
21         private static long earlestDate = 0;
22
23         /*
24          * query: execution time for the period from date1 till date2
25          */
26         public List<DataBase> extractExecutionTime(String date1, String date2) {
27                 if (null == date1) {
28                         date1 = "1970/1/1";
29                 }
30                 if (null == date2) {
31                         date1 = "2100/1/1";
32                 }
33                 if (!isThisDateValid(date1) || !isThisDateValid(date2)) {
34                         System.out.println("Wrong date: point 3");
35                         return null;
36                 }
37                 SetDateRange();
38                 int nbins = 5;
39                 long dateStart = DateParsing(date1);
40                 long dateEnd = DateParsing(date2);
41                 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
42                         return null;
43                 if (dateStart < earlestDate)
44                         dateStart = earlestDate;
45                 if (dateEnd > currentDate)
46                         dateStart = currentDate;
47
48                 Calendar start = Calendar.getInstance();
49                 start.setTime(new Date(dateStart));
50                 Calendar end = Calendar.getInstance();
51                 end.setTime(new Date(dateEnd));
52                 query = new ArrayList<DataBase>();
53                 List<Integer> totalTime = new ArrayList<Integer>();
54                 for (int i = 0; i < nbins; i++)
55                         totalTime.add(i, 0);
56                 List<Pair<String, String>> res = DBInstance.ReadProteinDataTable();
57                 List<Pair<Date, Long>> numres = new ArrayList<Pair<Date, Long>>();
58
59                 for (Pair<String, String> entry : res) {
60                         SimpleDateFormat dateformatter = new SimpleDateFormat("yyyy/MM/dd");
61                         try {
62                                 Date jobstartdate = dateformatter.parse(entry.getElement0());
63                                 long date = jobstartdate.getTime();
64                                 if (dateStart <= date && date <= dateEnd) {
65                                         SimpleDateFormat datetimeformatter = new SimpleDateFormat("yyyy/MM/dd:H:m:s");
66                                         Date jobstarttime = datetimeformatter.parse(entry.getElement0());
67                                         Date jobendtime = datetimeformatter.parse(entry.getElement1());
68                                         long diff = (jobendtime.getTime() - jobstarttime.getTime()) / 1000;
69                                         Pair<Date, Long> pair = new Pair<Date, Long>(jobstartdate, Long.valueOf(diff));
70                                         numres.add(pair);
71                                 }
72                         } catch (ParseException e) {
73                                 e.printStackTrace();
74                         }
75                 }
76
77                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
78                         List<Integer> timeResult = new ArrayList<Integer>();
79                         for (int i = 0; i < nbins; i++)
80                                 timeResult.add(i, 0);
81                         for (Pair<Date, Long> p : numres) {
82                                 if (date.equals(p.getElement0())) {
83                                         long lenResult = p.getElement1().longValue();
84                                         if (lenResult <= 30)
85                                                 timeResult.set(0, timeResult.get(0) + 1);
86                                         else if (lenResult > 30 && lenResult <= 60)
87                                                 timeResult.set(1, timeResult.get(1) + 1);
88                                         else if (lenResult > 60 && lenResult <= 120)
89                                                 timeResult.set(2, timeResult.get(2) + 1);
90                                         else if (lenResult > 120 && lenResult <= 600)
91                                                 timeResult.set(3, timeResult.get(3) + 1);
92                                         else {
93                                                 timeResult.set(4, timeResult.get(4) + 1);
94                                         }
95                                 }
96                         }
97                         for (int i = 0; i < nbins; i++)
98                                 totalTime.set(i, totalTime.get(i) + timeResult.get(i));
99                         DataBase db = new DataBase();
100                         db.setTimeRez(timeResult);
101                         db.setDate(DateFormat(date.getTime()));
102                         query.add(db);
103                 }
104
105                 DataBase db = new DataBase();
106                 db.setTimeTotalExec(totalTime);
107                 query.add(db);
108                 System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
109                 return query;
110         }
111         
112         /*
113          * query: total number of jobs  for the period from date1 till date2
114          */
115         public List<DataBase> countJobs(String date1, String date2) {
116                 if (null == date1) {
117                         date1 = "1970/1/1";
118                 }
119                 if (null == date2) {
120                         date1 = "2100/1/1";
121                 }
122                 if (!isThisDateValid(date1) || !isThisDateValid(date2)) {
123                         System.out.println("Wrong date: point 3");
124                         return null;
125                 }
126                 SetDateRange();
127                 long dateStart = DateParsing(date1);
128                 long dateEnd = DateParsing(date2);
129                 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
130                         return null;
131                 if (dateStart < earlestDate)
132                         dateStart = earlestDate;
133                 if (dateEnd > currentDate)
134                         dateStart = currentDate;
135
136                 Calendar start = Calendar.getInstance();
137                 start.setTime(new Date(dateStart));
138                 Calendar end = Calendar.getInstance();
139                 end.setTime(new Date(dateEnd));
140                 query = new ArrayList<DataBase>();
141                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
142                         Integer res = DBInstance.ReadDateTable(date.getTime());
143                         if (res == null)
144                                 continue;
145                         DataBase db = new DataBase();
146                         db.setTotal(res);
147                         db.setDate(DateFormat(date.getTime()));
148                         query.add(db);
149                 }                       
150                 System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
151                 return query;
152         }
153         /* 
154          * query: protein sequence
155          * */
156         public List<DataBase> readProteins(String protIn, String flag) {
157                 query = new ArrayList<DataBase>();
158                 List<StructureProteinPrediction> res;
159                 if (flag.equals("whole")) 
160                         res = DBInstance.ReadWholeSequence(protIn);
161                  else 
162                         res = DBInstance.ReadPartOfSequence(protIn);
163                 for (StructureProteinPrediction entry : res) {
164                         Map<String,String> pred = entry.getPrediction();
165                         Iterator it = pred.entrySet().iterator();
166                         while (it.hasNext()) {
167                                 DataBase db = new DataBase();
168                                 db.setProt(entry.getSequence());
169                                 Map.Entry pairs = (Map.Entry)it.next();
170                                 db.setId(entry.getJobid());
171                                 db.setJpred(pairs.getValue().toString());
172                                 if (flag.equals("part"))
173                                         db.setSubProt(CreateSubprot (entry.getSequence(), protIn));                             
174                                 query.add(db);
175                         }
176                 }
177                 return query;
178         }
179         
180         /* 
181          * query  protein sequences with number of jobs
182          * */
183         public List<DataBase> readProteinByCounter(int counter) {
184                 query = new ArrayList<DataBase>();
185         //      List<Pair<String, String>> res = DBInstance.ReadProteinDataByCounter(counter);
186                 
187                 return query;
188         }
189         
190         
191         private static List<String> CreateSubprot (String protein, String subprot) {
192                 List<String> sub = new ArrayList<String>();
193                 String subStr = protein;
194                 while (subStr.length() > 0 && subStr.contains(subprot)) {
195                         String first = subStr.substring(0, subStr.indexOf(subprot));
196                         if (first.length() > 0)
197                                 sub.add(first);
198                         sub.add(subprot);
199                         subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
200                 }
201                 if (subStr.length() > 0)
202                         sub.add(subStr);
203                 return sub;
204         }
205         /*
206          * convert String date into long date (miliseconds since the epoch start)
207          */
208         private static long DateParsing(String datInput) {
209                 if (datInput == null) {
210                         return 0;
211                 }
212                 long dateWorkSt = 0;
213                 SimpleDateFormat formatter = new SimpleDateFormat("yyyy/MM/dd");
214                 try {
215                         dateWorkSt = formatter.parse(datInput).getTime();
216                 } catch (ParseException e) {
217                         e.printStackTrace();
218                 }
219                 return dateWorkSt;
220         }
221
222         // convert long to date in string format
223         private static String DateFormat(long inDate) {
224                 SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
225                 return datformat.format(new Date(inDate));
226         }
227
228         /*
229          * set earlest date and current dates. earlestDate is static and should be
230          * set at the 1st call currentDate should be re-calculated every time
231          */
232         private static void SetDateRange() {
233                 Calendar cal = Calendar.getInstance();
234                 currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH));
235                 if (0 == earlestDate) {
236                         CassandraRequester cr = new CassandraRequester();
237                         earlestDate = cr.earliestDate();
238                         System.out.println("Set earlest Date = " + earlestDate);
239                 }
240         }
241
242         public boolean isThisDateValid(String dateToValidate) {
243                 if (dateToValidate == null || dateToValidate.equals("")) {
244                         System.out.println("Undefined date");
245                         return false;
246                 }
247                 SimpleDateFormat sdf = new SimpleDateFormat("yyyy/MM/dd");
248                 try {
249                         // if not valid, this will throw ParseException
250                         sdf.setLenient(false);
251                         Date date = sdf.parse(dateToValidate);
252                 } catch (ParseException e) {
253                         e.printStackTrace();
254                         return false;
255                 }
256                 return true;
257         }
258
259         /*
260          * find the earliest date in the database
261          */
262         public long earliestDate() {
263                 earlestDate = DBInstance.getEarliestDateInDB();
264                 return earlestDate;
265         }
266         
267 }