1 package compbio.statistic;
3 import java.text.ParseException;
4 import java.text.SimpleDateFormat;
5 import java.util.ArrayList;
6 import java.util.Calendar;
8 import java.util.Iterator;
12 import compbio.cassandra.CassandraNativeConnector;
13 import compbio.cassandra.CassandraReader;
14 import compbio.cassandra.DataBase;
15 import compbio.cassandra.Pair;
16 import compbio.cassandra.StructureJobLog;
17 import compbio.cassandra.StructureProteinPrediction;
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;
26 * query: execution time for the period from date1 till date2
28 public List<DataBase> extractExecutionTime(String date1, String date2) {
35 if (!isThisDateValid(date1) || !isThisDateValid(date2)) {
36 System.out.println("Wrong date: point 3");
41 long dateStart = DateParsing(date1);
42 long dateEnd = DateParsing(date2);
43 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
45 if (dateStart < earlestDate)
46 dateStart = earlestDate;
47 if (dateEnd > currentDate)
48 dateStart = currentDate;
50 Calendar start = Calendar.getInstance();
51 start.setTime(new Date(dateStart));
52 Calendar end = Calendar.getInstance();
53 end.setTime(new Date(dateEnd));
54 query = new ArrayList<DataBase>();
55 List<Integer> totalTime = new ArrayList<Integer>();
56 for (int i = 0; i < nbins; i++)
58 List<Pair<String, String>> res = db.ReadProteinDataTable();
59 List<Pair<Date, Long>> numres = new ArrayList<Pair<Date, Long>>();
61 for (Pair<String, String> entry : res) {
62 SimpleDateFormat dateformatter = new SimpleDateFormat("yyyy/MM/dd");
64 Date jobstartdate = dateformatter.parse(entry.getElement0());
65 long date = jobstartdate.getTime();
66 if (dateStart <= date && date <= dateEnd) {
67 SimpleDateFormat datetimeformatter = new SimpleDateFormat("yyyy/MM/dd:H:m:s");
68 Date jobstarttime = datetimeformatter.parse(entry.getElement0());
69 Date jobendtime = datetimeformatter.parse(entry.getElement1());
70 long diff = (jobendtime.getTime() - jobstarttime.getTime()) / 1000;
71 Pair<Date, Long> pair = new Pair<Date, Long>(jobstartdate, Long.valueOf(diff));
74 } catch (ParseException e) {
79 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
80 List<Integer> timeResult = new ArrayList<Integer>();
81 for (int i = 0; i < nbins; i++)
83 for (Pair<Date, Long> p : numres) {
84 if (date.equals(p.getElement0())) {
85 long lenResult = p.getElement1().longValue();
87 timeResult.set(0, timeResult.get(0) + 1);
88 else if (lenResult > 30 && lenResult <= 60)
89 timeResult.set(1, timeResult.get(1) + 1);
90 else if (lenResult > 60 && lenResult <= 120)
91 timeResult.set(2, timeResult.get(2) + 1);
92 else if (lenResult > 120 && lenResult <= 600)
93 timeResult.set(3, timeResult.get(3) + 1);
95 timeResult.set(4, timeResult.get(4) + 1);
99 for (int i = 0; i < nbins; i++)
100 totalTime.set(i, totalTime.get(i) + timeResult.get(i));
101 DataBase db = new DataBase();
102 db.setTimeRez(timeResult);
103 db.setDate(DateFormat(date.getTime()));
107 DataBase db = new DataBase();
108 db.setTimeTotalExec(totalTime);
110 System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
115 * query: total number of jobs for the period from date1 till date2
117 public List<DataBase> countJobs(String date1, String date2) {
124 if (!isThisDateValid(date1) || !isThisDateValid(date2)) {
125 System.out.println("Wrong date: point 3");
129 long dateStart = DateParsing(date1);
130 long dateEnd = DateParsing(date2);
131 if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
133 if (dateStart < earlestDate)
134 dateStart = earlestDate;
135 if (dateEnd > currentDate)
136 dateStart = currentDate;
138 Calendar start = Calendar.getInstance();
139 start.setTime(new Date(dateStart));
140 Calendar end = Calendar.getInstance();
141 end.setTime(new Date(dateEnd));
142 query = new ArrayList<DataBase>();
143 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
144 long res = db.ReadDateTable(date.getTime());
145 DataBase db = new DataBase();
146 db.setTotal((int)res);
147 db.setDate(DateFormat(date.getTime()));
150 System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
154 * query: protein sequence
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 = db.ReadWholeSequence(protIn);
162 res = db.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));
181 * query protein sequences with number of jobs
183 public List<DataBase> readProteinByCounter(int minimalcounter) {
184 query = new ArrayList<DataBase>();
185 Map<String, Integer> map = db.ReadProteinSequenceByCounter();
186 for (Map.Entry<String, Integer> entry : map.entrySet()) {
187 if (entry.getValue() > minimalcounter) {
188 DataBase db = new DataBase();
189 db.setTotalId(entry.getValue());
190 db.setProt(entry.getKey());
198 * query jobs log info
200 public DataBase readJobLog(String jobid) {
201 // query = new ArrayList<DataBase>();
202 StructureJobLog res = db.ReadJobLog(jobid);
203 DataBase query = new DataBase();
204 query.setLogInfo(res);
209 * create list of parts of protein sequence;
211 private static List<String> CreateSubprot (String protein, String subprot) {
212 List<String> sub = new ArrayList<String>();
213 String subStr = protein;
214 while (subStr.length() > 0 && subStr.contains(subprot)) {
215 String first = subStr.substring(0, subStr.indexOf(subprot));
216 if (first.length() > 0)
219 subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
221 if (subStr.length() > 0)
226 * convert String date into long date (miliseconds since the epoch start)
228 private static long DateParsing(String datInput) {
229 if (datInput == null) {
233 SimpleDateFormat formatter = new SimpleDateFormat("yyyy/MM/dd");
235 dateWorkSt = formatter.parse(datInput).getTime();
236 } catch (ParseException e) {
242 // convert long to date in string format
243 private static String DateFormat(long inDate) {
244 SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
245 return datformat.format(new Date(inDate));
249 * set earlest date and current dates. earlestDate is static and should be
250 * set at the 1st call currentDate should be re-calculated every time
252 private static void SetDateRange() {
253 Calendar cal = Calendar.getInstance();
254 currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH));
255 if (0 == earlestDate) {
256 CassandraRequester cr = new CassandraRequester();
257 earlestDate = cr.earliestDate();
258 System.out.println("Set earlest Date = " + earlestDate);
262 public boolean isThisDateValid(String dateToValidate) {
263 if (dateToValidate == null || dateToValidate.equals("")) {
264 System.out.println("Undefined date");
267 SimpleDateFormat sdf = new SimpleDateFormat("yyyy/MM/dd");
269 // if not valid, this will throw ParseException
270 sdf.setLenient(false);
271 Date date = sdf.parse(dateToValidate);
272 } catch (ParseException e) {
280 * find the earliest date in the database
282 public long earliestDate() {
283 earlestDate = CassandraNativeConnector.getEarliestDateInDB();