X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=server%2Fcompbio%2Fstatistic%2FCassandraRequester.java;h=0ce458baebe317b018a5ff65fcbc86716129bf7b;hb=c441a23b5f24da06199b9b5b830010e5a5833ab8;hp=1906c97e081ad88adf3f30d9792ee330d0117da2;hpb=87b9ef699282c308e5f259fb576b08eb25a1a25f;p=proteocache.git

diff --git a/server/compbio/statistic/CassandraRequester.java b/server/compbio/statistic/CassandraRequester.java
index 1906c97..0ce458b 100755
--- a/server/compbio/statistic/CassandraRequester.java
+++ b/server/compbio/statistic/CassandraRequester.java
@@ -4,42 +4,57 @@ import java.text.ParseException;
 import java.text.SimpleDateFormat;
 import java.util.ArrayList;
 import java.util.Calendar;
-import java.util.Collections;
 import java.util.Date;
-import java.util.Iterator;
+import java.util.HashMap;
 import java.util.List;
+import java.util.Map;
 
+import compbio.cassandra.DateBean;
+import compbio.cassandra.ProteinBean;
 import compbio.cassandra.CassandraNativeConnector;
+import compbio.cassandra.CassandraReaderOld;
 import compbio.cassandra.DataBase;
+import compbio.cassandra.Pair;
+import compbio.cassandra.JobBean;
+import compbio.cassandra.Total;
+import compbio.cassandra.TotalByCounterBean;
+import compbio.cassandra.TotalJobsStatisticBean;
+import compbio.cassandra.UserBean;
+import compbio.engine.JobStatus;
 
 public class CassandraRequester {
-	private CassandraNativeConnector DBInstance = new CassandraNativeConnector();
+	private CassandraReaderOld db = new CassandraReaderOld();
 	private ArrayList<DataBase> query;
 	private static long currentDate = 0;
 	private static long earlestDate = 0;
-
+	private final static SimpleDateFormat formatYYMMDD = new SimpleDateFormat("yyyy/MM/dd");
+	private final static SimpleDateFormat formatDDMMYY = new SimpleDateFormat("dd/MM/yyyy");
 
 	/*
 	 * query: execution time for the period from date1 till date2
-	 * */
+	 */
 	public List<DataBase> extractExecutionTime(String date1, String date2) {
-		if (!isThisDateValid(date1) || !isThisDateValid(date2)) {
-			System.out.println("Wrong date: point 3");
+		if (null == date1) {
+			date1 = "1970/1/1";
+		}
+		if (null == date2) {
+			date1 = "2100/1/1";
+		}
+		if (!isThisDateValid(date1, formatYYMMDD) || !isThisDateValid(date2, formatYYMMDD)) {
+			System.out.println("CassandraRequester.extractExecutionTime: wrong format for date1 " + date1 + "or date2 " + date2);
 			return null;
 		}
 		SetDateRange();
 		int nbins = 5;
-		long dateStart = DateParsing(date1);
-		long dateEnd = DateParsing(date2);
-		if ((dateStart < earlestDate && dateEnd < earlestDate) || (dateStart > currentDate && dateEnd > currentDate) || dateStart > dateEnd)
+		long dateStart = DateParsing(date1, formatYYMMDD);
+		long dateEnd = DateParsing(date2, formatYYMMDD);
+		if (dateEnd < earlestDate || dateStart > currentDate || dateStart > dateEnd)
 			return null;
 		if (dateStart < earlestDate)
 			dateStart = earlestDate;
 		if (dateEnd > currentDate)
 			dateStart = currentDate;
 
-		System.out.println("CassandraRequester.extractExecutionTime: earlestDate = " + earlestDate + ", currentDate = " + currentDate);
-
 		Calendar start = Calendar.getInstance();
 		start.setTime(new Date(dateStart));
 		Calendar end = Calendar.getInstance();
@@ -48,24 +63,34 @@ public class CassandraRequester {
 		List<Integer> totalTime = new ArrayList<Integer>();
 		for (int i = 0; i < nbins; i++)
 			totalTime.add(i, 0);
-		/*
+		List<Pair<String, String>> res = db.ReadProteinDataTable();
+		List<Pair<Date, Long>> numres = new ArrayList<Pair<Date, Long>>();
+
+		for (Pair<String, String> entry : res) {
+			SimpleDateFormat dateformatter = new SimpleDateFormat("yyyy/MM/dd");
+			try {
+				Date jobstartdate = dateformatter.parse(entry.getElement0());
+				long date = jobstartdate.getTime();
+				if (dateStart <= date && date <= dateEnd) {
+					SimpleDateFormat datetimeformatter = new SimpleDateFormat("yyyy/MM/dd:H:m:s");
+					Date jobstarttime = datetimeformatter.parse(entry.getElement0());
+					Date jobendtime = datetimeformatter.parse(entry.getElement1());
+					long diff = (jobendtime.getTime() - jobstarttime.getTime()) / 1000;
+					Pair<Date, Long> pair = new Pair<Date, Long>(jobstartdate, Long.valueOf(diff));
+					numres.add(pair);
+				}
+			} catch (ParseException e) {
+				e.printStackTrace();
+			}
+		}
+
 		for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
 			List<Integer> timeResult = new ArrayList<Integer>();
-			SliceQuery<Long, String, String> result = HFactory.createSliceQuery(DBInstance.GetKeyspace(), LongSerializer.get(),
-					StringSerializer.get(), StringSerializer.get());
-			result.setColumnFamily("ProteinData");
-			result.setKey(date.getTime());
-			result.setRange(null, null, false, Integer.MAX_VALUE);
-			QueryResult<ColumnSlice<String, String>> columnSlice = result.execute();
-			List<HColumn<String, String>> col = columnSlice.get().getColumns();
-			if (!col.isEmpty()) {
-				Iterator<HColumn<String, String>> itCol = col.iterator();
-				for (int i = 0; i < nbins; i++)
-					timeResult.add(i, 0);
-				// split all jobs into nbins bins
-				while (itCol.hasNext()) {
-					String id = itCol.next().getName();
-					long lenResult = CountID(id);
+			for (int i = 0; i < nbins; i++)
+				timeResult.add(i, 0);
+			for (Pair<Date, Long> p : numres) {
+				if (date.equals(p.getElement0())) {
+					long lenResult = p.getElement1().longValue();
 					if (lenResult <= 30)
 						timeResult.set(0, timeResult.get(0) + 1);
 					else if (lenResult > 30 && lenResult <= 60)
@@ -78,112 +103,241 @@ public class CassandraRequester {
 						timeResult.set(4, timeResult.get(4) + 1);
 					}
 				}
-				for (int i = 0; i < nbins; i++)
-					totalTime.set(i, totalTime.get(i) + timeResult.get(i));
-				DataBase db = new DataBase();
-				db.setTimeRez(timeResult);
-				db.setDate(DateFormat(date.getTime()));
-				query.add(db);
 			}
+			for (int i = 0; i < nbins; i++)
+				totalTime.set(i, totalTime.get(i) + timeResult.get(i));
+			DataBase db = new DataBase();
+			db.setTimeRez(timeResult);
+			db.setDate(DateFormat(date.getTime()));
+			query.add(db);
 		}
-		*/
+
 		DataBase db = new DataBase();
 		db.setTimeTotalExec(totalTime);
 		query.add(db);
-		System.out.println("StatisticsProt.readLength: total number of dates = " + query.size());
 		return query;
 	}
 
-	/* 
-	 * convert String date into long date (miliseconds since the epoch start)
+	/*
+	 * query: total number of jobs for the period from date1 till date2
 	 */
-	private static long DateParsing(String datInput) {
-		if (datInput == null) {
-			return 0;
+	public TotalJobsStatisticBean countJobs(String date1, String date2) {
+		/*
+		 * if (null == date1) { date1 = "1970/1/1"; } if (null == date2) { date1
+		 * = "2100/1/1"; } if (!isThisDateValid(date1, formatYYMMDD) ||
+		 * !isThisDateValid(date2, formatYYMMDD)) { System.out.println(
+		 * "CassandraRequester.countJobs: wrong format for date1 " + date1 +
+		 * "or date2 " + date2); return null; }
+		 */
+		SetDateRange();
+		long dateStart = DateParsing(date1, formatYYMMDD);
+		long dateEnd = DateParsing(date2, formatYYMMDD);
+		/*
+		 * if (dateEnd < earlestDate || dateStart > currentDate || dateStart >
+		 * dateEnd) return null; if (dateStart < earlestDate) dateStart =
+		 * earlestDate; if (dateEnd > currentDate) dateStart = currentDate;
+		 */
+		Calendar start = Calendar.getInstance();
+		start.setTime(new Date(dateStart));
+		Calendar end = Calendar.getInstance();
+		end.setTime(new Date(dateEnd));
+		TotalJobsStatisticBean query = new TotalJobsStatisticBean();
+		Total wholeTotal = new Total(0, 0, 0, 0, 0);
+		for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
+			Total res = db.ReadDateTable(date.getTime());
+			if (res == null)
+				continue;
+			query.setDateTotal(DateFormat(date.getTime()), res);
+			wholeTotal.setTotal(res.getTotal() + wholeTotal.getTotal());
+			wholeTotal.setTotalOK(res.getTotalOK() + wholeTotal.getTotalOK());
+			wholeTotal.setTotalStopped(res.getTotalStopped() + wholeTotal.getTotalStopped());
+			wholeTotal.setTotalError(res.getTotalError() + wholeTotal.getTotalError());
+			wholeTotal.setTotalTimeOut(res.getTotalTimeOut() + wholeTotal.getTotalTimeOut());
 		}
-		long dateWorkSt = 0;
-		SimpleDateFormat formatter = new SimpleDateFormat("yyyy/MM/dd");
-		try {
-			dateWorkSt = formatter.parse(datInput).getTime();
-		} catch (ParseException e) {
-			e.printStackTrace();
+		query.setWholeTotal(wholeTotal);
+		return query;
+	}
+
+	/*
+	 * query: jobs and sequence at date
+	 */
+	public DateBean readJobByDay(String date, JobStatus status) {
+		if (!isThisDateValid(date, formatDDMMYY)) {
+			System.out.println("CassandraRequester.readJobByDay: Wrong date format for " + date);
+			return null;
 		}
-		return dateWorkSt;
+		SetDateRange();
+		long day = DateParsing(date, formatDDMMYY);
+		if (day < earlestDate || day > currentDate)
+			return null;
+
+		if (status == JobStatus.OK) {
+			return db.ReadProteinData(day, date);
+		}
+		return db.ReadFailedJobs(day, date, status);
 	}
 
 	/*
-	 * convert String date:time into long date:time (miliseconds since the epoch start)
+	 * query: protein sequence
 	 */
-	private static long TimeConvert(String datInput) {
-		long dateWorkSt = 0;
-		if (datInput == null) {
-			return dateWorkSt;
+	public List<ProteinBean> readProteins(String protIn, String searchtype) {
+		List<ProteinBean> result;
+		if (searchtype.equals("whole"))
+			result = db.ReadWholeSequence(protIn);
+		else
+			result = db.ReadPartOfSequence(protIn);
+		if (result == null)
+			return null;
+
+		if (searchtype.equals("partial")) {
+			for (ProteinBean entry : result) {
+				entry.setSubProt(CreateSubprot(entry.getSequence(), protIn));
+			}
 		}
-		SimpleDateFormat formatter = new SimpleDateFormat("yyyy/MM/dd:hh:mm:ss");
-		try {
-			dateWorkSt = formatter.parse(datInput).getTime();
-		} catch (ParseException e) {
-			e.printStackTrace();
+		return result;
+	}
+
+	/*
+	 * query: protein feature
+	 */
+	public Map<String, String> readProteinsPrediction(String feature, int percent) {
+		Map<String, String> result = db.ReadProtein();
+		;
+		if (result == null)
+			return null;
+		Map<String, String> query = new HashMap<String, String>();
+		for (Map.Entry<String, String> entry : result.entrySet()) {
+			String pred = entry.getValue();
+			if (pred.replaceAll("[^" + feature + "]", "").length() > pred.length() * percent / 100 && (!entry.getKey().equals(""))) {
+				// if (!entry.getKey().equals(""))
+				query.put(entry.getKey(), pred);
+			}
 		}
-		return dateWorkSt;
+		return query;
 	}
 
-	// convert long to date in string format
-	private static String DateFormat(long inDate) {
-		SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
-		String dateString = datformat.format(new Date(inDate));
-		return dateString;
+	/*
+	 * query protein sequences with number of jobs
+	 */
+	public List<TotalByCounterBean> readProteinByCounter(int minimalcounter) {
+		List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
+		Map<String, Integer> map = db.ReadProteinSequenceByCounter();
+		if (map == null)
+			return null;
+		for (Map.Entry<String, Integer> entry : map.entrySet()) {
+			if (entry.getValue() > minimalcounter) {
+				TotalByCounterBean bean = new TotalByCounterBean();
+				bean.setTotaljobs(entry.getValue());
+				bean.setName(entry.getKey());
+				query.add(bean);
+			}
+		}
+		return query;
+	}
+
+	/*
+	 * query ip with number of jobs
+	 */
+	public List<TotalByCounterBean> readIpByCounter(Integer minimalcounter) {
+		List<TotalByCounterBean> query = new ArrayList<TotalByCounterBean>();
+		Map<String, Integer> map = db.ReadIpByCounter();
+		if (minimalcounter == null)
+			minimalcounter = 0;
+		if (map == null)
+			return null;
+		for (Map.Entry<String, Integer> entry : map.entrySet()) {
+			if (entry.getValue() > minimalcounter) {
+				TotalByCounterBean bean = new TotalByCounterBean();
+				bean.setTotaljobs(entry.getValue());
+				bean.setName(entry.getKey());
+				query.add(bean);
+			}
+		}
+		return query;
 	}
 
 	/*
-	 * convert ???
+	 * query jobs log info
 	 */
-	public static String DateFormatYYMMDD(long indate) {
-		SimpleDateFormat datformat = new SimpleDateFormat("yyyy/MM/dd");
-		String dateString = datformat.format(new Date(indate));
-		return dateString;
+	public JobBean readJobLog(String jobid) {
+		if (jobid == null)
+			return null;
+		return db.ReadJobLog(jobid);
 	}
 
 	/*
-	 * ???
+	 * query jobs by ipStructureJobLog
 	 */
-	public long CountID(String id) {
-		/*
-		SliceQuery<String, String, String> sliceQuery = HFactory.createSliceQuery(DBInstance.GetKeyspace(), StringSerializer.get(),
-				StringSerializer.get(), StringSerializer.get());
-		sliceQuery.setColumnFamily("ProteinLog").setKey(id).setRange("", "", false, 100);
-		QueryResult<ColumnSlice<String, String>> result = sliceQuery.execute();
-		String datBegin = result.get().getColumnByName("DataBegin").getValue();
-		String datEnd = result.get().getColumnByName("DataEnd").getValue();
+	public UserBean readIp(String ip) {
+		if (ip == null)
+			return null;
+		Map<String, String[]> res = db.ReadIpWithJobs(ip);
+		if (res == null)
+			return null;
+		UserBean query = new UserBean(ip);
+		query.setMainInfo(res);
+		return query;
+	}
+
+	/*
+	 * create list of parts of protein sequence;
+	 */
+	private static List<String> CreateSubprot(String protein, String subprot) {
+		List<String> sub = new ArrayList<String>();
+		String subStr = protein;
+		while (subStr.length() > 0 && subStr.contains(subprot)) {
+			String first = subStr.substring(0, subStr.indexOf(subprot));
+			if (first.length() > 0)
+				sub.add(first);
+			sub.add(subprot);
+			subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
+		}
+		if (subStr.length() > 0)
+			sub.add(subStr);
+		return sub;
+	}
+
+	/*
+	 * convert String date into long date (miliseconds since the epoch start)
+	 */
+	private static long DateParsing(String datInput, SimpleDateFormat formatter) {
+		if (datInput == null) {
+			return 0;
+		}
+		long dateWorkSt = 0;
+
+		try {
+			dateWorkSt = formatter.parse(datInput).getTime();
+		} catch (ParseException e) {
+			e.printStackTrace();
+		}
+		return dateWorkSt;
+	}
 
-		long datBeginLong = TimeConvert(datBegin);
-		long datEndLong = TimeConvert(datEnd);
-		return (datEndLong - datBeginLong) / 1000;
-		*/
-		return 0;
+	// convert long to date in string format
+	private static String DateFormat(long inDate) {
+		SimpleDateFormat datformat = new SimpleDateFormat("dd/MM/yyyy");
+		return datformat.format(new Date(inDate));
 	}
 
 	/*
-	 * set earlest date and current dates. 
-	 * earlestDate is static and should be set at the 1st call
-	 * currentDate should be re-calculated every time
+	 * set earlest date and current dates. earlestDate is static and should be
+	 * set at the 1st call currentDate should be re-calculated every time
 	 */
 	private static void SetDateRange() {
+		Calendar cal = Calendar.getInstance();
+		currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH),
+				formatYYMMDD);
 		if (0 == earlestDate) {
-			StatisticsProt sp = new StatisticsProt();
-			earlestDate = sp.earliestDate();
-			System.out.println("Set earlest Date = " + earlestDate);
+			CassandraRequester cr = new CassandraRequester();
+			earlestDate = cr.earliestDate();
 		}
-		Calendar cal = Calendar.getInstance();
-		currentDate = DateParsing(cal.get(Calendar.YEAR) + "/" + (cal.get(Calendar.MONTH) + 1) + "/" + cal.get(Calendar.DAY_OF_MONTH));
 	}
 
-	public boolean isThisDateValid(String dateToValidate) {
+	public boolean isThisDateValid(String dateToValidate, SimpleDateFormat sdf) {
 		if (dateToValidate == null || dateToValidate.equals("")) {
-			System.out.println("Undefined date");
 			return false;
 		}
-		SimpleDateFormat sdf = new SimpleDateFormat("yyyy/MM/dd");
 		try {
 			// if not valid, this will throw ParseException
 			sdf.setLenient(false);
@@ -199,31 +353,26 @@ public class CassandraRequester {
 	 * find the earliest date in the database
 	 */
 	public long earliestDate() {
-		/*
-		ArrayList<Long> dateSort = new ArrayList<Long>();
-		int row_count = 10000;
-		RangeSlicesQuery<Long, String, String> result = HFactory.createRangeSlicesQuery(DBInstance.GetKeyspace(), LongSerializer.get(),
-				StringSerializer.get(), StringSerializer.get());
-		result.setColumnFamily("ProteinData");
-		result.setRange(null, null, false, Integer.MAX_VALUE);
-		result.setRowCount(row_count);
-		Long last_key = null;
-		while (true) {
-			result.setKeys(last_key, null);
-			QueryResult<OrderedRows<Long, String, String>> columnSlice = result.execute();
-			OrderedRows<Long, String, String> rows = columnSlice.get();
-			Iterator<Row<Long, String, String>> rowsIterator = rows.iterator();
-			while (rowsIterator.hasNext()) {
-				Row<Long, String, String> row = rowsIterator.next();
-				last_key = row.getKey();
-				dateSort.add(last_key);
-			}
-			if (rows.getCount() < row_count)
-				break;
+		earlestDate = CassandraNativeConnector.getEarliestDateInDB();
+		return earlestDate;
+	}
+
+	/**
+	 * prepares an example of either job id or IP for the DB
+	 * 
+	 * @param exampletype
+	 *            defines which example you need (an existing job from the DB -
+	 *            jobid, an IP - "ip")
+	 * @return a string representation of the requested example, if the example
+	 *         type is not known empty string is returned
+	 */
+	public String getExample(String exampletype) {
+		if (exampletype.equals("jobid")) {
+			return "jp_NzBOJKo";
+		} else if (exampletype.equals("ip")) {
+			return "127.0.0.1";
 		}
-		Collections.sort(dateSort);
-		return dateSort.get(0);
-		*/
-		return 0;
+		return "";
 	}
+
 }