Programming Languages Suite/ProgLangD.iso/VCAFE.3.0A/Sample.bin/Report.java

Text File | 1998-11-05 | 23KB | 558 lines

// Copyright (c) 1997, 1998 Symantec, Inc. All Rights Reserved.

import java.io.Serializable;
import java.util.Date;
import java.util.Calendar;
import java.util.Hashtable;
import java.util.Vector;
import java.util.NoSuchElementException;
import java.util.Enumeration;

/*
This class encapsulates a single web log "report".
It contains:
1) The report options (specified/changed by New/Edit Report dialogs),
   including a LogFile object representing the log file specified in the
   report options.
2) The report results (available after analyzing the log file).
3) Working fields (used during log file analysis).
4) Methods to analyze the log file and generate the results.
Statistics are gathered over a day long interval (i.e. hits per day, etc) and
also gathered over an interval chosen to be appropriate for the timespan of
the report.
It was written manually.
It has no UI code.
*/
public class Report implements Serializable {
    //  --  The Report Options
    // possible timeframe choice values
    public static final int TIMEFRAME_ALL = 0;
    public static final int TIMEFRAME_TODAY = 1;
    public static final int TIMEFRAME_YESTERDAY = 2;
    public static final int TIMEFRAME_PAST_2_DAYS = 3;
    public static final int TIMEFRAME_PAST_7_DAYS = 4;
    public static final int TIMEFRAME_PAST_14_DAYS = 5;
    public static final int TIMEFRAME_PAST_30_DAYS = 6;
    public static final int TIMEFRAME_PAST_60_DAYS = 7;
    public static final int TIMEFRAME_PAST_90_DAYS = 8;
    public static final int TIMEFRAME_SPECIFIC = 9;
    String title;
    int timeframe;  // TIMEFRAME_
    Date dateStart; // valid if timeframe == TIMEFRAME_SPECIFIC
    Date dateEnd;   // valid if timeframe == TIMEFRAME_SPECIFIC
    LogFile log = new LogFile();
    String homePageFile;
    boolean optUsageChart;
    boolean optGeneralStatistics;
    boolean optMostActiveDomains;
    //  --  The Report Results
    Date analyzedDate;              // date report last run on
    long startTimestamp;            // actual starting time (millisecs)
    long endTimestamp;              // actual ending time (millisecs)
    int totalHits;                  // total number of hits
    int homePageHits;               // hits on the home page
    int totalUserSessionCount;      // total number of user sessions
    long avgUserSessionLength = 0;  // average session length (millisecs)
    DomainRecord sortedDomains[];   // all domains, sorted by number of hits
    // "per day" statistics
    int hitsVsDay[];                // array to track hits VS day
    int userSessionsVsDay[];        // array to sessions VS day
    double avgHitsPerDay;              // average hits per day
    double avgUserSessionsPerDay;      // average sessions per day
    // "per <time interval>" statistics
    long hitTimeInterval;           // interval statistics are gathered over (millisecs)
    int hitsVsTimeInterval[];       // array to track hits VS hitTimeInterval
    int userSessionsVsTimeInterval[];//array to sessions VS hitTimeInterval
    double avgHitsPerHitTimeInterval;  // average number of hits per hitTimeInterval
    double avgUserSessionsPerHitTimeInterval;  // average sessions per hitTimeInterval
    //SessionRecord sortedSessions[];
    //  --  Working fields. Used during log file analysis
    String normalizedHomePageFile;  // home page file with "/" added at start as needed and "\" changed to "/" as needed
    long newUserSessionThreshold;   // user idle time to create a new session (millisecs)
    long timeTillFirstInterval;     // millisecs till first full interval of time being analyzed
    long timeTillFirstDayInterval;  // millisecs till first full day of time being analyzed
    long totalUserSessionLength;    // total length in time of all user sessions
    Hashtable domains;              // of DomainRecord
    Hashtable sessions;             // of SessionRecord
    SessionRecord sessionListHead = null;   // list of all SessionRecords

    /**
     * Constructs a Report.
     */
    public Report() {
            //{{INIT_CONTROLS
        //}}
}

    /**
     * Frees up memory used to analyze (but not report) a web log.
     */
    void freeUpMem() {
        // Free up mem used to analyze, but not report, log info
        domains = null;
        sessions = null;
        sessionListHead = null;
        log.freeUpMem();
    }

    // -- ACCESS

    // Returns true if this report has been run (analyzed) before.
    public boolean hasBeenRun() {
        return analyzedDate != null;
    }

    // -- Handy shortcuts to get/set LogFile attributes
    public String getLogFileURL() {
        return log.getLogFileURL();
    }
    public void setLogFileURL(String logFileURL) {
        log.setLogFileURL(logFileURL);
    }
    public int getLogFileFormat() {
        return log.getLogFileFormat();
    }
    public void setLogFileFormat(int logFileFormat) {
        log.setLogFileFormat(logFileFormat);
    }

    // -- ANALYSIS

    /*
    Runs the report, analyzing the log file and gathering statistics.
    The TrackProgress interface is used to report analysis status to UI code.
    */
    public void run(TrackProgress trackProgress) {
        try {

            //  --  Setup analysis

            // Reading and parse the log file, getting the results
            Analyzer.throwExceptionIfCurrentThreadCancelled();
            LogRecord records[] = log.readAndParseLogFile(trackProgress);
            if(records == null || records.length == 0) {
                trackProgress.okAlert("No valid HTTP records to process (wrong log file format?).");
                trackProgress.step("ABORTED", 0);
                // Note that we're done in the UI
                trackProgress.done();
                return;
            }
            // determine starting and ending time for this report
            Analyzer.throwExceptionIfCurrentThreadCancelled();
            startTimestamp = log.getEarliestTimestamp();
            endTimestamp = log.getLatestTimestamp();
            if(timeframe != TIMEFRAME_ALL) {
                startTimestamp = dateStart.getTime();
                endTimestamp = dateEnd.getTime();
            }
            //
            totalHits = 0;
            homePageHits = 0;
            totalUserSessionCount = 0;
            avgUserSessionLength = 0;
            sortedDomains = null;
            // init per day stats
            timeTillFirstDayInterval = determineTimeTillFirstFullInterval(WLAUtil.MILLISECS_PER_DAY, CAL_HOUR);
            // table: hitsVsDay
            int numTimeIntervals = 1 + timestampToDayIntervalIndex(endTimestamp);
            hitsVsDay = new int[numTimeIntervals];
            // table: userSessionsVsDay
            userSessionsVsDay = new int[numTimeIntervals];
            avgHitsPerDay = 0;
            avgUserSessionsPerDay = 0;
            // init per <time interval> stats
            // determine the <time interval> units: min, hour, day, week, or year
            long millisecs = endTimestamp - startTimestamp;
            if(millisecs <= WLAUtil.MILLISECS_PER_HOUR) {
                hitTimeInterval = WLAUtil.MILLISECS_PER_MINUTE;
                timeTillFirstInterval = determineTimeTillFirstFullInterval(WLAUtil.MILLISECS_PER_MINUTE, CAL_SECOND);
            } else if(millisecs <= 2*WLAUtil.MILLISECS_PER_DAY) {
                hitTimeInterval = WLAUtil.MILLISECS_PER_HOUR;
                timeTillFirstInterval = determineTimeTillFirstFullInterval(WLAUtil.MILLISECS_PER_HOUR, CAL_MINUTE);
            } else if(millisecs <= 2*WLAUtil.MILLISECS_PER_WEEK) {
                hitTimeInterval = WLAUtil.MILLISECS_PER_DAY;
                timeTillFirstInterval = determineTimeTillFirstFullInterval(WLAUtil.MILLISECS_PER_DAY, CAL_HOUR);
            } else if(millisecs <= 2*WLAUtil.MILLISECS_PER_YEAR) {
                hitTimeInterval = WLAUtil.MILLISECS_PER_WEEK;
                timeTillFirstInterval = determineTimeTillFirstFullInterval(WLAUtil.MILLISECS_PER_WEEK, CAL_DAY_OF_WEEK);
            } else {
                hitTimeInterval = WLAUtil.MILLISECS_PER_YEAR;
                timeTillFirstInterval = determineTimeTillFirstFullInterval(WLAUtil.MILLISECS_PER_YEAR, CAL_WEEK_OF_YEAR);
            }
            // table: hitsVsTimeInterval
            numTimeIntervals = 1 + timestampToIntervalIndex(endTimestamp);
            hitsVsTimeInterval = new int[numTimeIntervals];
            // table: userSessionsVsTimeInterval
            userSessionsVsTimeInterval = new int[numTimeIntervals];
            avgHitsPerHitTimeInterval = 0;
            avgUserSessionsPerHitTimeInterval = 0;
            // "normalize" home page file name
            int lim = homePageFile.length();
            if(lim == 0) {
                normalizedHomePageFile = homePageFile;
            } else {
                StringBuffer buf = new StringBuffer(lim + 1);
                int idx = 0;
                char ch = homePageFile.charAt(0);
                // if no leading slash, add one
                buf.append('/');
                if(ch == '/' || ch == java.io.File.separatorChar) {
                    idx = 1;
                }
                while(idx < lim) {
                    ch = homePageFile.charAt(idx++);
                    if(ch == java.io.File.separatorChar) {
                        buf.append('/');
                    } else {
                        buf.append(ch);
                    }
                }
                normalizedHomePageFile = buf.toString();
            }
            // init remaining working fields
            newUserSessionThreshold = Data.getDataInstance().userSessionIdleLimit * WLAUtil.MILLISECS_PER_MINUTE;
            totalUserSessionLength = 0;
            domains = new Hashtable();      // of DomainRecord
            sessions = new Hashtable();      // of SessionRecord
            sessionListHead = null;
            // init local info
            int hitCount = 0;
            int totalLogHits = log.getTotalHits();
            // Done with analysis setup

            if(totalLogHits > 0) {

                //  --  SCAN LOG LINES AND ACCUMULATE DATA

                for(int r = 0; r < totalLogHits; r++) {
                    // percent for this phase ranges from 96 -> 100
                    int percent = 96 + ((4 * hitCount)/totalLogHits);
                    trackProgress.step("Processing log file...", percent);
                    // Note the line's stats
                    noteLogRecord(records[r]);
                }

                //  --  CALC STATISTICS

                // Calc averages
                double timespan = endTimestamp - startTimestamp;
                timespan = timespan / WLAUtil.MILLISECS_PER_DAY;
                avgHitsPerDay = (double)totalHits / timespan;
                avgUserSessionsPerDay = (double)totalUserSessionCount / timespan;
                timespan = endTimestamp - startTimestamp;
                timespan = timespan / hitTimeInterval;
                avgHitsPerHitTimeInterval = (double)totalHits / timespan;
                avgUserSessionsPerHitTimeInterval = (double)totalUserSessionCount / timespan;

                // Calc total length of user sessions
                Analyzer.throwExceptionIfCurrentThreadCancelled();
                SessionRecord sr = sessionListHead;
                while(sr != null) {
                    long delta = sr.lastTimestamp - sr.firstTimestamp;
                    totalUserSessionLength += delta;
                    sr = sr.next;
                }

                // Calc avg user session length
                if(totalUserSessionCount > 0) {
                    avgUserSessionLength = totalUserSessionLength / totalUserSessionCount;
                }

                // Sort domains in order of most hits to least hits
                sortedDomains = new DomainRecord[domains.size()];
                // copy into array...
                int idx = 0;
                for(Enumeration e = domains.elements(); e.hasMoreElements();) {
                    DomainRecord dr = (DomainRecord)e.nextElement();
                    sortedDomains[idx++] = dr;
                }
                // ...and sort
                Analyzer.throwExceptionIfCurrentThreadCancelled();
                WLAUtil.quickSort(sortedDomains);
            }

            //  --  DONE OK, note date/time report run

            analyzedDate = new Date();  // date report last run on
            trackProgress.step("Done", 100);

        } catch(NoSuchElementException x) {
            trackProgress.okAlert(x.toString());
            trackProgress.step("ABORTED", 0);
        } catch(ParseLogException x) {
            // These have already been reported.
        } catch(OutOfMemoryError x) {
            // free the main memory occupiers
            Data.freeUpMem();
            trackProgress.okAlert(x.toString());
            trackProgress.step("ABORTED", 0);
        }

        // Note that we're done in the UI
        trackProgress.done();
    }

    /*
    Called by the run() method for each log record to analyze.
    Accumulates report statistics.
    */
    void noteLogRecord(LogRecord logRec) throws ParseLogException {
        Analyzer.throwExceptionIfCurrentThreadCancelled();

        //  --  Filter out undesired log records

        // Does this record occur within the timeframe we're analyzing??
        if(startTimestamp > logRec.requestTimestamp || logRec.requestTimestamp > endTimestamp) {
            // No. Ignore it.
            return;
        }
        // HTTP error on this request?
        if(logRec.httpStatusCode >= 500) {
            // Yes. Ignore it.
            return;
        }
        // Determine type of HTTP request
        int idx = logRec.clientRequest.indexOf(' ');
        if(idx < 0) {
            // Unexpected clientRequest. Ignore
            return;
        }
        String str = logRec.clientRequest.substring(0, idx);
        if(!str.equals("GET")) {
            // Ignore non-GET HTTP requests
            return;
        }

        //  --  Use this log record

        // Note info for General Statistics
        totalHits++;

        // Note info for Hits VS Time chart
        int hitInterval = timestampToIntervalIndex(logRec.requestTimestamp);
        hitsVsTimeInterval[hitInterval]++;

        // Note info for Hits VS Day for day-related averages
        hitInterval = timestampToDayIntervalIndex(logRec.requestTimestamp);
        hitsVsDay[hitInterval]++;

        // Request for home page?
        int idx2 = logRec.clientRequest.indexOf(' ', ++idx);
        if(idx2 > 0) {
            str = logRec.clientRequest.substring(idx, idx2);
            if(str.equals("/") || str.equalsIgnoreCase(normalizedHomePageFile)) {
                homePageHits++;
            }
        }

        // Track # of hits for each domain
        DomainRecord dr = (DomainRecord)domains.get(logRec.remoteHost);
        if(dr == null) {
            dr = new DomainRecord(logRec.remoteHost);
            domains.put(logRec.remoteHost, dr);
        }
        // update the domain record
        dr.hits++;

        //  -- Track sessions

        String sessionStr = logRec.remoteHost + " " + logRec.remoteUserLogname + " " + logRec.authenticatedUserName;
        SessionRecord sr = (SessionRecord)sessions.get(sessionStr);

        if(sr == null
            || (logRec.requestTimestamp - sr.lastTimestamp >= newUserSessionThreshold)) {

            // New user session
            sr = new SessionRecord(sessionStr, logRec);
            // Add to linked list of all Session Records
            sr.next = sessionListHead;
            sessionListHead = sr;
            // Update the hashtable which tracks current "live" sessions
            sessions.put(sessionStr, sr);
            totalUserSessionCount++;      // total number of user sessions

            // Track userSessions Vs TimeInterval and Vs Day

            // Note info for Hits VS Time chart
            hitInterval = timestampToIntervalIndex(logRec.requestTimestamp);
            userSessionsVsTimeInterval[hitInterval]++;

            // Note info for Hits VS Day for day-related averages
            hitInterval = timestampToDayIntervalIndex(logRec.requestTimestamp);
            userSessionsVsDay[hitInterval]++;
        }
        // update the session record
        sr.hits++;
        sr.lastTimestamp = logRec.requestTimestamp;
    }

    /* determineTimeTillFirstFullInterval()
    Time intervals are aligned on natural boundries (i.e. midnight for a
    day long interval). This method determines the length of the first
    interval such that it ends on a natural boundry. Typically it is smaller
    than a full interval.
    */
    private static final int CAL_SECOND         = 0;
    private static final int CAL_MINUTE         = 1;
    private static final int CAL_HOUR           = 2;
    private static final int CAL_DAY_OF_WEEK    = 3;
    private static final int CAL_WEEK_OF_YEAR   = 4;
    private static final int CALENDAR_UNIT_VALUES[] = {
        Calendar.SECOND,
        Calendar.MINUTE,
        Calendar.HOUR_OF_DAY,
        Calendar.DAY_OF_WEEK,
        Calendar.WEEK_OF_YEAR
    };
    private long determineTimeTillFirstFullInterval(long MILLISECS_PER_, int CAL_) {
        // Get a calendar and set its time to the report starting time + interval length
        Calendar cal = Calendar.getInstance();
        cal.setTime(new Date(startTimestamp+MILLISECS_PER_));
        // zero the appropriate fields, truncating time to interval natural boundry
        do {
            cal.set(CALENDAR_UNIT_VALUES[CAL_], 0);
        } while(--CAL_ >= 0);
        // Get resultant time, and calculate delta to that time from report starting time
        Date date = cal.getTime();
        long deltaTime = date.getTime() - startTimestamp;
        // done
        return deltaTime;
    }

    /*
    Calculates the day interval index that the given time falls into.
    */
    private int timestampToDayIntervalIndex(long timestamp) {
        int intervalIndex = 0;
        long ms = timestamp - startTimestamp;
        if(ms >= timeTillFirstDayInterval) {
            ms -= timeTillFirstDayInterval;
            intervalIndex = (int)(ms / WLAUtil.MILLISECS_PER_DAY) + 1;
        }
        return intervalIndex;
    }

    /*
    Calculates the <time interval> index that the given time falls into.
    */
    private int timestampToIntervalIndex(long timestamp) {
        int intervalIndex = 0;
        long ms = timestamp - startTimestamp;
        if(ms >= timeTillFirstInterval) {
            ms -= timeTillFirstInterval;
            intervalIndex = (int)(ms / hitTimeInterval) + 1;
        }
        return intervalIndex;
    }


    /*
    This class tracks info on a single user session.
    All SessionRecords are kept on a linked-list (sessionListHead refers
    to the most recently created SessionRecord).
    This class implements the Sortable interface to allow the quicksort
    utility routine to sort it.
    */
    class SessionRecord implements Serializable, Sortable {
        // The next SessionRecord in the linked-list.
        SessionRecord next = null;
        // This session's identifying string (a concatenation of the domain
        // name, remote user log name, and authenticated user name)
        String sessionStr;
        // The number of hits by this session
        int hits;
        // The session's starting time
        long firstTimestamp;
        // The time of this session's last request
        long lastTimestamp;

        /*
        Constructs a new session.
        */
        SessionRecord(String sessionStr, LogRecord logRec) {
            this.sessionStr = sessionStr;
            firstTimestamp = logRec.requestTimestamp;
            lastTimestamp = logRec.requestTimestamp;
            hits = 0;
        }

        // -- Implement the Sortable interface

        // A Sortable interface method.
        // Returns true if this "is before" the given obj
        public boolean isLessThan(Object obj) {
            if(obj instanceof SessionRecord) {
                SessionRecord r = (SessionRecord)obj;
                if(hits != r.hits) {
                    return hits < r.hits;
                }
                return sessionStr.compareTo(r.sessionStr) > 0;
            }
            return false;
        }

        // A Sortable interface method.
        // Returns true if this "is before or equal to" the given obj
        public boolean isLessThanOrEqual(Object obj) {
            if(obj instanceof SessionRecord) {
                SessionRecord r = (SessionRecord)obj;
                if(hits != r.hits) {
                    return hits < r.hits;
                }
                return sessionStr.compareTo(r.sessionStr) >= 0;
            }
            return false;
        }
    }


    /*
    This class tracks info on a single user session.
    This class implements the Sortable interface to allow the quicksort
    utility routine to sort it.
    */
    class DomainRecord implements Serializable, Sortable {
        // The domain name
        String domain;
        // The number of hits this domain has generated
        int hits;

        // Creates a new DomainRecord with the given name.
        DomainRecord(String domain) {
            this.domain = domain;
        }

        // -- Implement the Sortable interface

        // A Sortable interface method.
        // Returns true if this "is before" the given obj
        public boolean isLessThan(Object obj) {
            if(obj instanceof DomainRecord) {
                DomainRecord r = (DomainRecord)obj;
                if(hits != r.hits) {
                    return hits < r.hits;
                }
                return domain.compareTo(r.domain) < 0;
            }
            return false;
        }

        // A Sortable interface method.
        // Returns true if this "is before or equal to" the given obj
        public boolean isLessThanOrEqual(Object obj) {
            if(obj instanceof DomainRecord) {
                DomainRecord r = (DomainRecord)obj;
                if(hits != r.hits) {
                    return hits < r.hits;
                }
                return domain.compareTo(r.domain) <= 0;
            }
            return false;
        }
    }

    //{{DECLARE_CONTROLS
    //}}
}