# Analog config for analysing Tomcat with DSpace logs

# leave debugging on
DEBUG ON

# define the directory that analog looks in as being
# "logs" in the directory from which it is run
LOGFILE logs\access.*.log

# standard output
OUTFILE Report.html

#set up the output parameters
HOSTNAME "Edinburgh Research Archive"
HOSTURL http://www.era.lib.ed.ac.uk/
OUTPUT HTML
ERRFILE errors.txt

# exclude common directories and files
FILEEXCLUDE /styles.css.jsp
FILEEXCLUDE /eul-image/*
FILEEXCLUDE /image/*
FILEEXCLUDE /favicon.ico

# turn on all reports first, then turn off the ones that
# we want
ALL ON

YEARLY OFF
QUARTERLY OFF
HOURLYREP OFF
WEEKHOUR OFF
QUARTERREP OFF
QUARTERSUM OFF
FIVEREP OFF
FIVESUM OFF
DIRECTORY OFF
DAILYREP OFF

# all time related reports should contain the following cols
TIMECOLS BbRr

# all graphs should be based on requests
ALLGRAPH R

# set the columns that each report displays
DOMCOLS NBbRr
ORGCOLS NBbRr
HOSTCOLS NBbRr
REDIRHOSTCOLS NRr
FAILHOSTCOLS NRr
REDIRREFCOLS NRr
FAILREFCOLS NRr
REFCOLS NRr
REFSITECOLS NRr
SEARCHQUERYCOLS NRr
SEARCHWORDCOLS NRr
BROWREPCOLS NBbRr
BROWSUMCOLS NBbRr
OSCOLS NBbRr
STATUSCOLS NRr
SIZECOLS NBbRr
TYPECOLS NRrB
REDIRCOLS NRr
FAILCOLS NRr
REQCOLS NBbRr

# set the sort order for each of the reports
DOMSORTBY REQUESTS
ORGSORTBY REQUESTS
HOSTSORTBY REQUESTS
REDIRHOSTSORTBY REQUESTS
FAILHOSTSORTBY REQUESTS
REDIRREFSORTBY REQUESTS
FAILREFSORTBY REQUESTS
REFSORTBY REQUESTS
REFSITESORTBY REQUESTS
SEARCHQUERYSORTBY REQUESTS
SEARCHWORDSORTBY REQUESTS
BROWREPSORTBY REQUESTS
BROWSUMSORTBY REQUESTS
OSSORTBY REQUESTS
STATUSSORTBY REQUESTS
TYPESORTBY REQUESTS
REDIRSORTBY REQUESTS
FAILSORTBY REQUESTS
REQSORTBY REQUESTS

# set the floor for being displayed for each of the reports
HOSTFLOOR 100r
DOMFLOOR 1r
ORGFLOOR 20r
REQFLOOR 20r
TYPEFLOOR 1r
REFFLOOR 10r
REFSITEFLOOR 2r
SEARCHQUERYFLOOR 5r
SEARCHWORDFLOOR 5r
REDIRREFFLOOR 1r
OSFLOOR 1r
STATUSFLOOR 1r

# expand the charts for the Unix operating system and for
# uk domains
OSCHARTEXPAND Unix
DOMCHARTEXPAND .uk

# general analog settings to make it look OK.
GOTOS FEW
GENSUMLINES ALL
GENSUMLINES -CDEKL
PNGIMAGES ON
NOROBOTS ON
BARSTYLE c

# switch on all the charts, then switch a couple of
# irrelevant ones off
ALLCHART ON
SEARCHQUERYCHART OFF
SEARCHWORDCHART OFF

# exclude some of the common search words.  Ammend this over
# time to strip out unwanted content
SEARCHWORDEXCLUDE of
SEARCHWORDEXCLUDE and
SEARCHWORDEXCLUDE pdf
SEARCHWORDEXCLUDE to
SEARCHWORDEXCLUDE the
SEARCHWORDEXCLUDE in
SEARCHWORDEXCLUDE a
SEARCHWORDEXCLUDE or
SEARCHWORDEXCLUDE its
SEARCHWORDEXCLUDE on
SEARCHWORDEXCLUDE by
SEARCHWORDEXCLUDE for
SEARCHWORDEXCLUDE is
SEARCHWORDEXCLUDE with
SEARCHWORDEXCLUDE that
SEARCHWORDEXCLUDE how
SEARCHWORDEXCLUDE what
SEARCHWORDEXCLUDE it
SEARCHWORDEXCLUDE as
SEARCHWORDEXCLUDE an
SEARCHWORDEXCLUDE &
SEARCHWORDEXCLUDE do
SEARCHWORDEXCLUDE *

# set up some useful UK and Edinburgh sub-domains
SUBDOMAIN ac.uk 'UK Academic'
SUBDOMAIN ed.ac.uk 'University of Edinburgh'


# pre-defined analog stuff that is set by default

# REQINCLUDE pages
REQLINKINCLUDE pages
REFLINKINCLUDE *
REDIRREFLINKINCLUDE *
FAILREFLINKINCLUDE *
SUBBROW */*
SUBTYPE *.gz,*.Z
# OSCHARTEXPAND Windows
# Add whichever of these types of pages you have on your server, or others.
# PAGEINCLUDE *.shtml
# PAGEINCLUDE *.asp
# PAGEINCLUDE *.jsp
# PAGEINCLUDE *.cfm
# PAGEINCLUDE *.pl
# PAGEINCLUDE *.php
# More SEARCHENGINE commands can be found at
#   http://www.analog.cx/helpers/#conffiles
SEARCHENGINE http://*google.*/* q,as_q,as_epq,as_oq
SEARCHENGINE http://*altavista.*/* q
SEARCHENGINE http://*yahoo.*/* p
SEARCHENGINE http://*lycos.*/* query,wfq
SEARCHENGINE http://*aol.*/* query
SEARCHENGINE http://*excite.*/* search
SEARCHENGINE http://*go2net.*/* general
SEARCHENGINE http://*metacrawler.*/* general
SEARCHENGINE http://*msn.*/* q,MT
SEARCHENGINE http://*netscape.*/* search
SEARCHENGINE http://*looksmart.*/* key
SEARCHENGINE http://*webcrawler.*/* qkw,search,searchText
SEARCHENGINE http://*overture.*/* Keywords
SEARCHENGINE http://*teoma.*/* q
SEARCHENGINE http://*infospace.*/* qkw
SEARCHENGINE http://*alltheweb.*/* q
SEARCHENGINE http://*dogpile.*/* q
SEARCHENGINE http://*ask.*/* ask
SEARCHENGINE http://*alltheweb.*/* query
SEARCHENGINE http://*northernlight.*/* qr
SEARCHENGINE http://*nlsearch.*/* qr
SEARCHENGINE http://*dmoz.*/* search
SEARCHENGINE http://*/netfind* query
SEARCHENGINE http://*/pursuit query
ROBOTINCLUDE REGEXPI:robot
ROBOTINCLUDE REGEXPI:spider
ROBOTINCLUDE REGEXPI:crawler
ROBOTINCLUDE Googlebot*
ROBOTINCLUDE Infoseek*
ROBOTINCLUDE Scooter*
ROBOTINCLUDE *Slurp*
ROBOTINCLUDE *Validator*
ROBOTINCLUDE Ultraseek*
ROBOTINCLUDE 207.46*
TYPEALIAS .html    ".html [Hypertext Markup Language]"
TYPEALIAS .htm     ".htm  [Hypertext Markup Language]"
TYPEALIAS .shtml   ".shtml [Server-parsed HTML]"
TYPEALIAS .ps      ".ps   [PostScript]"
TYPEALIAS .gz      ".gz   [Gzip compressed files]"
TYPEALIAS .tar.gz  ".tar.gz [Compressed archives]"
TYPEALIAS .jpg     ".jpg  [JPEG graphics]"
TYPEALIAS .jpeg    ".jpeg [JPEG graphics]"
TYPEALIAS .gif     ".gif  [GIF graphics]"
TYPEALIAS .png     ".png  [PNG graphics]"
TYPEALIAS .txt     ".txt  [Plain text]"
TYPEALIAS .cgi     ".cgi  [CGI scripts]"
TYPEALIAS .pl      ".pl   [Perl scripts]"
TYPEALIAS .css     ".css  [Cascading Style Sheets]"
TYPEALIAS .class   ".class [Java class files]"
TYPEALIAS .pdf     ".pdf  [Adobe Portable Document Format]"
TYPEALIAS .zip     ".zip  [Zip archives]"
TYPEALIAS .hqx     ".hqx  [Macintosh BinHex files]"
TYPEALIAS .exe     ".exe  [Executables]"
TYPEALIAS .wav     ".wav  [WAV sound files]"
TYPEALIAS .avi     ".avi  [AVI movies]"
TYPEALIAS .arc     ".arc  [Compressed archives]"
TYPEALIAS .mid     ".mid  [MIDI sound files]"
TYPEALIAS .mp3     ".mp3  [MP3 sound files]"
TYPEALIAS .doc     ".doc  [Microsoft Word document]"
TYPEALIAS .rtf     ".rtf  [Rich Text Format]"
TYPEALIAS .mov     ".mov  [Quick Time movie]"
TYPEALIAS .mpg     ".mpg  [MPEG movie]"
TYPEALIAS .mpeg    ".mpeg [MPEG movie]"
TYPEALIAS .asp     ".asp  [Active Server Pages]"
TYPEALIAS .jsp     ".jsp  [Java Server Pages]"
TYPEALIAS .cfm     ".cfm  [Cold Fusion]"
TYPEALIAS .php     ".php  [PHP]"
TYPEALIAS .js      ".js   [JavaScript code]"