#!/bin/bash
#
# gen_hist   - generate a v2.20 history file from html pages
#
# Description:
#
#  This script will scan a directory of Webalizer usage reports and
# create a history file from them for use with v2.20 of the Webalizer.
# It uses the GNU tools 'grep', 'sed', and 'sort', and should work
# on any GNU/linux based system.  This script is only required if you
# have more than 12 months of reports and would like the new version
# to display them in the main index (since it can now display up to
# 10 years worth of history).
#
# Usage:
#
#  Copy this script to whatever directory you have your webalizer
# reports in.. this is the directory where all the usage_YYYYMM.html
# files are (among others).  Then simply type:
#
# ./gen_hist > new.hist
#
# to generate a new history file named "new.hist".  Examine the file
# to verify everything went as expected.  If it did, you can then
# either copy or rename it to webalizer.hist.  The next time you run
# the webalizer, the new history file will be used and the months in
# it will appear on the main index page that gets generated.  It is
# of course recomended that you backup up your existing history file,
# just in case something does go wrong.
# 
# Note: If your reports do not display the daily totals, the first
# and last day of the month values will be forced to 1 and 31, which
# will cause incorrect average values.  If this applies to you, you
# probably should edit the generated history file and put in the real
# numbers unless the averages are unimportant to you.
#
#
# Modification History
#
# Date         Who  What
# -----------  ---  -----------------------------------------
# 09-Jun-2008  BLB  Original script
# 11-Jun-2008  BLB  Add check for DAYSTATS (use 1/31 if none)
#
#

# print descriptive header
echo "# Webalizer V2.20 History Data - generated by 'gen_hist'"

# Loop through files
for i in usage_*; do

#strip out our values
YEAR=`echo $i | sed "s/usage_//g" | sed "s/..\.html//g"`
MONTH=`echo $i | sed "s/usage_....//g" | sed "s/\.html//g"`
HITS=`grep "Total Hits" -A1 $i | tail -1 | sed "s/.*<B>//g" | sed "s/<\/B.*//g"`
FILES=`grep "Total Files" -A1 $i | tail -1 | sed "s/.*<B>//g" | sed "s/<\/B.*//g"`
PAGES=`grep "Total Pages" -A1 $i | tail -1 | sed "s/.*<B>//g" | sed "s/<\/B.*//g"`
VISITS=`grep "Total Visits" -A1 $i | tail -1 | sed "s/.*<B>//g" | sed "s/<\/B.*//g"`
XFER=`grep "Total KBytes" -A1 $i | tail -1 | sed "s/.*<B>//g" | sed "s/<\/B.*//g"`
SITES=`grep "Total Unique Sites" -A1 $i | tail -1 | sed "s/.*<B>//g" | sed "s/<\/B.*//g"`

FDAY=`grep "DAYSTAT" $i`
if [ "$FDAY" = "" ]; then
  FDAY=1; LDAY=31;
else
  FDAY=`grep "Daily Statistics for" -A10 $i | tail -1 | sed "s/.*<B>//g" | sed "s/<\/B.*//g"`
  LDAY=`grep "NAME=.HOURSTAT" -B16 $i | head -1 | sed "s/.*<B>//g" | sed "s/<\/B.*//g"`
fi

# spit out a line of history in proper format
echo "$MONTH $YEAR $HITS $FILES $SITES $XFER $FDAY $LDAY $PAGES $VISITS"

# when done, sort-n-exit
done | sort -k2nr -k1nr