#!/bin/bash
# webaccess - Analyzes an Apache-format access_log file, extracting
#    useful and interesting statistics.

setglobal bytes_in_gb = '1048576'
# You will want to change the following to match your own host name 
# to help weed out internally referred hits in the referrer analysis.
setglobal host = '"intuitive.com'"

if test $Argc -eq 0  {
  echo "Usage: $[basename $0] logfile" > !2
  exit 1
}

if test ! -r $1  {
  echo "Error: log file $1 not found." > !2
  exit 1
}

setglobal firstdate = $[head -1 $1 | awk '{print $4}' | sed 's/\[//]
setglobal lastdate = $[tail -1 $1 | awk '{print $4}' | sed 's/\[//]

echo "Results of analyzing log file $1"
echo ""
echo "  Start date: $[echo $firstdate|sed 's/:/ at /]"
echo "    End date: $[echo $lastdate|sed 's/:/ at /]"

setglobal hits = $[wc -l < $1 | sed 's/[^[:digit:]]//g]

echo "        Hits: $[nicenumber $hits] (total accesses)"

setglobal pages = $[grep -ivE '(.txt|.gif|.jpg|.png)' $1 | wc -l | sed 's/[^[:digit:]]//g]

echo "   Pageviews: $[nicenumber $pages] (hits minus graphics)"

setglobal totalbytes = $[awk '{sum+=$10} END {print sum}' $1]

/bin/echo -n " Transferred: $[nicenumber $totalbytes] bytes "

if test $totalbytes -gt $bytes_in_gb  {
  echo "($[scriptbc $totalbytes / $bytes_in_gb] GB)"
} elif test $totalbytes -gt 1024  {
  echo "($[scriptbc $totalbytes / 1024] MB)"
} else {
  echo ""
}

# Now let's scrape the log file for some useful data:

echo ""
echo "The ten most popular pages were:"

awk '{print $7}' $1 | grep -ivE '(.gif|.jpg|.png)' | \
  sed 's/\/$//g' | sort | \
  uniq -c | sort -rn | head -10

echo ""

echo "The ten most common referrer URLs were:"

awk '{print $11}' $1 | \
  grep -vE "(^\"-\"$|/www.$host|/$host)" | \
  sort | uniq -c | sort -rn | head -10

echo ""
exit 0