#####################################################################
# SCRIPT: SS_WebLogParser
#
# This script is a sample Web Log Parser. It takes a web log file and and
# reports pages, hit counts, referrers and referrer counts.
#
# The name of the file containing the web log is passed using FVA (Forward
# Variable Assignment) for variable $logfile.
#
# The output is written to screen. If your logfile is very large, redirect the
# output using > filename.txt when calling the script.
#
# This script can be stored, and edited as needed, in a file called
# SS_WebLogParser.txt. The script can then be called as
#
#	script "SS_WebLogParser.txt" logfile("C:/logs/web/20080820")
#
#####################################################################

# Declare FVA variables.
var str logfile

# Get the contents of the log file into a str variable
var str content
repro $logfile > $content

# We keep the page list in the following variable - it has the format of <r><page><f><count>
# <r> is the marker for the record, <f> is the marker for the field.
var str page_list

# We keep the referrer list in the following variable - it has the format of <r><referrer><f><count>
# <r> is the marker for the record, <f> is the marker for the field.
var str referrer_list

# Parse lines one by one
while ( $content <> "")
do
  # Get the next line
  var str line
  lex -e "1" $content > $line

  # Make sure that line is not empty.
  if ($line <> "")
  do

    # The line has the following form.
    # 1.2.3.4 x y [20/Oct/2008:01:02:03 -0100] "GET xxx HTTP/1.2" 1 2 "yyy" ...

    # The page URL has the following form "GET xxx "
    var str page
    stex -p -c "^GET ^[" $line > $page
    stex "[^ ^" $page > null

    # The referrer has the following form "yyy", this is the third double quote in the line
    var str referrer
    stex "^\"^3]" $line > null
    stex -r "]^(\"?)^" $line > $referrer

    # Is this page already in the page_list ? If so, we will just increment the count.
    # If not, we will add it and set the count to 1.
    var str sen_arg
    set $sen_arg = "^"+$page+"^"
    if ( { sen $sen_arg $page_list } > 0 )
    do
      # This page is already in page_list.

      # Create dynamic argument for the stex command.
      var str stex_arg
      set $stex_arg = "[^"+$page+"^"

      # Get the page record.
      var str page_record
      stex -p $stex_arg $page_list > $page_record
      stex "[^<r>^" $page_record > null

      # Extract the count, which is after <f>.
      var str count_str
      stex -p "^<f>^[" $page_record > $count_str
    
      # Get numeric count
      var int count
      set $count = makeint(str($count_str))

      # Increment count
      set $count = $count+1

      # Create the updated page record
      var str updated_page_record
      set $updated_page_record = $page+"<f>"+makestr(int($count))

      # Replace page_record with updated_page_record in page_list
      var str sal_arg
      set $sal_arg = "^"+$page_record+"^"
      sal $sal_arg $updated_page_record $page_list > null

      done
    else
    do
      # This page is not in the list. Add to the list.
      set $page_list = $page_list + "<r>" + $page + "<f>1"
      done
    endif # Completes if ( { sen $sen_arg $page_list } > 0 )

    # The following code is very similar to the above code for page.
    # Is this referrer already in the referrer_list ? If so, we will just increment the count.
    # If not, we will add it and set the count to 1.
    set $sen_arg = "^"+$referrer+"^"
    if ( { sen $sen_arg $referrer_list } > 0 )
    do
      # This referrer is already in referrer_list.

      # Create dynamic argument for the stex command.
      var str stex_arg
      set $stex_arg = "[^"+$referrer+"^"

      # Get the referrer record.
      var str referrer_record
      stex -p $stex_arg $referrer_list > $referrer_record
      stex "[^<r>^" $referrer_record > null

      # Extract the count, which is after <f>.
      var str count_str
      stex -p "^<f>^[" $referrer_record > $count_str
    
      # Get numeric count
      var int count
      set $count = makeint(str($count_str))

      # Increment count
      set $count = $count+1

      # Create the updated referrer record
      var str updated_referrer_record
      set $updated_referrer_record = $referrer+"<f>"+makestr(int($count))

      # Replace referrer_record with updated_referrer_record in referrer_list
      var str sal_arg
      set $sal_arg = "^"+$referrer_record+"^"
      sal $sal_arg $updated_referrer_record $referrer_list > null

      done
    else
    do
      # This referrer is not in the list. Add to the list.
      set $referrer_list = $referrer_list + "<r>" + $referrer + "<f>1"
      done
    endif # Completes if ( { sen $sen_arg $referrer_list } > 0 )
    done

  endif # Completes if ($line <> "")

  done # completes do after while ( $content <> "")

# Print the page_list
echo "Pages"
while ( { sen "^<r>^" $page_list } > 0 )
  sal "^<r>^" "\n" $page_list > null
while ( { sen "^<f>^" $page_list } > 0 )
  sal "^<f>^" "\t" $page_list > null
echo $page_list


# Print the referrer_list
echo "Referrers"
while ( { sen "^<r>^" $referrer_list } > 0 )
  sal "^<r>^" "\n" $referrer_list > null
while ( { sen "^<f>^" $referrer_list } > 0 )
  sal "^<f>^" "\t" $referrer_list > null
echo $referrer_list