#! /bin/bash

# script analyses ALL GUI production.log files for most visited pages
# results will be in the /tmp/log_analysis

. /usr/src/m2/framework/bash_functions.sh
. /usr/src/m2/framework/settings.sh


# ---- VARS -----

VERSION="1.1.1"
SCRIPT_NAME="M2 GUI Log Analysis for Most-visited-pages"

SRC_DIR="/home/m2/log"
DIR="/tmp/log_analysis"
LOG=$DIR/p

TMP_FILE1="$DIR/p1"

RESULT_FILE="$DIR/gui_most_visited_pages_result.log"

# ---- MAIN -----

k_start

if [ -d $DIR ];then
    report "$DIR present" 0
else
    report "$DIR not present" 2
    mkdir $DIR
    if [ -d $DIR ];then
        report "$DIR created" 4
    else
        report "Failed to create $DIR" 1
        exit
    fi
fi


if [ -d $SRC_DIR ];then
    report "$SRC_DIR present" 0
else
    report "$SRC_DIR not present. No GUI on server?" 2
    exit 1
fi


rm -fr $TMP_FILE1

#FILES="$SRC_DIR/production.log"  # one file for testing
FILES="$SRC_DIR/production.log*"   # all files
for f in $FILES
do
    report "Processing $f file..." 3
    cp -fr $f $LOG

    # getting links, removing parts after ?, digits at the end of string, / at the end
    cat $LOG | zgrep "Started GET" | grep -v "assets" | grep -v "images" | grep -v "hourly_actions" | grep -v "paypal/payment" | cut -d\" -f2 | awk -F '?' '{print $1}' | sed 's/[0-9]\+$//' | sed 's/\/$//' >> $TMP_FILE1

done

rm -fr $LOG


rm -fr $RESULT_FILE

report "Finished gathering data" 0
report "Calculating most visited pages..." 3

cat $TMP_FILE1 | sort | uniq -c | sort -k1nr > $RESULT_FILE

report "Finished calculating" 0
report "Results in: $RESULT_FILE" 3

echo
echo "Most popular pages:"
echo
cat $RESULT_FILE | head -30
echo

rm -fr $TMP_FILE1

report "Script completed" 3