#!/bin/sh
#	Provide some data from nginx log files.
#
# List the number of entries in each log file.
# List the number of requests for robots.txt.
# List the number of blocked IP addresses for the log files.
# nginx keeps 52 log files (1 year's weekly logs).
#
# Log files are rotated weekly, but not at a fixed time -- Sat/Sun/Mon.
# So they will cover 6-8 days. Week-to-week comparisons will not be exact.
# There's also variable down time when the network connection is lost.
#
# Send the output through /home/bchivers/commands/sed.html.ttbl
#
#	Brent Chivers		2020/Dec/27

LOGDIR=/var/log/nginx
LOGNAME=access

echo "      Log	Reqs	Robots	Mail	Bad IPs"

for FILE in `ls -tr $LOGDIR | grep $LOGNAME`
    do
	case $FILE in
	    *.gz)
		REQS=`zcat $LOGDIR/$FILE | wc -l`
		ROBOTS=`zcat $LOGDIR/$FILE | grep robots.txt | wc -l`
		FEEDBACKS=`zcat $LOGDIR/$FILE | grep /cgi-bin/mailer.bin | wc -l`
		;;
	    *)
		REQS=`wc -l < $LOGDIR/$FILE`		# redirect to avoid printing filename
		ROBOTS=`grep robots.txt $LOGDIR/$FILE | wc -l`
		FEEDBACKS=`grep /cgi-bin/mailer.bin $LOGDIR/$FILE | wc -l`
		;;
	    esac
	DATE=`ls -tr --full-time $LOGDIR/$FILE | awk '{print $6}'`
	case $FILE in
	    $LOGNAME.log)
		BLOCKFILES=`ls $LOGDIR/deny* | fgrep -v '.'`
#		BLOCKS="-"	# if logs have just rotated, don't use previous count
		;;
	    *)
		BLOCKFILES=`ls $LOGDIR/deny* | grep $DATE`
		;;
	    esac
	if [ -z "$BLOCKFILES" ] ; then
		BLOCKS="-"
	   else
#		BLOCKS=`sort $LOGDIR/deny*.$DATE | sort | uniq | wc -l`
		BLOCKS=`sort $BLOCKFILES | sort | uniq | wc -l`
	   fi
	echo "$DATE	$REQS	$ROBOTS	$FEEDBACKS	$BLOCKS"
    done
echo "      Log	Reqs	Robots	Mail	Bad IPs"