path: root/chili_change_detector

#! /bin/bash

check_every="37s" # Put `m' or `s' on the end because this will be used with the
		 # `sleep' command.
seen_issues_log="seen_issues_log"
atom_feed_log="atom_feed_log"
atom_url="https://labs.parabola.nu/activity.atom"
bot_ipc="/tmp/un-provoked-message-store"
# fauno wants him to deliver messages spaced out by two minutes so I need nother
# fifo
another_fifo="/tmp/pbot-ng_chili_change_detector_fifo"

# If the fifo doesn't exist then create it.
[[ -p ${another_fifo} ]] || mkfifo "${another_fifo}"

# OK this code will forward no more than two messages onto pbot-ng every two
# minutes (fauno doesn't want flood). It runs in the background. It could be
# sensible to split this out into a seperate file so it can tidily be used by
# other pbot-ng modules but I cba. This code is quite l33t actually because it
# doesn't delay the messages at all unless two messages get delivered in two
# minutes.
tail -f "${another_fifo}" | while true
do
    read smile && echo "${smile}" >> "${bot_ipc}"
    read -t 120 smile && echo "${smile}" >> "${bot_ipc}" ||
    continue
    sleep 2m
done &

declare -A bugs
declare -A old_bugs

function get_feed_make_array
{
    # Download the feed.
    feed=$( mktemp )
    header_dump=$( mktemp )

    if [[ -n ${etag} ]]
    then
	curl --user-agent "pbot-ng" -D "${header_dump}" -H "If-None-Match: \"${etag}\"" "${atom_url}" 2> /dev/null > "${feed}"
	ret_val="${?}"
    else
	curl --user-agent "pbot-ng" -D "${header_dump}" "${atom_url}" 2> /dev/null > "${feed}"
	ret_val="${?}"
    fi

    # Get the first line of the headers.
    read http_response < "${header_dump}"

    # If curl's exit status was zero and the http response was 200.
    if (( ! ret_val )) && [[ ${http_response} == *' 200 '* ]]
    then
	# Is there an ETag?
	if etag_line=$( grep 'ETag:' "${header_dump}" )
	then
	    etag_line="${etag_line#*\"}"
	    etag="${etag_line%\"*}" # Now we have the ETag.
	else
	    # It seems that the server no longer sends Etags
	    unset etag
	fi
    else
	# This indicates either a failure to download the page or that the page
	# has not changed since the last time it was downloaded.
	rm "${feed}"
	rm "${header_dump}"
	return 1
    fi

    unset bugs
    declare -A -g bugs

    # We want an array of all the bug titles and their urls.
    stage=0
    while read -r line
    do
	case "${stage}" in
	    0 )
		[[ ${line} == *'<entry>'* ]] &&
		{
		    title=""
		    url=""
		    updated=""
		    stage=1
		}

		;;
	    1 )
		case "${line}" in
		    *'<title>'*'</title>'* )
			booboo="${line#*>}"
			title="${booboo%<*}"
			;;
		    *'<link '* )
			yumyum="${line##*href=\"}"
			url="${yumyum%%\"*}"
			;;
		    *'<updated>'*'</updated>'* )
			foobar="${line#*>}"
			updated="${foobar%<*}"
			;;
		    *'</entry>'* )
			# We don't want revisions.
			if [[ -n ${title} ]] && [[ -n ${url} ]] && [[ -n ${updated} ]] && [[ "${title}" != *' - Revision '* ]]
			then
			    # Add this to the array, fixing the url if it's
			    # broken.
			    bugs["${title}🐵${updated}"]="${url/chili/labs.parabola.nu}"
			fi
			stage=0
			;;
		esac
		;;
	esac
    done < "${feed}"

    rm "${feed}"
    rm "${header_dump}"

    return 0
}

function slide_and_write
{
    # We need to make the new array the old one.
    unset old_bugs
    declare -A -g old_bugs

    for key in "${!bugs[@]}"
    do
	ii=$( sed 's/([^)]\+)//' <<< "${key}" )
	old_bugs[${ii}]=${bugs[${key}]}
    done

    # We now write this to the file. We seperate the fields with some sort of
    # cat face: 🐱
    for i in "${!old_bugs[@]}"
    do
	echo "${i}🐱 ${old_bugs[${i}]}"
    done > ${atom_feed_log}
}

if [[ -f ${atom_feed_log} ]]
then
    # Get the array that stores the info of all bug titles and their urls out of
    # the file it's stored in.
    while read -r line
    do
	old_bugs["${line%%🐱 *}"]="${line##*🐱 }"
    done < "${atom_feed_log}"
else
    # The log does not exist so we create it!
    get_feed_make_array
    slide_and_write

    # Also we need to create the log of seen issues.
    for i in "${!bugs[@]}"
    do
	ii=$( sed 's/([^)]\+)//' <<< "${i}" )
	[[ "${ii%%🐵*}" =~ \#[[:digit:]]+ ]] # match the bug number e.g. #390
        echo "${BASH_REMATCH}"
    done > "${seen_issues_log}"
fi

while true
do
    # Download the feed and make an array of the bugs in it. If it fails or
    # hasn't changed then we just wait until it's time to check again and then
    # continue with the next iteration of the loop.
    get_feed_make_array || { sleep "${check_every}" ; continue ; }

    temp_changes=$( mktemp )

    # Compare this array to the previous.  If any new have appeared since last
    # time then we check if it's in a log of seen issues, and if not, we add it,
    # and we know that this is a creation, not a change.
    for i in "${!bugs[@]}"
    do
	ii=$( sed 's/([^)]\+)//' <<< "${i}" )
	# If this bug is new since last time.
	if [[ -z ${old_bugs[${ii}]} ]]
	then
	    # Check the log of seen issues to find out if this is a creation or
	    # a change.
            [[ "${ii%%🐵*}" =~ \#[[:digit:]]+ ]] # match the bug number e.g. #390

	    if grep "${BASH_REMATCH}" "${seen_issues_log}" > /dev/null 2> /dev/null
	    then
		creation_or_change="changed"
	    else
		creation_or_change="created"

		# Add this issue to the log.
		echo "${BASH_REMATCH}" >> "${seen_issues_log}"
	    fi

	    real_title="${i%%🐵*}"
            real_title_a="${real_title%%:*}"
            real_title_b="${real_title#*:}"
            cat="${real_title_a%% - *}" # e.g. `pbot-ng'
            num="${real_title_a#* - }"
	    echo "${num} ${creation_or_change}: ( ${cat} -${real_title_b} ) ${bugs[${i}]}" >> "${temp_changes}"
	fi
    done

    if (( $( wc -l "${temp_changes}" 2> /dev/null | cut -d ' ' -f 1 ) > 12 ))
    then
	echo "More than 12 changes have been detected on the bug tracker. Ignoring." >> "${bot_ipc}"
    else
	while read line
	do
	    echo "${line}" >> "${another_fifo}"
	done < "${temp_changes}"
    fi

    rm -f "${temp_changes}"

    # Make this array the old one and write it to the file in case we need to
    # retrieve it later.
    slide_and_write

    sleep "${check_every}"
done