343 lines
11 KiB
Bash
Executable file
343 lines
11 KiB
Bash
Executable file
#!/bin/bash
|
|
|
|
# Header
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
printf "miniflux-filter - git.mgrote.net/mg/miniflux-filter\n"
|
|
fi
|
|
|
|
### wartezeit zwischen laeufen
|
|
MF_SLEEP="${MF_SLEEP:=30}"
|
|
### MF_DEBUG output
|
|
# standardmäßig 0 = aus
|
|
MF_DEBUG="${MF_DEBUG=0}"
|
|
|
|
# Functions
|
|
function output_help {
|
|
cat <<EOF
|
|
miniflux-filter
|
|
|
|
https://git.mgrote.net/mg/miniflux-filter
|
|
|
|
Usage:
|
|
- filter.sh [OPTIONS]
|
|
|
|
Options:
|
|
-h, --help displays this text
|
|
|
|
EOF
|
|
}
|
|
function check_dependencies {
|
|
# pruefe ob jq installiert ist
|
|
# https://stackoverflow.com/questions/592620/how-can-i-check-if-a-program-exists-from-a-bash-script
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] check dependencies"
|
|
fi
|
|
if ! command -v jq &> /dev/null
|
|
then
|
|
echo "[ERROR] jq could not be found!"
|
|
exit 5
|
|
fi
|
|
if ! command -v curl &> /dev/null
|
|
then
|
|
echo "[ERROR] curl could not be found!"
|
|
exit 6
|
|
fi
|
|
if ! command -v xargs &> /dev/null
|
|
then
|
|
echo "[ERROR] xargs could not be found!"
|
|
exit 8
|
|
fi
|
|
if ! command -v sed &> /dev/null
|
|
then
|
|
echo "[ERROR] sed could not be found!"
|
|
exit 9
|
|
fi
|
|
if ! command -v sort &> /dev/null
|
|
then
|
|
echo "[ERROR] sort could not be found!"
|
|
exit 10
|
|
fi
|
|
if ! command -v awk &> /dev/null
|
|
then
|
|
echo "[ERROR] awk could not be found!"
|
|
exit 11
|
|
fi
|
|
}
|
|
function check_vars {
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] check if vars are (correctly) set"
|
|
fi
|
|
# pruefe ob alle vars gesetzt sind
|
|
# -z = ob laenge gleich null ist
|
|
if [[ -z "${MF_AUTH_TOKEN}" ]]; then
|
|
# shellcheck disable=SC2016
|
|
# shellcheck disable=SC2102
|
|
echo [ERROR] '"$MF_AUTH_TOKEN"' not set.
|
|
exit 2
|
|
fi
|
|
if [[ -z "${MF_API_URL}" ]]; then
|
|
# shellcheck disable=SC2016
|
|
# shellcheck disable=SC2102
|
|
echo [ERROR] '"$MF_API_URL"' not set.
|
|
exit 3
|
|
fi
|
|
# pruefe ob filter-datei NICHT existiert
|
|
if [[ -z "${MF_FILTERLIST}" ]]; then
|
|
# shellcheck disable=SC2016
|
|
# shellcheck disable=SC2102
|
|
echo [ERROR] '"$MF_FILTERLIST"' not set.
|
|
exit 1
|
|
fi
|
|
}
|
|
function check_ping_connectivity {
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] check if miniflux can be reached with ping"
|
|
fi
|
|
# pruefe ob miniflux erreichbar ist, wenn ja setze abbruchbedingung, sonst warte
|
|
mf_ping_connectivity=0
|
|
while [[ $mf_ping_connectivity -eq 0 ]]; do
|
|
# hole fqdn + port aus MF_API_URL und pinge an
|
|
mf_fqdn=$(echo "$MF_API_URL" | cut -d'/' -f3 | cut -d':' -f1-2 | cut -d":" -f1)
|
|
if ping -c 1 "$mf_fqdn" > /dev/null 2>&1 ; then
|
|
mf_ping_connectivity=1
|
|
else
|
|
mf_ping_connectivity=0
|
|
sleep 10
|
|
echo "[INFO] wait for miniflux (ping)..."
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] miniflux could not be pinged, wait 10s"
|
|
fi
|
|
fi
|
|
done
|
|
}
|
|
function check_api_connectivity {
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] check if miniflux-api can be reached with curl"
|
|
fi
|
|
# pruefe ob miniflux erreichbar ist, wenn ja setze abbruchbedingung, sonst warte
|
|
mf_api_connectivity=0
|
|
while [[ $mf_api_connectivity -eq 0 ]]; do
|
|
http_status_code=$(curl --silent --header "X-Auth-Token: $MF_AUTH_TOKEN" "$MF_API_URL/me" -i | grep HTTP/2 | awk '{print $2}')
|
|
if [[ $http_status_code -eq 200 ]] ; then
|
|
mf_api_connectivity=1
|
|
else
|
|
mf_api_connectivity=0
|
|
sleep 10
|
|
echo "[INFO] wait for miniflux (api)..."
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] api could not be reached, wait 10s"
|
|
fi
|
|
fi
|
|
done
|
|
}
|
|
function debug_output {
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo -----------------------------------------------------
|
|
echo [DEBUG] Sleep-Intervall: "$MF_SLEEP"
|
|
echo [DEBUG] Auth-Token: "$MF_AUTH_TOKEN"
|
|
echo [DEBUG] MF-URL: "$MF_API_URL"
|
|
echo -----------------------------------------------------
|
|
fi
|
|
}
|
|
function get_unread_entries {
|
|
# hole alle ungelesenen entries und speichere sie in der variable unread_entries
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] get unread entries from miniflux"
|
|
fi
|
|
unread_entries="$(curl --silent --header "X-Auth-Token: $MF_AUTH_TOKEN" "$MF_API_URL/entries?status=unread&limit=0")"
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] show unread entries from miniflux"
|
|
echo -----------------------------------------------------
|
|
echo "$unread_entries"
|
|
echo -----------------------------------------------------
|
|
fi
|
|
}
|
|
function filter_entries {
|
|
echo "[INFO] Filtering entries..."
|
|
# fuer jede Zeile in $MF_FILTERLIST
|
|
echo "$MF_FILTERLIST" | while read -r line; do
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] set search values"
|
|
fi
|
|
# setze $url auf den Wert vor dem Trennzeichen/Delimiter, ersetze alle Grossschreibungen durch Kleinschreibung
|
|
url=$(echo "$line" | tr '[:upper:]' '[:lower:]' | awk --field-separator="::" '{print $1}')
|
|
# setze $suchbegriff auf den Wert vor dem Trennzeichen/Delimiter, ersetze alle Grossschreibungen durch Kleinschreibung
|
|
suchbegriff=$(echo "$line" | tr '[:upper:]' '[:lower:]' | awk --field-separator="::" '{print $2}')
|
|
# in jq uebergebe shell-variablen an jq selber
|
|
# entferne die erste ebene
|
|
# suche jeden eintrag wo die feed_url == $url, konvertiere in kleinschreibung, dasselbe fuer den title
|
|
# gebe dann nur die id aus
|
|
# die id, wird dann an die variable marked_entries angehangen
|
|
# z.B. 53443 52332 48787 [...]
|
|
if [[ -n "$url" ]]; then
|
|
# abfangen der letzten zeile die leer ist; sonst wird alles gefiltert
|
|
if [[ -n "$suchbegriff" ]]; then
|
|
# abfangen der letzten zeile die leer ist; sonst wird alles gefiltert
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo [DEBUG] url:"$url" - value:"$suchbegriff"
|
|
fi
|
|
# das leerzeichen am anfang ist notwendig, trennt die zahlenwerte
|
|
# suche in titel
|
|
titles+=" $(echo "$unread_entries" | jq --arg url "$url" --arg suchbegriff "$suchbegriff" '.entries[] | select(.feed.site_url | ascii_downcase | contains($url)) | select(.title | ascii_downcase | contains($suchbegriff)) | .id' )"
|
|
# suche in content
|
|
content+=" $(echo "$unread_entries" | jq --arg url "$url" --arg suchbegriff "$suchbegriff" '.entries[] | select(.feed.site_url | ascii_downcase | contains($url)) | select(.content | ascii_downcase | contains($suchbegriff)) | .id' )"
|
|
fi
|
|
fi
|
|
# sortiere und forme marked_entries um
|
|
marked_entries="$title $content"
|
|
# https://unix.stackexchange.com/questions/353321/remove-all-duplicate-word-from-string-using-shell-script
|
|
# entfernt doppelte eintraege innerhalb einer zeile
|
|
marked_entries=$(echo marked_entries | xargs -n1 | sort -u | xargs | sed -r 's/\s/\, /g')
|
|
echo "$marked_entries" #d
|
|
done
|
|
}
|
|
function mark_as_read {
|
|
# https://stackoverflow.com/questions/3869072/test-for-non-zero-length-string-in-bash-n-var-or-var
|
|
# wenn variable NICHT leer...
|
|
# sende in put request mit curl
|
|
# der wert muss escaped werden, aber NICHT die variable die uebergeben wird
|
|
if [[ $MF_DEBUG -eq 1 ]]; then
|
|
echo "[DEBUG] mark entries as read"
|
|
echo "[DEBUG] marked entry ids:"
|
|
echo "$marked_entries"
|
|
fi
|
|
# wenn NICHT leer
|
|
# sed wandelt 123 345 456 in 123, 245, 345 um.
|
|
if [[ -n "$marked_entries" ]]; then
|
|
curl --request PUT --silent --header "X-Auth-Token: $MF_AUTH_TOKEN" --header "Content-Type: application/json" --data "{"entry_ids":"$marked_entries"],"status":"read"}" "$MF_API_URL/entries"
|
|
# gebe entry-titel aus
|
|
for i in "$marked_entries" ; do
|
|
# gebe aus welcher eintrag gefiltert wurde, cut begrenzt die maximale laenge auf 100 zeichen
|
|
# jq "XXX", fügt XXX in ausgabe hinzu
|
|
echo [INFO] Filtered entry "$i" - "$(curl --silent --header "X-Auth-Token: $MF_AUTH_TOKEN" "$MF_API_URL"/entries/"$i" | jq --join-output '"url: ",.feed.site_url," - title: ", .title' | cut -c -100)".
|
|
done
|
|
fi
|
|
# setze variablen auf leer
|
|
marked_entries=""
|
|
}
|
|
|
|
|
|
# debug
|
|
MF_FILTERLIST="
|
|
golem.de::Anzeige
|
|
golem.de::Aus dem Verlag:
|
|
golem.de::Elon Musk
|
|
golem.de::Fifa
|
|
golem.de::Fussball
|
|
golem.de::Fußball
|
|
golem.de::(g+)
|
|
golem.de::Golem Karrierewelt
|
|
golem.de::PODCAST BESSER WISSEN:
|
|
golem.de::Sonst noch was?
|
|
golem.de::Star Trek:
|
|
golem.de::Tech Talks:
|
|
golem.de::Twitter
|
|
golem.de::Wochenrückblick
|
|
heise.de::Dienstag:
|
|
heise.de::Die Produktwerker
|
|
heise.de::Elon Musk
|
|
heise.de::Ferris Talk
|
|
heise.de::FIFA
|
|
heise.de::Freitag
|
|
heise.de::Fußball-WM
|
|
heise.de::heise+
|
|
"
|
|
# Doing
|
|
case "$1" in
|
|
--help | -h | help)
|
|
output_help
|
|
;;
|
|
*)
|
|
check_dependencies
|
|
# fuehre script durchgaengig aus
|
|
while true; do
|
|
check_vars
|
|
debug_output
|
|
check_ping_connectivity
|
|
check_api_connectivity
|
|
get_unread_entries
|
|
filter_entries
|
|
mark_as_read
|
|
# warte zeit x
|
|
sleep "$MF_SLEEP"
|
|
done
|
|
esac
|
|
|
|
|
|
|
|
# curl --request PUT --header "X-Auth-Token: hLPp23gO8QeZWPYXXXXXXXMs0jG0XTYgNZBgRA=" --header "Content-Type: application/json" --data '{"entry_ids":[388229, 388211], "status":"read"}' https://miniflux.mgrote.net/v1/entries
|
|
|
|
|
|
|
|
lese suchwörter per feedurl ein
|
|
suche pro feed url nach suchwortern
|
|
markiere als gelesen
|
|
|
|
|
|
|
|
|
|
get all feeds where are unread entries+
|
|
check if for that feeds a filter is defined
|
|
search for keywords for this feedsunread entries
|
|
mark entries as read
|
|
|
|
|
|
token: hLPp23gO8QeZWPYzK3jYgmGbABvMs0jG0XTYgNZBgRA= # später tauschen
|
|
|
|
|
|
curl -H "X-Auth-Token: hLPp23gO8QeZWPYzK3jYgmGbABvMs0jG0XTYgNZBgRA=" https://miniflux.mgrote.net/v1/me
|
|
|
|
|
|
#!/bin/bash
|
|
# get all feeds where are unread entries+
|
|
|
|
token="X-Auth-Token: hLPp23gO8QeZWPYzK3jYgmGbABvMs0jG0XTYgNZBgRA="
|
|
url="https://miniflux.mgrote.net/v1/"
|
|
for feed in $(curl --silent -H $token $url/feeds | jq .[].id) ; do
|
|
unread_entries=$(curl --silent -H $token $url/feeds/$feed/entries?status=unread)
|
|
done
|
|
|
|
|
|
|
|
|
|
----
|
|
#!/bin/bash
|
|
# get all feeds where are unread entries+
|
|
|
|
MF_FILTERLIST="
|
|
golem.de::Anzeige
|
|
golem.de::Aus dem Verlag:
|
|
golem.de::Elon Musk
|
|
golem.de::Fifa
|
|
golem.de::Fussball
|
|
golem.de::Fußball
|
|
golem.de::(g+)
|
|
"
|
|
|
|
token="X-Auth-Token: hLPp23gO8QeZWPYzK3jYgmGbABvMs0jG0XTYgNZBgRA="
|
|
url="https://miniflux.mgrote.net/v1"
|
|
|
|
# für jeden feed
|
|
for feed in $(curl --silent -H "$token" "$url/feeds" | jq .[].id) ; do
|
|
echo for1
|
|
unread_entries=$(curl --silent -H "$token" "$url/feeds/$feed/entries?status=unread" | jq .total)
|
|
# der ungelesen einträge hat
|
|
if [[ $unread_entries -gt 0 ]] ; then
|
|
echo if1
|
|
fi
|
|
done
|
|
|
|
echo "$MF_FILTERLIST" | while read -r line ; do
|
|
echo while1
|
|
filter_url=$(echo "$line" | tr '[:upper:]' '[:lower:]' | awk --field-separator="::" '{print $1}')
|
|
echo $filter_url
|
|
mf_site_url=$(curl --silent -H "$token" "$url/feeds/$feed" | jq --raw-output .site_url)
|
|
echo $mf_site_url
|
|
# prufe ob ein filter dafur existiert
|
|
echo "$filter_url" | grep "$mf_site_url"
|
|
if [[ $(echo "$filter_url" | grep --silent "$mf_site_url") ]] ; then
|
|
echo for2
|
|
echo "id: $feed"
|
|
echo "$mf_site_url $filter_url"
|
|
fi
|
|
done
|