fediverse-blocklists/fedisync.sh

86 lines
4.2 KiB
Bash
Raw Normal View History

#!/bin/bash
set -e
# generate files
#exec > >(tee /opt/fediblockhole/logs/fedisync.log) 2>&1
exec 2> >(tee /opt/fediblockhole/logs/fedisync.log)
echo "Pulling down all blocklist sources..."
cd /opt/fediblockhole
GIT=`which git`
${GIT} fetch
#${GIT} fetchall
2023-05-11 19:53:02 +00:00
PULLCONFIG="/opt/fediblockhole/config/pull.conf.toml" # default, pull sources WITHOUT public_description
#PULLCONFIG="/opt/fediblockhole/config/pull.all.conf.toml" # pull down all sources WITH public_description
SYNC=`which fediblock-sync`
2023-05-11 19:53:02 +00:00
${SYNC} -c $PULLCONFIG # pull down all files, create max list
# rename and remove intermediate files
rm -f /opt/fediblockhole/blocklists/file:---opt-fediblockhole-blocklists-__allowlist.csv.csv
mv /opt/fediblockhole/blocklists/https:--seirdy.one-pb-tier0.csv.csv /opt/fediblockhole/blocklists/tier0.csv
mv /opt/fediblockhole/blocklists/https:--seirdy.one-pb-pleroma.envs.net.csv.csv /opt/fediblockhole/blocklists/pleroma.envs.net.csv
mv /opt/fediblockhole/blocklists/https:--raw.githubusercontent.com-gardenfence-blocklist-main-gardenfence-fediblocksync.csv.csv /opt/fediblockhole/blocklists/gardenfence.csv
mv /opt/fediblockhole/blocklists/https:--artisan.chat-api-v1-instance-domain_blocks.csv /opt/fediblockhole/blocklists/artisan.chat.csv
mv /opt/fediblockhole/blocklists/https:--mastodon.art-api-v1-instance-domain_blocks.csv /opt/fediblockhole/blocklists/mastodon.art.csv
mv /opt/fediblockhole/blocklists/https:--mastodon.online-api-v1-instance-domain_blocks.csv /opt/fediblockhole/blocklists/mastodon.online.csv
mv /opt/fediblockhole/blocklists/https:--mastodon.social-api-v1-instance-domain_blocks.csv /opt/fediblockhole/blocklists/mastodon.social.csv
mv /opt/fediblockhole/blocklists/https:--rage.love-api-v1-instance-domain_blocks.csv /opt/fediblockhole/blocklists/rage.love.csv
mv /opt/fediblockhole/blocklists/https:--sunny.garden-api-v1-instance-domain_blocks.csv /opt/fediblockhole/blocklists/sunny.garden.csv
mv /opt/fediblockhole/blocklists/https:--toot.wales-api-v1-instance-domain_blocks.csv /opt/fediblockhole/blocklists/toot.wales.csv
mv /opt/fediblockhole/blocklists/https:--union.place-api-v1-instance-domain_blocks.csv /opt/fediblockhole/blocklists/union.place.csv
mv /opt/fediblockhole/blocklists/https:--solarpunk.moe-api-v1-instance-domain_blocks.csv /opt/fediblockhole/blocklists/solarpunk.moe.csv
# destroy the 'max' list
rm -f /opt/fediblockhole/blocklists/_unified_max_blocklist.csv
# create fedi council tier0 file and 100% file
${SYNC} -c /opt/fediblockhole/config/council.tier0.conf.toml && \
${SYNC} -c /opt/fediblockhole/config/100.percent.conf.toml
# create unified tier0 file
${SYNC} -c /opt/fediblockhole/config/tier0.conf.toml
# create missing from mastodon.social blocklist file, and the inverse file
#${SYNC} -c /opt/fediblockhole/config/ms-noblock.conf.toml && \
#${SYNC} -c /opt/fediblockhole/config/ms-we-block.conf.toml
# create intermediate working tier files
${SYNC} -c /opt/fediblockhole/config/tier1-threshold.conf.toml -m min && \
${SYNC} -c /opt/fediblockhole/config/tier2-threshold.conf.toml -m min && \
${SYNC} -c /opt/fediblockhole/config/tier3-threshold.conf.toml -m min
# merge in associated tier file with tier0 blocks
${SYNC} -c /opt/fediblockhole/config/tier1.conf.toml -m min && \
${SYNC} -c /opt/fediblockhole/config/tier2.conf.toml -m min && \
${SYNC} -c /opt/fediblockhole/config/tier3.conf.toml -m min && \
cp /opt/fediblockhole/blocklists/_unified_tier0_blocklist.csv /opt/fediblockhole/blocklists/_unified_min_blocklist.csv
# cleanup working tier files
rm -f /opt/fediblockhole/blocklists/_working_*.csv
2023-05-11 17:58:20 +00:00
# copy all files to mastodon subfolder
cp /opt/fediblockhole/blocklists/*.csv /opt/fediblockhole/blocklists/mastodon/ && \
cp /opt/fediblockhole/blocklists/README.md /opt/fediblockhole/blocklists/mastodon/README.md
2023-05-11 17:58:20 +00:00
echo "Creating mastodon versions of files..."
# convert files to Mastodon header
cd /opt/fediblockhole/blocklists/mastodon
rm -f __allowlist.csv
# replace first line with mastodon-format header row
headerrow="#domain,#severity,#reject_media,#reject_reports,#public_comment,#obfuscate"
for file in *.csv; do
2023-05-11 17:58:20 +00:00
sed -i "1s/.*/$headerrow/" $file
2023-05-11 18:47:55 +00:00
# replace bad line terminators
sed -i "s/\r//g" $file
# lowercase file to fix booleans
sed -i "s/\(.*\)/\L\1/" $file
2023-05-11 17:58:20 +00:00
done
echo "Pull complete."