From 19ffd00d88c1e605f0bd246ba896d1b7a76f8b5e Mon Sep 17 00:00:00 2001 From: Lillian Skinner Date: Tue, 7 Apr 2026 18:53:21 -0400 Subject: [PATCH] Delete SCRAPE_JOB.SH --- SCRAPE_JOB.SH | 127 -------------------------------------------------- 1 file changed, 127 deletions(-) delete mode 100644 SCRAPE_JOB.SH diff --git a/SCRAPE_JOB.SH b/SCRAPE_JOB.SH deleted file mode 100644 index 74470f7..0000000 --- a/SCRAPE_JOB.SH +++ /dev/null @@ -1,127 +0,0 @@ -#!/usr/bin/env bash -start_timer() { - START_SECONDS=$(date +%s) - START_READABLE=$(date "+%Y-%m-%d %H:%M:%S") -} -end_timer() { - END_SECONDS=$(date +%s) - END_READABLE=$(date "+%Y-%m-%d %H:%M:%S") - ELAPSED_SECONDS=$((END_SECONDS - START_SECONDS)) - ELAPSED_READABLE=$(printf "%02d:%02d:%02d" \ - $((ELAPSED_SECONDS/3600)) \ - $(( (ELAPSED_SECONDS%3600)/60 )) \ - $((ELAPSED_SECONDS%60))) -} -push_log() { - PROCURL=$(basename $(echo $1)) - echo " \ - View \ - $2 \ - $START_READABLE \ - $END_READABLE \ - $ELAPSED_READABLE \ - " >> $CRON_LOG_INDEX -} -push_webhook() { - PROCURL=$(basename "$2") - WEBHOOK_URL="https://discord.com/api/webhooks/1472056322886209600/8EtHDzTdVYuaU2mn0-fY6BZZwxW4ZMkNnGzFyTCJhcS6FMHYagjxeyw0rw9o5S-TNRRA" - WEBHOOK_JSON=$(cat <> $CRON_LOG_SCRIPT -end_timer && push_log "$CRON_LOG_SCRIPT" "BACK UP SCRIPT" - -start_timer -./SCRAPE_MEET.SH >> $CRON_LOG_MEET -end_timer && push_log "$CRON_LOG_MEET" "SCRAPE_MEET.SH" && push_webhook "SCRAPE_MEET.SH" "$CRON_LOG_MEET" "Done processing city meetings." "$START_READABLE" "$ELAPSED_READABLE" - -#start_timer -#./SCRAPE_PLAN.SH >> $CRON_LOG_PLAN -#end_timer && push_log "$CRON_LOG_PLAN" "SCRAPE_PLAN.SH" && push_webhook "SCRAPE_PLAN.SH" #"$CRON_LOG_PLAN" "Done processing planning applications." "$START_READABLE" "$ELAPSED_READABLE" - -start_timer -./SCRAPE_ESCRIBE.SH >> $CRON_LOG_JSON -end_timer && push_log "$CRON_LOG_JSON" "SCRAPE_JSON.SH" && push_webhook "SCRAPE_JSON.SH" "$CRON_LOG_JSON" "Done backing up eScribe meeting lists." "$START_READABLE" "$ELAPSED_READABLE" - -start_timer -./SCRAPE_LTC.SH >> $CRON_LOG_LTC -end_timer && push_log "$CRON_LOG_LTC" "SCRAPE_LTC.SH" && push_webhook "SCRAPE_LTC.SH" "$CRON_LOG_LTC" "Done processing LTC meetings." "$START_READABLE" "$ELAPSED_READABLE" - -start_timer -./SCRAPE_LPS.SH >> $CRON_LOG_LPS -end_timer && push_log "$CRON_LOG_LPS" "SCRAPE_LTC.SH" && push_webhook "SCRAPE_LPS.SH" "$CRON_LOG_LPS" "Done processing LPS meetings." "$START_READABLE" "$ELAPSED_READABLE" - -start_timer -aws s3 sync ./LondonArchive "s3://public-file-browser-files-0261cd08327d/" --profile london --no-progress --size-only >> $CRON_LOG_S3 -end_timer && push_log "$CRON_LOG_S3" "AWS S3 SYNC" && push_webhook "AWS S3 SYNC" "$CRON_LOG_S3" "Done syncing files to S3." "$START_READABLE" "$ELAPSED_READABLE" - -rm -rf "./LondonArchive" -mkdir "./LondonArchive" - -start_timer -# Make/upload sitemap AFTER clearing the work dir. Otherwise everything gets uploaded again. -./MAKE_SITEMAP.SH >> $CRON_LOG_SITEMAP -end_timer && push_log "$CRON_LOG_SITEMAP" "MAKE_SITEMAP.SH" && push_webhook "MAKE_SITEMAP.SH" "$CRON_LOG_SITEMAP" "Done updating archive sitemap, requested YaCy indexing." "$START_READABLE" "$ELAPSED_READABLE" - -rm -rf "./LondonArchive" - -END_SECONDS_ALL=$(date +%s) -ELAPSED_SECONDS_ALL=$((END_SECONDS_ALL - START_SECONDS_ALL)) -ELAPSED_READABLE_ALL=$(printf "%02d:%02d:%02d" \ - $((ELAPSED_SECONDS_ALL/3600)) \ - $(( (ELAPSED_SECONDS_ALL%3600)/60 )) \ - $((ELAPSED_SECONDS_ALL%60))) -push_webhook "Finished scrape job" "$CRON_LOG_INDEX" "Archive is now fully updated." "$START_READABLE_ALL" "$ELAPSED_READABLE_ALL"