|
| 1 | +#!/bin/bash |
| 2 | +twitterUN="${1}" |
| 3 | +outFolder="${2}/${1}" |
| 4 | +interval=${3} |
| 5 | +spacesPath="${outFolder}/spaces" |
| 6 | +COOKIES_PATH="${2}/cookies.txt" |
| 7 | +echo "$outFolder" |
| 8 | + |
| 9 | +#If downloading media was interrupted, reget new media |
| 10 | +if [ -s "${outFolder}/${twitterUN}-tweets.json.new" ]; then |
| 11 | + /app/scripts/getMedia.sh "${outFolder}/${twitterUN}-tweets.json.new" "${outFolder}" |
| 12 | + mv -f "${outFolder}/${twitterUN}-tweets.json.new" "${outFolder}/${twitterUN}-tweets.json" |
| 13 | + if [ -f "${outFolder}/${twitterUN}-tweets.json.bkup" ]; then |
| 14 | + cat "${outFolder}/${twitterUN}-tweets.json.bkup" >> "${outFolder}/${twitterUN}-tweets.json" |
| 15 | + fi |
| 16 | +fi |
| 17 | + |
| 18 | + |
| 19 | +while true |
| 20 | +do |
| 21 | + #Make output folder if not existing |
| 22 | + if ! [ -d "${outFolder}" ]; then |
| 23 | + mkdir -p "${outFolder}" |
| 24 | + fi |
| 25 | + |
| 26 | + #Start download of any available spaces |
| 27 | + if [ ! -f "$COOKIES_PATH" ]; then |
| 28 | + echo "Starting without cookies" |
| 29 | + twspace_dl -U "https://twitter.com/${twitterUN}" --write-url "${spacesPath}/master_urls.txt" -m -p -o "${spacesPath}/[%(creator_screen_name)s]-%(title)s|%(start_date)s" |
| 30 | + else |
| 31 | + twspace_dl -U "https://twitter.com/${twitterUN}" --write-url "${spacesPath}/master_urls.txt" --input-cookie-file "$COOKIES_PATH" -o "${spacesPath}/[%(creator_screen_name)s]-%(title)s|%(start_date)s" |
| 32 | + fi |
| 33 | + |
| 34 | + #Archive tweets |
| 35 | + if ! [ -f "${outFolder}/${twitterUN}-tweets.json" ]; then |
| 36 | + #if no existing archive |
| 37 | + snscrape --jsonl twitter-profile "${twitterUN}" > "${outFolder}/${twitterUN}-tweets.json.new" |
| 38 | + /app/scripts/getMedia.sh "${outFolder}/${twitterUN}-tweets.json.new" "${outFolder}" |
| 39 | + mv -f "${outFolder}/${twitterUN}-tweets.json.new" "${outFolder}/${twitterUN}-tweets.json" |
| 40 | + else |
| 41 | + #If existing, only grab after latest date |
| 42 | + dateOne=$(jq -r '[.date] | @tsv' <<< $(sed -n '1{p;q;}' "${outFolder}/${twitterUN}-tweets.json")) |
| 43 | + dateTwo=$(jq -r '[.date] | @tsv' <<< $(sed -n '2{p;q;}' "${outFolder}/${twitterUN}-tweets.json")) |
| 44 | +# echo "Date one: $dateOne" |
| 45 | +# echo "Date one: $dateTwo" |
| 46 | + dateA=$(date -d "$dateOne" +"%s") |
| 47 | + dateB=$(date -d "$dateTwo" +"%s") |
| 48 | + dateSince=0 |
| 49 | + if [ "$dateA" -ge "$dateB" ]; then |
| 50 | + dateSince=$dateA |
| 51 | + else |
| 52 | + dateSince=$dateB |
| 53 | + fi |
| 54 | + dateSince=$(date -d @$(($dateSince)) +"%Y-%m-%d %H:%M:%S %z") |
| 55 | + snscrape --since "${dateSince}" --jsonl twitter-profile "${twitterUN}" > "${outFolder}/${twitterUN}-tweets.json.new" |
| 56 | + |
| 57 | + if [ -s "${outFolder}/${twitterUN}-tweets.json.new" ]; then |
| 58 | + cp -f "${outFolder}/${twitterUN}-tweets.json" "${outFolder}/${twitterUN}-tweets.json.bkup" |
| 59 | + /app/scripts/getMedia.sh "${outFolder}/${twitterUN}-tweets.json.new" "${outFolder}" |
| 60 | + mv -f "${outFolder}/${twitterUN}-tweets.json.new" "${outFolder}/${twitterUN}-tweets.json" |
| 61 | + cat "${outFolder}/${twitterUN}-tweets.json.bkup" >> "${outFolder}/${twitterUN}-tweets.json" |
| 62 | + fi |
| 63 | + fi |
| 64 | + sleep $(($interval)) |
| 65 | +done |
0 commit comments