From: Benjamin Mako Hill Date: Wed, 1 Apr 2020 20:15:11 +0000 (-0500) Subject: add two small shellscripts for automation X-Git-Url: https://code.communitydata.science/covid19.git/commitdiff_plain/141871eda6754b95f1617e354fe1d9728d0fd8e1?hp=04e00f363bec0c5aa349e6c85f9a78d6ee70c3fd add two small shellscripts for automation - Added two bash scripts usable as cronjobs to automate the production of revisions and view data. These commands automate the process of running code and copying material --- diff --git a/cron-wikipedia_revisions.sh b/cron-wikipedia_revisions.sh new file mode 100644 index 0000000..71ee752 --- /dev/null +++ b/cron-wikipedia_revisions.sh @@ -0,0 +1,17 @@ +#!/bin/bash -x + +TZ="UTC" +date_string=$(date +%Y%m%d) + +./wikipedia/scripts/wikiproject_scraper.py 2> >(tee wikipedia/logs/enwp-wikiproject_scraper-${date_string}.log) + +revs_log="enwp-revisions-${date_string}.log" +./wikipedia/scripts/fetch_enwiki_revisions.py 2> >(tee wikipedia/logs/${rev_log}) +cp wikipedia/logs/${rev_log} /var/www/covid19/wikipedia/logs/ + +revs_tsv="digobs_covid19-wikipedia-enwiki_revisions-${date_string}.tsv" +cp wikipedia/data/${revs_tsv} /var/www/covid19/wikipedia + +revs_json="digobs_covid19-wikipedia-enwiki_revisions-${date_string}.json" +xz wikipedia/data/${revs_json} +cp wikipedia/data/${revs_json}.xz /var/www/covid19/wikipedia diff --git a/cron-wikipedia_views.sh b/cron-wikipedia_views.sh new file mode 100644 index 0000000..ec3d4ec --- /dev/null +++ b/cron-wikipedia_views.sh @@ -0,0 +1,16 @@ +#!/bin/bash -x + +TZ="UTC" +date_string=$(date +%Y%m%d) + +./wikipedia/scripts/wikiproject_scraper.py 2> >(tee wikipedia/logs/enwp-wikiproject_scraper-${date_string}.log) + +# get the list of files +view_log="enwp-daily_views-${date_string}.log" +./wikipedia/scripts/fetch_enwiki_daily_views.py 2> >(tee wikipedia/logs/${view_log}) +cp wikipedia/logs/${view_log} /var/www/covid19/wikipedia/logs/${view_log} +cp wikipedia/data/digobs_covid19-wikipedia-enwiki_dailyviews-${date_string}.tsv /var/www/covid19/wikipedia/ + +# xz wikipedia/data/digobs_covid19-wikipedia-enwiki_dailyviews-${date_string}.json +cp wikipedia/data/digobs_covid19-wikipedia-enwiki_dailyviews-${date_string}.json /var/www/covid19/wikipedia/ +