monitor pages from dsaez's wikidata crawler

[covid19.git] / cron-wikipedia_revisions.sh

diff --git a/cron-wikipedia_revisions.sh b/cron-wikipedia_revisions.sh
index 71ee7527d904921601e62109f1a1126aa3da2be5..555fd46197838b327b9b2a5a2acf6a78f7bf088b 100644 (file)
--- a/cron-wikipedia_revisions.sh
+++ b/cron-wikipedia_revisions.sh
@@ -3,15 +3,28 @@
 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/
+./wikipedia/scripts/wikiproject_scraper.py 2> >(tee wikipedia/logs/${revs_log})
+
+wd_log="wd-page-crawler-${date_string}.log"
+python3 ./real-time-wiki-covid-tracker/PageCrawler.py -a "./wikipedia/resources/enwp_wikiproject_covid19_articles.txt" 2> >(tee wikipedia/logs/${wd_log})
+
+./wikipedia/scripts/fetch_revisions.py 2> >(tee -a wikipedia/logs/${revs_log})
+mv wikipedia/logs/${revs_log} /var/www/covid19/wikipedia/logs/
+
+python3 ./wikipedia/scripts/copy_revisions_data.py ${date_string}
+
+cd wikipedia/data
+xz */${date_string}/*revisions*.json
+
+find */${date_string}/*revisions*.xz | while read line; do
+    mkdir -p /var/www/covid9/wikipedia/$line
+    mv $line /var/www/covid19/wikipedia/$line
+done
 
-revs_tsv="digobs_covid19-wikipedia-enwiki_revisions-${date_string}.tsv"
-cp wikipedia/data/${revs_tsv} /var/www/covid19/wikipedia
+find */${date_string}/*revisions*.tsv | while read line; do
+    mkdir -p /var/www/covid19/wikipedia/$line
+    mv $line /var/www/covid19/wikipedia/$line
+done
 
-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
+cd ../..