From b123a324d9ac5faad3167052710e71edaadf1bfe Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Fri, 10 Apr 2020 17:20:40 -0700
Subject: pipeline scripts

---
 bin/run_pipeline.sh | 61 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 61 insertions(+)
 create mode 100755 bin/run_pipeline.sh

(limited to 'bin/run_pipeline.sh')

diff --git a/bin/run_pipeline.sh b/bin/run_pipeline.sh
new file mode 100755
index 0000000..7ace3b3
--- /dev/null
+++ b/bin/run_pipeline.sh
@@ -0,0 +1,61 @@
+#!/usr/bin/env bash
+
+
+# invoke this script like:
+#
+#   pipenv shell
+#   CORDDATE=2020-04-03 ./bin/run_pipeline.sh
+
+set -e -u -o pipefail
+
+export TODAY="`TZ=UTC date --iso-8601=date`"
+
+echo "TODAY=$TODAY"
+echo "CORDDATE=$CORDDATE"
+
+# versbose:
+set -x
+
+mkdir -p metadata fulltext_web
+
+if [ ! -f metadata/cord19.$CORDDATE.csv ]; then
+    wget https://archive.org/download/s2-cord19-dataset/cord19.$CORDDATE.csv -O metadata/cord19.$CORDDATE.csv --no-clobber || true
+fi
+
+if [ ! -f metadata/cord19.$CORDDATE.json ]; then
+    ./covid19_tool.py parse-cord19 metadata/cord19.$CORDDATE.csv > metadata/cord19.$CORDDATE.json
+fi
+
+if [ ! -f metadata/cord19.$CORDDATE.enrich.json ]; then
+    cat metadata/cord19.$CORDDATE.json | parallel -j10 --linebuffer --round-robin --pipe ./covid19_tool.py enrich-fatcat - | pv -l > metadata/cord19.$CORDDATE.enrich.json
+fi
+
+if [ ! -f metadata/cord19.$CORDDATE.missing.json ]; then
+    cat metadata/cord19.$CORDDATE.enrich.json | jq 'select(.release_id == null) | .cord19_paper' -c > metadata/cord19.$CORDDATE.missing.json
+fi
+
+if [ ! -f metadata/fatcat_hits.$TODAY.enrich.json ]; then
+    ./covid19_tool.py query-fatcat | pv -l > metadata/fatcat_hits.$TODAY.enrich.json
+fi
+
+if [ ! -f metadata/combined.$TODAY.enrich.json ]; then
+    cat metadata/fatcat_hits.$TODAY.enrich.json metadata/cord19.$CORDDATE.enrich.json | ./covid19_tool.py dedupe | pv -l > metadata/combined.$TODAY.enrich.json
+fi
+
+if [ ! -f fatcat_web_$TODAY.log ]; then
+    cat metadata/combined.$TODAY.enrich.json | jq .fatcat_release -c | parallel -j20 --linebuffer --round-robin --pipe ./bin/deliver_file2disk.py --disk-dir fulltext_web - | pv -l > fatcat_web_$TODAY.log
+fi
+
+if [ ! -f metadata/derivatives.$TODAY.stamp ]; then
+    ./bin/make_dir_derivatives.sh fulltext_web
+    touch metadata/derivatives.$TODAY.stamp
+fi
+
+if [ ! -f metadata/combined.$TODAY.fulltext.json ]; then
+    ./covid19_tool.py enrich-derivatives metadata/combined.$TODAY.enrich.json --base-dir fulltext_web/ | pv -l > metadata/combined.$TODAY.fulltext.json
+fi
+
+echo "## Fulltext Inclusion Counts"
+cat metadata/combined.$TODAY.fulltext.json | jq .fulltext_status -r | sort | uniq -c | sort -nr
+
+./covid19_tool.py transform-es metadata/combined.$TODAY.fulltext.json | pv -l | esbulk -verbose -size 1000 -id fatcat_ident -w 8 -index covid19_fatcat_fulltext -type release
-- 
cgit v1.2.3