From 6cad89bfbd343338137da604abc76a0e3fb23d8b Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Fri, 11 Jun 2021 15:13:39 -0700 Subject: sitemap: new access URL format --- extra/sitemap/access_urls_query.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'extra/sitemap/access_urls_query.sh') diff --git a/extra/sitemap/access_urls_query.sh b/extra/sitemap/access_urls_query.sh index eb34564..7f1c3ef 100755 --- a/extra/sitemap/access_urls_query.sh +++ b/extra/sitemap/access_urls_query.sh @@ -10,7 +10,7 @@ SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )" # NOTE: filter out 12-digit access URLs (vs. 14-digit) fatcat-cli search scholar 'doc_type:work (fulltext.access_type:ia_file OR fulltext.access_type:wayback) (NOT biblio.arxiv_id:*) (NOT biblio.pmcid:*) (NOT biblio.publisher_type:big5) (year:<1926 OR tags:*)' --index-json --limit 0 \ | pv -l \ - | jq '[.fulltext.access_type, .fulltext.access_url] | @tsv' -r \ + | jq '[.work_ident, .fulltext.access_type, .fulltext.access_url] | @tsv' -r \ | rg -v '^null' \ | rg -v 'web.archive.org/web/\d{12}/' \ | $SCRIPT_DIR/transform_access_url.py \ -- cgit v1.2.3