aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2022-01-20 15:03:18 -0800
committerBryan Newbold <bnewbold@archive.org>2022-01-20 15:03:18 -0800
commit8b4388eed8b0424d8d83b191a059037ee32d73b6 (patch)
tree48d0ec82b9bcaa96e83dac2112602aa79de0465e
parent3fe6a191cf002520181f48f47a997b681603dd27 (diff)
downloadfatcat-scholar-8b4388eed8b0424d8d83b191a059037ee32d73b6.tar.gz
fatcat-scholar-8b4388eed8b0424d8d83b191a059037ee32d73b6.zip
sitemap generation: move public domain wall; remove 'is_biblio' filter
-rwxr-xr-xextra/sitemap/access_urls_query.sh3
-rwxr-xr-xextra/sitemap/work_urls_query.sh3
2 files changed, 2 insertions, 4 deletions
diff --git a/extra/sitemap/access_urls_query.sh b/extra/sitemap/access_urls_query.sh
index 6238a9d..e2439d0 100755
--- a/extra/sitemap/access_urls_query.sh
+++ b/extra/sitemap/access_urls_query.sh
@@ -8,8 +8,7 @@ SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )"
# query for specific works; about 8.6 million circa 2021-04-29
# 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) OR year:<1926 OR tags:oa)' --index-json --limit 0 \
- | jq -c 'select(.biblio.release_ident == .fulltext.release_ident)' \
+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) OR year:<=1926 OR tags:oa)' --index-json --limit 0 \
| jq -r '[.work_ident, .fulltext.access_type, .fulltext.access_url] | @tsv' \
| rg -v '^null' \
| rg -v 'web.archive.org/web/\d{12}/' \
diff --git a/extra/sitemap/work_urls_query.sh b/extra/sitemap/work_urls_query.sh
index 813596b..432bfa6 100755
--- a/extra/sitemap/work_urls_query.sh
+++ b/extra/sitemap/work_urls_query.sh
@@ -7,8 +7,7 @@ set -o pipefail # fail if part of a '|' command fails
# query for specific works; about 8.6 million circa 2021-04-29
# 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) OR year:<1926 OR tags:oa)' --index-json --limit 0 \
- | jq -c 'select(.biblio.release_ident == .fulltext.release_ident)' \
+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) OR year:<=1926 OR tags:oa)' --index-json --limit 0 \
| rg -v '^null' \
| rg -v 'web.archive.org/web/\d{12}/' \
| jq .key -r \