From 81087a5e3b83bde710f44dc08219d0c7af1ec63d Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Fri, 30 Apr 2021 11:55:23 -0700 Subject: updates to sitemap generation --- extra/sitemap/README.md | 3 ++- extra/sitemap/work_urls_query.sh | 5 ++--- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/extra/sitemap/README.md b/extra/sitemap/README.md index 242378a..1e1938a 100644 --- a/extra/sitemap/README.md +++ b/extra/sitemap/README.md @@ -8,7 +8,8 @@ installed. Run these commands on a production machine. export DATE=`date --iso-8601` /srv/fatcat_scholar/src/extra/sitemap/work_urls_query.sh $DATE rm *.txt.gz - /srv/fatcat/src/extra/sitemap/generate_sitemap_indices.py + gzip sitemap-*.txt + /srv/fatcat_scholar/src/extra/sitemap/generate_sitemap_indices.py ## Background diff --git a/extra/sitemap/work_urls_query.sh b/extra/sitemap/work_urls_query.sh index c02eb74..2971f47 100755 --- a/extra/sitemap/work_urls_query.sh +++ b/extra/sitemap/work_urls_query.sh @@ -9,9 +9,8 @@ set -o pipefail # fail if part of a '|' command fails # eg, 2020-08-19 DATE="$1" -# query for specific works - -fatcat-cli search scholar "doc_type:work (fulltext.access_type:ia_file OR fulltext.access_type:wayback) (year:<1925 OR publisher_type:longtail OR publisher_type:oa)" --index-json --limit 0 \ +# query for specific works; about 8.6 million circa 2021-04-29 +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 .key -r \ | tr '_' '/' \ -- cgit v1.2.3