diff options
| author | Bryan Newbold <bnewbold@robocracy.org> | 2020-02-07 14:38:13 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@robocracy.org> | 2020-02-07 14:38:13 -0800 |
| commit | 83387210e6775751e5eb690a7d8b56fe99dbe380 (patch) | |
| tree | 7fff4ed06981731594f4282f704911dff59c4090 | |
| parent | 3655bbe6c539fdeccfbfaa19b6fc93a4859e0ca7 (diff) | |
| download | fatcat-83387210e6775751e5eb690a7d8b56fe99dbe380.tar.gz fatcat-83387210e6775751e5eb690a7d8b56fe99dbe380.zip | |
ES files: don't remove archive.org domains/hosts
| -rw-r--r-- | python/fatcat_tools/transforms/elasticsearch.py | 5 |
1 files changed, 0 insertions, 5 deletions
diff --git a/python/fatcat_tools/transforms/elasticsearch.py b/python/fatcat_tools/transforms/elasticsearch.py index f8bc05fb..e00d7830 100644 --- a/python/fatcat_tools/transforms/elasticsearch.py +++ b/python/fatcat_tools/transforms/elasticsearch.py @@ -509,9 +509,4 @@ def file_to_elasticsearch(entity): t['in_ia'] = bool('archive.org' in t['domains']) t['in_ia_petabox'] = bool('archive.org' in t['hosts']) - # ok, but actually remove archive.org hosts, because they make other - # aggregations hard and are a waste of storage - t['hosts'] = [h for h in t['hosts'] if h not in ('archive.org', 'web.archive.org')] - t['domains'] = [h for h in t['domains'] if h not in ('archive.org')] - return t |