diff options
| author | Bryan Newbold <bnewbold@robocracy.org> | 2019-05-23 16:34:08 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@robocracy.org> | 2019-05-23 16:34:08 -0700 |
| commit | 537db73cce0d0a71fafeeb20423491c0cb75ea7a (patch) | |
| tree | 5d83b11ae8aedf7a4f54df5ae38f66e13051dc51 /python | |
| parent | 104f1fd882216dc79bd7515e0b66834a779c1bf7 (diff) | |
| parent | 9fbab0defca5016c0eb5b851ff73e03efac4fac8 (diff) | |
| download | fatcat-537db73cce0d0a71fafeeb20423491c0cb75ea7a.tar.gz fatcat-537db73cce0d0a71fafeeb20423491c0cb75ea7a.zip | |
Merge branch 'bnewbold-import-tweaks'
Diffstat (limited to 'python')
| -rw-r--r-- | python/fatcat_tools/importers/arxiv.py | 5 | ||||
| -rw-r--r-- | python/fatcat_tools/transforms/elasticsearch.py | 1 | ||||
| -rw-r--r-- | python/tests/import_arxiv.py | 3 |
3 files changed, 8 insertions, 1 deletions
diff --git a/python/fatcat_tools/importers/arxiv.py b/python/fatcat_tools/importers/arxiv.py index 182d0471..71b2d134 100644 --- a/python/fatcat_tools/importers/arxiv.py +++ b/python/fatcat_tools/importers/arxiv.py @@ -198,6 +198,7 @@ class ArxivRawImporter(EntityImporter): # group-title # arxiv: comments, categories, etc extra_arxiv['base_id'] = base_id + extra['superceded'] = True extra['arxiv'] = extra_arxiv versions = [] @@ -223,11 +224,13 @@ class ArxivRawImporter(EntityImporter): license_slug=license_slug, abstracts=abstracts, contribs=contribs, - extra=extra, + extra=extra.copy(), )) # TODO: assert that versions are actually in order? assert versions + versions[-1].extra.pop('superceded') + # only apply DOI to most recent version (HACK) if doi: versions[-1].ext_ids.doi = doi diff --git a/python/fatcat_tools/transforms/elasticsearch.py b/python/fatcat_tools/transforms/elasticsearch.py index f287fe10..8589d364 100644 --- a/python/fatcat_tools/transforms/elasticsearch.py +++ b/python/fatcat_tools/transforms/elasticsearch.py @@ -292,6 +292,7 @@ def container_to_elasticsearch(entity, force_bool=True): if extra.get('ia'): if extra['ia'].get('sim'): any_ia_sim = True + t['is_superceded'] = bool(extra.get('superceded')) t['in_doaj'] = bool(in_doaj) t['in_road'] = bool(in_road) diff --git a/python/tests/import_arxiv.py b/python/tests/import_arxiv.py index bbad8fa6..1e649616 100644 --- a/python/tests/import_arxiv.py +++ b/python/tests/import_arxiv.py @@ -71,6 +71,7 @@ def test_arxiv_xml_parse(arxiv_importer): assert len(r1.contribs) == 4 assert r1.extra['arxiv']['categories'] == ['cond-mat.stat-mech', 'physics.bio-ph', 'physics.data-an'] assert r1.extra['arxiv']['base_id'] == '1810.09584' + assert r1.extra['superceded'] == True assert r1.contribs[0].raw_name == "Raphael Chetrite" assert r1.contribs[0].role == "author" @@ -92,6 +93,8 @@ def test_arxiv_xml_parse(arxiv_importer): assert r1.extra['arxiv']['comments'] == "7 pages, 2 figures" assert r1.extra['arxiv']['categories'] == ["cond-mat.stat-mech", "physics.bio-ph", "physics.data-an"] + assert not r2.extra.get('superceded') + r2.extra['superceded'] = True assert r1.extra == r2.extra assert not r1.refs |