diff options
| author | Martin Czygan <martin.czygan@gmail.com> | 2020-12-02 00:58:20 +0100 |
|---|---|---|
| committer | Martin Czygan <martin.czygan@gmail.com> | 2020-12-02 00:58:20 +0100 |
| commit | ba68fa4d91fbcd1dda3363b78bc24ca64ca2546b (patch) | |
| tree | b49b02f5c148d6bd7045580a7f9919e7ec364935 /fuzzycat | |
| parent | 34c441b7828694492904108ac307d4aa640ac64d (diff) | |
| download | fuzzycat-ba68fa4d91fbcd1dda3363b78bc24ca64ca2546b.tar.gz fuzzycat-ba68fa4d91fbcd1dda3363b78bc24ca64ca2546b.zip | |
add case
Diffstat (limited to 'fuzzycat')
| -rw-r--r-- | fuzzycat/common.py | 5 | ||||
| -rw-r--r-- | fuzzycat/verify.py | 7 |
2 files changed, 10 insertions, 2 deletions
diff --git a/fuzzycat/common.py b/fuzzycat/common.py index 34508b7..5cef684 100644 --- a/fuzzycat/common.py +++ b/fuzzycat/common.py @@ -17,7 +17,10 @@ class OK(str, Enum): Reason for assuming we have a match. """ ARXIV_VERSION = 'ok.arxiv_version' + CUSTOM_BSI_UNDATED = 'ok.custom_bsi_undated' + CUSTOM_IEEE_ARXIV = 'ok.custom_ieee_arxiv' DATACITE_RELATED_ID = 'ok.datacite_related_id' + DATACITE_VERSION = 'ok.datacite_version' DOI = 'ok.doi' DUMMY = 'ok.dummy' FIGSHARE_VERSION = 'ok.figshare_version' @@ -25,8 +28,6 @@ class OK(str, Enum): SLUG_TITLE_AUTHOR_MATCH = 'ok.slug_title_author_match' TITLE_AUTHOR_MATCH = 'ok.title_author_match' TOKENIZED_AUTHORS = 'ok.tokenized_authors' - CUSTOM_IEEE_ARXIV = 'ok.custom_ieee_arxiv' - CUSTOM_BSI_UNDATED = 'ok.custom_bsi_undated' class Miss(str, Enum): diff --git a/fuzzycat/verify.py b/fuzzycat/verify.py index e688d49..5977f8e 100644 --- a/fuzzycat/verify.py +++ b/fuzzycat/verify.py @@ -160,6 +160,13 @@ def compare(a, b): return (Status.AMBIGUOUS, Miss.BLACKLISTED_FRAGMENT) try: + if a_title and a_title == b_title and glom(a, "extra.datacite.metadataVersion") != glom( + b, "extra.datacite.metadataVersion"): + return (Status.EXACT, OK.DATACITE_VERSION) + except PathAccessError: + pass + + try: a_doi = glom(a, "ext_ids.doi") b_doi = glom(b, "ext_ids.doi") if a_doi.startswith("10.14288/") and b_doi.startswith("10.14288/") and a_doi != b_doi: |