diff options
| author | Martin Czygan <martin.czygan@gmail.com> | 2020-11-21 00:34:56 +0100 |
|---|---|---|
| committer | Martin Czygan <martin.czygan@gmail.com> | 2020-11-21 00:34:56 +0100 |
| commit | e3a8b7e17b7a34d148aca4e3296afca81b845339 (patch) | |
| tree | 6c1c271a1fd49268da9cd69c6595b2d179edba27 /fuzzycat | |
| parent | a5b9e0fee9f97f0205dfc96d35fc4b2cc823554b (diff) | |
| download | fuzzycat-e3a8b7e17b7a34d148aca4e3296afca81b845339.tar.gz fuzzycat-e3a8b7e17b7a34d148aca4e3296afca81b845339.zip | |
wip: handle empty lists
Diffstat (limited to 'fuzzycat')
| -rw-r--r-- | fuzzycat/verify.py | 15 |
1 files changed, 9 insertions, 6 deletions
diff --git a/fuzzycat/verify.py b/fuzzycat/verify.py index 609d617..5e41e1c 100644 --- a/fuzzycat/verify.py +++ b/fuzzycat/verify.py @@ -190,7 +190,7 @@ def compare(a, b): # TODO: figshare versions, "xxx.v1" FIGSHARE_PREFIX = "10.6084" - if a.get("doi").startswith(FIGSHARE_PREFIX + "/") and b.get("doi").startswith(FIGSHARE_PREFIX + + if a.get("doi") and b.get("doi") and a.get("doi").startswith(FIGSHARE_PREFIX + "/") and b.get("doi").startswith(FIGSHARE_PREFIX + "/"): a_doi_v_stripped = re.sub(r"[.]v[0-9]+$", "", a.get("doi")) b_doi_v_stripped = re.sub(r"[.]v[0-9]+$", "", a.get("doi")) @@ -217,7 +217,7 @@ def compare(a, b): continue if rel.get("relatedIdentifierType") != "DOI": continue - doi = reg.get("relatedIdentifier") + doi = rel.get("relatedIdentifier") if not doi: continue dois.add(doi) @@ -345,10 +345,11 @@ def compare(a, b): for _, g in itertools.groupby(scores, key=lambda s: s.a): sorted_scores = sorted(g, key=lambda s: s.value, reverse=True) if len(sorted_scores) > 0: - top_scores.append(sorted_scores[0]) - avg_score = sum(top_scores) / len(top_scores) - if avg_score > 0.5: - return (Status.STRONG, OK.TOKENIZED_AUTHORS) + top_scores.append(sorted_scores[0].value) + if len(top_scores) > 0: + avg_score = sum(top_scores) / len(top_scores) + if avg_score > 0.5: + return (Status.STRONG, OK.TOKENIZED_AUTHORS) # TODO: This misses spelling differences, e.g. # https://fatcat.wiki/release/7nbcgsohrrak5cuyk6dnit6ega and @@ -370,6 +371,8 @@ def jaccard(a, b): """ Jaccard of sets a and b. """ + if len(a | b) == 0: + return 0 return len(a & b) / len(a | b) |