diff options
24 files changed, 1 insertions, 202 deletions
diff --git a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/0.yaml b/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/0.yaml deleted file mode 100644 index 2df8d9a..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/0.yaml +++ /dev/null @@ -1,13 +0,0 @@ -title: titles are case insensitive -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Michael Adams" - } - ], - "title": "digital libraries", - "ext_ids": {} - } -expected: 2 diff --git a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/1.yaml b/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/1.yaml deleted file mode 100644 index 1070408..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/1.yaml +++ /dev/null @@ -1,13 +0,0 @@ -title: another vanilla query -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Poul-Henning Kamp" - } - ], - "title": "The hyperdimensional tar pit", - "ext_ids": {} - } -expected: 2 diff --git a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/2.yaml b/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/2.yaml deleted file mode 100644 index 882e746..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/2.yaml +++ /dev/null @@ -1,16 +0,0 @@ -title: order of contribs does not matter -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Maurice Florence" - }, - { - "raw_name": "Tuomo Tiisala" - } - ], - "title": "Foucault", - "ext_ids": {} - } -expected: 1 diff --git a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/3.yaml b/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/3.yaml deleted file mode 100644 index 0a2ad12..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/3.yaml +++ /dev/null @@ -1,16 +0,0 @@ -title: order of contribs does not matter -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Tuomo Tiisala" - }, - { - "raw_name": "Maurice Florence" - } - ], - "title": "Foucault", - "ext_ids": {} - } -expected: 1 diff --git a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/4.yaml b/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/4.yaml deleted file mode 100644 index 36ea0fe..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_exact_contrib/4.yaml +++ /dev/null @@ -1,16 +0,0 @@ -title: short version of name should not work -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Tuomo Tiisala" - }, - { - "raw_name": "M. Florence" - } - ], - "title": "Foucault", - "ext_ids": {} - } -expected: 0 diff --git a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/0.yaml b/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/0.yaml deleted file mode 100644 index 07230e8..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/0.yaml +++ /dev/null @@ -1,14 +0,0 @@ -title: titles are case insensitive -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Michael Adams" - } - ], - "title": "digital libraries", - "ext_ids": {} - } -jaccard_index_threshold: 1.0 -expected: 2 diff --git a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/1.yaml b/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/1.yaml deleted file mode 100644 index 62e9586..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/1.yaml +++ /dev/null @@ -1,14 +0,0 @@ -title: another vanilla query -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Poul-Henning Kamp" - } - ], - "title": "The hyperdimensional tar pit", - "ext_ids": {} - } -jaccard_index_threshold: 1.0 -expected: 2 diff --git a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/2.yaml b/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/2.yaml deleted file mode 100644 index b89e825..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/2.yaml +++ /dev/null @@ -1,17 +0,0 @@ -title: order of contribs does not matter -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Maurice Florence" - }, - { - "raw_name": "Tuomo Tiisala" - } - ], - "title": "Foucault", - "ext_ids": {} - } -jaccard_index_threshold: 1.0 -expected: 1 diff --git a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/3.yaml b/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/3.yaml deleted file mode 100644 index 3de7262..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/3.yaml +++ /dev/null @@ -1,17 +0,0 @@ -title: order of contribs does not matter -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Tuomo Tiisala" - }, - { - "raw_name": "Maurice Florence" - } - ], - "title": "Foucault", - "ext_ids": {} - } -jaccard_index_threshold: 1.0 -expected: 1 diff --git a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/4.yaml b/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/4.yaml deleted file mode 100644 index 39fb065..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/4.yaml +++ /dev/null @@ -1,17 +0,0 @@ -title: short version of name should not work -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Tuomo Tiisala" - }, - { - "raw_name": "M. Florence" - } - ], - "title": "Foucault", - "ext_ids": {} - } -jaccard_index_threshold: 1.0 -expected: 0 diff --git a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/5.yaml b/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/5.yaml deleted file mode 100644 index fff19fa..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/5.yaml +++ /dev/null @@ -1,17 +0,0 @@ -title: here, Iz Beltagy is missing from author, but still retrieved -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Arman Cohan" - }, - { - "raw_name": "Kyle Lo" - } - ], - "title": "SciBERT: A Pretrained Language Model for Scientific Text", - "ext_ids": {} - } -jaccard_index_threshold: 0.5 -expected: 3 diff --git a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/6.yaml b/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/6.yaml deleted file mode 100644 index d4e0025..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/6.yaml +++ /dev/null @@ -1,14 +0,0 @@ -title: here, 2/3 authors are missing, we fail with jaccard index 0.5 -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Arman Cohan" - } - ], - "title": "SciBERT: A Pretrained Language Model for Scientific Text", - "ext_ids": {} - } -jaccard_index_threshold: 0.5 -expected: 0 diff --git a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/7.yaml b/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/7.yaml deleted file mode 100644 index 23d5a8d..0000000 --- a/tests/files/fuzzy_release_match_release_exact_title_partial_contrib/7.yaml +++ /dev/null @@ -1,17 +0,0 @@ -title: match, despite trailing whitespace -date: 2021-11-08 -input: > - { - "contribs": [ - { - "raw_name": "Arman Cohan" - }, - { - "raw_name": "Kyle Lo" - } - ], - "title": "SciBERT: A Pretrained Language Model for Scientific Text ", - "ext_ids": {} - } -jaccard_index_threshold: 0.5 -expected: 3 diff --git a/tests/files/simple_fuzzy_release_matcher/0.yaml b/tests/files/fuzzy_release_matcher/0.yaml index 71fc992..71fc992 100644 --- a/tests/files/simple_fuzzy_release_matcher/0.yaml +++ b/tests/files/fuzzy_release_matcher/0.yaml diff --git a/tests/files/simple_fuzzy_release_matcher/1.yaml b/tests/files/fuzzy_release_matcher/1.yaml index df6a954..df6a954 100644 --- a/tests/files/simple_fuzzy_release_matcher/1.yaml +++ b/tests/files/fuzzy_release_matcher/1.yaml diff --git a/tests/files/simple_fuzzy_release_matcher/2.yaml b/tests/files/fuzzy_release_matcher/2.yaml index df6a954..df6a954 100644 --- a/tests/files/simple_fuzzy_release_matcher/2.yaml +++ b/tests/files/fuzzy_release_matcher/2.yaml diff --git a/tests/files/simple_fuzzy_release_matcher/3.yaml b/tests/files/fuzzy_release_matcher/3.yaml index 1ab761b..1ab761b 100644 --- a/tests/files/simple_fuzzy_release_matcher/3.yaml +++ b/tests/files/fuzzy_release_matcher/3.yaml diff --git a/tests/files/simple_fuzzy_release_matcher/4.yaml b/tests/files/fuzzy_release_matcher/4.yaml index 9419406..9419406 100644 --- a/tests/files/simple_fuzzy_release_matcher/4.yaml +++ b/tests/files/fuzzy_release_matcher/4.yaml diff --git a/tests/files/simple_fuzzy_release_matcher/5.yaml b/tests/files/fuzzy_release_matcher/5.yaml index 1eb435b..1eb435b 100644 --- a/tests/files/simple_fuzzy_release_matcher/5.yaml +++ b/tests/files/fuzzy_release_matcher/5.yaml diff --git a/tests/files/simple_fuzzy_release_matcher/6.yaml b/tests/files/fuzzy_release_matcher/6.yaml index ae52b23..ae52b23 100644 --- a/tests/files/simple_fuzzy_release_matcher/6.yaml +++ b/tests/files/fuzzy_release_matcher/6.yaml diff --git a/tests/files/simple_fuzzy_release_matcher/7.yaml b/tests/files/fuzzy_release_matcher/7.yaml index 2330f0d..2330f0d 100644 --- a/tests/files/simple_fuzzy_release_matcher/7.yaml +++ b/tests/files/fuzzy_release_matcher/7.yaml diff --git a/tests/files/simple_fuzzy_release_matcher/8.yaml b/tests/files/fuzzy_release_matcher/8.yaml index b43e53a..b43e53a 100644 --- a/tests/files/simple_fuzzy_release_matcher/8.yaml +++ b/tests/files/fuzzy_release_matcher/8.yaml diff --git a/tests/files/simple_fuzzy_release_matcher/9.yaml b/tests/files/fuzzy_release_matcher/9.yaml index b43e53a..b43e53a 100644 --- a/tests/files/simple_fuzzy_release_matcher/9.yaml +++ b/tests/files/fuzzy_release_matcher/9.yaml diff --git a/tests/test_matching.py b/tests/test_matching.py index b9d7fae..eb54751 100644 --- a/tests/test_matching.py +++ b/tests/test_matching.py @@ -55,7 +55,7 @@ def test_simple_fuzzy_release_matcher(es_client, caplog): """ matcher = FuzzyReleaseMatcher(es=es_client) Case = collections.namedtuple("Case", "about input release_year_padding expected") - cases = yaml_to_cases(Case, "tests/files/simple_fuzzy_release_matcher/*.yaml") + cases = yaml_to_cases(Case, "tests/files/fuzzy_release_matcher/*.yaml") for i, c in enumerate(cases): matcher.release_year_padding = c.release_year_padding entity = entity_from_json(c.input, ReleaseEntity) |