From 1657965ce70a51532a3879dc4a7afed4ea076a8a Mon Sep 17 00:00:00 2001 From: Martin Czygan Date: Wed, 9 Dec 2020 23:11:43 +0100 Subject: add a few more dummy cases --- fuzzycat/verify.py | 9 +++++++++ notes/2020_11_testruns.md | 13 +++++++++++++ tests/data/release/eomug3y4afbynhcyzrtcvki55u | 20 ++++++++++++++++++++ tests/data/release/zcu2pugta5fttnqltgersan52q | 20 ++++++++++++++++++++ tests/data/verify.csv | 1 + 5 files changed, 63 insertions(+) create mode 100644 tests/data/release/eomug3y4afbynhcyzrtcvki55u create mode 100644 tests/data/release/zcu2pugta5fttnqltgersan52q diff --git a/fuzzycat/verify.py b/fuzzycat/verify.py index 4daed97..f4a5444 100644 --- a/fuzzycat/verify.py +++ b/fuzzycat/verify.py @@ -332,6 +332,15 @@ def compare(a, b): except PathAccessError: pass + try: + if glom(a, "extra.crossref.type") == "component" and glom(b, "extra.crossref.type"): + a_doi = glom(a, "ext_ids.doi") + b_doi = glom(b, "ext_ids.doi") + if a_doi != b_doi: + return (Status.DIFFERENT, Miss.COMPONENT) + except PathAccessError: + pass + # https://fatcat.wiki/release/knzhequchfcethcyyi3gsp5gry, some title contain newlines a_slug_title = slugify_string(a.get("title", "")).replace("\n", " ") b_slug_title = slugify_string(b.get("title", "")).replace("\n", " ") diff --git a/notes/2020_11_testruns.md b/notes/2020_11_testruns.md index 3ee4340..aa27be4 100644 --- a/notes/2020_11_testruns.md +++ b/notes/2020_11_testruns.md @@ -206,3 +206,16 @@ BSI, one is a subdocument of another. The subdocument has a subtitle. That's mor * [ ] https://fatcat.wiki/release/3d73i34irjedjnhulyhmehnrvq https://fatcat.wiki/release/winjvljpu5al5deaz7v5qi7kta Status.AMBIGUOUS OK.DUMMY * [ ] https://fatcat.wiki/release/jna6aodi2rfq7gwzxhb4s45kqy https://fatcat.wiki/release/3qst2qrdbvdhrdvu22kyqt22xy Status.AMBIGUOUS OK.DUMMY +---- + +* [ ] https://fatcat.wiki/release/eomug3y4afbynhcyzrtcvki55u https://fatcat.wiki/release/zcu2pugta5fttnqltgersan52q Status.AMBIGUOUS OK.DUMMY +* [ ] https://fatcat.wiki/release/5rcu6myqx5ezjjytzpvsauyut4 https://fatcat.wiki/release/zvsffdeufjb5dbchww7ydqdq3a Status.AMBIGUOUS OK.DUMMY +* [ ] https://fatcat.wiki/release/f5ebjc63j5dzpct5hsme5j3ote https://fatcat.wiki/release/zeoquc2f4nbmdbmbcbkmkxmtzi Status.AMBIGUOUS OK.DUMMY +* [ ] https://fatcat.wiki/release/cd5aik2whrd5jlvleyvdq6iwja https://fatcat.wiki/release/kfttghqcsbddvofqd7l4bhtavy Status.AMBIGUOUS OK.DUMMY +* [ ] https://fatcat.wiki/release/hwnqyz7n65eabhlivvkipkytji https://fatcat.wiki/release/cwqujxztefdghhssb7ysxj7b5m Status.AMBIGUOUS OK.DUMMY +* [ ] https://fatcat.wiki/release/yespzqkm2zed7n4vhjpkddap5e https://fatcat.wiki/release/5yixxzyl3vh4xd56lwcraowgty Status.AMBIGUOUS OK.DUMMY +* [ ] https://fatcat.wiki/release/47opwjqugjecjmiqgukahw6p2m https://fatcat.wiki/release/real7tmfxjan7j3fgkilt7fze4 Status.AMBIGUOUS OK.DUMMY +* [ ] https://fatcat.wiki/release/gaf7gjwetrbpzntrp4bt4nxaiy https://fatcat.wiki/release/htsa3mrirndbdjtdangr4mzrdu Status.AMBIGUOUS OK.DUMMY +* [ ] https://fatcat.wiki/release/pobnow7sxfhnxhltgwpru5k7oi https://fatcat.wiki/release/uplqxenmk5axjes6zokml6q73y Status.AMBIGUOUS OK.DUMMY +* [ ] https://fatcat.wiki/release/tm3gaiumkvb3xc7t3i6suna6u4 https://fatcat.wiki/release/r6dj63wh3zcrrolisn6xuacnve Status.AMBIGUOUS OK.DUMMY + diff --git a/tests/data/release/eomug3y4afbynhcyzrtcvki55u b/tests/data/release/eomug3y4afbynhcyzrtcvki55u new file mode 100644 index 0000000..6decd7b --- /dev/null +++ b/tests/data/release/eomug3y4afbynhcyzrtcvki55u @@ -0,0 +1,20 @@ +{ + "abstracts": [], + "contribs": [], + "ext_ids": { + "doi": "10.1021/ic0620891.s007" + }, + "extra": { + "crossref": { + "type": "component" + } + }, + "ident": "eomug3y4afbynhcyzrtcvki55u", + "publisher": "American Chemical Society (ACS)", + "refs": [], + "release_type": "component", + "revision": "f8552fc2-377e-465a-aa87-b966a6e59c1d", + "state": "active", + "title": "Single-Molecule Magnet Behavior in Heterometallic MII-MnIII2-MII Tetramers (MII = Cu, Ni) Containing MnIII Salen-Type Dinuclear Core", + "work_id": "yk5lasti45dlles3e5mfwwurn4" +} diff --git a/tests/data/release/zcu2pugta5fttnqltgersan52q b/tests/data/release/zcu2pugta5fttnqltgersan52q new file mode 100644 index 0000000..e661321 --- /dev/null +++ b/tests/data/release/zcu2pugta5fttnqltgersan52q @@ -0,0 +1,20 @@ +{ + "abstracts": [], + "contribs": [], + "ext_ids": { + "doi": "10.1021/ic0620891.s014" + }, + "extra": { + "crossref": { + "type": "component" + } + }, + "ident": "zcu2pugta5fttnqltgersan52q", + "publisher": "American Chemical Society (ACS)", + "refs": [], + "release_type": "component", + "revision": "a2b8063e-67bc-474b-a396-22d9b499a8d1", + "state": "active", + "title": "Single-Molecule Magnet Behavior in Heterometallic MII-MnIII2-MII Tetramers (MII = Cu, Ni) Containing MnIII Salen-Type Dinuclear Core", + "work_id": "cjzr63koybhvhd5ofsswvwip2m" +} diff --git a/tests/data/verify.csv b/tests/data/verify.csv index d3c8dbe..18a899b 100644 --- a/tests/data/verify.csv +++ b/tests/data/verify.csv @@ -138,3 +138,4 @@ voruupqxhvggfex4zlczcmjxxu,jg72qhdvmncfdfxg5l47hw3uba,Status.AMBIGUOUS,OK.DUMMY arqtphat7fashokettncepu7xe,v6p7xct6kfgwtdbh57zfjqmuua,Status.AMBIGUOUS, zwru5ugcsfcyzeuqlygfw46vwq,b3uhit7b4vhvliocdzwxr7peyy,Status.AMBIGUOUS,Miss.CUSTOM_PREFIX_10_7916 s7a4o5v5gfg4tbzna6poyg7nzy,tcro5wr6brhqnf5wettyiauw34,Status.STRONG,OK.CUSTOM_BSI_SUBDOC +eomug3y4afbynhcyzrtcvki55u,zcu2pugta5fttnqltgersan52q,Status.DIFFERENT,Miss.COMPONENT -- cgit v1.2.3