diff options
-rw-r--r-- | fuzzycat/utils.py | 4 | ||||
-rw-r--r-- | fuzzycat/verify.py | 15 | ||||
-rw-r--r-- | tests/data/release/iitldffmnncijgnf6ujb6zmdfu | 30 | ||||
-rw-r--r-- | tests/data/release/ppnzru2opnhxlai7pcmo7phe4i | 30 | ||||
-rw-r--r-- | tests/data/verify.csv | 1 |
5 files changed, 71 insertions, 9 deletions
diff --git a/fuzzycat/utils.py b/fuzzycat/utils.py index 4d1325d..1cac668 100644 --- a/fuzzycat/utils.py +++ b/fuzzycat/utils.py @@ -1,14 +1,13 @@ import io import itertools -import re import string +import re printable_no_punct = string.digits + string.ascii_letters + string.whitespace # More correct: https://www.johndcook.com/blog/2016/02/04/regular-expression-to-match-a-chemical-element/ CHEM_FORMULA = re.compile(r"([A-Z]{1,2}[0-9]{1,2})+") - def slugify_string(s: str) -> str: """ Keeps ascii chars and single whitespace only. @@ -90,3 +89,4 @@ def contains_chemical_formula(s): for token in s.split(): if CHEM_FORMULA.search(token): return True + diff --git a/fuzzycat/verify.py b/fuzzycat/verify.py index ab26603..2bb4adb 100644 --- a/fuzzycat/verify.py +++ b/fuzzycat/verify.py @@ -328,6 +328,14 @@ TITLE_FRAGMENT_BLACKLIST = set([ "untersuchung einzelner abdominaler regionen und organe", ]) +CONTAINER_NAME_BLACKLIST = set([ + "crossref listing of deleted dois", +]) + +PUBLISHER_BLACKLIST = set([ + "test accounts", +]) + # There titles appear too often, so ignore them for now. TITLE_BLACKLIST = set([ "", @@ -3526,10 +3534,3 @@ TITLE_BLACKLIST = set([ "週刊ダイヤモンド = diamond weekly 別冊", ]) -CONTAINER_NAME_BLACKLIST = set([ - "crossref listing of deleted dois", -]) - -PUBLISHER_BLACKLIST = set([ - "test accounts", -]) diff --git a/tests/data/release/iitldffmnncijgnf6ujb6zmdfu b/tests/data/release/iitldffmnncijgnf6ujb6zmdfu new file mode 100644 index 0000000..f2c1b84 --- /dev/null +++ b/tests/data/release/iitldffmnncijgnf6ujb6zmdfu @@ -0,0 +1,30 @@ +{ + "abstracts": [], + "container_id": "qea6koy5crhvfgkjwgnwojjkrm", + "contribs": [], + "ext_ids": { + "doi": "10.1017/s000497270001981x" + }, + "extra": { + "crossref": { + "alternative-id": [ + "S000497270001981X" + ], + "type": "journal-article" + } + }, + "ident": "iitldffmnncijgnf6ujb6zmdfu", + "issue": "03", + "language": "en", + "pages": "f1", + "publisher": "Cambridge University Press (CUP)", + "refs": [], + "release_stage": "published", + "release_type": "article-journal", + "release_year": 2001, + "revision": "39087ec0-0c3f-4353-9656-961c5f11be42", + "state": "active", + "title": "BAZ volume 64 issue 3 Cover and Front matter", + "volume": "64", + "work_id": "7bjigtuhhvh3nk3rvckd6w7emy" +} diff --git a/tests/data/release/ppnzru2opnhxlai7pcmo7phe4i b/tests/data/release/ppnzru2opnhxlai7pcmo7phe4i new file mode 100644 index 0000000..9f68d90 --- /dev/null +++ b/tests/data/release/ppnzru2opnhxlai7pcmo7phe4i @@ -0,0 +1,30 @@ +{ + "abstracts": [], + "container_id": "qea6koy5crhvfgkjwgnwojjkrm", + "contribs": [], + "ext_ids": { + "doi": "10.1017/s0004972700005785" + }, + "extra": { + "crossref": { + "alternative-id": [ + "S0004972700005785" + ], + "type": "journal-article" + } + }, + "ident": "ppnzru2opnhxlai7pcmo7phe4i", + "issue": "03", + "language": "en", + "pages": "f1", + "publisher": "Cambridge University Press (CUP)", + "refs": [], + "release_stage": "published", + "release_type": "article-journal", + "release_year": 1982, + "revision": "e6f9dedb-b6c0-4bf4-b3f4-d3427aa30c62", + "state": "active", + "title": "BAZ volume 26 issue 3 Cover and Front matter", + "volume": "26", + "work_id": "4wwtp45b3nchpo2bs7pifwsb44" +} diff --git a/tests/data/verify.csv b/tests/data/verify.csv index 46df18e..732d401 100644 --- a/tests/data/verify.csv +++ b/tests/data/verify.csv @@ -83,3 +83,4 @@ jdtngtiz3bdqboypujoni2x3ry,byh7xr5qhjca3bw53ivdotck3e,Status.EXACT, 5lk635o65nc2tnkus3pkf2ggeq,hqrvhbvocvaabg6nr5p43tl3uq,TODO, 5lk635o65nc2tnkus3pkf2ggeq,zfwf3tefajc6zdxa47vgilm7wm,TODO, hqrvhbvocvaabg6nr5p43tl3uq,zfwf3tefajc6zdxa47vgilm7wm,TODO, +ppnzru2opnhxlai7pcmo7phe4i,iitldffmnncijgnf6ujb6zmdfu,Status.DIFFERENT,Miss.NUM_DIFF |