OA: fix URLS and precommit setup

dags/open_access/open_access.py

@@ -14,8 +14,11 @@ def oa_dag():
     @task()
     def fetch_data_task(query, **kwargs):
         year = kwargs["params"].get("year")
-        base_query = f"(affiliation:CERN+or+595:'For+annual+report')+\
-        and+year:{year}+not+980:ConferencePaper+not+980:BookChapter"
+        base_query = (
+            r"(affiliation:CERN+or+595:'For+annual+report')"
+            + rf"and+year:{year}+not+980:ConferencePaper+"
+            + r"not+980:BookChapter"
+        )
         type_of_query = [*query][0]
         url = utils.get_url(f"{base_query}+{query[type_of_query]}")
         data = utils.get_data(url)

dags/open_access/utils.py

@@ -5,16 +5,17 @@
 
 
 def get_url(query, current_collection="Published+Articles"):
-    url = f"https://cds.cern.ch/search?ln=en&cc={current_collection}&p={query}\
-    &action_search=Search&op1=a&m1=a&p1=&f1=&c=\
-    Published+Articles&c=&sf=&so=d&rm=&rg=100&sc=0&of=xm"
+    url = (
+        rf"https://cds.cern.ch/search?ln=en&cc={current_collection}&p={query}"
+        + r"&action_search=Search&op1=a&m1=a&p1=&f1=&c="
+        + r"Published+Articles&c=&sf=&so=d&rm=&rg=100&sc=0&of=xm"
+    )
     return url
 
 
 def get_total_results_count(data):
     TOTAL_RECORDS_COUNT = re.compile(
-        r"Search-Engine-Total-Number-Of-Results\
-        :\s(\d*)\s"
+        r"Search-Engine-Total-Number-Of-Results" + r":\s(\d*)\s"
     )
     comment_line = data.split("\n")[1]
     match = TOTAL_RECORDS_COUNT.search(comment_line)
@@ -25,13 +26,17 @@ def get_total_results_count(data):
         return 0
 
 
-closed_access_query = "not+540__a:'CC+BY'+not+540__a:'CC-BY'+\
-            not+540__f:Bronze+not+540__3:preprint"
-bronze_access_query = "540__f:'Bronze'"
-green_access_query = "not+540__a:'CC+BY'+not+540__a:'CC-BY'+not+540__a:\
-        'arXiv+nonexclusive-distrib'+not+540__f:'Bronze'"
-gold_access_query = "540__3:'publication'+and+\
-        (540__a:'CC-BY'+OR++540__a:'CC+BY')"
+closed_access_query = (
+    r"not+540__a:'CC+BY'+not+540__a:'CC-BY'+" + r"not+540__f:Bronze+not+540__3:preprint"
+)
+bronze_access_query = r"540__f:'Bronze'"
+green_access_query = (
+    r"not+540__a:'CC+BY'+not+540__a:'CC-BY'+not+540__a:"
+    + r"'arXiv+nonexclusive-distrib'+not+540__f:'Bronze'"
+)
+gold_access_query = (
+    r"540__3:'publication'+and+" + r"(540__a:'CC-BY'+OR++540__a:'CC+BY')"
+)
 
 
 @backoff.on_exception(

setup.cfg

@@ -0,0 +1,8 @@
+[flake8]
+max-line-length = 120
+
+[pycodestyle]
+max-line-length = 120
+
+[isort]
+line_length = 88