fmt (black): fatcat_tools/

3 files changed, 45 insertions, 39 deletions

diff --git a/python/fatcat_tools/cleanups/__init__.py b/python/fatcat_tools/cleanups/__init__.py
index 587c7b9b..0aeec977 100644
--- a/python/fatcat_tools/cleanups/__init__.py
+++ b/python/fatcat_tools/cleanups/__init__.py
@@ -1,3 +1,2 @@
-
 from .common import EntityCleaner
 from .files import FileCleaner
diff --git a/python/fatcat_tools/cleanups/common.py b/python/fatcat_tools/cleanups/common.py
index d0fcc761..26ca7bd6 100644
--- a/python/fatcat_tools/cleanups/common.py
+++ b/python/fatcat_tools/cleanups/common.py
@@ -1,4 +1,3 @@
-
 import copy
 import json
 import subprocess
@@ -30,16 +29,19 @@ class EntityCleaner:
 
     def __init__(self, api, entity_type, **kwargs):
 
-        eg_extra = kwargs.get('editgroup_extra', dict())
-        eg_extra['git_rev'] = eg_extra.get('git_rev',
-            subprocess.check_output(["git", "describe", "--always"]).strip()).decode('utf-8')
-        eg_extra['agent'] = eg_extra.get('agent', 'fatcat_tools.EntityCleaner')
+        eg_extra = kwargs.get("editgroup_extra", dict())
+        eg_extra["git_rev"] = eg_extra.get(
+            "git_rev", subprocess.check_output(["git", "describe", "--always"]).strip()
+        ).decode("utf-8")
+        eg_extra["agent"] = eg_extra.get("agent", "fatcat_tools.EntityCleaner")
 
         self.api = api
         self.entity_type = entity_type
-        self.dry_run_mode = kwargs.get('dry_run_mode', True)
-        self.edit_batch_size = kwargs.get('edit_batch_size', 50)
-        self.editgroup_description = kwargs.get('editgroup_description', "Generic Entity Cleaner Bot")
+        self.dry_run_mode = kwargs.get("dry_run_mode", True)
+        self.edit_batch_size = kwargs.get("edit_batch_size", 50)
+        self.editgroup_description = kwargs.get(
+            "editgroup_description", "Generic Entity Cleaner Bot"
+        )
         self.editgroup_extra = eg_extra
         self.reset()
         self.ac = ApiClient()
@@ -48,7 +50,7 @@ class EntityCleaner:
             print("Running in dry-run mode!")
 
     def reset(self):
-        self.counts = Counter({'lines': 0, 'cleaned': 0, 'updated': 0})
+        self.counts = Counter({"lines": 0, "cleaned": 0, "updated": 0})
         self._edit_count = 0
         self._editgroup_id = None
         self._entity_queue = []
@@ -63,23 +65,23 @@ class EntityCleaner:
 
         Returns nothing.
         """
-        self.counts['lines'] += 1
-        if (not record):
-            self.counts['skip-null'] += 1
+        self.counts["lines"] += 1
+        if not record:
+            self.counts["skip-null"] += 1
             return
 
         entity = entity_from_dict(record, self.entity_type, api_client=self.ac)
 
-        if entity.state != 'active':
-            self.counts['skip-inactive'] += 1
+        if entity.state != "active":
+            self.counts["skip-inactive"] += 1
             return
 
         cleaned = self.clean_entity(copy.deepcopy(entity))
         if entity == cleaned:
-            self.counts['skip-clean'] += 1
+            self.counts["skip-clean"] += 1
             return
         else:
-            self.counts['cleaned'] += 1
+            self.counts["cleaned"] += 1
 
         if self.dry_run_mode:
             entity_dict = entity_to_dict(entity, api_client=self.ac)
@@ -87,11 +89,13 @@ class EntityCleaner:
             return
 
         if entity.ident in self._idents_inflight:
-            raise ValueError("Entity already part of in-process update: {}".format(entity.ident))
+            raise ValueError(
+                "Entity already part of in-process update: {}".format(entity.ident)
+            )
 
         updated = self.try_update(cleaned)
         if updated:
-            self.counts['updated'] += updated
+            self.counts["updated"] += updated
             self._edit_count += updated
             self._idents_inflight.append(entity.ident)
 
@@ -132,9 +136,8 @@ class EntityCleaner:
 
         if not self._editgroup_id:
             eg = self.api.create_editgroup(
-                Editgroup(
-                    description=self.editgroup_description,
-                    extra=self.editgroup_extra))
+                Editgroup(description=self.editgroup_description, extra=self.editgroup_extra)
+            )
             self._editgroup_id = eg.editgroup_id
 
         return self._editgroup_id
diff --git a/python/fatcat_tools/cleanups/files.py b/python/fatcat_tools/cleanups/files.py
index 0d275ba6..d378a91f 100644
--- a/python/fatcat_tools/cleanups/files.py
+++ b/python/fatcat_tools/cleanups/files.py
@@ -1,4 +1,3 @@
-
 from fatcat_openapi_client.models import FileEntity
 from fatcat_openapi_client.rest import ApiException
 
@@ -12,14 +11,19 @@ class FileCleaner(EntityCleaner):
 
     def __init__(self, api, **kwargs):
 
-        eg_desc = kwargs.pop('editgroup_description', None) or "Automated cleanup of file entities (eg, remove bad URLs)"
-        eg_extra = kwargs.pop('editgroup_extra', dict())
-        eg_extra['agent'] = eg_extra.get('agent', 'fatcat_tools.FileCleaner')
-        super().__init__(api,
+        eg_desc = (
+            kwargs.pop("editgroup_description", None)
+            or "Automated cleanup of file entities (eg, remove bad URLs)"
+        )
+        eg_extra = kwargs.pop("editgroup_extra", dict())
+        eg_extra["agent"] = eg_extra.get("agent", "fatcat_tools.FileCleaner")
+        super().__init__(
+            api,
             entity_type=FileEntity,
             editgroup_description=eg_desc,
             editgroup_extra=eg_extra,
-            **kwargs)
+            **kwargs
+        )
 
     def clean_entity(self, entity):
         """
@@ -27,24 +31,24 @@ class FileCleaner(EntityCleaner):
         """
 
         # URL has ://web.archive.org/web/None/ link => delete URL
-        entity.urls = [u for u in entity.urls if '://web.archive.org/web/None/' not in u.url]
+        entity.urls = [u for u in entity.urls if "://web.archive.org/web/None/" not in u.url]
 
         # URL has ://archive.org/ link with rel=repository => rel=archive
         for u in entity.urls:
-            if '://archive.org/' in u.url and u.rel == 'repository':
-                u.rel = 'archive'
+            if "://archive.org/" in u.url and u.rel == "repository":
+                u.rel = "archive"
 
         # URL has short wayback date ("2017") and another url with that as prefix => delete URL
         stub_wayback_urls = []
         full_wayback_urls = []
         for u in entity.urls:
-            if '://web.archive.org/web/' in u.url:
-                if len(u.url.split('/')[4]) <= 8:
+            if "://web.archive.org/web/" in u.url:
+                if len(u.url.split("/")[4]) <= 8:
                     stub_wayback_urls.append(u.url)
                 else:
-                    full_wayback_urls.append('/'.join(u.url.split('/')[5:]))
+                    full_wayback_urls.append("/".join(u.url.split("/")[5:]))
         for stub in stub_wayback_urls:
-            target = '/'.join(stub.split('/')[5:])
+            target = "/".join(stub.split("/")[5:])
             if target in full_wayback_urls:
                 entity.urls = [u for u in entity.urls if u.url != stub]
 
@@ -57,14 +61,14 @@ class FileCleaner(EntityCleaner):
         except ApiException as err:
             if err.status != 404:
                 raise err
-            self.counts['skip-not-found'] += 1
+            self.counts["skip-not-found"] += 1
             return 0
 
-        if existing.state != 'active':
-            self.counts['skip-existing-inactive'] += 1
+        if existing.state != "active":
+            self.counts["skip-existing-inactive"] += 1
             return 0
         if existing.revision != entity.revision:
-            self.counts['skip-revision'] += 1
+            self.counts["skip-revision"] += 1
             return 0
 
         self.api.update_file(self.get_editgroup_id(), entity.ident, entity)