Merge pull request #1 from chaoss/dev

Updating my dev branch from the main dev branch at chaoss/augur

schema/generate/05-seed_data.sql

@@ -117,13 +117,6 @@ INSERT INTO "augur_data"."chaoss_metric_status" VALUES (80, 'experimental', 'lib
 INSERT INTO "augur_data"."chaoss_metric_status" VALUES (81, 'experimental', 'librariesio', 'metric', 'implemented', 'implemented', 'f', '/api/unstable/<owner>/<repo>/dependents', NULL, 'Dependents', 'experimental', '"dependents"', 'Insight Worker', '0.0.1', 'librariesio', '2019-06-20 22:51:25', NULL);
 COMMIT;
 
-
-
-BEGIN;
-INSERT INTO "augur_data"."contributor_affiliations"("ca_id", "ca_domain", "ca_affiliation", "ca_start_date", "ca_active", "ca_last_used", "tool_source", "tool_version", "data_source", "data_collection_date") VALUES (25152, 'goggins.com', 'goggins', '2016-03-03', 1, '2019-10-29 12:22:25', NULL, NULL, NULL, '2019-10-29 17:22:30');
-COMMIT; 
-
-
 -- SPDX Schema Required Metadata 
 
 

schema/generate/15-schema_update_17.sql

@@ -3,8 +3,8 @@ ALTER TABLE "augur_data"."repo"
   ALTER COLUMN "forked_from" TYPE varchar USING "forked_from"::varchar;
 
 ALTER TABLE "augur_data"."repo" 
-  ADD COLUMN "repo_archived" int4,
-  ADD COLUMN "repo_archived_date_collected" timestamptz(0),
+  ADD COLUMN IF NOT EXISTS "repo_archived" int4,
+  ADD COLUMN IF NOT EXISTS "repo_archived_date_collected" timestamptz(0),
   ALTER COLUMN "forked_from" TYPE varchar USING "forked_from"::varchar;
 
 update "augur_operations"."augur_settings" set value = 17 where setting = 'augur_data_version'; 

workers/pull_request_worker/pull_request_worker.py

@@ -108,6 +108,9 @@ def find_root_of_subject(data, key_subject):
 
                     if 'errors' in data:
                         self.logger.info("Error!: {}".format(data['errors']))
+                        if j['errors'][0]['type'] == 'NOT_FOUND':
+                            self.logger.warning("Github repo was not found or does not exist for endpoint: {}\n".format(url))
+                            break
                         if data['errors'][0]['type'] == 'RATE_LIMITED':
                             self.update_gh_rate_limit(response)
                             num_attempts -= 1
@@ -133,7 +136,7 @@ def find_root_of_subject(data, key_subject):
 
                 if not success:
                     self.logger.info('GraphQL query failed: {}'.format(query))
-                    continue
+                    break
 
                 before_parameters.update({
                     data_subject: ', before: \"{}\"'.format(page_info['startCursor'])