implemented timecheck for gsql

gh_gsql_req.py

@@ -5,12 +5,14 @@ import os
 
 key = os.environ.get('KKEXKEY')
 
-def main(vcs):
+def main(vcs, early_cutoff):
+    gsql_dict = {}
     vcs_list = vcs[0].split('/')
     repo_name = '"' + vcs_list[-1] + '"'
     repo_owner = '"' + vcs_list[-2] + '"'
-    returned_content = get_discussion_gql(repo_owner, repo_name)
-    return returned_content
+    gsql_dict["original_returned_content"] = get_discussion_gql(repo_owner, repo_name)
+    gsql_dict["time_cleaned_comm"] = within_time(gsql_dict["original_returned_content"].content, early_cutoff)
+    return gsql_dict
 
 def get_discussion_gql(repo_owner, repo_name):
     url = "https://api.github.com/graphql"
@@ -61,6 +63,16 @@ def get_discussion_gql(repo_owner, repo_name):
     print(r.content)
     return r
 
+def within_time(comment_content, early_cutoff):
+    list_of_comments = json.loads(comment_content)["data"]["repository"]["discussions"]["edges"]
+    valid_comments = []
+    for comment in list_of_comments:
+        if dt.datetime.fromisoformat(comment['node']['createdAt'][:-1]) < early_cutoff:
+            break
+        else:
+            valid_comments.append(comment)
+    return valid_comments
+    
 
 if __name__ == "__main__":
     get_discussion_gql()

main.py

@@ -9,7 +9,7 @@ import gh_gsql_req as ghs
 
 def main():
     # we should discuss whether we're using the 93 day window that seems to be widely used or if we want a longer window
-    early_cutoff = dt.datetime(2023,6, 17)
+    early_cutoff = dt.datetime(2023,10, 11)
     print("Earliest date examined: " + str(early_cutoff))
     #placeholder for now
     manifest = '../kaylea_dissertation/lifecycle/package_metadata/jupyter-notebook_manifest.yaml'
@@ -35,9 +35,8 @@ def main():
             new_formality = compute_formality_score(new_mmt, gha_obj['milestone_count'], perceval_obj['age_of_project'])
             print(new_formality)
             # testing out beneath:
-            # TODO: incorporate early cutoff to this, probably in a cleaning function
-            ghs_obj = ghs.main(vcs_path)
-            print(ghs_obj)
+            ghs_obj = ghs.main(vcs_path, early_cutoff)
+            print(ghs_obj["time_cleaned_comm"])
         except yaml.YAMLOError as err:
             print(err)