data collection script

src/helper_scripts/repo_collection.py

@@ -0,0 +1,36 @@
+import os 
+import csv 
+import datetime
+
+from pyrepohistory import repo_lifecycle
+
+file_location = "/data/users/mgaughan/mw-repo-lifecycles/commit_data/"
+repo_location = "/data/users/mgaughan/mw-repo-lifecycles/repo_artifacts/"
+
+cst = datetime.timezone(datetime.timedelta(hours=-6))
+repos = {
+        "core" : {
+            "url": "https://gerrit.wikimedia.org/r/mediawiki/core",
+            "from_date": datetime.datetime(2022, 9, 10, 00, 00, 00, tzinfo=cst),
+            "to_date": datetime.datetime(2022, 10, 10, 00, 00, 00, tzinfo=cst)
+            },
+        "parsoid" :{
+            "url": "https://gerrit.wikimedia.org/r/mediawiki/services/parsoid",
+            "from_date": datetime.datetime(2024, 6, 10, 00, 00, 00, tzinfo=cst),
+            "to_date": datetime.datetime(2024, 10, 10, 00, 00, 00, tzinfo=cst)
+            },
+        "visualeditor": {
+            "url": "https://gerrit.wikimedia.org/r/VisualEditor/VisualEditor",
+            "from_date": datetime.datetime(2013, 5, 1, 00, 00, 00, tzinfo=cst),
+            "to_date": datetime.datetime(2013, 10, 10, 00, 00, 00, tzinfo=cst)
+            }
+        }
+
+for repo in repos.keys():
+    print(repos[repo])
+    print(repos[repo]['url'])
+    print(repos[repo]["from_date"])
+    print(repos[repo]["to_date"])
+    repo_info = repos[repo]
+    repo_lifecycle(repo_info['url'], repo_location + repo + "/tmp", repo_info["from_date"], repo_info["to_date"], to_save=True)
+