update repo scraper

.gitignore

@@ -166,6 +166,8 @@ prediction*.json
 **/service_account_key.json
 corruptmnist_v1/
 corruptmnist_v2/
+tools/repo_stats/report.py
+tools/repo_stats/README.md
 
 # vscode
 .vscode/

tools/repo_stats/leaderboard.py

@@ -71,7 +71,7 @@ def main() -> None:
     """Main function for the leaderboard."""
     download_data("repo_stats.json")
     dataframe = load_data("repo_stats.json")
-    dataframe["num_warnings"] = dataframe["num_warnings"].apply(lambda x: 27 - x if pd.notnull(x) else x)
+    dataframe["num_warnings"] = dataframe["num_warnings"].apply(lambda x: 31 - x if pd.notnull(x) else x)
     dataframe["activity_matrix"] = dataframe["activity_matrix"].apply(
         lambda x: activity_to_image(x) if x is not None else x
     )

tools/repo_stats/models.py

@@ -102,7 +102,9 @@ def check_answers(self) -> int | None:
                 stdin=PIPE,
             )
             output = p.stderr.read()
-            return len(output.decode("utf-8").split("\n")[:-1:2])
+            output = output.decode("utf-8").split("\n")
+            lines = [line for line in output if "WARNING" in line]
+            return len(lines) if len(lines) else None
         return None
 
 

tools/repo_stats/scraper.py

@@ -58,13 +58,20 @@ def load_data(file_name: str) -> list[GroupInfo]:
     return content
 
 
-def create_activity_matrix(commits: list, max_delta: int = 5, normalize: bool = True) -> list[list[int]]:
+def create_activity_matrix(
+    commits: list,
+    max_delta: int = 5,
+    min_delta: int = 1,
+) -> list[list[int]]:
     """Creates an activity matrix from the commits."""
     commit_times = [datetime.datetime.fromisoformat(commit["commit"]["committer"]["date"][:-1]) for commit in commits]
     commit_times.sort()
 
     start_time = commit_times[0]
-    end_time = min(start_time + datetime.timedelta(weeks=max_delta), commit_times[-1])
+    end_time = max(
+        start_time + datetime.timedelta(weeks=min_delta),
+        min(start_time + datetime.timedelta(weeks=max_delta), commit_times[-1]),
+    )
 
     num_days = (end_time - start_time).days + 1  # include last day
 
@@ -124,7 +131,7 @@ def main():
                             contributor.commits_pr += 1
                 commits += pr_commits
 
-            activity_matrix = create_activity_matrix(commits, max_delta=3)
+            activity_matrix = create_activity_matrix(commits, max_delta=3, min_delta=1)
 
             average_commit_length = sum([len(c) for c in commit_messages]) / len(commit_messages)