feat: add repository-level CODEOWNERS coverage chart (hiero-hackers#265)

* feat: add repository-level CODEOWNERS coverage chart

Signed-off-by: aashu2006 <akshatp439@gmail.com>

* fix lint by ruff check

Signed-off-by: aashu2006 <akshatp439@gmail.com>

* fix: aggregate duplicate repository records in CODEOWNERS summary

Signed-off-by: aashu2006 <akshatp439@gmail.com>

---------

Signed-off-by: aashu2006 <akshatp439@gmail.com>

Author: aashu2006

outputs/charts/org/hiero-ledger/org_codeowner_by_repo.png

@@ -30,6 +30,32 @@ def prepare_repo_level_codeowner_summary(codeowners: list[CodeOwnersRecord]) ->
     )
 
 
+def prepare_stacked_codeowner_summary(codeowners: list[CodeOwnersRecord]) -> pd.DataFrame:
+    """Aggregates CODEOWNERS presence per repository for stacked bar chart visualization.
+
+    If duplicate repository entries exist, a presence-wins policy is applied (i.e.
+    if any entry for a repository is marked as Present, the repository is resolved
+    as Present).
+    """
+    if not codeowners:
+        return pd.DataFrame(columns=["repo", "Present", "Missing"])
+
+    repo_status: dict[str, bool] = {}
+    for r in codeowners:
+        repo_status[r.repo] = repo_status.get(r.repo, False) or r.status
+
+    rows = [
+        {
+            "repo": repo,
+            "Present": 1 if has_owners else 0,
+            "Missing": 0 if has_owners else 1,
+        }
+        for repo, has_owners in repo_status.items()
+    ]
+
+    return pd.DataFrame(rows)
+
+
 def runner_records_to_dataframe(runners: list[RunnerRecord]) -> pd.DataFrame:
     """Converts a list of RunnerRecords into DataFrame."""
     return records_to_dataframe(

outputs/charts/org/hiero-ledger/org_codeowner_summary.png

@@ -158,3 +158,9 @@
     "text_light": "#FFFFFF",
     "tick": "#64748B",
 }
+
+# Compliance / Codeowners status colors.
+CODEOWNER_STATUS_COLORS = {
+    "Present": "#2A9D8F",  # teal
+    "Missing": "#E76F51",  # coral
+}

outputs/charts/org/hiero-ledger/org_runner_chart.png

@@ -211,6 +211,7 @@
                     "description": "CODEOWNERS coverage and the CI runners configured across repos.",
                     "files": [
                         ("Code-owner coverage", "org_codeowner_summary.png"),
+                        ("Code-owner coverage by repo", "org_codeowner_by_repo.png"),
                         ("Runners", "org_runner_chart.png"),
                     ],
                 },
@@ -671,6 +672,9 @@
         "Code-Review, Branch-Protection), so you can see which practices contribute."
     ),
     "org_codeowner_summary.png": ("How many repositories have a CODEOWNERS file (Present) versus none (Missing)."),
+    "org_codeowner_by_repo.png": (
+        "CODEOWNERS file presence per repository. Teal indicates Present, while coral indicates Missing."
+    ),
     "org_runner_chart.png": (
         "GitHub Actions runner usage per repository, stacked by type: self-hosted, standard "
         "(GitHub-hosted), or indeterminate (could not be classified)."
@@ -793,4 +797,8 @@
         "Sum the weighted score per repository per month — each event counts once.",
         "Rank repositories by total, show the busiest 25, and colour each cell by its monthly score.",
     ],
+    "org_codeowner_by_repo.png": [
+        "Check each repository for the presence of a CODEOWNERS file in standard locations (root, .github/, or docs/).",
+        "Represent presence/absence per repository as a stacked bar chart showing 100% compliance status (Present vs. Missing).",
+    ],
 }

src/hiero_analytics/analysis/codeowner_workflow_analysis.py

@@ -5,9 +5,11 @@
 from hiero_analytics.analysis.codeowner_workflow_analysis import (
     prepare_org_codeowners_summary,
     prepare_repo_level_codeowner_summary,
+    prepare_stacked_codeowner_summary,
     prepare_stacked_runner_summary,
     runner_records_to_dataframe,
 )
+from hiero_analytics.config.charts import CODEOWNER_STATUS_COLORS
 from hiero_analytics.config.logging_config import setup_logging
 from hiero_analytics.config.paths import ORG, ensure_org_dirs
 from hiero_analytics.data_sources.cache import load_records_cache, save_records_cache
@@ -145,7 +147,7 @@ def main() -> None:
             y_col="count",
             title="Organization Wide Codeowners File Summary",
             output_path=org_charts_dir / "org_codeowner_summary.png",
-            colors={"Present": "#2A9D8F", "Missing": "#E76F51"},
+            colors=CODEOWNER_STATUS_COLORS,
         )
 
     codeowners_repo_df = prepare_repo_level_codeowner_summary(codeowners)
@@ -154,6 +156,19 @@ def main() -> None:
 
     generate_codeowners_markdown_report(records=codeowners, output_path=org_data_dir / "codeowners_report.md")
 
+    codeowners_stacked_df = prepare_stacked_codeowner_summary(codeowners)
+    if not codeowners_stacked_df.empty:
+        plot_stacked_bar(
+            df=codeowners_stacked_df,
+            x_col="repo",
+            stack_cols=["Present", "Missing"],
+            labels=["Present", "Missing"],
+            title="Repository Wide Codeowners Status Breakdown",
+            output_path=org_charts_dir / "org_codeowner_by_repo.png",
+            colors=CODEOWNER_STATUS_COLORS,
+            annotate_totals=False,
+        )
+
     runners = get_workflow_for_repos(client, ORG, repos)
 
     runners_df = runner_records_to_dataframe(runners)

src/hiero_analytics/config/charts.py

@@ -5,6 +5,7 @@
 from hiero_analytics.analysis.codeowner_workflow_analysis import (
     prepare_org_codeowners_summary,
     prepare_repo_level_codeowner_summary,
+    prepare_stacked_codeowner_summary,
     prepare_stacked_runner_summary,
     runner_records_to_dataframe,
 )
@@ -115,3 +116,55 @@ def test_prepare_stacked_runner_summary_columns_exist():
     assert df.iloc[0]["Standard"] == 1
     assert df.iloc[0]["Self-Hosted"] == 0
     assert df.iloc[0]["Indeterminate"] == 0
+
+
+def test_prepare_stacked_codeowner_summary_empty():
+    """Test empty list returns empty DataFrame with correct columns."""
+    df = prepare_stacked_codeowner_summary([])
+    assert isinstance(df, pd.DataFrame)
+    assert list(df.columns) == ["repo", "Present", "Missing"]
+    assert df.empty
+
+
+def test_prepare_stacked_codeowner_summary_aggregation():
+    """Verify correct mapping of boolean status to Present/Missing columns."""
+    records = [
+        CodeOwnersRecord(repo="repo-a", status=True),
+        CodeOwnersRecord(repo="repo-b", status=False),
+    ]
+    df = prepare_stacked_codeowner_summary(records)
+
+    assert len(df) == 2
+
+    # repo-a check (status=True -> Present=1, Missing=0)
+    row_a = df[df["repo"] == "repo-a"].iloc[0]
+    assert row_a["Present"] == 1
+    assert row_a["Missing"] == 0
+
+    # repo-b check (status=False -> Present=0, Missing=1)
+    row_b = df[df["repo"] == "repo-b"].iloc[0]
+    assert row_b["Present"] == 0
+    assert row_b["Missing"] == 1
+
+
+def test_prepare_stacked_codeowner_summary_duplicates():
+    """Verify that duplicate repo entries are aggregated with a presence-wins policy."""
+    records = [
+        CodeOwnersRecord(repo="repo-a", status=False),
+        CodeOwnersRecord(repo="repo-a", status=True),
+        CodeOwnersRecord(repo="repo-b", status=False),
+        CodeOwnersRecord(repo="repo-b", status=False),
+    ]
+    df = prepare_stacked_codeowner_summary(records)
+
+    assert len(df) == 2
+
+    # repo-a check (status False and True -> resolved as Present=1, Missing=0)
+    row_a = df[df["repo"] == "repo-a"].iloc[0]
+    assert row_a["Present"] == 1
+    assert row_a["Missing"] == 0
+
+    # repo-b check (status False and False -> resolved as Present=0, Missing=1)
+    row_b = df[df["repo"] == "repo-b"].iloc[0]
+    assert row_b["Present"] == 0
+    assert row_b["Missing"] == 1

src/hiero_analytics/dashboard_spec.py

@@ -0,0 +1,137 @@
+"""Integration tests for the codeowner and runner analytics pipeline."""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+from unittest.mock import MagicMock
+
+import matplotlib
+import pytest
+
+matplotlib.use("Agg")
+
+import hiero_analytics.run_codeowner_and_runner as runner
+from hiero_analytics.data_sources.models import (
+    CodeOwnersRecord,
+    RepositoryRecord,
+    RunnerRecord,
+)
+
+
+@pytest.fixture
+def mock_github_client():
+    """Mock GitHubClient."""
+    return MagicMock()
+
+
+@pytest.fixture
+def synthetic_repos():
+    """Synthetic repository data."""
+    return [
+        RepositoryRecord(
+            full_name="hiero-ledger/repo1",
+            name="repo1",
+            owner="hiero-ledger",
+        ),
+        RepositoryRecord(
+            full_name="hiero-ledger/repo2",
+            name="repo2",
+            owner="hiero-ledger",
+        ),
+    ]
+
+
+@pytest.fixture
+def synthetic_codeowners():
+    """Synthetic CODEOWNERS data."""
+    return [
+        CodeOwnersRecord(repo="repo1", status=True),
+        CodeOwnersRecord(repo="repo2", status=False),
+    ]
+
+
+@pytest.fixture
+def synthetic_runners():
+    """Synthetic runner data."""
+    return [
+        RunnerRecord(
+            repo="repo1",
+            workflow_file="ci.yml",
+            job_name="test-job-1",
+            runner="self-hosted",
+            is_self_hosted=True,
+        ),
+        RunnerRecord(
+            repo="repo2",
+            workflow_file="ci.yml",
+            job_name="test-job-2",
+            runner="ubuntu-latest",
+            is_self_hosted=False,
+        ),
+    ]
+
+
+def test_main_creates_output_files(
+    tmp_path: Path,
+    monkeypatch: pytest.MonkeyPatch,
+    mock_github_client,
+    synthetic_repos,
+    synthetic_codeowners,
+    synthetic_runners,
+):
+    """Running main() should create expected chart and data files."""
+    # Redirect paths to tmp_path
+    monkeypatch.setattr(
+        "hiero_analytics.run_codeowner_and_runner.ensure_org_dirs",
+        lambda _org: (tmp_path / "data", tmp_path / "charts"),
+    )
+
+    # Mock fetch functions
+    monkeypatch.setattr(
+        "hiero_analytics.run_codeowner_and_runner.fetch_org_repos",
+        lambda _client, _org: synthetic_repos,
+    )
+    monkeypatch.setattr(
+        "hiero_analytics.run_codeowner_and_runner.get_codeowners_for_repos",
+        lambda _client, _org, _repos: synthetic_codeowners,
+    )
+    monkeypatch.setattr(
+        "hiero_analytics.run_codeowner_and_runner.get_workflow_for_repos",
+        lambda _client, _org, _repos: synthetic_runners,
+    )
+
+    # Mock GitHubClient initialization
+    monkeypatch.setattr(
+        "hiero_analytics.run_codeowner_and_runner.GitHubClient",
+        lambda: mock_github_client,
+    )
+
+    # Run the main function
+    runner.main()
+
+    # Assert expected output files exist
+    charts_dir = tmp_path / "charts"
+    data_dir = tmp_path / "data"
+
+    expected_charts = [
+        "org_codeowner_summary.png",
+        "org_codeowner_by_repo.png",
+        "org_runner_chart.png",
+    ]
+    expected_data = [
+        "repo_wise_codeowner_status.csv",
+        "org_runner_status.csv",
+        "codeowners_report.md",
+        "runner_report.md",
+    ]
+
+    for chart_file in expected_charts:
+        chart_path = charts_dir / chart_file
+        assert chart_path.exists(), f"Chart {chart_file} not created"
+        assert os.path.getsize(chart_path) > 0, f"Chart {chart_file} is empty"
+
+    for data_file in expected_data:
+        data_path = data_dir / data_file
+        assert data_path.exists(), f"Data file {data_file} not created"
+        assert os.path.getsize(data_path) > 0, f"Data file {data_file} is empty"