feat(prs): stacked bar (#7)

notebook/dashboard.ipynb

@@ -159,17 +159,18 @@
     "        \"archived\": repo.archived,\n",
     "        \"fork\": repo.fork,\n",
     "        \"forks\": repo.forks_count,\n",
-    "        \"issues\": len(open_issues),\n",
+    "        \"issues\": open_issues,\n",
     "        \"topics\": repo.get_topics(),\n",
     "        \"languages\": repo.get_languages(),\n",
     "        \"license\": license_name,\n",
-    "        \"prs\": len(open_prs),\n",
+    "        \"prs\": open_prs,\n",
     "        \"created_at\": repo.created_at,\n",
     "        \"updated_at\": repo.updated_at,\n",
     "        \"coverage\": coverage,\n",
     "        \"readthedocs\": readthedocs_project,\n",
     "        \"has_readthedocs\": readthedocs_project is not None,\n",
     "        \"has_readme\": readme_file is not None,\n",
+    "        \"_repo\": repo,\n",
     "    })\n",
     "\n",
     "df = pd.DataFrame(repo_data)\n",
@@ -186,10 +187,11 @@
     "print(f'Total Repositories: {len(repo_data)}')\n",
     "print(f'Archived Repositories: {df[\"archived\"].sum()}')\n",
     "print(f'Forked Repositories: {df[\"fork\"].sum()}')\n",
-    "print(f'Total Open Issues: {df[\"issues\"].sum()}')\n",
-    "print(f'Total Open PRs: {df[\"prs\"].sum()}')\n",
-    "print(f'Open issues in active repositories: {df_repos[\"issues\"].sum()}')\n",
-    "print(f'Open PRs in active repositories: {df_repos[\"prs\"].sum()}')"
+    "\n",
+    "print(f'Total Open Issues: {df[\"issues\"].apply(len).sum()}')\n",
+    "print(f'Total Open PRs: {df[\"prs\"].apply(len).sum()}')\n",
+    "print(f'Open issues in active repositories: {df_repos[\"issues\"].apply(len).sum()}')\n",
+    "print(f'Open PRs in active repositories: {df_repos[\"prs\"].apply(len).sum()}')"
    ],
    "id": "initial_id",
    "outputs": [],
@@ -293,17 +295,26 @@
    },
    "source": [
     "# Open Issues\n",
+    "\n",
+    "# Calculate the count of issues for each repository\n",
+    "df_repos.loc[:, 'issue_count'] = df_repos['issues'].apply(len)\n",
+    "\n",
+    "# Sort by issue count in descending order\n",
     "df_issues = df_repos.sort_values(\n",
-    "    by='issues',\n",
+    "    by='issue_count',\n",
     "    ascending=False,\n",
     ")\n",
-    "df_issues['log_issues'] = np.log1p(df_issues['issues'])\n",
+    "\n",
+    "# Calculate the log of issue count\n",
+    "df_issues.loc[:, 'log_issues'] = np.log1p(df_issues['issue_count'])\n",
+    "\n",
+    "# Visualize data using a bar chart\n",
     "fig = px.bar(\n",
     "    df_issues,\n",
     "    x='repo',\n",
     "    y='log_issues',\n",
     "    title='Open Issues',\n",
-    "    text='issues',\n",
+    "    text='issue_count',\n",
     ")\n",
     "fig.update_traces(\n",
     "    texttemplate='%{text}',\n",
@@ -334,25 +345,46 @@
    },
    "source": [
     "# Open PRs\n",
-    "df_prs = df_repos.sort_values(\n",
-    "    by='prs',\n",
-    "    ascending=False,\n",
-    ")\n",
-    "df_prs['log_prs'] = np.log1p(df_prs['prs'])\n",
+    "pr_data = []\n",
+    "for repo in df_repos.to_dict('records'):\n",
+    "    draft_prs = 0\n",
+    "    non_draft_prs = 0\n",
+    "    dependabot_prs = 0\n",
+    "\n",
+    "    for pr in repo['prs']:\n",
+    "        pr_details = repo['_repo'].get_pull(pr.number)\n",
+    "        if pr_details.user.login == 'dependabot[bot]' or pr_details.user.login == 'renovate[bot]':\n",
+    "            dependabot_prs += 1\n",
+    "        elif pr_details.draft:\n",
+    "            draft_prs += 1\n",
+    "        else:\n",
+    "            non_draft_prs += 1\n",
+    "\n",
+    "    pr_data.append({\n",
+    "        \"repo\": repo['repo'],\n",
+    "        \"Draft\": draft_prs,\n",
+    "        \"Ready for review\": non_draft_prs,\n",
+    "        \"Dependency\": dependabot_prs,\n",
+    "    })\n",
+    "\n",
+    "df_prs = pd.DataFrame(pr_data)\n",
+    "df_prs['total_prs'] = df_prs[['Draft', 'Ready for review', 'Dependency']].sum(axis=1)\n",
+    "\n",
+    "# Sort by total PRs in descending order\n",
+    "df_prs = df_prs.sort_values(by='total_prs', ascending=False)\n",
+    "\n",
+    "# Visualize data using a stacked bar chart\n",
     "fig = px.bar(\n",
     "    df_prs,\n",
     "    x='repo',\n",
-    "    y='log_prs',\n",
-    "    title='Open PRs',\n",
-    "    text='prs',\n",
-    ")\n",
-    "fig.update_traces(\n",
-    "    texttemplate='%{text}',\n",
-    "    textposition='inside',\n",
+    "    y=['Draft', 'Ready for review', 'Dependency'],\n",
+    "    title='Open Pull Requests',\n",
+    "    labels={'value': 'Count', 'variable': 'PR Type'},\n",
+    "    barmode='stack'\n",
     ")\n",
     "fig.update_layout(\n",
-    "    yaxis_title=None,\n",
-    "    yaxis_showticklabels=False,\n",
+    "    yaxis_title='Count of PRs',\n",
+    "    xaxis_title='Repository',\n",
     ")\n",
     "fig.show()"
    ],