Refactor update_profile.py to enhance Gitea data fetching and README generation

update_profile.py

@@ -1,34 +1,129 @@
-import datetime
 import os
-import re
+import requests
+import datetime
 
-def generate_stats():
+# --- Configuration ---
-    now = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+GITEA_URL = "https://gitea.nathan-falvey.synology.me"
-    return f"✅ My Gitea instance is alive and well!\n🕒 Last update: {now}."
+USERNAME = "nathan"
+BOT_NAME = "Gitea Action"  # The name used in your git config in the .yaml
+GITEA_TOKEN = os.getenv("GITEA_TOKEN")
+
+def format_bytes(size_bytes):
+    """Converts bytes to a human-readable string."""
+    if size_bytes == 0: return "0 B"
+    for unit in ['B', 'KB', 'MB', 'GB', 'TB']:
+        if size_bytes < 1024.0:
+            return f"{size_bytes:.2f} {unit}"
+        size_bytes /= 1024.0
+    return f"{size_bytes:.2f} PB"
+
+def get_uptime():
+    """Reads Linux system uptime."""
+    try:
+        with open("/proc/uptime", "r") as f:
+            seconds = float(f.readline().split()[0])
+            return str(datetime.timedelta(seconds=int(seconds)))
+    except: return "Running in Container"
+
+def fetch_gitea_data():
+    headers = {"Authorization": f"token {GITEA_TOKEN}"}
+    
+    # 1. System & User Info
+    ver = requests.get(f"{GITEA_URL}/api/v1/version", headers=headers).json().get("version", "N/A")
+    repos = requests.get(f"{GITEA_URL}/api/v1/user/repos?type=owner", headers=headers).json()
+    
+    # 2. Activity Feed (Filtered)
+    all_feeds = requests.get(f"{GITEA_URL}/api/v1/users/{USERNAME}/feeds", headers=headers).json()
+    # Filter out pushes from the bot
+    human_feeds = [f for f in all_feeds if f['act_user']['login'] != BOT_NAME][:10]
+    # Specific list for "Recent Pushes"
+    recent_pushes = [f for f in all_feeds if f['op_type'] == 'push_repo' and f['act_user']['login'] != BOT_NAME][:5]
+
+    repo_stats = []
+    lang_totals = {}
+    total_instance_size = 0
+
+    for r in repos:
+        name, owner = r['name'], r['owner']['login']
+        
+        # Languages & Size
+        langs = requests.get(f"{GITEA_URL}/api/v1/repos/{owner}/{name}/languages", headers=headers).json()
+        for l, b in langs.items():
+            lang_totals[l] = lang_totals.get(l, 0) + b
+        
+        # Commits (Gitea returns total count in the header of the commits endpoint)
+        c_req = requests.get(f"{GITEA_URL}/api/v1/repos/{owner}/{name}/commits?limit=1", headers=headers)
+        commit_count = c_req.headers.get("X-Total", "0")
+        
+        repo_size = r['size'] * 1024
+        total_instance_size += repo_size
+        
+        repo_stats.append({
+            "name": name, "url": r['html_url'], "commits": commit_count,
+            "issues": r['open_issues_count'], "size": repo_size
+        })
+
+    # Identify most active repo (by commit count)
+    most_active = max(repo_stats, key=lambda x: int(x['commits'])) if repo_stats else None
+
+    return {
+        "ver": ver, "repos": repo_stats, "langs": lang_totals,
+        "feeds": human_feeds, "pushes": recent_pushes,
+        "total_size": total_instance_size, "uptime": get_uptime(),
+        "most_active": most_active
+    }
+
+def build_readme(data):
+    now = datetime.datetime.now().strftime("%Y-%m-%d %H:%M")
+    
+    # Header
+    md = f"# 🛠️ {USERNAME}'s Dev Hub\n\n"
+    md += f"**Gitea Instance:** `v{data['ver']}` | **Uptime:** `{data['uptime']}` | **Last Updated:** `{now}`\n\n"
+
+    # Top Languages & Total Space
+    sorted_langs = sorted(data['langs'].items(), key=lambda x: x[1], reverse=True)
+    lang_line = ", ".join([f"**{l[0]}** ({format_bytes(l[1])})" for l in sorted_langs[:5]])
+    
+    md += "### 📊 Stats Summary\n"
+    md += f"* **Total Storage Used:** `{format_bytes(data['total_size'])}` across {len(data['repos'])} repositories.\n"
+    md += f"* **Language Footprint:** {lang_line}\n"
+    if data['most_active']:
+        md += f"* **Most Active Project:** [{data['most_active']['name']}]({data['most_active']['url']}) with **{data['most_active']['commits']}** commits.\n"
+    md += "\n---\n"
+
+    # Repo Table
+    md += "### 📂 Repository Breakdown\n"
+    md += "| Repository | Commits | Issues | Space |\n| :--- | :--- | :--- | :--- |\n"
+    for r in data['repos']:
+        md += f"| [{r['name']}]({r['url']}) | {r['commits']} | {r['issues']} | {format_bytes(r['size'])} |\n"
+    md += "\n"
+
+    # Recent Pushes (Human Only)
+    md += "### 🚀 Recent Human Pushes\n"
+    if not data['pushes']: md += "*No recent manual pushes.*\n"
+    for p in data['pushes']:
+        date = p['created'].split('T')[0]
+        md += f"* **{date}** - Pushed to `{p['repo']['name']}`\n"
+    md += "\n"
+
+    # Global Activity
+    md += "### 🕒 Latest Changes\n"
+    for f in data['feeds']:
+        date = f['created'].split('T')[0]
+        action = f['op_type'].replace('_', ' ')
+        md += f"* **{date}**: {action} in `{f['repo']['name']}`\n"
+
+    return md
 
 def main():
-    readme_path = "README.md"
+    try:
-    
+        data = fetch_gitea_data()
-    if not os.path.exists(readme_path):
+        markdown = build_readme(data)
-        print(f"{readme_path} not found!")
+        with open("README.md", "w", encoding="utf-8") as f:
-        return
+            f.write(markdown)
-
+        print("README.md rebuilt successfully.")
-    with open(readme_path, "r", encoding="utf-8") as f:
+    except Exception as e:
-        content = f.read()
+        print(f"Error: {e}")
-
-    # Generate new dynamic content
-    stats_content = generate_stats()
-
-    # Replace content between the markers
-    pattern = r".*?"
-    replacement = f"\n{stats_content}\n"
-    new_content = re.sub(pattern, replacement, content, flags=re.DOTALL)
-
-    # 4. Save the file back to disk
-    with open(readme_path, "w", encoding="utf-8") as f:
-        f.write(new_content)
-    
-    print("README.md updated locally.")
 
 if __name__ == "__main__":
     main()