Create app.py

app.py

@@ -0,0 +1,70 @@
+import gradio as gr
+from internetarchive import search_items, get_item
+import requests
+import time
+
+# --- VirusTotal helper functions ---
+def scan_url_vt(url, api_key):
+    headers = {"x-apikey": api_key}
+    # submit URL
+    resp = requests.post(
+        "https://www.virustotal.com/api/v3/urls", headers=headers, data={"url": url}
+    )
+    resp.raise_for_status()
+    analysis_id = resp.json()["data"]["id"]
+    # poll until complete
+    while True:
+        time.sleep(5)
+        status_resp = requests.get(
+            f"https://www.virustotal.com/api/v3/analyses/{analysis_id}",
+            headers=headers
+        )
+        status_resp.raise_for_status()
+        attr = status_resp.json()["data"]["attributes"]
+        if attr.get("status") == "completed":
+            stats = attr.get("stats", {})
+            return stats.get("malicious", 0) == 0
+
+# --- Core search & scan logic ---
+def fetch_clean_videos(keywords, api_key, scan_enabled):
+    query = " OR ".join([f"{kw.strip().replace(' ', '+')}" for kw in keywords.split(",")])
+    ia_query = f"mediatype:(movies) AND ({query})"
+    results = list(search_items(ia_query, rows=50))
+    clean_urls = []
+    for res in results:
+        item = get_item(res['identifier'])
+        for f in item.files:
+            fmt = f.get('format','').lower()
+            if fmt.startswith(('mpeg','mp4','avi','mov','webm')):
+                url = f"https://archive.org/download/{res['identifier']}/{f['name']}"
+                if scan_enabled and api_key:
+                    try:
+                        is_clean = scan_url_vt(url, api_key)
+                    except Exception as e:
+                        continue
+                else:
+                    is_clean = True
+                if is_clean:
+                    clean_urls.append(url)
+    return clean_urls
+
+# --- Gradio UI setup ---
+def gui(keywords, api_key, scan_enabled):
+    """Perform search and optional VT scan, return list of clean URLs."""
+    urls = fetch_clean_videos(keywords, api_key, scan_enabled)
+    if not urls:
+        return "No clean videos found."
+    return "\n".join(urls)
+
+with gr.Blocks() as demo:
+    gr.Markdown("# 📼 IA Drone‑Strike Video Finder with VT Scan")
+    with gr.Row():
+        kw_input = gr.Textbox(label="Search keywords (comma-separated)", value="drone strike, military uav, kamikaze drone")
+        key_input = gr.Textbox(label="VirusTotal API Key", type="password")
+    scan_toggle = gr.Checkbox(label="Enable VirusTotal scan", value=True)
+    run_btn = gr.Button("Search & Scan")
+    output = gr.Textbox(label="Results (URLs)", lines=10)
+    run_btn.click(fn=gui, inputs=[kw_input, key_input, scan_toggle], outputs=output)
+
+if __name__ == "__main__":
+    demo.launch()  # on HuggingFace Spaces, this binds to 0.0.0.0 automatically