Upload app.py

app.py

@@ -0,0 +1,116 @@
+import os
+import requests
+import gradio as gr
+from fastapi import FastAPI, Request
+from fastapi.responses import JSONResponse
+import uvicorn
+
+# Configuration
+API_URL = "https://router.huggingface.co/v1/chat/completions"
+headers = {
+    "Authorization": f"Bearer {os.environ.get('HF_TOKEN', '')}",
+}
+MODEL = "deepseek-ai/DeepSeek-V3.2:novita"
+
+app = FastAPI()
+
+def query_ai(prompt):
+    payload = {
+        "messages": [{"role": "user", "content": prompt}],
+        "model": MODEL
+    }
+    try:
+        response = requests.post(API_URL, headers=headers, json=payload, timeout=30)
+        return response.json()
+    except Exception as e:
+        return {"error": str(e)}
+
+def search_osm_logic(query, location="Nigeria"):
+    params = {
+        "q": f"{query} in {location}",
+        "format": "json",
+        "addressdetails": 1,
+        "limit": 10
+    }
+    try:
+        resp = requests.get("https://nominatim.openstreetmap.org/search", params=params, headers={"User-Agent": "Internly/1.0"}, timeout=10)
+        return resp.json()
+    except Exception as e:
+        return []
+
+@app.get("/api/search")
+async def api_search(q: str, loc: str = "Nigeria"):
+    osm_results = search_osm_logic(q, loc)
+    enriched_results = []
+    
+    for item in osm_results:
+        name = item.get("display_name").split(",")[0]
+        address = item.get("display_name")
+        lat = item.get("lat")
+        lon = item.get("lon")
+        osm_id = item.get("osm_id")
+        
+        prompt = (
+            f"Analyze '{name}' at '{address}' for SIWES internship in Nigeria. "
+            "Return ONLY a JSON object with: 'orgType' (Private/Government/Unknown), "
+            "'siwesCategory' (e.g. Healthcare, Engineering), 'summary' (1 sentence), 'relevance' (1 sentence)."
+        )
+        ai_resp = query_ai(prompt)
+        
+        result = {
+            "osmId": str(osm_id),
+            "name": name,
+            "address": address,
+            "location": loc,
+            "coordinates": {"lat": float(lat) if lat else 0, "lon": float(lon) if lon else 0},
+            "type": item.get("type", "unknown")
+        }
+        
+        try:
+            content = ai_resp["choices"][0]["message"]["content"]
+            # Basic JSON extraction if model wraps in markdown
+            if "```json" in content:
+                content = content.split("```json")[1].split("```")[0]
+            import json
+            ai_data = json.loads(content.strip())
+            result.update(ai_data)
+        except:
+            result.update({
+                "orgType": "Unknown",
+                "siwesCategory": "Unknown",
+                "summary": "AI enrichment failed.",
+                "relevance": "Unknown"
+            })
+            
+        enriched_results.append(result)
+            
+    return enriched_results
+
+# Gradio Interface for testing
+def gradio_search(query, location):
+    import asyncio
+    import json
+    # Use the same logic for the UI
+    results = search_osm_logic(query, location)
+    formatted = []
+    for r in results:
+        formatted.append(f"### {r.get('display_name')}\n")
+    return "\n".join(formatted) if formatted else "No results found."
+
+with gr.Blocks(title="Internly API Service") as demo:
+    gr.Markdown("# Internly API Service")
+    gr.Markdown("This space serves as the backend API for Internly. Search is exposed at `/api/search`.")
+    
+    with gr.Row():
+        q_input = gr.Textbox(label="Query")
+        l_input = gr.Textbox(label="Location", value="Nigeria")
+    
+    btn = gr.Button("Test Search")
+    output = gr.Markdown()
+    
+    btn.click(fn=gradio_search, inputs=[q_input, l_input], outputs=output)
+
+app = gr.mount_gradio_app(app, demo, path="/")
+
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=7860)