Deploy Full Engine

README.md

@@ -8,6 +8,15 @@ app_port: 7860
 pinned: false
 ---
 
+---
+title: Awn AI Service
+emoji: 🚀
+colorFrom: blue
+colorTo: purple
+sdk: docker
+app_port: 7860
+---
+
 # Awn AI Service
 
 AI-powered backend service for the Awn platform — connecting families in need with donors and charitable organizations.

app/api/admin_ops.py

@@ -6,6 +6,10 @@ from app.core.auth import verify_api_key
 from app.services.chat.api.llm_router import llm_router, circuit_registry
 import app.services.chat.api.llm_router as router_module
 from app.core.redis_client import redis_client
+from app.core.config import settings, BASE_DIR
+from app.services.chat.api.model_registry import model_registry
+import os
+import re
 
 logger = logging.getLogger(__name__)
 
@@ -28,8 +32,80 @@ class ModelBanReq(BaseModel):
 class KillSwitchReq(BaseModel):
     active: bool
 
+class EnvUpdateReq(BaseModel):
+    provider: str
+    models_string: str
+
+
+# ─── Permanent Configuration (.env) ───────────────────────────────────────────
+
+def _update_env_file(key: str, new_value: str):
+    """Safely updates a key inside the physical .env file. Fails gracefully if Read-Only."""
+    env_path = BASE_DIR / ".env"
+    
+    try:
+        if not env_path.exists():
+            # Fallback creation if it doesn't exist
+            with open(env_path, "w", encoding="utf-8") as f:
+                f.write(f"{key}={new_value}\n")
+            return
+
+        with open(env_path, "r", encoding="utf-8") as f:
+            lines = f.readlines()
+
+        updated = False
+        with open(env_path, "w", encoding="utf-8") as f:
+            for line in lines:
+                if line.startswith(f"{key}="):
+                    f.write(f"{key}={new_value}\n")
+                    updated = True
+                else:
+                    f.write(line)
+            if not updated:
+                # Key wasn't found, append it
+                if lines and not lines[-1].endswith("\n"):
+                    f.write("\n")
+                f.write(f"{key}={new_value}\n")
+    except IOError as e:
+        logger.warning(f"File system is restricted (e.g. Hugging Face Space). Could not save {key} to disk. Falling back to hot RAM reload only. Error: {e}")
+        pass
+
+@router.get("/env")
+async def get_env_models():
+    """Retrieve current permanent models defined in the core system."""
+    return {
+        "groq": settings.LLM_MODELS_GROQ,
+        "gemini": settings.LLM_MODELS_GEMINI,
+        "openai": settings.LLM_MODELS_OPENAI,
+        "openrouter": settings.LLM_MODELS_OPENROUTER,
+    }
 
-# ─── Provider Controls ────────────────────────────────────────────────────────
+@router.post("/env")
+async def update_env_models(payload: EnvUpdateReq):
+    """Securely writes default fallback models to physical .env file and hot-reloads."""
+    provider = payload.provider.strip().lower()
+    mapping = {
+        "groq": "LLM_MODELS_GROQ",
+        "gemini": "LLM_MODELS_GEMINI",
+        "openai": "LLM_MODELS_OPENAI",
+        "openrouter": "LLM_MODELS_OPENROUTER",
+    }
+    
+    if provider not in mapping:
+        raise HTTPException(status_code=400, detail="Invalid provider ID")
+    
+    key = mapping[provider]
+    
+    # Write to physical file securely
+    _update_env_file(key, payload.models_string)
+    
+    # Write to local settings memory so it's instantly available everywhere
+    setattr(settings, key, payload.models_string)
+    
+    # Reload model_registry defaults instantly to trigger Hot-Reload
+    model_registry._load_defaults()
+    
+    return {"status": "success", "msg": f"Updated {key} to {payload.models_string} safely."}
 
 @router.post("/provider/{name}/disable")
 async def disable_provider(name: str):

app/services/chat/api/model_registry.py

@@ -79,9 +79,9 @@ class ModelRegistry:
     def _load_defaults(self):
         """
         Default model catalogue.
-        In the future, this can be loaded from a config file or env variable
-        so models can be updated without touching source code.
+        Loads from config strings and hot-reloads dynamically.
         """
+        self._models.clear()
         defaults = {
             "groq": [(m.strip(), i) for i, m in enumerate(settings.LLM_MODELS_GROQ.split(",")) if m.strip()],
             "gemini": [(m.strip(), i) for i, m in enumerate(settings.LLM_MODELS_GEMINI.split(",")) if m.strip()],

app/services/rag/vector_store.py

@@ -54,8 +54,18 @@ class VectorStore:
             return False
 
         if not settings.QDRANT_URL:
-            logger.info("QDRANT_URL not configured — vector store disabled.")
-            return False
+            logger.info("QDRANT_URL not configured. Attempting :memory: fallback for RAG...")
+            try:
+                self.client = QdrantClient(":memory:")
+                self._connected = True
+                self._ensure_collection()
+                logger.warning("🟢 RAG running in Ephemeral :memory: mode. Vectors will be wiped on restart.")
+                return True
+            except Exception as exc:
+                logger.warning("Failed to fallback to memory Qdrant: %s — vector store disabled.", exc)
+                self.client = None
+                self._connected = False
+                return False
 
         try:
             self.client = QdrantClient(

app/static/dashboard.html

@@ -286,6 +286,47 @@
                 <div class="panel"><div class="kpi-lbl">Success Rate Discrepancy (%)</div><div class="chart-wrap"><canvas id="chartSr"></canvas></div></div>
             </div>
 
+            <!-- .env Configuration Core -->
+            <div class="section-title" style="margin-top:40px; color:var(--accent)">
+                <span>⚙️ System Configuration Core (.env IDE)</span>
+                <span class="tag" style="font-size:11px; background:rgba(59,130,246,0.1); color:var(--accent)">Live Hot-Reloading Enabled</span>
+            </div>
+            <div class="panel" style="border-color: rgba(59,130,246,0.3); background: rgba(19,22,32,0.95)">
+                <div style="color:var(--text-dim); font-size:13px; margin-bottom:20px;">
+                    These text areas write directly to the <b>.env</b> core file. Changes are hot-reloaded into memory instantly. Separate models by commas.
+                </div>
+                <div class="grid-2">
+                    <div style="background:rgba(0,0,0,0.3); padding:16px; border-radius:8px; border:1px solid var(--border);">
+                        <div class="flex justify-between items-center" style="margin-bottom:12px;">
+                            <strong style="font-family:var(--font-mono); font-size:13px; color:#a78bfa">LLM_MODELS_OPENROUTER</strong>
+                            <button class="btn btn-ok" style="padding:4px 8px; font-size:11px" onclick="saveEnvConfig('openrouter')">💾 Save & Reload</button>
+                        </div>
+                        <textarea id="env_openrouter" class="form-input" style="height:60px; resize:vertical; font-size:12px;"></textarea>
+                    </div>
+                    <div style="background:rgba(0,0,0,0.3); padding:16px; border-radius:8px; border:1px solid var(--border);">
+                        <div class="flex justify-between items-center" style="margin-bottom:12px;">
+                            <strong style="font-family:var(--font-mono); font-size:13px; color:#f472b6">LLM_MODELS_GROQ</strong>
+                            <button class="btn btn-ok" style="padding:4px 8px; font-size:11px" onclick="saveEnvConfig('groq')">💾 Save & Reload</button>
+                        </div>
+                        <textarea id="env_groq" class="form-input" style="height:60px; resize:vertical; font-size:12px;"></textarea>
+                    </div>
+                    <div style="background:rgba(0,0,0,0.3); padding:16px; border-radius:8px; border:1px solid var(--border);">
+                        <div class="flex justify-between items-center" style="margin-bottom:12px;">
+                            <strong style="font-family:var(--font-mono); font-size:13px; color:#34d399">LLM_MODELS_GEMINI</strong>
+                            <button class="btn btn-ok" style="padding:4px 8px; font-size:11px" onclick="saveEnvConfig('gemini')">💾 Save & Reload</button>
+                        </div>
+                        <textarea id="env_gemini" class="form-input" style="height:60px; resize:vertical; font-size:12px;"></textarea>
+                    </div>
+                    <div style="background:rgba(0,0,0,0.3); padding:16px; border-radius:8px; border:1px solid var(--border);">
+                        <div class="flex justify-between items-center" style="margin-bottom:12px;">
+                            <strong style="font-family:var(--font-mono); font-size:13px; color:#fbbf24">LLM_MODELS_OPENAI</strong>
+                            <button class="btn btn-ok" style="padding:4px 8px; font-size:11px" onclick="saveEnvConfig('openai')">💾 Save & Reload</button>
+                        </div>
+                        <textarea id="env_openai" class="form-input" style="height:60px; resize:vertical; font-size:12px;"></textarea>
+                    </div>
+                </div>
+            </div>
+
         </div>
     </main>
 
@@ -301,7 +342,7 @@
         if(API_KEY) { document.getElementById('authOverlay').classList.remove('active'); setTimeout(() => document.getElementById('authOverlay').style.display='none', 200); }
         function saveKey() {
             const v = document.getElementById('apiKeyIn').value.trim();
-            if(v) { API_KEY = v; localStorage.setItem('awn_api_key', v); document.getElementById('authOverlay').classList.remove('active'); setTimeout(() => document.getElementById('authOverlay').style.display='none', 200); fetchLoop(); }
+            if(v) { API_KEY = v; localStorage.setItem('awn_api_key', v); document.getElementById('authOverlay').classList.remove('active'); setTimeout(() => document.getElementById('authOverlay').style.display='none', 200); fetchLoop(); fetchEnvConfig(); }
         }
         function logout() {
             localStorage.removeItem('awn_api_key'); API_KEY = null; 
@@ -524,7 +565,29 @@
             loopTimer = setTimeout(fetchLoop, 5000);
         }
 
-        document.addEventListener('DOMContentLoaded', () => { initCharts(); if(API_KEY) fetchLoop(); });
+        // .env Editor Functions
+        async function fetchEnvConfig() {
+            if(!API_KEY) return;
+            try {
+                const res = await fetch('/api/ai/admin/env', {headers: {'X-API-Key': API_KEY}});
+                if(res.ok) {
+                    const data = await res.json();
+                    if(document.activeElement.tagName !== "TEXTAREA") {
+                        document.getElementById('env_groq').value = data.groq || '';
+                        document.getElementById('env_gemini').value = data.gemini || '';
+                        document.getElementById('env_openai').value = data.openai || '';
+                        document.getElementById('env_openrouter').value = data.openrouter || '';
+                    }
+                }
+            } catch(e) {}
+        }
+
+        async function saveEnvConfig(provider) {
+            const str = document.getElementById(`env_${provider}`).value;
+            adminCmd('POST', '/api/ai/admin/env', {provider: provider, models_string: str});
+        }
+
+        document.addEventListener('DOMContentLoaded', () => { initCharts(); if(API_KEY) { fetchLoop(); fetchEnvConfig(); } });
     </script>
 </body>
 </html>