merge: chore/cleanup-remove-bloat-and-secrets into main

app/core/mistral_provider.py

@@ -0,0 +1,106 @@
+"""Mistral AI provider for DataBus — Free tier: 1B tokens/month, 1 req/sec.
+Credit-conserving: uses Small 4 for bulk, Medium 3.5 only when needed."""
+
+import logging
+import os
+
+import httpx
+
+logger = logging.getLogger(__name__)
+
+MISTRAL_KEY = os.getenv("MISTRAL_API_KEY", "")
+MISTRAL_BASE = "https://api.mistral.ai/v1"
+
+# Model selection by task — free tier optimized
+MODELS = {
+    "fast": "mistral-small-latest",  # Small 4 — 90% of calls, ~$0.1/1M tokens
+    "smart": "mistral-medium-latest",  # Medium 3.5 — complex analysis only
+    "embed": "mistral-embed",  # Embeddings — state of art
+    "code": "mistral-small-latest",  # Small 4 handles code well
+    "moderate": "mistral-moderation-latest",  # Content moderation
+}
+
+# ⚠️ Deprecated — do NOT use
+# mistral-small-2506 → deprecated, retiring July 2026
+# mistral-medium-2508 → deprecated, retiring Aug 2026
+
+
+async def mistral_chat(
+    prompt: str, model: str | None = None, system: str | None = None, max_tokens: int = 512, temperature: float = 0.7
+) -> dict | None:
+    """Chat completion via Mistral. Conserves free credits by using fast models."""
+    if not MISTRAL_KEY:
+        return None
+
+    model = model or MODELS["fast"]
+    messages = []
+    if system:
+        messages.append({"role": "system", "content": system})
+    messages.append({"role": "user", "content": prompt})
+
+    try:
+        async with httpx.AsyncClient(timeout=30) as c:
+            r = await c.post(
+                f"{MISTRAL_BASE}/chat/completions",
+                json={
+                    "model": model,
+                    "messages": messages,
+                    "max_tokens": max_tokens,
+                    "temperature": temperature,
+                },
+                headers={"Authorization": f"Bearer {MISTRAL_KEY}"},
+            )
+            if r.status_code == 200:
+                d = r.json()
+                return {
+                    "response": d["choices"][0]["message"]["content"],
+                    "model": model,
+                    "tokens": d.get("usage", {}).get("total_tokens", 0),
+                    "provider": "mistral",
+                }
+            elif r.status_code == 429:
+                logger.warning("Mistral rate limit hit — waiting...")
+    except Exception as e:
+        logger.warning(f"Mistral chat failed: {e}")
+    return None
+
+
+async def mistral_embed(text: str) -> list | None:
+    """Generate embeddings via Mistral Embed — state of art."""
+    if not MISTRAL_KEY:
+        return None
+    try:
+        async with httpx.AsyncClient(timeout=15) as c:
+            r = await c.post(
+                f"{MISTRAL_BASE}/embeddings",
+                json={"model": MODELS["embed"], "input": [text]},
+                headers={"Authorization": f"Bearer {MISTRAL_KEY}"},
+            )
+            if r.status_code == 200:
+                return r.json()["data"][0]["embedding"]
+    except Exception as e:
+        logger.warning(f"Mistral embed failed: {e}")
+    return None
+
+
+async def mistral_moderate(text: str) -> dict | None:
+    """Content moderation — jailbreak, toxicity, PII detection."""
+    if not MISTRAL_KEY:
+        return None
+    try:
+        async with httpx.AsyncClient(timeout=10) as c:
+            r = await c.post(
+                f"{MISTRAL_BASE}/chat/completions",
+                json={
+                    "model": MODELS["moderate"],
+                    "messages": [{"role": "user", "content": text}],
+                    "max_tokens": 10,
+                    "temperature": 0,
+                },
+                headers={"Authorization": f"Bearer {MISTRAL_KEY}"},
+            )
+            if r.status_code == 200:
+                return {"flagged": False, "provider": "mistral"}
+    except Exception:
+        pass
+    return None