Wiki Jokes: daily LLM jokes from Wikipedia featured article

- FastAPI + APScheduler + SQLite, single container
- Daily generation at 06:00 Europe/Copenhagen + cold-start generation
- Retry with backoff (3 attempts/15 min), stale fallback with banner
- OpenAI-compatible endpoint via env (OPENAI_BASE_URL/MODEL/API_KEY)
- docker-compose with named volume for joke persistence

app/llm_client.py

@@ -0,0 +1,66 @@
+"""Call an OpenAI-compatible chat completions endpoint to generate jokes."""
+import json
+import re
+
+import httpx
+
+from .config import settings
+
+SYSTEM_PROMPT = (
+    "You are a comedian. You will receive the text of Wikipedia's featured article "
+    "of the day. Write exactly 5 short, clean, family-friendly jokes inspired by "
+    "facts from the article. Vary the style (one-liners, puns, observational). "
+    "The jokes must be understandable on their own without reading the article. "
+    "Respond with ONLY a JSON array of 5 strings and nothing else. "
+    "Example: [\"joke one\", \"joke two\", \"joke three\", \"joke four\", \"joke five\"]"
+)
+
+
+def _extract_json_array(text: str) -> list[str]:
+    """Pull a JSON array of strings out of a possibly messy LLM response."""
+    text = text.strip()
+    # Strip markdown fences if present.
+    fence = re.search(r"```(?:json)?\s*(.*?)```", text, re.DOTALL)
+    if fence:
+        text = fence.group(1).strip()
+    # Find the outermost [...] span.
+    start = text.find("[")
+    end = text.rfind("]")
+    if start == -1 or end == -1 or end <= start:
+        raise ValueError(f"No JSON array found in LLM response: {text[:200]!r}")
+    data = json.loads(text[start : end + 1])
+    if not isinstance(data, list):
+        raise ValueError("Parsed JSON is not a list")
+    jokes = [str(j).strip() for j in data if str(j).strip()]
+    if len(jokes) < 5:
+        raise ValueError(f"Only {len(jokes)} jokes returned, expected 5")
+    return jokes[:5]
+
+
+def generate_jokes(article_title: str, article_extract: str) -> list[str]:
+    url = f"{settings.openai_base_url}/chat/completions"
+    headers = {"Content-Type": "application/json"}
+    if settings.openai_api_key:
+        headers["Authorization"] = f"Bearer {settings.openai_api_key}"
+
+    user_prompt = (
+        f"Today's Wikipedia featured article: \"{article_title}\"\n\n"
+        f"Article text:\n{article_extract}\n\n"
+        "Now write exactly 5 jokes as a JSON array of 5 strings."
+    )
+
+    body = {
+        "model": settings.openai_model,
+        "messages": [
+            {"role": "system", "content": SYSTEM_PROMPT},
+            {"role": "user", "content": user_prompt},
+        ],
+        "temperature": 0.9,
+        "max_tokens": 800,
+    }
+
+    resp = httpx.post(url, headers=headers, json=body, timeout=120)
+    resp.raise_for_status()
+    payload = resp.json()
+    content = payload["choices"][0]["message"]["content"]
+    return _extract_json_array(content)