lazyflat/web/llm.py

"""Minimal Anthropic Messages API wrapper for flat enrichment.

Uses tool-use forced output so Haiku returns structured JSON instead of free
text we'd have to regex. No SDK — plain `requests` is enough here.
"""
from __future__ import annotations

import logging
from typing import Any, Optional

import requests

from settings import ANTHROPIC_API_KEY, ANTHROPIC_MODEL

logger = logging.getLogger("web.llm")

API_URL = "https://api.anthropic.com/v1/messages"
API_VERSION = "2023-06-01"

TOOL_NAME = "record_flat_details"
TOOL_SCHEMA: dict[str, Any] = {
    "type": "object",
    "properties": {
        "address": {"type": ["string", "null"],
                    "description": "Full street address incl. postcode+city if present"},
        "rooms": {"type": ["number", "null"], "description": "Number of rooms (decimal ok)"},
        "size_sqm": {"type": ["number", "null"], "description": "Size in m²"},
        "rent_cold": {"type": ["number", "null"], "description": "Kaltmiete in €"},
        "rent_total": {"type": ["number", "null"], "description": "Warm/Gesamtmiete in €"},
        "utilities": {"type": ["number", "null"], "description": "Nebenkosten in €"},
        "deposit": {"type": ["number", "null"], "description": "Kaution in €"},
        "available_from": {"type": ["string", "null"], "description": "Bezugsfrei ab (text)"},
        "floor": {"type": ["string", "null"], "description": "Etage (text, z.B. '3. OG')"},
        "heating": {"type": ["string", "null"]},
        "energy_certificate": {"type": ["string", "null"]},
        "energy_value": {"type": ["string", "null"]},
        "year_built": {"type": ["string", "null"]},
        "wbs_required": {"type": ["boolean", "null"]},
        "wbs_type": {"type": ["string", "null"], "description": "WBS-Typ, z.B. '160' oder null"},
        "description": {
            "type": ["string", "null"],
            "description": "Kurze 2–3-Satz-Beschreibung der Wohnung auf Deutsch. Fakten, keine Werbesprache.",
        },
        "features": {
            "type": "array", "items": {"type": "string"},
            "description": "Ausstattungsmerkmale (z.B. 'Balkon', 'Einbauküche', 'Parkett')",
        },
        "pros": {
            "type": "array", "items": {"type": "string"},
            "description": "2–4 konkrete Vorteile aus Bewerbersicht (keine Werbung)",
        },
        "cons": {
            "type": "array", "items": {"type": "string"},
            "description": "2–4 mögliche Nachteile / Punkte zum Beachten",
        },
    },
    "required": [],
    "additionalProperties": False,
}

SYSTEM_PROMPT = (
    "Du extrahierst strukturierte Wohnungsdaten aus deutschem HTML-Quelltext von "
    "Berliner Wohnungsbaugesellschaften (howoge, gewobag, degewo, gesobau, wbm, "
    "stadt-und-land). Antworte AUSSCHLIESSLICH über den bereitgestellten Tool-Call. "
    "Fehlende Werte → null. Keine Erfindungen — wenn etwas nicht klar aus dem HTML "
    "hervorgeht, lass das Feld null. Zahlen bitte als Zahlen (nicht als String), "
    "Beschreibung/Pros/Cons auf Deutsch."
)


def extract_flat_details(html: str, url: str,
                         max_html_chars: int = 60_000,
                         timeout: int = 60) -> Optional[dict]:
    """Call Haiku; return the structured dict or None on failure."""
    if not ANTHROPIC_API_KEY:
        logger.info("skipping enrichment: ANTHROPIC_API_KEY not set")
        return None

    user_content = (
        f"URL: {url}\n\n"
        f"HTML-Quellcode (ggf. gekürzt):\n---\n{html[:max_html_chars]}\n---"
    )
    body = {
        "model": ANTHROPIC_MODEL,
        "max_tokens": 1500,
        "system": SYSTEM_PROMPT,
        "tools": [{
            "name": TOOL_NAME,
            "description": "Persist the extracted flat details.",
            "input_schema": TOOL_SCHEMA,
        }],
        "tool_choice": {"type": "tool", "name": TOOL_NAME},
        "messages": [{"role": "user", "content": user_content}],
    }
    try:
        r = requests.post(
            API_URL,
            headers={
                "x-api-key": ANTHROPIC_API_KEY,
                "anthropic-version": API_VERSION,
                "content-type": "application/json",
            },
            json=body,
            timeout=timeout,
        )
    except requests.RequestException as e:
        logger.warning("anthropic request failed: %s", e)
        return None

    if r.status_code >= 400:
        logger.warning("anthropic %s: %s", r.status_code, r.text[:300])
        return None

    data = r.json()
    for block in data.get("content", []):
        if block.get("type") == "tool_use" and block.get("name") == TOOL_NAME:
            return block.get("input") or {}
    logger.warning("anthropic returned no tool_use block: %s", data)
    return None