"""Prepare the listing text for the text-only extractor: concatenate description, features and broker details,
and MASK every price, tax and municipal-assessment amount so that the text model cannot read value information
(the WP14 leakage lesson: DuProprio descriptions embed the asking price).
"""
from __future__ import annotations

import json
import re

MASKED_KEYS = re.compile(r"(évaluation municipale|evaluation municipale|municipal assessment|taxes?|prix|price|revenus|"
                         r"asking|hypoth|mortgage|frais de condo|condo fees|img_audited|transaction)", re.I)
MONEY = re.compile(r"(\$\s?\d[\d\s.,]*\d|\d[\d\s]{2,}(?:[.,]\d+)?\s?(?:\$|dollars|k\$|CAD))", re.I)
ASSESS_SENT = re.compile(r"(municipal assessment|évaluation municipale|evaluation municipale|asking price|prix demandé|"
                         r"taxes municipales|taxes scolaires|school taxes|municipal taxes)[^.\n]{0,80}", re.I)


def mask_money(s: str) -> str:
    s = ASSESS_SENT.sub("[masked]", s)
    return MONEY.sub("[amount]", s)


def build_text(row) -> tuple[str, dict]:
    """Return (masked text, stats). `row` must have description, features (JSON list), details (JSON dict)."""
    parts = []
    desc = row.get("description") or ""
    parts.append("DESCRIPTION:\n" + mask_money(desc))
    try:
        feats = json.loads(row.get("features") or "[]")
    except Exception:
        feats = []
    if isinstance(feats, list) and feats:
        feats = [mask_money(str(f)) for f in feats[:80] if not MASKED_KEYS.search(str(f).split(":")[0])]
        parts.append("FEATURES: " + "; ".join(feats))
    try:
        det = json.loads(row.get("details") or "{}")
    except Exception:
        det = {}
    kept, dropped = [], 0
    if isinstance(det, dict):
        for k, v in det.items():
            if MASKED_KEYS.search(str(k)):
                dropped += 1
                continue
            if k == "pieces" and isinstance(v, list):
                rooms = "; ".join(f"{p.get('nom','')} ({p.get('niveau','')}, {p.get('dimensions','')}, {p.get('revetement','')})"
                                  for p in v[:25] if isinstance(p, dict))
                kept.append(f"ROOMS: {rooms}")
            elif isinstance(v, (str, int, float)):
                if k == "remarques_proprio" and str(v).strip()[:200] == desc.strip()[:200]:
                    continue  # duplicate of the description
                kept.append(f"{k}: {mask_money(str(v))}")
    if kept:
        parts.append("BROKER FIELDS:\n" + "\n".join(kept))
    text = "\n\n".join(parts)
    stats = {"desc_chars": len(desc), "n_features": len(feats) if isinstance(feats, list) else 0,
             "n_detail_kept": len(kept), "n_detail_masked_keys": dropped,
             "n_amounts_masked": len(MONEY.findall(desc)) + len(ASSESS_SENT.findall(desc))}
    return text[:12000], stats