"""Поиск по каталогу: полнотекстовый запрос с учётом морфологии."""
from __future__ import annotations

import json
import re

from .textutil import stem

SAFE = re.compile(r"[^\w\-Ѐ-ӿ]+", re.U)

# Веса колонок для bm25: заголовок и ключевые слова важнее тела документа.
BM25_WEIGHTS = (10.0, 8.0, 4.0, 1.0, 0.8, 2.0)


def build_query(text: str) -> str:
    """Превращает пользовательский запрос в выражение FTS5.

    Каждое слово ищется и как есть, и по основе с префиксом — чтобы
    «инвертор» находил «инверторы» и «инвертора».
    """
    phrases = re.findall(r'"([^"]+)"', text)
    rest = re.sub(r'"[^"]*"', " ", text)

    parts = ['"%s"' % SAFE.sub(" ", p).strip() for p in phrases if p.strip()]
    for raw in rest.split():
        neg = raw.startswith("-")
        word = SAFE.sub("", raw.lstrip("-+")).strip()
        if len(word) < 2:
            continue
        st = stem(word)
        variants = {'"%s"' % word.lower()}
        if st and st != word.lower():
            variants.add('"%s"*' % st)
        else:
            variants.add('"%s"*' % word.lower())
        clause = "(" + " OR ".join(sorted(variants)) + ")"
        parts.append(("NOT " + clause) if neg else clause)
    return " AND ".join(p for p in parts if p).replace("AND NOT", "NOT")


def search(con, query: str, *, limit: int = 20, ext: str | None = None,
           folder: str | None = None, lang: str | None = None,
           year_from: int | None = None, kind: str | None = None) -> list[dict]:
    fts_query = build_query(query)
    if not fts_query:
        return []

    sql = [
        "SELECT d.id, d.title, d.author, d.year, d.pages, d.ext, d.lang, d.kind,",
        "       d.rel_path, d.path, d.folder, d.description, d.keywords, d.theses,",
        "       d.scanned, u.keywords AS user_keywords,",
        "       snippet(docs_fts, 3, '«', '»', ' … ', 14) AS snip,",
        "       bm25(docs_fts, %s) AS rank" % ", ".join(str(w) for w in BM25_WEIGHTS),
        "FROM docs_fts JOIN docs d ON d.id = docs_fts.rowid",
        "     LEFT JOIN user_keywords u ON u.path = d.path",
        "WHERE docs_fts MATCH ?",
    ]
    params: list = [fts_query]
    if ext:
        sql.append("AND d.ext = ?")
        params.append("." + ext.lower().lstrip("."))
    if folder:
        sql.append("AND d.folder LIKE ?")
        params.append("%" + folder + "%")
    if lang:
        sql.append("AND d.lang = ?")
        params.append(lang)
    if kind:
        sql.append("AND d.kind = ?")
        params.append(kind)
    if year_from:
        sql.append("AND d.year >= ?")
        params.append(year_from)
    sql.append("ORDER BY rank LIMIT ?")
    params.append(limit)

    rows = con.execute("\n".join(sql), params).fetchall()
    out = []
    for r in rows:
        d = dict(r)
        d["keywords"] = json.loads(d.get("keywords") or "[]")
        d["theses"] = json.loads(d.get("theses") or "[]")
        d["user_keywords"] = json.loads(d.get("user_keywords") or "[]")
        d["score"] = round(-d.pop("rank"), 2)
        out.append(d)
    return out