backend/define.py

from __future__ import annotations

"""Serve short glosses from the local Kaikki dump baked into atlas.sqlite."""

import json
import sqlite3

MAX_SENSES = 4
MAX_GLOSSES = 4


def _trim_senses(raw: object) -> list[dict]:
    if isinstance(raw, str):
        try:
            raw = json.loads(raw)
        except json.JSONDecodeError:
            return []
    if not isinstance(raw, list):
        return []
    senses: list[dict] = []
    for block in raw:
        if not isinstance(block, dict):
            continue
        pos = str(block.get("pos") or "Sense").strip() or "Sense"
        glosses: list[str] = []
        for g in block.get("glosses") or []:
            text = " ".join(str(g).split()).strip()
            if text:
                glosses.append(text)
            if len(glosses) >= MAX_GLOSSES:
                break
        if glosses:
            senses.append({"pos": pos, "glosses": glosses})
        if len(senses) >= MAX_SENSES:
            break
    return senses


def _candidate_langs(lang: str, iso_639_3: str | None, lang_meta: dict[str, dict] | None) -> list[str]:
    out: list[str] = []
    key = (lang or "").strip()
    if key:
        out.append(key)
        folded = key.casefold()
        if folded != key:
            out.append(folded)
    iso = (iso_639_3 or "").strip().casefold()
    if lang_meta and iso:
        for lk, meta in lang_meta.items():
            if (meta.get("iso_639_3") or "").casefold() == iso:
                out.append(lk)
    if iso:
        out.append(iso)
    seen: set[str] = set()
    uniq: list[str] = []
    for c in out:
        if c and c not in seen:
            seen.add(c)
            uniq.append(c)
    return uniq


def lookup_definitions(
    conn: sqlite3.Connection,
    term: str,
    lang: str = "",
    ety: str | None = None,
    iso_639_3: str | None = None,
    lang_meta: dict[str, dict] | None = None,
) -> dict:
    term = (term or "").strip()
    if not term:
        return {"term": term, "senses": [], "error": "empty_term", "source": "kaikki"}

    candidates = _candidate_langs(lang, iso_639_3, lang_meta)

    row = None
    matched_lang = None
    matched_ety = None
    for cand in candidates:
        if ety is not None:
            row = conn.execute(
                "SELECT lang, ety, senses_json FROM definitions WHERE term = ? AND lang = ? AND ety = ?",
                (term, cand, ety),
            ).fetchone()
        else:
            row = conn.execute(
                """
                SELECT lang, ety, senses_json FROM definitions
                WHERE term = ? AND lang = ?
                ORDER BY (ety = '') DESC, ety
                LIMIT 1
                """,
                (term, cand),
            ).fetchone()
        if row is not None:
            matched_lang = row["lang"] if isinstance(row, sqlite3.Row) else row[0]
            matched_ety = row["ety"] if isinstance(row, sqlite3.Row) else row[1]
            break

    fallback = False
    if row is None:
        rows = conn.execute(
            "SELECT lang, ety, senses_json FROM definitions WHERE term = ? ORDER BY lang, ety LIMIT 8",
            (term,),
        ).fetchall()
        if rows:
            prefer = {c.casefold() for c in candidates}
            chosen = None
            for r in rows:
                rlang = r["lang"] if isinstance(r, sqlite3.Row) else r[0]
                if rlang.casefold() in prefer or rlang.casefold() == "english":
                    chosen = r
                    break
            row = chosen or rows[0]
            matched_lang = row["lang"] if isinstance(row, sqlite3.Row) else row[0]
            matched_ety = row["ety"] if isinstance(row, sqlite3.Row) else row[1]
            fallback = bool(candidates) and matched_lang not in candidates

    if row is None:
        return {
            "term": term,
            "lang": lang,
            "ety": ety or "",
            "matched_code": None,
            "senses": [],
            "fallback": False,
            "source": "kaikki",
            "error": "not_found",
        }

    senses_json = row["senses_json"] if isinstance(row, sqlite3.Row) else row[2]
    senses = _trim_senses(senses_json)
    return {
        "term": term,
        "lang": lang,
        "ety": matched_ety or "",
        "matched_code": matched_lang,
        "senses": senses,
        "fallback": fallback,
        "source": "kaikki",
        "error": None if senses else "no_senses",
    }


def fetch_definitions(
    term: str,
    lang: str = "",
    ety: str | None = None,
    iso_639_3: str | None = None,
    *,
    atlas=None,
) -> dict:
    """Lookup glosses from the baked index. ``atlas`` must be a loaded Atlas."""
    if atlas is None or getattr(atlas, "conn", None) is None:
        return {
            "term": term,
            "lang": lang,
            "ety": ety or "",
            "matched_code": None,
            "senses": [],
            "fallback": False,
            "source": "kaikki",
            "error": "index_unavailable",
        }
    return lookup_definitions(
        atlas.conn,
        term=term,
        lang=lang,
        ety=ety,
        iso_639_3=iso_639_3,
        lang_meta=getattr(atlas, "lang_meta", None),
    )