#!/usr/bin/env python3
"""Seed a small, representative content set for local dev / CI / demos.

The real Search-UI databases are multi-GB and built from JW.org downloads,
and the semantic index needs a sentence-transformers model that may not be
available offline. That makes it hard to exercise the search experience
end-to-end without the production data.

This script seeds just enough **keyword-searchable** content to drive the
app for real:

  * writes ``json/<lang>/all_media_items.json`` so title search and result
    decoration work,
  * inserts subtitle text straight into the ``subtitles_fts`` FTS5 table
    (bypassing embeddings, so no model download is required),
  * indexes scripture references from that text (pure regex, model-free).

Keyword search, title search, and scripture search all return real results
afterwards. Semantic/hybrid search still needs the embedding model and is
intentionally out of scope here.

The sample set is fixed and deterministic so a golden snapshot taken
against it is stable across runs.

Usage::

    # seed into the default throwaway ./sample-data dir (gitignored)
    python scripts/seed_sample_data.py
    # seed a specific throwaway location
    python scripts/seed_sample_data.py --data-root /tmp/searchui-sample
    # re-clear the sample rows first (scoped to sample keys only)
    python scripts/seed_sample_data.py --clear

Safety: the default data root is ./sample-data, never your real database.
If you point --data-root at a location that already holds real (non-sample)
content, the script REFUSES to run unless you pass --force — it will not
overwrite a production media catalog or mix fake videos into a real index.
Nothing it writes is committed to git (sample-data/, json/ and *.db are
gitignored).
"""

from __future__ import annotations

import argparse
import json
import os
import sys
from pathlib import Path

REPO_ROOT = Path(__file__).resolve().parent.parent
sys.path.insert(0, str(REPO_ROOT / "backend"))


# A fixed, representative slice of JW.org-style VOD content. Subtitle text is
# written to read naturally and to contain the vocabulary + scripture
# references the golden queries probe (kingdom, field service, New World
# Translation, John 3:16, ...). natural_key mirrors the real
# languageAgnosticNaturalKey style (pub_track_VIDEO).
SAMPLE_ITEMS: list[dict] = [
    {
        "natural_key": "pub-jwbcov_E_1_VIDEO",
        "title": "What Is God's Kingdom?",
        "primaryCategory": "VODBibleTeachings",
        "subtitles": (
            "God's Kingdom is a real government in heaven. Jesus taught his "
            "followers to pray for the Kingdom to come. At Daniel 2:44 we read "
            "that the Kingdom will never be destroyed and will crush all human "
            "governments. The Kingdom will bring peace to the whole earth."
        ),
    },
    {
        "natural_key": "pub-jwbcov_E_2_VIDEO",
        "title": "Why Do We Share in the Field Ministry?",
        "primaryCategory": "VODBibleTeachings",
        "subtitles": (
            "Jehovah's Witnesses share in the field ministry because Jesus "
            "commanded his disciples to preach the good news of the Kingdom. "
            "Field service gives everyone an opportunity to help others learn "
            "the truth. At Matthew 28:19, 20 Jesus said to make disciples of "
            "people of all the nations."
        ),
    },
    {
        "natural_key": "pub-jwbcov_E_3_VIDEO",
        "title": "How Can You Trust the Bible?",
        "primaryCategory": "VODBibleTeachings",
        "subtitles": (
            "The Bible has been translated into thousands of languages. The "
            "New World Translation is accurate and easy to understand. Second "
            "Timothy 3:16 explains that all Scripture is inspired of God and "
            "beneficial for teaching. You can trust the Bible because its "
            "prophecies have come true."
        ),
    },
    {
        "natural_key": "pub-jwbcov_E_4_VIDEO",
        "title": "God's Love for Us",
        "primaryCategory": "VODBibleTeachings",
        "subtitles": (
            "John 3:16 tells us that God loved the world so much that he gave "
            "his only-begotten Son. God's love moves us to want to know him "
            "better and to show love to our neighbors. Love is the greatest "
            "quality, as explained at 1 Corinthians 13."
        ),
    },
    {
        "natural_key": "pub-jwbcov_E_5_VIDEO",
        "title": "Enjoy Life Forever in Paradise",
        "primaryCategory": "VODBibleTeachings",
        "subtitles": (
            "The Bible promises that the earth will become a paradise. "
            "Revelation 21:4 says God will wipe out every tear, and death will "
            "be no more. Psalm 37:29 promises that the righteous will possess "
            "the earth and live forever on it."
        ),
    },
    {
        "natural_key": "pub-jwbvod24_E_201_VIDEO",
        "title": "JW Broadcasting—Bible Principles for Family Life",
        "primaryCategory": "VODProgramsEvents",
        "subtitles": (
            "Applying Bible principles strengthens family life. Husbands and "
            "wives who follow the Bible's counsel build strong marriages. "
            "Ephesians 5:33 counsels husbands to love their wives and wives to "
            "respect their husbands. Children who honor their parents are "
            "happier."
        ),
    },
    {
        "natural_key": "pub-jwbvod24_E_202_VIDEO",
        "title": "Cart Witnessing in the Community",
        "primaryCategory": "VODProgramsEvents",
        "subtitles": (
            "Public witnessing with carts lets people approach at their own "
            "pace. Many take a free Bible study or a magazine. The field "
            "ministry reaches people in busy public places. Volunteers stand "
            "ready to answer Bible questions about the Kingdom hope."
        ),
    },
    {
        "natural_key": "pub-jwbvod24_E_203_VIDEO",
        "title": "Memorial of Jesus' Death",
        "primaryCategory": "VODProgramsEvents",
        "subtitles": (
            "Each year millions gather for the Memorial of Jesus' death. Jesus "
            "said at Luke 22:19 to keep doing this in remembrance of him. The "
            "emblems of unleavened bread and red wine represent his body and "
            "blood. The Memorial reminds us of God's love and the ransom."
        ),
    },
    {
        "natural_key": "pub-ihelp_E_7_VIDEO",
        "title": "Coping With Anxiety",
        "primaryCategory": "VODBibleTeachings",
        "subtitles": (
            "Everyone feels anxious sometimes. Prayer can help us cope with "
            "anxiety. Philippians 4:6, 7 encourages us to bring our worries to "
            "God in prayer, and the peace of God will guard our hearts. "
            "Casting our anxiety on Jehovah brings real comfort."
        ),
    },
    {
        "natural_key": "pub-ihelp_E_8_VIDEO",
        "title": "Finding True Happiness",
        "primaryCategory": "VODBibleTeachings",
        "subtitles": (
            "True happiness comes from helping others and knowing God. Jesus "
            "said at Acts 20:35 that there is more happiness in giving than in "
            "receiving. Spiritual things bring lasting joy that money cannot "
            "buy."
        ),
    },
]


SAMPLE_KEYS: frozenset[str] = frozenset(item["natural_key"] for item in SAMPLE_ITEMS)


def _build_media_record(item: dict) -> dict:
    """Shape one item the way all_media_items.json entries look."""
    return {
        "title": item["title"],
        "languageAgnosticNaturalKey": item["natural_key"],
        "primaryCategory": item.get("primaryCategory", "VODBibleTeachings"),
        "_category_key": "VideoOnDemand",
        "_subcategory": item.get("primaryCategory", "VODBibleTeachings"),
        "duration": 300,
        "firstPublished": "2024-01-01T00:00:00Z",
    }


def _describe_real_data(instance, language: str, media_path: Path) -> str | None:
    """Return a human description of REAL (non-sample) content found at the
    target, or None if the target is empty / sample-only.

    Used to refuse seeding over a developer's production database — the
    sample keys are a fixed known set, so anything else is real data.
    """
    conn = instance._get_db_connection()
    try:
        rows = conn.execute(
            "SELECT DISTINCT natural_key FROM subtitles_fts WHERE language = ?",
            (language,),
        ).fetchall()
    finally:
        conn.close()
    real_subs = [r[0] for r in rows if r[0] not in SAMPLE_KEYS]
    if real_subs:
        return f"{len(real_subs)} non-sample subtitle row(s) already indexed for language {language}"

    if media_path.exists():
        try:
            existing = json.loads(media_path.read_text(encoding="utf-8"))
            real_media = [k for k in existing if k not in SAMPLE_KEYS]
            if real_media:
                return f"{len(real_media)} non-sample entrie(s) in existing {media_path.name}"
        except (json.JSONDecodeError, OSError):
            return f"an existing {media_path.name} that could not be parsed (refusing to overwrite blindly)"
    return None


def seed(*, data_root: Path, language: str, clear: bool, force: bool = False) -> dict:
    """Seed the sample content. Returns a summary dict.

    Refuses to run if the target data root already holds real (non-sample)
    content, unless ``force=True`` — seeding is for throwaway dev/CI/demo
    roots, never for a production database.
    """
    # Point the runtime path helpers at the chosen data root before importing
    # anything that resolves DB/JSON locations. Restore it afterwards so a
    # caller in the same process (e.g. a test) doesn't inherit it.
    _prev_data_root = os.environ.get("SEARCH_UI_DATA_ROOT")
    os.environ["SEARCH_UI_DATA_ROOT"] = str(data_root)
    try:
        return _seed_inner(
            data_root=data_root, language=language, clear=clear, force=force
        )
    finally:
        if _prev_data_root is None:
            os.environ.pop("SEARCH_UI_DATA_ROOT", None)
        else:
            os.environ["SEARCH_UI_DATA_ROOT"] = _prev_data_root


def _seed_inner(*, data_root: Path, language: str, clear: bool, force: bool) -> dict:
    import search  # noqa: E402  (imported after env setup, by design)

    search_db_path = os.path.join(str(data_root), "database.db")
    instance = search.SubtitleSearch(db_path=search_db_path)

    media_path = data_root / "json" / language / "all_media_items.json"
    if not force:
        real = _describe_real_data(instance, language, media_path)
        if real:
            raise RuntimeError(
                f"Refusing to seed: {data_root} already contains real data "
                f"({real}). Seeding would overwrite the media catalog and mix "
                f"fake videos into a real index.\n"
                f"Point --data-root at a throwaway directory (the default is "
                f"the gitignored ./sample-data), or pass --force if you really "
                f"mean to seed this location."
            )

    if clear:
        # Scope the clear to the SAMPLE keys only. Using the global
        # clear_language() here would wipe a developer's entire real
        # subtitle index for this language when run against the default
        # backend/ data root — a destructive footgun. Removing just the
        # sample rows gives a clean re-seed without touching real content.
        for item in SAMPLE_ITEMS:
            instance.remove_subtitle(item["natural_key"], language)
            instance.remove_scripture_references_for_video(item["natural_key"], language)
        print(f"Cleared {len(SAMPLE_ITEMS)} sample subtitle rows for language {language}.")

    conn = instance._get_db_connection()
    indexed_subtitles = 0
    try:
        for item in SAMPLE_ITEMS:
            conn.execute(
                "DELETE FROM subtitles_fts WHERE natural_key = ? AND language = ?",
                (item["natural_key"], language),
            )
            conn.execute(
                "INSERT INTO subtitles_fts (natural_key, language, content) VALUES (?, ?, ?)",
                (item["natural_key"], language, item["subtitles"]),
            )
            indexed_subtitles += 1
        conn.commit()
    finally:
        conn.close()

    # Scripture references (model-free regex parse + index).
    scripture_refs = 0
    for item in SAMPLE_ITEMS:
        scripture_refs += instance.index_scripture_from_text(
            item["natural_key"], language, item["subtitles"]
        )

    # Media metadata for title search + result decoration. (The guard above
    # has already ensured media_path holds no real catalog entries.)
    media = {item["natural_key"]: _build_media_record(item) for item in SAMPLE_ITEMS}
    media_path.parent.mkdir(parents=True, exist_ok=True)
    with open(media_path, "w", encoding="utf-8") as handle:
        json.dump(media, handle, indent=2)

    return {
        "data_root": str(data_root),
        "language": language,
        "media_items": len(media),
        "subtitles_indexed": indexed_subtitles,
        "scripture_references": scripture_refs,
        "search_db": search_db_path,
        "media_json": str(media_path),
    }


def main() -> int:
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument(
        "--data-root",
        default=str(REPO_ROOT / "sample-data"),
        help=(
            "Where to write the sample DB + json. Defaults to the gitignored "
            "./sample-data so it never touches a real database. Pass an "
            "explicit path to target elsewhere."
        ),
    )
    parser.add_argument("--language", default="E", help="Language code (default: E).")
    parser.add_argument(
        "--clear",
        action="store_true",
        help="Remove the sample rows first (scoped to sample keys only).",
    )
    parser.add_argument(
        "--force",
        action="store_true",
        help=(
            "Seed even if the target data root already contains real "
            "(non-sample) content. Dangerous — only for deliberate use."
        ),
    )
    args = parser.parse_args()

    data_root = Path(args.data_root).resolve()
    data_root.mkdir(parents=True, exist_ok=True)

    summary = seed(
        data_root=data_root,
        language=args.language,
        clear=args.clear,
        force=args.force,
    )

    print("Seeded sample content:")
    for key, value in summary.items():
        print(f"  {key:22} {value}")
    print(
        "\nKeyword, title, and scripture search now return results. "
        "Semantic/hybrid search still needs the embedding model."
    )
    return 0


if __name__ == "__main__":
    raise SystemExit(main())