stack/tests/bib/test_store.py

"""Tests for bib.store — SQLite-backed bibliography store."""

from __future__ import annotations

from pathlib import Path

import pytest

from bib.client import connect
from bib.item import (
    Download,
    Manual,
    Regulation,
    Rule,
    Source,
)
from bib.store import Store, _generate_key

# ── client.connect ─────────────────────────────────────────────────


class TestConnect:
    def test_connect_returns_store(self) -> None:
        """client.py line 55: connect() returns Store."""
        s = connect(":memory:")
        assert isinstance(s, Store)
        s.close()


# ── Key generation ──────────────────────────────────────────────


class TestGenerateKey:
    def test_length(self) -> None:
        key = _generate_key()
        assert len(key) == 8

    def test_alphanumeric(self) -> None:
        key = _generate_key()
        assert key.isalnum()

    def test_uppercase(self) -> None:
        key = _generate_key()
        assert key == key.upper()

    def test_unique(self) -> None:
        keys = {_generate_key() for _ in range(100)}
        assert len(keys) > 90


# ── Store initialization ────────────────────────────────────────


class TestStoreInit:
    def test_in_memory(self) -> None:
        s = Store(":memory:")
        assert s._db_path == ":memory:"
        s.close()

    def test_file_based(self, tmp_path: Path) -> None:
        db = tmp_path / "bib.sqlite"
        s = Store(db)
        assert Path(s._db_path) == db
        assert db.exists()
        s.close()

    def test_storage_dir_default(self, tmp_path: Path) -> None:
        db = tmp_path / "bib.sqlite"
        s = Store(db)
        assert s._storage == tmp_path / "storage"
        s.close()

    def test_storage_dir_custom(self, tmp_path: Path) -> None:
        db = tmp_path / "bib.sqlite"
        custom = tmp_path / "my_files"
        s = Store(db, storage_dir=custom)
        assert s._storage == custom
        s.close()

    def test_storage_dir_memory_default(self) -> None:
        s = Store(":memory:")
        assert s._storage == Path("storage")
        s.close()

    def test_schema_created(self, tmp_path: Path) -> None:
        db = tmp_path / "bib.sqlite"
        s = Store(db)
        con = s._con()
        tables = con.execute(
            "SELECT name FROM sqlite_master WHERE type='table' ORDER BY name"
        ).fetchall()
        names = {r["name"] for r in tables}
        expected = {
            "items",
            "tags",
            "item_tags",
            "collections",
            "collection_items",
            "attachments",
            "notes",
            "creators",
            "item_creators",
            "bib_meta",
        }
        assert expected.issubset(names)
        s.close()

    def test_close_idempotent(self, tmp_path: Path) -> None:
        db = tmp_path / "bib.sqlite"
        s = Store(db)
        s.close()
        s.close()  # should not raise


# ── Item CRUD ───────────────────────────────────────────────────


class TestCreate:
    def test_create_returns_key(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="CY 2026 PFS Final Rule"))
        assert isinstance(key, str)
        assert len(key) == 8
        s.close()

    def test_create_uses_supplied_key(self) -> None:
        s = Store(":memory:")
        item = Rule(key="MYKEY123", title="Test Rule")
        key = s.create(item)
        assert key == "MYKEY123"
        s.close()

    def test_create_with_tags(self) -> None:
        s = Store(":memory:")
        key = s.create(
            Rule(title="Test"),
            tags=["module:pfs", "year:2026"],
        )
        item = s.get(key)
        assert "module:pfs" in item.tags
        assert "year:2026" in item.tags
        s.close()

    def test_create_stamps_access(self) -> None:
        s = Store(":memory:")
        rule = Rule(title="Test")
        assert rule.access_date == ""
        key = s.create(rule)
        item = s.get(key)
        assert item.access_date != ""
        s.close()


class TestGet:
    def test_get_round_trip(self) -> None:
        s = Store(":memory:")
        key = s.create(
            Rule(
                title="PFS Final Rule",
                fr_volume="90",
                fr_page="98452",
            )
        )
        item = s.get(key)
        assert isinstance(item, Rule)
        assert item.title == "PFS Final Rule"
        assert item.fr_volume == "90"
        s.close()

    def test_get_missing_raises(self) -> None:
        s = Store(":memory:")
        with pytest.raises(KeyError, match="Item not found"):
            s.get("NOEXIST1")
        s.close()

    def test_get_preserves_item_type(self) -> None:
        s = Store(":memory:")
        for cls, kwargs in [
            (Rule, {"title": "R"}),
            (Manual, {"title": "M"}),
            (Regulation, {"title": "Reg"}),
            (Download, {"title": "D"}),
            (Source, {"title": "S"}),
        ]:
            key = s.create(cls(**kwargs))
            item = s.get(key)
            assert isinstance(item, cls)
        s.close()


class TestUpdate:
    def test_update_title(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="Old Title"))
        s.update(key, title="New Title")
        assert s.get(key).title == "New Title"
        s.close()

    def test_update_no_fields_noop(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="Test"))
        s.update(key)  # should not raise
        assert s.get(key).title == "Test"
        s.close()

    def test_update_tags(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="Test"), tags=["old"])
        s.update(key, tags=["new1", "new2"])
        item = s.get(key)
        assert "new1" in item.tags
        assert "new2" in item.tags
        assert "old" not in item.tags
        s.close()

    def test_update_missing_raises(self) -> None:
        s = Store(":memory:")
        with pytest.raises(KeyError):
            s.update("NOEXIST1", tags=["x"])
        s.close()


class TestDelete:
    def test_delete_removes_item(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="To Delete"))
        s.delete(key)
        with pytest.raises(KeyError):
            s.get(key)
        s.close()

    def test_delete_nonexistent_no_error(self) -> None:
        s = Store(":memory:")
        s.delete("NOEXIST1")  # should not raise
        s.close()


class TestUpsert:
    def test_upsert_creates_new(self) -> None:
        s = Store(":memory:")
        item = Rule(
            title="New Rule",
            url="https://example.com/rule1",
        )
        key = s.upsert(item)
        assert s.get(key).title == "New Rule"
        s.close()

    def test_upsert_updates_existing_by_url(self) -> None:
        s = Store(":memory:")
        item1 = Rule(
            title="First",
            url="https://example.com/rule1",
        )
        key1 = s.upsert(item1)
        item2 = Rule(
            title="Updated",
            url="https://example.com/rule1",
        )
        key2 = s.upsert(item2)
        assert key1 == key2
        assert s.get(key2).title == "Updated"
        s.close()

    def test_upsert_no_url_always_creates(self) -> None:
        s = Store(":memory:")
        k1 = s.upsert(Rule(title="A"))
        k2 = s.upsert(Rule(title="B"))
        assert k1 != k2
        s.close()

    def test_upsert_with_tags(self) -> None:
        s = Store(":memory:")
        item = Rule(
            title="Test",
            url="https://example.com/t",
        )
        key = s.upsert(item, tags=["module:pfs"])
        got = s.get(key)
        assert "module:pfs" in got.tags
        s.close()

    def test_upsert_with_collection(self) -> None:
        s = Store(":memory:")
        coll_map = s.ensure_collections({"Rules": {}})
        coll_key = coll_map["Rules"]
        item = Rule(
            title="Test",
            url="https://example.com/c",
        )
        key = s.upsert(item, collection=coll_key)
        got = s.get(key)
        assert coll_key in got.collections
        s.close()


# ── Query / list ────────────────────────────────────────────────


class TestListItems:
    @pytest.fixture
    def populated_store(self) -> Store:
        s = Store(":memory:")
        s.create(
            Rule(title="PFS Rule"),
            tags=["module:pfs", "year:2026"],
        )
        s.create(
            Manual(title="CPM Chapter 12"),
            tags=["module:pfs"],
        )
        s.create(
            Download(title="RVU Files"),
            tags=["module:pfs", "file:rvu"],
        )
        s.create(
            Source(title="Guidance Doc"),
            tags=["year:2025"],
        )
        return s

    def test_list_all(self, populated_store: Store) -> None:
        items = populated_store.list_items()
        assert len(items) == 4
        populated_store.close()

    def test_list_by_tag(self, populated_store: Store) -> None:
        items = populated_store.list_items(tag="year:2026")
        assert len(items) == 1
        assert items[0].title == "PFS Rule"
        populated_store.close()

    def test_list_by_item_type(self, populated_store: Store) -> None:
        items = populated_store.list_items(item_type="rule")
        assert len(items) == 1
        assert isinstance(items[0], Rule)
        populated_store.close()

    def test_list_by_query_title(self, populated_store: Store) -> None:
        items = populated_store.list_items(query="RVU")
        assert len(items) == 1
        assert items[0].title == "RVU Files"
        populated_store.close()

    def test_list_no_results(self, populated_store: Store) -> None:
        items = populated_store.list_items(tag="nonexistent")
        assert items == []
        populated_store.close()

    def test_count(self, populated_store: Store) -> None:
        assert populated_store.count() == 4
        assert populated_store.count(tag="module:pfs") == 3
        populated_store.close()

    def test_to_dataframe(self, populated_store: Store) -> None:
        import polars as pl

        df = populated_store.to_dataframe()
        assert isinstance(df, pl.DataFrame)
        assert len(df) == 4
        assert "key" in df.columns
        assert "title" in df.columns
        assert "tags" in df.columns
        populated_store.close()

    def test_to_dataframe_with_filter(self, populated_store: Store) -> None:
        df = populated_store.to_dataframe(tag="year:2026")
        assert len(df) == 1
        populated_store.close()


# ── Tags ────────────────────────────────────────────────────────


class TestTags:
    def test_add_tag(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="Test"))
        s.add_tag(key, "new-tag")
        item = s.get(key)
        assert "new-tag" in item.tags
        s.close()

    def test_add_tag_idempotent(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="Test"), tags=["t1"])
        s.add_tag(key, "t1")
        item = s.get(key)
        assert item.tags.count("t1") == 1
        s.close()

    def test_add_tag_missing_item(self) -> None:
        s = Store(":memory:")
        with pytest.raises(KeyError, match="Item not found"):
            s.add_tag("NOEXIST1", "tag")
        s.close()

    def test_remove_tag(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="Test"), tags=["keep", "remove"])
        s.remove_tag(key, "remove")
        item = s.get(key)
        assert "keep" in item.tags
        assert "remove" not in item.tags
        s.close()

    def test_remove_tag_nonexistent_noop(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="Test"))
        s.remove_tag(key, "nope")  # should not raise
        s.close()

    def test_list_tags(self) -> None:
        s = Store(":memory:")
        s.create(Rule(title="A"), tags=["module:pfs"])
        s.create(Rule(title="B"), tags=["module:pfs"])
        s.create(
            Rule(title="C"),
            tags=["module:aco", "year:2026"],
        )
        tags = s.list_tags()
        names = {t["name"] for t in tags}
        assert "module:pfs" in names
        assert "module:aco" in names
        assert "year:2026" in names
        pfs_tag = next(t for t in tags if t["name"] == "module:pfs")
        assert pfs_tag["count"] == 2
        s.close()

    def test_list_tags_namespace(self) -> None:
        s = Store(":memory:")
        s.create(
            Rule(title="A"),
            tags=["module:pfs", "year:2026"],
        )
        s.create(Rule(title="B"), tags=["module:aco"])
        tags = s.list_tags(namespace="module")
        names = [t["name"] for t in tags]
        assert "module:pfs" in names
        assert "module:aco" in names
        # year tag should not appear
        assert all("year" not in n for n in names)
        s.close()

    def test_list_tags_empty(self) -> None:
        s = Store(":memory:")
        assert s.list_tags() == []
        s.close()


# ── Collections ─────────────────────────────────────────────────


class TestCollections:
    def test_ensure_collections(self) -> None:
        s = Store(":memory:")
        result = s.ensure_collections(
            {
                "Federal Register": {},
                "CMS Manuals": {},
            }
        )
        assert "Federal Register" in result
        assert "CMS Manuals" in result
        assert len(result["Federal Register"]) == 8
        s.close()

    def test_ensure_collections_idempotent(self) -> None:
        s = Store(":memory:")
        r1 = s.ensure_collections({"Rules": {}})
        r2 = s.ensure_collections({"Rules": {}})
        assert r1["Rules"] == r2["Rules"]
        s.close()

    def test_ensure_nested_collections(self) -> None:
        s = Store(":memory:")
        result = s.ensure_collections(
            {
                "CMS": {
                    "Rules": {},
                    "Manuals": {},
                },
            }
        )
        assert "CMS" in result
        assert "Rules" in result
        assert "Manuals" in result
        s.close()

    def test_list_collections(self) -> None:
        s = Store(":memory:")
        s.ensure_collections({"Alpha": {}, "Beta": {}})
        colls = s.list_collections()
        names = [c["name"] for c in colls]
        assert "Alpha" in names
        assert "Beta" in names
        s.close()

    def test_list_collections_empty(self) -> None:
        s = Store(":memory:")
        assert s.list_collections() == []
        s.close()

    def test_list_collections_item_count(self) -> None:
        s = Store(":memory:")
        coll_map = s.ensure_collections({"Rules": {}})
        coll_key = coll_map["Rules"]
        s.create(
            Rule(title="R1"),
            collection=coll_key,
        )
        colls = s.list_collections()
        rules_coll = next(c for c in colls if c["name"] == "Rules")
        assert rules_coll["item_count"] == 1
        s.close()

    def test_list_items_by_collection(self) -> None:
        s = Store(":memory:")
        coll_map = s.ensure_collections({"Rules": {}})
        coll_key = coll_map["Rules"]
        s.create(Rule(title="In"), collection=coll_key)
        s.create(Rule(title="Out"))
        items = s.list_items(collection=coll_key)
        assert len(items) == 1
        assert items[0].title == "In"
        s.close()


# ── Attachments ─────────────────────────────────────────────────


class TestAttachments:
    def test_attach_file(self, tmp_path: Path) -> None:
        db = tmp_path / "bib.sqlite"
        s = Store(db, storage_dir=tmp_path / "storage")
        key = s.create(Rule(title="Test"))
        src = tmp_path / "doc.pdf"
        src.write_bytes(b"fake pdf content")
        att_key = s.attach_file(key, src)
        assert isinstance(att_key, str)
        assert len(att_key) == 8
        dest = tmp_path / "storage" / att_key / "doc.pdf"
        assert dest.exists()
        s.close()

    def test_attach_file_missing_item(self, tmp_path: Path) -> None:
        db = tmp_path / "bib.sqlite"
        s = Store(db)
        src = tmp_path / "doc.pdf"
        src.write_bytes(b"data")
        with pytest.raises(KeyError, match="Item not found"):
            s.attach_file("NOEXIST1", src)
        s.close()


# ── Notes ───────────────────────────────────────────────────────


class TestNotes:
    def test_attach_note(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="Test"))
        note_id = s.attach_note(key, "Some notes here", title="My Note")
        assert isinstance(note_id, int)
        assert note_id > 0
        s.close()

    def test_attach_note_missing_item(self) -> None:
        s = Store(":memory:")
        with pytest.raises(KeyError, match="Item not found"):
            s.attach_note("NOEXIST1", "content")
        s.close()


# ── Citation formatting ─────────────────────────────────────────


class TestCitationFormatting:
    def test_format_citation_returns_string(self) -> None:
        s = Store(":memory:")
        key = s.create(
            Rule(
                title="CY 2026 PFS Final Rule",
                institution="CMS",
                date_published="2025-11-01",
            )
        )
        citation = s.format_citation(key)
        assert isinstance(citation, str)
        assert len(citation) > 0
        s.close()

    def test_format_bibliography(self) -> None:
        s = Store(":memory:")
        k1 = s.create(
            Rule(
                title="Rule One",
                institution="CMS",
            )
        )
        k2 = s.create(
            Manual(
                title="Manual One",
                institution="CMS",
            )
        )
        bib = s.format_bibliography([k1, k2])
        assert isinstance(bib, str)
        assert "Rule One" in bib
        assert "Manual One" in bib
        s.close()


# ── __del__ ────────────────────────────────────────────────────────


class TestStoreDel:
    def test_del_closes_connection(self) -> None:
        s = Store(":memory:")
        s._con()  # force connection open
        assert s._connection is not None
        s.__del__()
        assert s._connection is None

    def test_del_on_closed_store(self) -> None:
        s = Store(":memory:")
        s.close()
        s.__del__()  # should not raise

    def test_del_catches_close_exception(self) -> None:
        """Lines 90-91: __del__ catches exceptions from close()."""
        from unittest.mock import patch

        s = Store(":memory:")
        s._con()  # force connection open
        with patch.object(s, "close", side_effect=RuntimeError("boom")):
            s.__del__()  # should not raise


class TestSyncTagsEmptyName:
    def test_empty_tag_name_skipped(self) -> None:
        """Line 399: _sync_tags skips empty tag names."""
        s = Store(":memory:")
        key = s.create(Rule(title="Test"), tags=["keep", "", "also-keep"])
        item = s.get(key)
        assert "keep" in item.tags
        assert "also-keep" in item.tags
        assert "" not in item.tags
        s.close()


# ── Upsert URL dedup with tags/collections ─────────────────────────


class TestUpsertDedup:
    def test_upsert_adds_tags_on_existing(self) -> None:
        """Lines 222-225: upsert with tags= on existing URL."""
        from bib.tag import Tag

        s = Store(":memory:")
        item1 = Rule(
            title="First",
            url="https://example.com/dup",
        )
        key = s.upsert(item1, tags=[Tag.module("pfs")])

        # Upsert same URL with Tag objects in tags=
        item2 = Rule(
            title="Updated",
            url="https://example.com/dup",
        )
        key2 = s.upsert(item2, tags=[Tag.year(2026)])
        assert key == key2

        got2 = s.get(key2)
        # item2 had no initial tags, then Tag.year(2026) was added via tags=
        assert "year:2026" in got2.tags
        assert got2.title == "Updated"
        s.close()

    def test_upsert_adds_collection_on_existing(self) -> None:
        """Line 227: upsert with collection= on existing URL."""
        s = Store(":memory:")
        coll_map = s.ensure_collections({"Rules": {}})
        coll_key = coll_map["Rules"]

        item1 = Rule(title="First", url="https://example.com/dup2")
        key = s.upsert(item1)

        item2 = Rule(title="Updated", url="https://example.com/dup2")
        key2 = s.upsert(item2, collection=coll_key)
        assert key == key2

        got = s.get(key2)
        assert coll_key in got.collections
        s.close()

    def test_upsert_with_string_tags_on_existing(self) -> None:
        """Ensure tags= with plain strings works on dedup path."""
        s = Store(":memory:")
        item1 = Rule(title="First", url="https://example.com/dup3")
        key = s.upsert(item1)

        item2 = Rule(title="Updated", url="https://example.com/dup3")
        key2 = s.upsert(item2, tags=["plain-tag"])
        assert key == key2

        got = s.get(key2)
        assert "plain-tag" in got.tags
        s.close()


# ── _sync_collections with unknown key ─────────────────────────────


class TestSyncCollections:
    def test_unknown_collection_key_skipped(self) -> None:
        s = Store(":memory:")
        key = s.create(Rule(title="Test"))
        # Get item_id
        con = s._con()
        row = con.execute("SELECT id FROM items WHERE key = ?", (key,)).fetchone()
        item_id = row["id"]

        # Should not raise when collection key doesn't exist
        s._sync_collections(item_id, ["NOEXIST1"])

        # Verify no collection_items were created
        ci = con.execute(
            "SELECT count(*) FROM collection_items WHERE item_id = ?",
            (item_id,),
        ).fetchone()
        assert ci[0] == 0
        s.close()