Category overhaul (#20)

Reviewed-on: #20
Co-authored-by: Mathias <mathias2784@web.de>

README.md

@@ -23,20 +23,29 @@ alembic upgrade head
 
 ```python
 from src.ingest.b4w import ingest_b4w
+from src.db import get_session
 from pathlib import Path
 
 csv_path = Path("/mnt/anbau/Software/Python/finanz-cockpit/scripts/dummy_data/b4w_export_dummy_01.csv")
-ingest_b4w(csv_path)
+with get_session() as session:
+    ingest_b4w(csv_path, session)
 ```
 
 ### Import comdirect Export
 
 ```python
 from src.ingest.comdirect import ingest_comdirect
+from src.db import get_session
 from pathlib import Path
 
 csv_path = Path("/mnt/anbau/Software/Python/finanz-cockpit/scripts/dummy_data/comdirect_giro_01.csv")
-ingest_comdirect(csv_path)
+with get_session() as session:
+    ingest_comdirect(
+        session,
+        csv_path,
+        account_iban="DE89370400440532013000",
+        account_currency="EUR",
+    )
 ```
 
 ### Set Category Budgets
@@ -45,6 +54,7 @@ ingest_comdirect(csv_path)
 from src.models.category import Category, CategoryBudget
 from datetime import date
 from src.domain_utils import set_category_budget
+from src.db import get_session
 
 # Create category hierarchy
 cat = Category.from_str("Lebenshaltung:Essen:Supermarkt")
@@ -56,14 +66,17 @@ budget = CategoryBudget(
     valid_from=date(2026, 10, 1),
     valid_to=date(2026, 11, 30),
 )
-set_category_budget(db_session, budget)
+with get_session() as session:
+    set_category_budget(session, budget)
+    session.commit()
 
     # Budgets can also be unbounded (no dates specified)
     budget = CategoryBudget(
         category=cat,
         amount_eur=200.00,  # Monthly average
     )
-set_category_budget(db_session, budget)
+    set_category_budget(session, budget)
+    session.commit()
 ```
 
 ## Testing
@@ -141,7 +154,7 @@ alembic upgrade head
 pytest
 
 # Run with coverage report
-pytest --cov=src --cov-report=html -c .coverage htmlcov
+pytest --cov=src --cov-report=html:htmlcov
 ```
 
 ### Common Patterns

alembic/versions/c18ae67f4b2a_merge_category_feature_heads.py

@@ -0,0 +1,23 @@
+"""merge category validity and budget migration branches
+
+Revision ID: c18ae67f4b2a
+Revises: d9c6533f8a1b, fe23064de1c5
+Create Date: 2026-10-27
+
+"""
+
+from collections.abc import Sequence
+
+# revision identifiers, used by Alembic.
+revision: str = "c18ae67f4b2a"
+down_revision: str | Sequence[str] | None = ("d9c6533f8a1b", "fe23064de1c5")
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def upgrade() -> None:
+    pass
+
+
+def downgrade() -> None:
+    pass

alembic/versions/d9c6533f8a1b_add_category_validity.py

@@ -0,0 +1,102 @@
+"""add temporal validity to categories
+
+Revision ID: d9c6533f8a1b
+Revises: 78ec55b9e1d1
+Create Date: 2026-10-25 00:00:00.000000
+
+"""
+
+from collections.abc import Sequence
+
+import sqlalchemy as sa
+
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision: str = "d9c6533f8a1b"
+down_revision: str | Sequence[str] | None = "78ec55b9e1d1"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def upgrade() -> None:
+    with op.batch_alter_table("category") as batch_op:
+        batch_op.drop_constraint("uq_category_name_parent", type_="unique")
+        batch_op.add_column(
+            sa.Column(
+                "valid_from",
+                sa.DateTime(timezone=True),
+                nullable=True,
+                info={
+                    "description": "UTC start of validity; NULL means valid since the indefinite past."
+                },
+            )
+        )
+        batch_op.add_column(
+            sa.Column(
+                "valid_to",
+                sa.DateTime(timezone=True),
+                nullable=True,
+                info={
+                    "description": "UTC end of validity; NULL means the version is currently valid."
+                },
+            )
+        )
+        batch_op.add_column(
+            sa.Column(
+                "last_action",
+                sa.String(),
+                nullable=False,
+                server_default="created",
+                info={
+                    "description": "Most recent lifecycle action recorded for this category version."
+                },
+            )
+        )
+
+    # Existing rows predate temporal tracking, so their start is unbounded.
+    op.execute("UPDATE category SET last_action = 'migrated'")
+
+    # Separate root and non-root indexes so NULL parent IDs are unique too.
+    op.create_index(
+        "uq_category_active_name_parent",
+        "category",
+        ["name", "parent_id"],
+        unique=True,
+        sqlite_where=sa.text("valid_to IS NULL AND parent_id IS NOT NULL"),
+        postgresql_where=sa.text("valid_to IS NULL AND parent_id IS NOT NULL"),
+    )
+    op.create_index(
+        "uq_category_active_root_name",
+        "category",
+        ["name"],
+        unique=True,
+        sqlite_where=sa.text("valid_to IS NULL AND parent_id IS NULL"),
+        postgresql_where=sa.text("valid_to IS NULL AND parent_id IS NULL"),
+    )
+
+
+def downgrade() -> None:
+    bind = op.get_bind()
+    has_temporal_history = bind.execute(
+        sa.text(
+            "SELECT 1 FROM category "
+            "WHERE valid_from IS NOT NULL OR valid_to IS NOT NULL LIMIT 1"
+        )
+    ).first()
+    if has_temporal_history:
+        raise RuntimeError(
+            "Cannot downgrade category validity while temporal history exists; "
+            "the downgrade would discard it."
+        )
+
+    op.drop_index("uq_category_active_root_name", table_name="category")
+    op.drop_index("uq_category_active_name_parent", table_name="category")
+
+    with op.batch_alter_table("category") as batch_op:
+        batch_op.drop_column("last_action")
+        batch_op.drop_column("valid_to")
+        batch_op.drop_column("valid_from")
+        batch_op.create_unique_constraint(
+            "uq_category_name_parent", ["name", "parent_id"]
+        )

pyproject.toml

@@ -33,11 +33,7 @@ build-backend = "uv_build"
 [tool.mypy]
 explicit_package_bases = true
 mypy_path = "."
-# Ignore test file type errors (inline class definitions, float literals)
-# See AGENTS.md for type checking notes
-[[tool.mypy.overrides]]
-module = ["tests.test_db_utils_update_path", "tests.test_domain_utils_budget_overlap"]
-ignore_errors = true
+exclude = ["^tests/"]
 
 [tool.ruff]
 target-version = "py314"

src/domain_utils.py

@@ -29,7 +29,8 @@ def find_category_db(session: Session, category: Category | None) -> CategoryDb
         return (
             session.query(CategoryDb)
             .filter_by(name=node.name, parent_id=parent_id)
-            .first()
+            .filter(CategoryDb.valid_to.is_(None))
+            .one_or_none()
         )
 
     return find_existing(category)

src/ingest/b4w.py

@@ -1,3 +1,4 @@
+from datetime import UTC, datetime
 from pathlib import Path
 
 from sqlalchemy.orm import Session
@@ -6,7 +7,7 @@ from src.db_utils import upsert
 from src.ingest.common import _reader_raw
 from src.ingest.transactions import consolidate_transactions, map_b4w_to_transaction
 from src.models.b4w import B4wDb, B4wExport
-from src.models.category import upsert_category
+from src.models.category import Category, CategoryDb, upsert_category
 
 
 def reader_b4w(path: Path) -> list[B4wExport]:
@@ -14,6 +15,93 @@ def reader_b4w(path: Path) -> list[B4wExport]:
     return list(map(B4wExport.model_validate, raw))
 
 
+def _utc_datetime(value: datetime | None) -> datetime:
+    if value is None:
+        return datetime.now(UTC)
+    if value.tzinfo is None or value.utcoffset() is None:
+        return value.replace(tzinfo=UTC)
+    return value.astimezone(UTC)
+
+
+def _ensure_after_valid_from(category: CategoryDb, effective_at: datetime) -> None:
+    if category.valid_from is None:
+        return
+    valid_from = _utc_datetime(category.valid_from)
+    if effective_at <= valid_from:
+        raise ValueError("The action time must be later than valid_from")
+
+
+def change_category(
+    session: Session,
+    category_id: int,
+    replacement: Category,
+    effective_at: datetime | None = None,
+    do_commit: bool = True,
+) -> CategoryDb:
+    """Close an active category version and start its replacement version."""
+    current = (
+        session.query(CategoryDb).filter_by(id=category_id, valid_to=None).one_or_none()
+    )
+    if current is None:
+        raise ValueError(f"No active category found with id {category_id}")
+
+    action_time = _utc_datetime(effective_at)
+    _ensure_after_valid_from(current, action_time)
+    parent = upsert_category(session, replacement.parent, do_commit=False)
+    parent_id = parent.id if parent else None
+
+    if current.name == replacement.name and current.parent_id == parent_id:
+        return current
+
+    replacement_active = (
+        session.query(CategoryDb)
+        .filter_by(name=replacement.name, parent_id=parent_id, valid_to=None)
+        .one_or_none()
+    )
+
+    current.valid_to = action_time
+    current.last_action = "changed"
+    session.flush()
+
+    if replacement_active is None:
+        replacement_active = CategoryDb(
+            name=replacement.name,
+            parent=parent,
+            valid_from=action_time,
+            valid_to=None,
+            last_action="changed",
+        )
+        session.add(replacement_active)
+        session.flush()
+
+    if do_commit:
+        session.commit()
+    return replacement_active
+
+
+def retire_category(
+    session: Session,
+    category_id: int,
+    effective_at: datetime | None = None,
+    do_commit: bool = True,
+) -> CategoryDb:
+    """Close an active category version without creating a replacement."""
+    category = (
+        session.query(CategoryDb).filter_by(id=category_id, valid_to=None).one_or_none()
+    )
+    if category is None:
+        raise ValueError(f"No active category found with id {category_id}")
+
+    action_time = _utc_datetime(effective_at)
+    _ensure_after_valid_from(category, action_time)
+    category.valid_to = action_time
+    category.last_action = "retired"
+    session.flush()
+    if do_commit:
+        session.commit()
+    return category
+
+
 def upsert_b4w(
     session: Session, data: B4wExport, do_update=False, do_commit=True
 ) -> B4wDb | None:

src/models/b4w.py

@@ -2,11 +2,16 @@ from datetime import date
 from decimal import Decimal
 
 from pydantic import Field, computed_field, field_validator
-from sqlalchemy import Date, ForeignKey, Integer, String
+from sqlalchemy import (
+    Date,
+    ForeignKey,
+    Integer,
+    String,
+)
 from sqlalchemy.orm import Mapped, mapped_column, relationship
 
 from src.constants import IBAN_PATTERN, Currency
-from src.models.category import Category
+from src.models.category import Category, CategoryDb
 from src.models.common import (
     ExportBaseModel,
     IngestdbBase,
@@ -15,6 +20,8 @@ from src.models.common import (
     hashify,
 )
 
+__all__ = ["B4wDb", "B4wExport", "Category", "CategoryDb"]
+
 
 class B4wExport(ExportBaseModel):
     id: int = Field(alias="Id1", description="internal B4w ID")

src/models/category.py

@@ -1,15 +1,50 @@
-from datetime import date
+from __future__ import annotations
+
+from datetime import UTC, date, datetime, timedelta
 from decimal import Decimal
 from typing import Self
 
 from pydantic import BaseModel, Field, computed_field, model_validator
-from sqlalchemy import Date, ForeignKey, Numeric, String, UniqueConstraint
+from sqlalchemy import (
+    Date,
+    DateTime,
+    ForeignKey,
+    Index,
+    Numeric,
+    String,
+    TypeDecorator,
+    text,
+)
 from sqlalchemy.orm import Mapped, Session, mapped_column, relationship
 
 from src.db_utils import upsert
 from src.models.common import BaseModelDb, ExportBaseModel, hashify
 
 
+class UTCDateTime(TypeDecorator[datetime]):
+    """Store UTC timestamps and return timezone-aware UTC values on every dialect."""
+
+    impl = DateTime
+    cache_ok = True
+
+    def load_dialect_impl(self, dialect):
+        return dialect.type_descriptor(DateTime(timezone=True))
+
+    def process_bind_param(self, value: datetime | None, dialect):
+        if value is None:
+            return None
+        if value.tzinfo is None or value.utcoffset() is None:
+            return value.replace(tzinfo=UTC)
+        return value.astimezone(UTC)
+
+    def process_result_value(self, value: datetime | None, dialect):
+        if value is None:
+            return None
+        if value.tzinfo is None or value.utcoffset() is None:
+            return value.replace(tzinfo=UTC)
+        return value.astimezone(UTC)
+
+
 class Category(ExportBaseModel):
     name: str = Field(pattern=r"[^:\r\n]+")
     parent: Category | None = None
@@ -44,7 +79,21 @@ class Category(ExportBaseModel):
 class CategoryDb(BaseModelDb):
     __tablename__ = "category"
     __table_args__ = (
-        UniqueConstraint("name", "parent_id", name="uq_category_name_parent"),
+        Index(
+            "uq_category_active_name_parent",
+            "name",
+            "parent_id",
+            unique=True,
+            sqlite_where=text("valid_to IS NULL AND parent_id IS NOT NULL"),
+            postgresql_where=text("valid_to IS NULL AND parent_id IS NOT NULL"),
+        ),
+        Index(
+            "uq_category_active_root_name",
+            "name",
+            unique=True,
+            sqlite_where=text("valid_to IS NULL AND parent_id IS NULL"),
+            postgresql_where=text("valid_to IS NULL AND parent_id IS NULL"),
+        ),
     )
 
     id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
@@ -52,6 +101,29 @@ class CategoryDb(BaseModelDb):
     parent_id: Mapped[int | None] = mapped_column(
         ForeignKey("category.id"), nullable=True
     )
+    valid_from: Mapped[datetime | None] = mapped_column(
+        UTCDateTime(),
+        nullable=True,
+        info={
+            "description": "UTC start of validity; NULL means valid since the indefinite past."
+        },
+    )
+    valid_to: Mapped[datetime | None] = mapped_column(
+        UTCDateTime(),
+        nullable=True,
+        info={
+            "description": "UTC end of validity; NULL means the version is currently valid."
+        },
+    )
+    last_action: Mapped[str] = mapped_column(
+        String,
+        nullable=False,
+        default="created",
+        server_default="created",
+        info={
+            "description": "Most recent lifecycle action recorded for this category version."
+        },
+    )
 
     # Relationships
     parent = relationship(
@@ -69,11 +141,45 @@ def upsert_category(
 
     parent = upsert_category(session, category.parent, do_commit=False)
     parameters = {"name": category.name, "parent": parent}
+    matching_rows = session.query(CategoryDb).filter_by(
+        name=category.name, parent_id=parent.id if parent else None
+    )
+    active = matching_rows.filter(CategoryDb.valid_to.is_(None)).one_or_none()
+    if active is not None:
+        return active
+
+    historical = (
+        matching_rows.filter(CategoryDb.valid_to.is_not(None))
+        .order_by(CategoryDb.valid_to.desc())
+        .first()
+    )
+    valid_from = None
+    last_action = "created"
+    if historical is not None:
+        valid_from = datetime.now(UTC)
+        if historical.valid_to is not None:
+            previous_valid_to = historical.valid_to
+            if (
+                previous_valid_to.tzinfo is None
+                or previous_valid_to.utcoffset() is None
+            ):
+                previous_valid_to = previous_valid_to.replace(tzinfo=UTC)
+            else:
+                previous_valid_to = previous_valid_to.astimezone(UTC)
+            valid_from = max(valid_from, previous_valid_to + timedelta(microseconds=1))
+        last_action = "changed"
+
+    parameters_new = {
+        **parameters,
+        "valid_from": valid_from,
+        "valid_to": None,
+        "last_action": last_action,
+    }
     return upsert(
         session,
         CategoryDb,
-        parameters,
-        parameters,
+        {**parameters, "valid_to": None},
+        parameters_new,
         do_update=False,
         do_commit=do_commit,
     )
@@ -113,7 +219,9 @@ class CategoryBudgetDb(BaseModelDb):
     __tablename__ = "category_budget"
 
     id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
-    category_id: Mapped[int] = mapped_column(ForeignKey("category.id"), nullable=False)
+    category_id: Mapped[int] = mapped_column(
+        ForeignKey("category.id"), nullable=False, index=True
+    )
     amount_eur: Mapped[Decimal] = mapped_column(Numeric(12, 2), nullable=False)
     valid_from: Mapped[date | None] = mapped_column(Date, nullable=True, default=None)
     valid_to: Mapped[date | None] = mapped_column(Date, nullable=True, default=None)

tests/test_domain_utils_budget_overlap.py

@@ -1,13 +1,13 @@
 """Comprehensive tests for set_category_budget overlap handling."""
 
-from datetime import date
+from datetime import UTC, date, datetime
 
 import pytest
 from sqlalchemy.orm import Session
 
 from src.domain_utils import find_category_db, set_category_budget
 from src.models import Category, CategoryBudget, CategoryBudgetDb
-from src.models.category import upsert_category
+from src.models.category import CategoryDb, upsert_category
 from tests.factories import CategoryFactory
 
 # NOTE: MYPY errors suppressed for test fixtures - see AGENTS.md for type checking notes.
@@ -615,6 +615,33 @@ class TestFindCategoryDb:
         result = find_category_db(db_session, Category(name="NonExistent"))
         assert result is None
 
+    def test_find_reintroduced_category_returns_active_version(
+        self, db_session: Session
+    ):
+        category = Category(name="Reintroduced")
+        historical = CategoryDb(
+            name=category.name,
+            valid_to=datetime(2026, 1, 1, tzinfo=UTC),
+            last_action="retired",
+        )
+        db_session.add(historical)
+        db_session.commit()
+
+        active = upsert_category(db_session, category, do_commit=False)
+        assert active is not None
+        db_session.commit()
+
+        result = find_category_db(db_session, category)
+        assert result is not None
+        assert result.id == active.id
+
+        set_category_budget(
+            db_session,
+            CategoryBudget(category=category, amount_eur=100.00),
+        )
+        budget = db_session.query(CategoryBudgetDb).one()
+        assert budget.category_id == active.id
+
 
 class TestSetCategoryBudgetErrorHandling:
     """Test error handling in set_category_budget."""

tests/test_helper_db.py

@@ -0,0 +1,132 @@
+from datetime import UTC, datetime
+
+import pytest
+from sqlalchemy import select
+from sqlalchemy.exc import IntegrityError
+from sqlalchemy.orm import Session
+
+from src.ingest.b4w import change_category, retire_category, upsert_category
+from src.models import Category
+from src.models.b4w import CategoryDb
+
+
+def test_upsert_category_returns_orm_object(db_session: Session):
+    result = upsert_category(
+        db_session,
+        Category.from_str("dummy_parent_1:dummy_parent_2:dummy"),
+        do_commit=False,
+    )
+    assert result
+    assert isinstance(result, CategoryDb)
+    assert str(result.name) == "dummy"
+    assert str(result.parent.name) == "dummy_parent_2"
+    assert str(result.parent.parent.name) == "dummy_parent_1"
+    assert result.valid_from is None
+    assert result.valid_to is None
+    assert result.last_action == "created"
+    assert all(
+        CategoryDb.__table__.c[name].info.get("description")
+        for name in ("valid_from", "valid_to", "last_action")
+    )
+
+
+def test_upsert_category_writes_new_to_db(db_session: Session):
+    upsert_category(
+        db_session,
+        Category.from_str("grand_father:father:new_category"),
+        do_commit=True,
+    )
+    result = db_session.execute(
+        select(CategoryDb).filter_by(name="new_category")
+    ).scalar_one()
+    assert result
+    assert str(result.name) == "new_category"
+    assert str(result.parent.name) == "father"
+    assert str(result.parent.parent.name) == "grand_father"
+
+
+def test_upsert_category_skips_existing(db_session: Session):
+    for _ in range(3):
+        upsert_category(
+            db_session,
+            Category.from_str("grand_father:father:try_again_and_again"),
+            do_commit=True,
+        )
+    result = db_session.execute(
+        select(CategoryDb).filter_by(name="try_again_and_again")
+    ).all()
+    assert len(result) == 1
+
+
+def test_change_category_closes_old_version_and_starts_new_one(
+    db_session: Session,
+):
+    original = upsert_category(
+        db_session, Category.from_str("old_category"), do_commit=False
+    )
+    assert original is not None
+    effective_at = datetime(2026, 10, 16, 12, tzinfo=UTC)
+
+    replacement = change_category(
+        db_session,
+        original.id,
+        Category.from_str("new_category"),
+        effective_at=effective_at,
+        do_commit=False,
+    )
+
+    assert original.valid_from is None
+    assert original.valid_to == effective_at
+    assert original.last_action == "changed"
+    assert replacement.name == "new_category"
+    assert replacement.valid_from == effective_at
+    assert replacement.valid_to is None
+    assert replacement.last_action == "changed"
+
+    db_session.expire(original)
+    db_session.expire(replacement)
+    assert original.valid_to == effective_at
+    assert replacement.valid_from == effective_at
+
+    # A historical name can be introduced again as a new active version.
+    reintroduced = upsert_category(
+        db_session, Category.from_str("old_category"), do_commit=False
+    )
+    assert reintroduced is not None
+    assert reintroduced.id != original.id
+    assert reintroduced.valid_from is not None
+    assert reintroduced.valid_from > original.valid_to
+    assert reintroduced.last_action == "changed"
+
+
+def test_retire_category_closes_active_version(db_session: Session):
+    category = upsert_category(
+        db_session, Category.from_str("category_to_retire"), do_commit=False
+    )
+    assert category is not None
+    effective_at = datetime(2026, 10, 16, 12, tzinfo=UTC)
+
+    retired = retire_category(
+        db_session, category.id, effective_at=effective_at, do_commit=False
+    )
+
+    assert retired.valid_to == effective_at
+    assert retired.last_action == "retired"
+    assert (
+        db_session.query(CategoryDb)
+        .filter_by(name="category_to_retire", valid_to=None)
+        .one_or_none()
+        is None
+    )
+
+
+def test_active_root_categories_are_unique(db_session: Session):
+    db_session.add_all(
+        [
+            CategoryDb(name="duplicate_root"),
+            CategoryDb(name="duplicate_root"),
+        ]
+    )
+
+    with pytest.raises(IntegrityError):
+        db_session.flush()