Extract Category model into src/models/category.py

src/ingest/b4w.py

@@ -4,7 +4,8 @@ from sqlalchemy.orm import Session
 
 from src.ingest._common import _reader_raw, _upsert
 from src.ingest.transactions import consolidate_transactions, map_b4w_to_transaction
-from src.models.b4w import B4wDb, B4wExport, Category, CategoryDb
+from src.models.b4w import B4wDb, B4wExport, Category
+from src.models.category import CategoryDb
 
 
 def reader_b4w(path: Path) -> list[B4wExport]:

src/ingest/comdirect.py

@@ -73,6 +73,10 @@ def _reader_raw_comdirect(path: Path) -> CsV2DictMulti:
     account_name: str | None = None
     header: list[str] = []
     result: CsV2DictMulti = {}
+    # Comdirect groups "new" transactions (marked with "neu") under a header
+    # line that carries only their Buchungstag; remember it until the next
+    # transaction row for the same account.
+    pending_book_date: dict[str, str] = {}
     for line in lines:
         if expect_next == EXPECT_LINE.ACCOUNT_NAME:
             matches = re.findall(PATTERN_ACCOUNT_NAME, line)
@@ -91,13 +95,18 @@ def _reader_raw_comdirect(path: Path) -> CsV2DictMulti:
                 expect_next = EXPECT_LINE.DATA
         elif expect_next == EXPECT_LINE.DATA:
             data = re.findall(PATTERN_DATA, line)
-            if data:
+            if not data:
-                assert account_name is not None
-                result[account_name].append(
-                    {column: value for column, value in zip(header, data)}
-                )
-            else:
                 expect_next = EXPECT_LINE.ACCOUNT_NAME
+            elif len(data) == 1:
+                # Group header line (single Buchungstag for the "neu" rows below).
+                assert account_name is not None
+                pending_book_date[account_name] = data[0]
+            else:
+                assert account_name is not None
+                row = {column: value for column, value in zip(header, data)}
+                if data[0] == "neu" and account_name in pending_book_date:
+                    row["Buchungstag"] = pending_book_date.pop(account_name)
+                result[account_name].append(row)
         else:
             expect_next = EXPECT_LINE.ACCOUNT_NAME
     return result

src/models/__init__.py

@@ -1,5 +1,5 @@
 from src.models.b4w import B4wExport as B4wExport
-from src.models.b4w import Category as Category
+from src.models.category import Category as Category
 from src.models.category_budget import CategoryBudgetDb as CategoryBudgetDb
 from src.models.comdirect import ComdirectCreditCard as ComdirectCreditCard
 from src.models.comdirect import ComdirectDepot as ComdirectDepot

src/models/b4w.py

@@ -2,18 +2,18 @@ from datetime import date
 from decimal import Decimal
 
 from pydantic import Field, computed_field, field_validator
-from sqlalchemy import Date, ForeignKey, Integer, String, UniqueConstraint
+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._common import (
-    BaseModelDb,
     ExportBaseModel,
     IngestdbBase,
     Sign,
     _hash,
     date2str,
 )
+from src.models.category import Category
 
 
 class B4wExport(ExportBaseModel):
@@ -87,57 +87,6 @@ class B4wExport(ExportBaseModel):
         return category
 
 
-class Category(ExportBaseModel):
-    name: str = Field(pattern=r"[^:\r\n]+")
-    parent: Category | None = None
-
-    @computed_field  # type: ignore[prop-decorator]
-    @property
-    def hash_id(self) -> str:
-        return _hash(self.name, self.parent)
-
-    @classmethod
-    def from_str(cls, s: str) -> Category | None:
-        if not s:
-            return None
-        components = s.split(":")
-        name = components[-1]
-        rest = ":".join(components[:-1])
-        return cls(name=name, parent=cls.from_str(rest))
-
-    def to_str(self) -> str | None:
-        if not self.parent:
-            return self.name
-        else:
-            return f"{self.parent.to_str()}:{self.name}"
-
-    def __str__(self):
-        return self.to_str()
-
-    def __repr__(self):
-        return f'Category.from_str("{self.__str__()}")'
-
-
-class CategoryDb(BaseModelDb):
-    __tablename__ = "category"
-    __table_args__ = (
-        UniqueConstraint("name", "parent_id", name="uq_category_name_parent"),
-    )
-
-    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
-    name: Mapped[str] = mapped_column(String, nullable=False)
-    parent_id: Mapped[int | None] = mapped_column(
-        ForeignKey("category.id"), nullable=True
-    )
-
-    # Relationships
-    parent = relationship(
-        "CategoryDb", remote_side="CategoryDb.id", back_populates="children"
-    )
-    children = relationship("CategoryDb", back_populates="parent")
-    budgets = relationship("CategoryBudgetDb", back_populates="category")
-
-
 class B4wDb(IngestdbBase):
     __tablename__ = "import_b4w"
 

src/models/category.py

@@ -0,0 +1,56 @@
+from pydantic import Field, computed_field
+from sqlalchemy import ForeignKey, String, UniqueConstraint
+from sqlalchemy.orm import Mapped, mapped_column, relationship
+
+from src.models._common import BaseModelDb, ExportBaseModel, _hash
+
+
+class Category(ExportBaseModel):
+    name: str = Field(pattern=r"[^:\r\n]+")
+    parent: Category | None = None
+
+    @computed_field  # type: ignore[prop-decorator]
+    @property
+    def hash_id(self) -> str:
+        return _hash(self.name, self.parent)
+
+    @classmethod
+    def from_str(cls, s: str) -> Category | None:
+        if not s:
+            return None
+        components = s.split(":")
+        name = components[-1]
+        rest = ":".join(components[:-1])
+        return cls(name=name, parent=cls.from_str(rest))
+
+    def to_str(self) -> str | None:
+        if not self.parent:
+            return self.name
+        else:
+            return f"{self.parent.to_str()}:{self.name}"
+
+    def __str__(self):
+        return self.to_str()
+
+    def __repr__(self):
+        return f'Category.from_str("{self.__str__()}")'
+
+
+class CategoryDb(BaseModelDb):
+    __tablename__ = "category"
+    __table_args__ = (
+        UniqueConstraint("name", "parent_id", name="uq_category_name_parent"),
+    )
+
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    name: Mapped[str] = mapped_column(String, nullable=False)
+    parent_id: Mapped[int | None] = mapped_column(
+        ForeignKey("category.id"), nullable=True
+    )
+
+    # Relationships
+    parent = relationship(
+        "CategoryDb", remote_side="CategoryDb.id", back_populates="children"
+    )
+    children = relationship("CategoryDb", back_populates="parent")
+    budgets = relationship("CategoryBudgetDb", back_populates="category")

tests/test_category_budget.py

@@ -4,7 +4,7 @@ from decimal import Decimal
 from sqlalchemy import inspect
 from sqlalchemy.orm import Session
 
-from src.models.b4w import CategoryDb
+from src.models.category import CategoryDb
 from src.models.category_budget import CategoryBudgetDb
 
 

tests/test_helper_db.py

@@ -3,7 +3,7 @@ from sqlalchemy.orm import Session
 
 from src.ingest.b4w import upsert_category
 from src.models import Category
-from src.models.b4w import CategoryDb
+from src.models.category import CategoryDb
 
 
 def test_upsert_category_returns_orm_object(db_session: Session):