Initial commit: intercompany ledger app

src/bank_importer/reader.py

@@ -0,0 +1,75 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import datetime
+from pathlib import Path
+from typing import Any
+
+
+class UnsupportedWorkbookError(RuntimeError):
+    pass
+
+
+@dataclass(frozen=True)
+class RawSheet:
+    name: str
+    rows: tuple[tuple[Any, ...], ...]
+
+
+def read_workbook(path: Path) -> tuple[RawSheet, ...]:
+    suffix = path.suffix.lower()
+    if suffix == ".xlsx":
+        return _read_xlsx(path)
+    if suffix == ".xls":
+        return _read_xls(path)
+    raise UnsupportedWorkbookError(f"Unsupported workbook type: {suffix}")
+
+
+def _read_xlsx(path: Path) -> tuple[RawSheet, ...]:
+    try:
+        from openpyxl import load_workbook
+    except ImportError as exc:
+        raise UnsupportedWorkbookError(
+            "Reading .xlsx files requires openpyxl. Install requirements.txt."
+        ) from exc
+
+    workbook = load_workbook(path, read_only=True, data_only=True)
+    try:
+        return tuple(
+            RawSheet(
+                name=worksheet.title,
+                rows=tuple(tuple(row) for row in worksheet.iter_rows(values_only=True)),
+            )
+            for worksheet in workbook.worksheets
+        )
+    finally:
+        workbook.close()
+
+
+def _read_xls(path: Path) -> tuple[RawSheet, ...]:
+    try:
+        import xlrd
+    except ImportError as exc:
+        raise UnsupportedWorkbookError(
+            "Reading .xls files requires xlrd. Install requirements.txt."
+        ) from exc
+
+    workbook = xlrd.open_workbook(path, on_demand=True)
+    sheets: list[RawSheet] = []
+    try:
+        for sheet_name in workbook.sheet_names():
+            worksheet = workbook.sheet_by_name(sheet_name)
+            rows: list[tuple[Any, ...]] = []
+            for row_index in range(worksheet.nrows):
+                values: list[Any] = []
+                for column_index in range(worksheet.ncols):
+                    cell = worksheet.cell(row_index, column_index)
+                    value = cell.value
+                    if cell.ctype == xlrd.XL_CELL_DATE:
+                        value = xlrd.xldate_as_datetime(value, workbook.datemode)
+                    values.append(value)
+                rows.append(tuple(values))
+            sheets.append(RawSheet(name=sheet_name, rows=tuple(rows)))
+    finally:
+        workbook.release_resources()
+    return tuple(sheets)