Add source validation for files sheets columns and month markers

src/taetigkeitsnachweis/validator.py

@@ -1 +1,234 @@
 """Source and business rule validation."""
+
+from __future__ import annotations
+
+from datetime import datetime
+from pathlib import Path
+from typing import Any
+
+from openpyxl import load_workbook
+from openpyxl.utils.cell import column_index_from_string
+
+from .models import ValidationIssue
+
+
+def _to_text(value: Any) -> str:
+    if value is None:
+        return ""
+    return str(value).strip()
+
+
+def _to_float_or_none(value: Any) -> float | None:
+    text = _to_text(value)
+    if not text:
+        return None
+    text = text.replace(",", ".")
+    try:
+        return float(text)
+    except ValueError:
+        return None
+
+
+def _parse_month_marker(value: Any, prefix: str, date_format: str) -> datetime | None:
+    marker = _to_text(value)
+    if not marker or not marker.lower().startswith(prefix.lower()):
+        return None
+
+    date_text = marker[len(prefix) :].strip()
+    try:
+        return datetime.strptime(date_text, date_format)
+    except ValueError:
+        return None
+
+
+def validate_sources(
+    config: dict[str, Any],
+    report_month: int,
+    report_year: int,
+    project_root: Path,
+) -> list[ValidationIssue]:
+    issues: list[ValidationIssue] = []
+
+    paths_cfg = config["paths"]
+    files_cfg = config["files"]
+    sheets_cfg = config["sheets"]
+    allocation_cfg = config["allocation"]
+    allocation_cols = allocation_cfg["columns"]
+    month_cfg = allocation_cfg["month_marker"]
+
+    input_dir = project_root / paths_cfg["input_dir"]
+    template_dir = project_root / paths_cfg["template_dir"]
+
+    allocation_path = input_dir / files_cfg["allocation_workbook"]
+    masterdata_path = input_dir / files_cfg["masterdata_workbook"]
+    template_path = template_dir / files_cfg["template_workbook"]
+
+    for required_path in [allocation_path, masterdata_path, template_path]:
+        if not required_path.exists():
+            issues.append(
+                ValidationIssue(
+                    code="MISSING_FILE",
+                    message=f"Required file not found: {required_path}",
+                    severity="error",
+                )
+            )
+
+    if any(issue.code == "MISSING_FILE" for issue in issues):
+        return issues
+
+    allocation_wb = load_workbook(allocation_path, data_only=True)
+    master_wb = load_workbook(masterdata_path, data_only=True)
+
+    allocation_sheet_name = sheets_cfg["allocation"]
+    master_sheet_name = sheets_cfg["masterdata"]
+
+    if allocation_sheet_name not in allocation_wb.sheetnames:
+        issues.append(
+            ValidationIssue(
+                code="MISSING_SHEET",
+                message=f"Sheet '{allocation_sheet_name}' missing in {allocation_path.name}",
+                severity="error",
+            )
+        )
+
+    if master_sheet_name not in master_wb.sheetnames:
+        issues.append(
+            ValidationIssue(
+                code="MISSING_SHEET",
+                message=f"Sheet '{master_sheet_name}' missing in {masterdata_path.name}",
+                severity="error",
+            )
+        )
+
+    if any(issue.code == "MISSING_SHEET" for issue in issues):
+        allocation_wb.close()
+        master_wb.close()
+        return issues
+
+    allocation_ws = allocation_wb[allocation_sheet_name]
+    master_ws = master_wb[master_sheet_name]
+
+    required_allocation_columns = [
+        allocation_cols["employee_name"],
+        allocation_cols["contract_hours_weekly"],
+        allocation_cols["project_key"],
+        allocation_cols["first_month_column"],
+        allocation_cols["remarks"],
+    ]
+    max_allocation_col = allocation_ws.max_column
+    for col_name in required_allocation_columns:
+        if column_index_from_string(col_name) > max_allocation_col:
+            issues.append(
+                ValidationIssue(
+                    code="MISSING_COLUMN",
+                    message=f"Allocation column '{col_name}' is not present.",
+                    severity="error",
+                )
+            )
+
+    master_cols = config["masterdata"]["columns"]
+    required_master_columns = [
+        master_cols["id"],
+        master_cols["antragsnummer"],
+        master_cols["name"],
+        master_cols["qualifikation"],
+        master_cols["stellenanteil"],
+        master_cols["funktion"],
+        master_cols["taetigkeit"],
+    ]
+    max_master_col = master_ws.max_column
+    for col_name in required_master_columns:
+        if column_index_from_string(col_name) > max_master_col:
+            issues.append(
+                ValidationIssue(
+                    code="MISSING_COLUMN",
+                    message=f"Master-data column '{col_name}' is not present.",
+                    severity="error",
+                )
+            )
+
+    header_row = allocation_cfg["header_month_marker_row"]
+    first_month_col = column_index_from_string(allocation_cols["first_month_column"])
+    month_matches: list[int] = []
+    parseable_markers = 0
+
+    for col_idx in range(first_month_col, allocation_ws.max_column + 1):
+        marker_value = allocation_ws.cell(row=header_row, column=col_idx).value
+        parsed = _parse_month_marker(
+            marker_value,
+            month_cfg["prefix"],
+            month_cfg["date_format"],
+        )
+        if parsed is None:
+            continue
+
+        parseable_markers += 1
+        if parsed.month == report_month and parsed.year == report_year:
+            month_matches.append(col_idx)
+
+    if parseable_markers == 0:
+        issues.append(
+            ValidationIssue(
+                code="INVALID_MONTH_MARKERS",
+                message="No parseable month marker found in allocation header row.",
+                severity="error",
+            )
+        )
+
+    if len(month_matches) == 0:
+        issues.append(
+            ValidationIssue(
+                code="MONTH_NOT_FOUND",
+                message=f"Requested month {report_month:02d}/{report_year} not found in allocation sheet.",
+                severity="error",
+            )
+        )
+    elif len(month_matches) > 1:
+        issues.append(
+            ValidationIssue(
+                code="MONTH_AMBIGUOUS",
+                message=f"Requested month {report_month:02d}/{report_year} appears in multiple columns.",
+                severity="error",
+            )
+        )
+
+    if len(month_matches) == 1:
+        month_column = month_matches[0]
+        contract_col = column_index_from_string(allocation_cols["contract_hours_weekly"])
+        project_col = column_index_from_string(allocation_cols["project_key"])
+
+        for row_idx in range(header_row + 1, allocation_ws.max_row + 1):
+            project_key = _to_text(allocation_ws.cell(row=row_idx, column=project_col).value)
+            if not project_key:
+                continue
+
+            contract_value = _to_float_or_none(
+                allocation_ws.cell(row=row_idx, column=contract_col).value
+            )
+            if contract_value is None or contract_value <= 0:
+                issues.append(
+                    ValidationIssue(
+                        code="INVALID_CONTRACT_HOURS",
+                        message=f"Invalid contract weekly hours at row {row_idx}.",
+                        severity="error",
+                    )
+                )
+
+            monthly_value = allocation_ws.cell(row=row_idx, column=month_column).value
+            monthly_text = _to_text(monthly_value)
+            if monthly_text in {"", "-", "0"}:
+                continue
+
+            monthly_hours = _to_float_or_none(monthly_value)
+            if monthly_hours is None:
+                issues.append(
+                    ValidationIssue(
+                        code="INVALID_HOUR_VALUE",
+                        message=f"Invalid funded hour value at row {row_idx}.",
+                        severity="error",
+                    )
+                )
+
+    allocation_wb.close()
+    master_wb.close()
+    return issues