feat: enhance error handling and configuration validation

- Added a global exception handler to log unhandled exceptions and return a generic 500 JSON response without exposing details to the client.
- Updated the configuration to validate the `DATABASE_URL` format, ensuring it starts with `sqlite://` or `postgresql://`, and log warnings for invalid formats.
- Introduced safe parsing for numeric environment variables (`HTTP_PORT`, `INIT_DATA_MAX_AGE_SECONDS`) with defaults on invalid values, including logging warnings for out-of-range values.
- Enhanced the duty schedule parser to enforce limits on the number of schedule rows and the length of full names and duty strings, raising appropriate errors when exceeded.
- Updated internationalization messages to include generic error responses for import failures and parsing issues, improving user experience.
- Added unit tests to verify the new error handling and configuration validation behaviors.

duty_teller/importers/duty_schedule.py

@@ -9,6 +9,11 @@ DUTY_MARKERS = frozenset({"б", "Б", "в", "В"})
 UNAVAILABLE_MARKER = "Н"
 VACATION_MARKER = "О"
 
+# Limits to avoid abuse and unreasonable input.
+MAX_SCHEDULE_ROWS = 500
+MAX_FULL_NAME_LENGTH = 200
+MAX_DUTY_STRING_LENGTH = 10000
+
 
 @dataclass
 class DutyScheduleEntry:
@@ -69,10 +74,24 @@ def parse_duty_schedule(raw_bytes: bytes) -> DutyScheduleResult:
     except ValueError as e:
         raise DutyScheduleParseError(f"Invalid meta.start_date: {start_str}") from e
 
+    # Reject dates outside current year ± 1.
+    today = date.today()
+    min_year = today.year - 1
+    max_year = today.year + 1
+    if not (min_year <= start_date.year <= max_year):
+        raise DutyScheduleParseError(
+            f"meta.start_date year must be between {min_year} and {max_year}"
+        )
+
     schedule = data.get("schedule")
     if not isinstance(schedule, list):
         raise DutyScheduleParseError("Missing or invalid 'schedule' (must be array)")
 
+    if len(schedule) > MAX_SCHEDULE_ROWS:
+        raise DutyScheduleParseError(
+            f"schedule has too many rows (max {MAX_SCHEDULE_ROWS})"
+        )
+
     max_days = 0
     entries: list[DutyScheduleEntry] = []
 
@@ -85,12 +104,20 @@ def parse_duty_schedule(raw_bytes: bytes) -> DutyScheduleResult:
         full_name = name.strip()
         if not full_name:
             raise DutyScheduleParseError("schedule item 'name' cannot be empty")
+        if len(full_name) > MAX_FULL_NAME_LENGTH:
+            raise DutyScheduleParseError(
+                f"schedule item 'name' must not exceed {MAX_FULL_NAME_LENGTH} characters"
+            )
 
         duty_str = row.get("duty")
         if duty_str is None:
             duty_str = ""
         if not isinstance(duty_str, str):
             raise DutyScheduleParseError("schedule item 'duty' must be string")
+        if len(duty_str) > MAX_DUTY_STRING_LENGTH:
+            raise DutyScheduleParseError(
+                f"schedule item 'duty' must not exceed {MAX_DUTY_STRING_LENGTH} characters"
+            )
 
         cells = [c.strip() for c in duty_str.split(";")]
         max_days = max(max_days, len(cells))
@@ -120,4 +147,9 @@ def parse_duty_schedule(raw_bytes: bytes) -> DutyScheduleResult:
     else:
         end_date = start_date + timedelta(days=max_days - 1)
 
+    if not (min_year <= end_date.year <= max_year):
+        raise DutyScheduleParseError(
+            f"Computed end_date year must be between {min_year} and {max_year}"
+        )
+
     return DutyScheduleResult(start_date=start_date, end_date=end_date, entries=entries)