parser.py

"""
parser.py — row parsing and validation
Module D: Files + Error Handling
"""


# ── Custom exception ──────────────────────────────────────────────────────────

class ParseError(Exception):
    """Raised when a CSV row cannot be parsed into a valid log entry."""

    def __init__(self, row: dict, reason: str):
        self.row = row
        self.reason = reason
        super().__init__(f"Parse failed on {row}: {reason}")


# ── Constants ─────────────────────────────────────────────────────────────────

REQUIRED_FIELDS = {"path", "status", "duration_ms", "method", "timestamp"}
VALID_METHODS   = {"GET", "POST", "PUT", "DELETE", "PATCH", "HEAD", "OPTIONS"}


# ── Parsing ───────────────────────────────────────────────────────────────────

def parse_row(row: dict) -> dict:
    """
    Parse a raw CSV row dict into a typed log entry.

    Args:
        row: Raw dict from csv.DictReader.

    Returns:
        Dict with typed fields: endpoint, status, ms, method, timestamp.

    Raises:
        ParseError: If any required field is missing or has an invalid type.
    """
    try:
        return {
            "endpoint":  row["path"],
            "status":    int(row["status"]),
            "ms":        float(row["duration_ms"]),
            "method":    row["method"].upper(),
            "timestamp": row["timestamp"],
        }
    except KeyError as e:
        raise ParseError(row, f"Missing field: {e}") from e
    except ValueError as e:
        raise ParseError(row, f"Invalid value: {e}") from e


def validate_row(parsed: dict) -> None:
    """
    Validate a parsed log entry for business rules.

    Args:
        parsed: Dict returned by parse_row().

    Raises:
        ParseError: If any field fails validation.
    """
    if not (100 <= parsed["status"] <= 599):
        raise ParseError(parsed, f"Invalid HTTP status: {parsed['status']}")

    if parsed["ms"] < 0:
        raise ParseError(parsed, f"Negative duration: {parsed['ms']}")

    if parsed["method"] not in VALID_METHODS:
        raise ParseError(parsed, f"Unknown HTTP method: {parsed['method']}")

    if not parsed["endpoint"].startswith("/"):
        raise ParseError(parsed, f"Endpoint must start with /: {parsed['endpoint']}")