Validate KYC CSV Records

Validate KYC CSV Records

When onboarding new businesses, Stripe verifies business identity by matching submitted application details against official registry records. You are tasked with writing a validation parser for a CSV export of onboarding records.

Write a function validate_kyc_records(csv_data: str) -> list[dict] that parses the CSV text and returns validation results for each record.

Input Format

csv_data is a multi-line string containing a CSV header followed by zero or more data rows. The header row is always:

record_id,business_name,tax_id,legal_owner_name,submitted_owner_name

Fields may contain leading or trailing whitespace that must be stripped.

Validation Rules

Evaluate each row against the following rules in order:

1. Missing Fields: All 5 fields must be present and non-empty after trimming whitespace. If any field is empty, flag error "MISSING_REQUIRED_FIELD".
2. Tax ID: A valid tax_id must consist of exactly 9 digits after stripping hyphens (-) and spaces. (e.g., "12-3456789" and " 123456789 " are valid; "12-345" or "12345678A" are invalid). If invalid, flag error "INVALID_TAX_ID".
3. Owner Name Match: Compare legal_owner_name and submitted_owner_name using token overlap:
   • Normalize both names by converting to lowercase, removing all punctuation/symbols (keep only letters, digits, and spaces), and splitting on whitespace into sets of unique words.
   • Let S be the set of unique words in submitted_owner_name and L be the set of unique words in legal_owner_name.
   • A name is considered a match if at least one word overlaps (len(S & L) >= 1) and the overlap covers at least 50% of the words submitted (len(S & L) / len(S) >= 0.5). If not, flag error "OWNER_NAME_MISMATCH".

A record may accumulate multiple distinct errors if it violates multiple rules (except MISSING_REQUIRED_FIELD skips rules 2 and 3 if the corresponding field is empty).

Output Format

Return a list of dictionaries, one per record in the order they appear:

[
    {
        "record_id": str,      # trimmed record_id (or "" if missing)
        "is_valid": bool,       # True if errors is empty, else False
        "errors": list[str]     # list of error strings in the order defined
    },
    ...
]

Examples

csv_input = """record_id,business_name,tax_id,legal_owner_name,submitted_owner_name
rec_1,Acme Corp,12-3456789,Jane Doe,Jane Doe
rec_2,Beta LLC,99-999,Robert Smith,Bob Smith
rec_3,Gamma Inc,11-2233445,Alice M. Walker,Alice Walker
"""

validate_kyc_records(csv_input)
# Returns:
# [
#     {"record_id": "rec_1", "is_valid": True, "errors": []},
#     {"record_id": "rec_2", "is_valid": False, "errors": ["INVALID_TAX_ID", "OWNER_NAME_MISMATCH"]},
#     {"record_id": "rec_3", "is_valid": True, "errors": []}
# ]

Constraints

• csv_data contains between 1 and 10,000 rows.
• Field values do not contain commas or escaped quotes.