lanka-validations - v2.1.0
    Preparing search index...

    CHANGELOG

    Changelog

    All notable changes to this project are documented in this file.

    The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

    Documentation release: no changes to validation behavior or the public API. Upgrading from 2.0.0 needs no code changes.

    • llms.txt ships in the package and on the docs site, with llms-full.txt (llms.txt, README, and changelog in one file), so AI coding assistants can read the API, its conventions, and its limits.
    • Every public function documents its parameters and return value and has a @example, shown in editor hovers and the API reference. Each entry point (lanka-validations/nic, /phone, and so on) has an overview with its import path.
    • README: a Quick Start, a "Which Function Do I Use?" task table, examples for Zod, React Hook Form, and Express, and an FAQ.
    • API reference: each area is its own module page; header links to GitHub, npm, and the changelog; the changelog is published on the site; search covers descriptions.
    • The npm homepage points to the API reference.
    • README, llms.txt, and TSDoc examples now run as tests, so documented results always match the library.
    • Shortened the README's input-rules and performance sections and moved Data Sources near the end.
    • formatNic documentation showed the wrong output for the spaced-old and spaced-new styles. The function itself is unchanged.
    • The README's calculateAge example depended on the current date.
    • The bundle-size badge, which could not measure the package, now shows the lanka-validations/nic size; a Socket badge was added.

    This release hardens input validation. Some inputs that 1.0.0 accepted are now rejected, and several public types are stricter. Read Breaking changes before upgrading.

    • Node.js 18 is no longer supported; the minimum is Node.js 20. Node 18 reached end-of-life in April 2025.
    • ValidationResult<T> is now a discriminated union of ValidationSuccess<T> and ValidationFailure. Checking isValid narrows the result: on success data is always present; on failure error, errors and code are always present. Code that builds its own { isValid: boolean, ... } object and types it as ValidationResult no longer compiles.
    • Province and district getters return immutable copies typed as ReadonlyProvinceInfo / readonly records. Mutating a returned record is a type error, and mutation no longer affects later lookups.
    • All exported reference tables (MOBILE_PREFIXES, AREA_CODES, FIXED_OPERATOR_DIGITS, SRI_LANKA_SHORT_CODES, POSTAL_PREFIXES, MAJOR_CITIES, DMT_LICENSE_CATEGORIES, VEHICLE_PROVINCES, and the nested entries of VALIDATION_MESSAGES and REFERENCE_DATA_METADATA) are deeply frozen and typed Readonly. Writing to them is a type error and throws in strict mode. Previously, any code in the process could change validation results, for example by adding a mobile prefix.
    • Validators reject non-scalar input (arrays, objects) instead of coercing it to a string. Only strings and finite numbers are accepted. For example, isValidPhone(['0771234567']) now returns false.
    • Strings longer than MAX_INPUT_LENGTH (1024 UTF-16 code units) are rejected before normalization.
    • Date of birth parsing accepts only Date objects and the documented formats YYYY-MM-DD, YYYY/MM/DD, DD-MM-YYYY and DD/MM/YYYY. Free-form strings such as 'March 5 2000' are rejected. Date-only comparisons and ages use the UTC calendar date.
    • isValidTIN / validateTIN no longer accept an NIC number as a TIN. IRD issues a separate 9- or 10-digit TIN to individuals who register using their NIC; no IRD source supports treating the NIC itself as a TIN. Validate NIC fields with isValidNic instead.
    • Phone validation rejects strings containing letters or other disallowed characters (for example, 'abc119xyz' is no longer read as short code 119).
    • Postal code validation rejects letters and non-integer numbers (for example, 'abc20000xyz' and 200.5).
    • LKR validation rejects non-finite values, amounts above MAX_LKR_AMOUNT, and malformed digit grouping such as '1,2,3'.
    • createSchemaValidator().validate() returns a SCHEMA_INVALID_ROOT failure for non-object input (for example, null) instead of reporting it as valid.
    • lanka-validations/privacy subpath: redactSensitiveValue, maskNic, maskPhone, maskBankAccount, validationResultForLog and REDACTED_VALUE.
    • lanka-validations/i18n subpath: translateValidationCode, localizeValidationResult, localizeSchemaResult and VALIDATION_MESSAGES.
    • lanka-validations/reference-data subpath: getReferenceDataMetadata and REFERENCE_DATA_METADATA, describing the source, version, coverage and limitations of each bundled data table.
    • Exact LKR minor-unit (cents) handling: parseLKRMinorUnits, formatLKRMinorUnits, validateLKRDecimal and MAX_LKR_MINOR_UNITS.
    • isValidContactPhone, which accepts full phone numbers and rejects short codes.
    • Schema rules rules.required(), rules.optional(), rules.nullable(), rules.contactPhone() and rules.mobile().
    • validateBankCode, returning a detailed ValidationResult.
    • referenceDate option for NIC parsing and validation (NicOptions), for deterministic age and future-date checks.
    • Schema validators infer their output type from the rules passed to createSchemaValidator.
    • MAX_INPUT_LENGTH and MAX_LKR_AMOUNT constants.
    • lanka-validations/package.json is exported for tools that read package metadata.
    • ESM consumers using moduleResolution: node16/nodenext now get ESM type declarations (.d.mts) instead of CommonJS ones, so default and named imports type-check correctly.
    • Subpath imports such as lanka-validations/nic now resolve types under the legacy moduleResolution: node (node10) setting.
    • Bank code validation and lookup no longer accept inherited object property names such as '__proto__' or 'constructor'.
    • maskNic, maskPhone and maskBankAccount revealed the full value when visibleDigits was 0, negative, or NaN. They now reveal nothing in those cases and round fractional values down.
    • Validators no longer throw on objects with a broken toString.
    • getLicenseCategoryInfo and getAllLicenseCategories return copies, so changing a returned record no longer affects later lookups.
    • getReferenceDataMetadata(table) is typed to return a single table's metadata, instead of a union that callers had to narrow.
    • lkrToWords rounds to cents once and carries into rupees: lkrToWords(1.999) returns 'Two Rupees Only' instead of 'One Rupee and One Hundred Cents Only'.
    • Schema validation records failures for special field names such as '__proto__' in errors as well as errorList.
    • Releases are published from GitHub Actions with npm provenance.
    • Province and district lookups use precomputed indexes instead of scanning arrays on every call.
    • Initial release: validators and parsers for Sri Lankan NIC, phone numbers, passports, driving licenses, vehicle plates, LKR currency (including amounts in words), date of birth, gender, postal codes, CBSL bank codes, TIN/VAT and administrative divisions, plus a schema/form validator.

    Settings
    Member Visibility

    On This Page
    Changelog