Duo Data Utilities

Encode a WGS-84 latitude and longitude to a geohash at a chosen precision, or decode a geohash back to its bounding box, centre point and error bounds in metres. Encoding also returns the eight neighbouring cells, which is what a caller needs to do proximity search over a geohash-indexed store without a spatial database.

VAT, tax ID and national ID validator: check a national identity, tax or company registration number against its country's official checksum rule. Covers 23 countries including NL BSN, ES NIF/NIE/CIF, IT codice fiscale, FR SIREN/SIRET, DE Steuer-ID, BE/PL/CZ national numbers, SE/NO/DK/FI personal numbers, BR CPF/CNPJ, AU ABN/TFN, UK NINo, IE PPS, US NPI/EIN. Structure and check digit only, no registry lookup; the number is never stored (logs keep only its length and character types).

Spell a number out in words in English, Dutch, German, French, Spanish, Italian, Portuguese, Polish, Russian, Turkish, Swedish or Japanese, in cardinal, ordinal or currency form. Currency form agrees the number with the currency noun and gives real minor-unit names for 19 currencies (EUR, USD, GBP, CHF, SEK, NOK, DKK, PLN, RUB, TRY, CNY, INR, BRL, MXN, CAD, AUD, NZD, CZK, HUF); any other ISO 4217 code uses a fraction such as 56/100. Values up to 10^15.

Convert an integer between any two numeric bases from 2 to 36, with arbitrary precision so values far beyond 2^53 are exact. Accepts an optional sign and common prefixes (0x, 0b, 0o), validates every digit against the source base, and returns the value in the target base, in decimal, and as its bit length and digit count.

Look up a country or territory by ISO 3166-1 alpha-2, alpha-3 or numeric code, or by name, and get the other codes, the official and common English names, the UN M49 region and sub-region, the currencies in use, the country-code top-level domain, and optionally the full ISO 3166-2 subdivision list with each subdivision's type. Also resolves an ISO 3166-2 subdivision code such as NL-NH directly. Names are matched case- and accent-insensitively.

Look up a currency by its ISO 4217 alphabetic or numeric code: the official English name, minor-unit exponent, fund flag, the smallest cash denomination where it differs from the minor unit, the territories that use it, and historic codes on request. Returns the correct rounding increment for cash amounts, which is what a caller needs to render or settle a price. This route returns currency facts only and never an exchange rate.

Look up a NUTS statistical region of the European Union: resolve a NUTS code to its name, country and level, walk up to its parents or down to its children, or search regions by name. Covers all four levels (country, major region, basic region, small region) for the 27 member states and 12 further reporting countries, from the NUTS 2024 classification. Returns the Latin-script and local-language names plus the urbanisation, coastal and mountain typologies.

Compute the Keccak-256 hash of a Solidity function or event signature and derive the 4-byte function selector and the 32-byte event topic. Normalises the signature first — strips parameter names, expands uint to uint256 and the other aliases, flattens tuples — so a signature copied from source code produces the same selector as the canonical form. Also hashes arbitrary text or hex bytes with Keccak-256.

Distance and direction between WGS-84 coordinates: geodesic distance by Vincenty with a haversine fallback, initial and final bearing, the midpoint, and the destination point reached from a start point on a given bearing and distance. Also computes the cumulative length of a supplied coordinate path. Results in metres, kilometres, nautical miles and miles.

Decode a JSON Web Token without verifying it: returns the header, the claims, the signature length and algorithm, and a computed view of the time claims — whether exp has passed, whether nbf is in the future, and the remaining lifetime in seconds against a supplied or current instant. Reports structural problems such as a missing part, invalid base64url or an alg of none. It does not and cannot verify the signature; do not use it to decide trust.

Compute a cryptographic digest of supplied data: SHA-256, SHA-384, SHA-512, SHA-1, SHA3-256, SHA3-512, Keccak-256, BLAKE2b-256 and BLAKE2b-512, plus HMAC with any of them. Input is UTF-8 text, base64 or hex bytes; output is hex, base64, base64url or base32. Returns the digest, the byte length of the input, and the exact algorithm identifier used.

Parse a semantic version into major, minor, patch, prerelease and build parts; compare two versions by precedence including prerelease ordering; or test whether a version satisfies a range expression using caret, tilde, hyphen, x-ranges and comparator sets. Returns the parsed structure, the comparison result and the reason a range did not match.

Compare two strings and return every standard similarity measure in one call: Levenshtein and Damerau-Levenshtein edit distance, normalised similarity, Jaro and Jaro-Winkler, longest common subsequence, Dice coefficient on character bigrams, and the Soundex, Metaphone and NYSIIS phonetic keys of each string. Optional case folding, accent folding and whitespace normalisation before comparison.

Convert a Gregorian date to other calendar systems: Islamic (both the tabular civil calendar and the Umm al-Qura variant), Hebrew, Japanese imperial era, Republic of China, Buddhist and Persian. Also computes Easter Sunday for a year in both the Gregorian and Julian reckonings. Uses the calendar support built into the runtime, with the algorithm named in every response.

Expand an RFC 5545 recurrence rule into concrete occurrence timestamps. Accepts an RRULE string with DTSTART, an IANA time zone, and optional EXDATE and RDATE lists, and returns up to 200 occurrences as both local wall-clock times and UTC instants. Supports FREQ, INTERVAL, COUNT, UNTIL, BYDAY, BYMONTHDAY, BYMONTH, BYSETPOS and WKST. Occurrences moved by a DST gap carry gap: true. Rules needing over 50,000 candidate evaluations are rejected.

Business-day arithmetic with a caller-supplied calendar: add or subtract N working days from a date, or count the working days between two dates. The weekend is a weekday mask (Friday-Saturday and Sunday-only weeks work) and holidays are an explicit date list, so nothing depends on a stale holiday table. Returns the result date, the day count and the skipped dates with reasons. The skipped list is capped at 1,000 entries (skipped_truncated flags it); skipped_total is always exact.

List the daylight-saving transitions for an IANA time zone over a date range: the exact instant of each shift, the offset before and after, and whether local clocks jump forward or back. Returns the current offset and DST state, and the next transition from a given instant. Zones without DST return an empty transition list rather than an error.

Construct an IBAN from a country code and the domestic account parts, computing the ISO 7064 mod-97-10 check digits. Takes either a single BBAN string or named parts (bank code, branch code, account number); digit fields are left-padded with zeros to the country's registered widths. Returns the IBAN in electronic and print formats with its bank, branch and account parts. Covers the 89 SWIFT registry countries. Named parts work only where they span the whole BBAN; otherwise pass the BBAN.

Validate the check digit of a financial security identifier: ISIN (ISO 6166), CUSIP, SEDOL and FIGI. Returns the verdict, the computed correct check character, and for an ISIN the issuing country prefix and the embedded NSIN. Also builds an ISIN from a country code plus a 9-character NSIN. Format and checksum only, no security master lookup.

IBAN validator: validate an IBAN's length for its country, permitted character classes per field, and the ISO 7064 mod-97-10 check digits. Returns the verdict with a specific failure reason, the electronic and print formats, and the decomposed bank code, branch code and account number according to that country's registered layout. Covers the 89 countries of the SWIFT IBAN registry. Structure and checksum only, it does not confirm the account exists.

Identify which blockchain an address belongs to and validate its encoding: EVM EIP-55 checksum, Bitcoin base58check and bech32/bech32m with witness version, Solana base58 and on-curve check, Cosmos bech32 with its HRP, Tron, XRP, Litecoin, Dogecoin, Cardano, Polkadot SS58 and Algorand. Returns every chain the address is valid for (ambiguous when several match), the checksummed form for EVM addresses, and the reason a check failed, including unrecognised_format.

Validate and convert GS1 identifiers: GTIN-8, GTIN-12 (UPC-A), GTIN-13 (EAN-13), GTIN-14, GLN and SSCC. Returns the check-digit verdict, the computed correct check digit, the GTIN-14 form with its indicator digit, and conversion between GTIN-8, GTIN-12, GTIN-13 and GTIN-14. No company-prefix split is returned: GS1's prefix-length table is no longer public data. Also computes a missing check digit from a partial code. Structure only, no product database.

Convert internationalised domain names between Unicode and ASCII punycode in either direction, label by label, and validate each label against IDNA2008: length, leading and trailing hyphens, the hyphen-hyphen rule at positions 3 and 4, disallowed code points and bidirectional rules. Returns both forms, the per-label breakdown and the first rule violated. Strict mode is IDNA2008 (UTS 46 non-transitional); it may reject names /v1/code/url displays.

Transliterate Cyrillic text into Latin using a named standard: ISO 9:1995 (one-to-one, with diacritics, covering the full standard: Russian, other Slavic, historic and non-Slavic Cyrillic letters) or BGN/PCGN, which is the Russian system only (Ukrainian, Serbian and other letters are passed through unmapped). Returns the transliteration, the source script, the standard, the mapped fraction and the unmapped characters. Other scripts are rejected with UNSUPPORTED_VALUE.

Encode coordinates as an Open Location Code (plus code) at a chosen length, or decode a full code back to its bounding box and centre. Also shortens a full code against a reference coordinate and recovers a shortened code given a reference, which is how plus codes are normally exchanged in text. Returns code length, precision in metres and validity.

Split a money amount into shares without losing or inventing a cent. Divide evenly into N parts, or by a list of ratios, and the remainder is distributed one minor unit at a time by a named, deterministic rule — largest-remainder, first-wins or last-wins — so the parts always sum exactly to the original. Uses the ISO 4217 minor-unit exponent of the currency, so zero- and three-decimal currencies work (JPY 0, KWD 3); metals and XXX are rejected.

Descriptive statistics for a supplied numeric array in one call: count, sum, mean, median, mode, variance and standard deviation with variance and standard deviation honouring ddof (default 1, the sample form; 0 gives the population form) and always-population *_population fields, min, max, range, the quartiles and interquartile range, skewness, kurtosis, and any requested percentiles. Percentiles use a named interpolation method so the result is reproducible rather than implementation-defined.

Look up an airport by IATA or ICAO code and get its name, city, country, ISO 3166-2 region code, WGS-84 coordinates, elevation and IANA time zone, or find the nearest airports to a coordinate within a radius. Covers every airport worldwide that holds a three-letter IATA code and scheduled service. Returns the code type that matched and the distance in kilometres for proximity results.

Geometric predicates on a GeoJSON geometry supplied in the request: point-in-polygon with correct hole handling, bounding box, centroid, signed area in square metres, perimeter length, and ring winding order. Accepts Polygon and MultiPolygon, and a GeoJSON Feature wrapping either. Area and length are geodesic on the WGS-84 ellipsoid, correct at any latitude; containment is planar in lon/lat.

Parse and canonicalise a URL, and extract its registrable domain using the Public Suffix List — the part an agent needs to decide whether two URLs belong to the same owner. Returns scheme, host, port, path, decoded query parameters and fragment, the public suffix, the registrable domain (eTLD+1), the subdomain, whether the suffix is an ICANN or a private entry, plus a canonical form with default port, dot-segments and percent-encoding normalised. unicode_host uses WHATWG/UTS 46 (non-strict).