Parse URL query string into a key-value object

It decodes percent escapes and form-style plus signs, preserves empty values, and collects repeated parameter names into arrays in their original order. Malformed escapes and invalid UTF-8 are rejected clearly instead of being replaced or passed through. Use it interactively when inspecting a link, or call the deterministic API for $0.002 per item when an application, test suite, or ingestion workflow needs the same result every time.

Choose the input form that matches your source

You can submit a complete absolute URL such as <code>https://example.com/search?q=red+shoes</code>, a raw query such as <code>q=red+shoes&amp;page=2</code>, or the same raw query with a leading question mark. For an absolute URL, only the query component is parsed; the scheme, authority, path, and fragment do not become fields. A URL without a query returns an empty parameter object. Raw input is treated entirely as query data after an optional initial question mark, so it is useful when a framework, access log, webhook, or browser API has already separated the query from the rest of the address. Each ampersand starts another pair, while the first equals sign divides its name from its value. A name without an equals sign is retained with an empty value, as is a name followed by an explicit empty assignment. Empty segments between consecutive ampersands are ignored. The result also reports distinct-key and parsed-pair counts, making it easy to distinguish repeated fields from unique names without recounting the object yourself.

Understand decoding and repeated keys

Names and values are decoded independently using the conventions normally applied to URL query forms. A plus sign becomes a space, while <code>%HH</code> sequences are interpreted as UTF-8 bytes. That means <code>city=San+Jos%C3%A9</code> produces a readable Unicode value, and an encoded separator such as <code>%26</code> remains inside one value rather than being mistaken for the start of another field. Decoding happens exactly once, so <code>%2520</code> becomes <code>%20</code>, not a space. When a key occurs once, its value is a string. When the same decoded key occurs again, the value becomes an array containing every occurrence in source order. This rule represents checkbox groups, filters, tags, and other multi-select controls without discarding information or inventing numbered property names. Repetition is determined after decoding, so equivalent encoded spellings of the same name are grouped together. Blank repeated values are preserved in those arrays. The parser does not infer booleans, numbers, dates, or nested bracket notation; all scalar values remain strings so downstream code can apply its own domain-specific conversion deliberately.

Reject corruption before it enters a workflow

Loose query parsers can silently keep a stray percent sign, accept a one-digit escape, or substitute a replacement character when escaped bytes do not form valid UTF-8. Those behaviors make damaged identifiers appear usable and can create difficult differences between a browser, a server, and a signature-verification routine. This capability checks every percent sign for exactly two hexadecimal digits and then verifies that each decoded byte sequence is valid UTF-8. If either condition fails, the request returns an invalid-input error rather than a partial object. The parser is pure and deterministic: it performs no network request, follows no redirect, reads no clock, uses no randomness, and retains no state between calls. Treat the returned strings as data rather than trusted HTML, code, file paths, or database expressions; decoding restores characters but does not make their later use safe. In automation, the API costs $0.002 per item and can serve as a stable normalization step before validation, routing, comparison, or storage. In the browser, the same parsing logic is suitable for quickly examining copied links without contacting their destination.