velt/net/url#

A WHATWG-style URL and URLSearchParams, plus JS's URI encoding functions (encodeURIComponent and the others): globals, as in Node; velt/net/url exports them too.

• new URL(input, base: string | null = null): throws UrlError (Invalid URL: …, or Invalid base URL: … when base does not parse), a TypeError as in JS (its name is TypeError, and e instanceof TypeError holds). URL.parse(input, base = null): URL | null. URL.canParse(input, base = null).
• Getters: href protocol username password host hostname port pathname search hash origin.
• Setters: everything except origin. Invalid values are ignored, as in JS, except href, which throws.
• searchParams (get) returns a copy. Write changes back with u.searchParams = params.
• toString() and toJSON() both return href.
• new URLSearchParams(init = ""): a leading ? is skipped.
  • size, get(name): string | null, getAll, has(name, value?)
  • append, set, delete(name, value?), sort() (stable)
  • keys(), values(), entries(): arrays, not iterators
  • toString(): form encoding (space becomes +)
• encodeURIComponent, encodeURI, decodeURIComponent, decodeURI: they throw the prelude's URIError("URI malformed"), as JS: the encoders on a lone surrogate, the decoders on a bad escape or invalid UTF-8.

import { URL, URLSearchParams, encodeURIComponent } from "velt/net/url";

const u = new URL("../api/items?page=2#top", "https://Example.com:443/app/v1/index.html");
console.log(u.href); // https://example.com/app/api/items?page=2#top
const params = u.searchParams;
params.set("q", "red shoes");
u.searchParams = params;
console.log(u.href); // https://example.com/app/api/items?page=2&q=red+shoes#top
console.log(new URLSearchParams("a=1&b=x+y").get("b"), encodeURIComponent("a&b/c"));

Notes / gaps:

• Special schemes (http https ws wss ftp file) get lowercased hosts, dropped default ports, \ treated as /, dot-segment removal, and IPv4 normalization (127.1 becomes 127.0.0.1).
• No IDNA/punycode: a non-ASCII host is invalid.
• IPv6 literals are validated loosely and not compressed.
• Windows drive letters in file: URLs follow the URL spec: a path starting with / keeps the base's drive letter (/x against file:///C:/p/m.ts is file:///C:/x), .. never removes it, a first segment C| is C: (also through the pathname setter), a relative path that starts with one takes none of the base's path (C|/x against file:///D:/p/m.ts is file:///C:/x), and one where the host would be starts the path (file://C:/x is file:///C:/x).