url/resources/toascii.json

[
  "This contains assorted IDNA tests that IdnaTestV2 might not cover.",
  "Feel free to deduplicate with a clear commit message.",
  "",
  "If the test only applies to the URL Standard's 'domain to ASCII', ",
  "and not to TR46's ToASCII, then tag it with `urlStandardOnly`",
  {
    "comment": "Label with hyphens in 3rd and 4th position",
    "input": "aa--",
    "output": "aa--"
  },
  {
    "input": "a†--",
    "output": "xn--a---kp0a"
  },
  {
    "input": "ab--c",
    "output": "ab--c"
  },
  {
    "comment": "Label with leading hyphen",
    "input": "-x",
    "output": "-x"
  },
  {
    "input": "-†",
    "output": "xn----xhn"
  },
  {
    "input": "-x.xn--zca",
    "output": "-x.xn--zca"
  },
  {
    "input": "-x.ß",
    "output": "-x.xn--zca"
  },
  {
    "comment": "Label with trailing hyphen",
    "input": "x-.xn--zca",
    "output": "x-.xn--zca"
  },
  {
    "input": "x-.ß",
    "output": "x-.xn--zca"
  },
  {
    "comment": "Empty labels",
    "input": "x..xn--zca",
    "output": "x..xn--zca"
  },
  {
    "input": "x..ß",
    "output": "x..xn--zca"
  },
  {
    "comment": "Invalid Punycode",
    "input": "xn--a",
    "output": null
  },
  {
    "input": "xn--a.xn--zca",
    "output": null
  },
  {
    "input": "xn--a.ß",
    "output": null
  },
  {
    "input": "xn--ls8h=",
    "output": null
  },
  {
    "comment": "Invalid Punycode (contains non-ASCII character)",
    "input": "xn--tešla",
    "output": null
  },
  {
    "comment": "Valid Punycode",
    "input": "xn--zca.xn--zca",
    "output": "xn--zca.xn--zca"
  },
  {
    "comment": "Mixed",
    "input": "xn--zca.ß",
    "output": "xn--zca.xn--zca"
  },
  {
    "input": "ab--c.xn--zca",
    "output": "ab--c.xn--zca"
  },
  {
    "input": "ab--c.ß",
    "output": "ab--c.xn--zca"
  },
  {
    "comment": "CheckJoiners is true",
    "input": "\u200D.example",
    "output": null
  },
  {
    "input": "xn--1ug.example",
    "output": null
  },
  {
    "comment": "CheckBidi is true",
    "input": "يa",
    "output": null
  },
  {
    "input": "xn--a-yoc",
    "output": null
  },
  {
    "comment": "processing_option is Nontransitional_Processing",
    "input": "ශ්‍රී",
    "output": "xn--10cl1a0b660p"
  },
  {
    "input": "نامه‌ای",
    "output": "xn--mgba3gch31f060k"
  },
  {
    "comment": "U+FFFD",
    "input": "\uFFFD.com",
    "output": null
  },
  {
    "comment": "U+FFFD character encoded in Punycode",
    "input": "xn--zn7c.com",
    "output": null
  },
  {
    "comment": "Label longer than 63 code points",
    "input": "x01234567890123456789012345678901234567890123456789012345678901x",
    "output": "x01234567890123456789012345678901234567890123456789012345678901x"
  },
  {
    "input": "x01234567890123456789012345678901234567890123456789012345678901†",
    "output": "xn--x01234567890123456789012345678901234567890123456789012345678901-6963b"
  },
  {
    "input": "x01234567890123456789012345678901234567890123456789012345678901x.xn--zca",
    "output": "x01234567890123456789012345678901234567890123456789012345678901x.xn--zca"
  },
  {
    "input": "x01234567890123456789012345678901234567890123456789012345678901x.ß",
    "output": "x01234567890123456789012345678901234567890123456789012345678901x.xn--zca"
  },
  {
    "comment": "Domain excluding TLD longer than 253 code points",
    "input": "01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.0123456789012345678901234567890123456789012345678.x",
    "output": "01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.0123456789012345678901234567890123456789012345678.x"
  },
  {
    "input": "01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.0123456789012345678901234567890123456789012345678.xn--zca",
    "output": "01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.0123456789012345678901234567890123456789012345678.xn--zca"
  },
  {
    "input": "01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.0123456789012345678901234567890123456789012345678.ß",
    "output": "01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.01234567890123456789012345678901234567890123456789.0123456789012345678901234567890123456789012345678.xn--zca"
  },
  {
    "comment": "IDNA ignored code points",
    "input": "a\u00ADb",
    "output": "ab"
  },
  {
    "comment": "Interesting UseSTD3ASCIIRules=false cases",
    "input": "≠",
    "output": "xn--1ch"
  },
  {
    "input": "≮",
    "output": "xn--gdh"
  },
  {
    "input": "≯",
    "output": "xn--hdh"
  },
  {
    "comment": "NFC normalization (forbidden < and > characters are normalized to valid ones)",
    "input": "=\u0338",
    "output": "xn--1ch"
  },
  {
    "input": "<\u0338",
    "output": "xn--gdh"
  },
  {
    "input": ">\u0338",
    "output": "xn--hdh"
  },
  {
    "comment": "Same with inserted IDNA ignored code point",
    "input": "=\u00AD\u0338",
    "output": "xn--1ch"
  },
  {
    "input": "<\u00AD\u0338",
    "output": "xn--gdh"
  },
  {
    "input": ">\u00AD\u0338",
    "output": "xn--hdh"
  },
  "Tests below are from WebKit (fast/url/idna2003.html & fast/url/idna2008.html; contributed by Chris Weber back in 2011).",
  {
    "input": "fa\u00DF.de",
    "output": "xn--fa-hia.de"
  },
  {
    "input": "\u03B2\u03CC\u03BB\u03BF\u03C2.com",
    "output": "xn--nxasmm1c.com"
  },
  {
    "input": "\u0DC1\u0DCA\u200D\u0DBB\u0DD3.com",
    "output": "xn--10cl1a0b660p.com"
  },
  {
    "input": "\u0646\u0627\u0645\u0647\u200C\u0627\u06CC.com",
    "output": "xn--mgba3gch31f060k.com"
  },
  {
    "input": "www.loo\u0138out.net",
    "output": "www.xn--looout-5bb.net"
  },
  {
    "input": "\u15EF\u15EF\u15EF.lookout.net",
    "output": "xn--1qeaa.lookout.net"
  },
  {
    "input": "www.lookout.\u0441\u043E\u043C",
    "output": "www.lookout.xn--l1adi"
  },
  {
    "input": "www\u2025lookout.net",
    "output": null
  },
  {
    "input": "www.lookout\u2027net",
    "output": "www.xn--lookoutnet-406e"
  },
  {
    "input": "www.lookout.net\u2A7480",
    "output": null,
    "urlStandardOnly": true
  },
  {
    "input": "www\u00A0.lookout.net",
    "output": null,
    "urlStandardOnly": true
  },
  {
    "input": "\u1680lookout.net",
    "output": null
  },
  {
    "input": "\u001flookout.net",
    "output": null,
    "urlStandardOnly": true
  },
  {
    "input": "look\u06DDout.net",
    "output": null
  },
  {
    "input": "look\u180Eout.net",
    "output": "lookout.net"
  },
  {
    "input": "look\u2060out.net",
    "output": "lookout.net"
  },
  {
    "input": "look\uFEFFout.net",
    "output": "lookout.net"
  },
  {
    "input": "look\uD83F\uDFFEout.net",
    "output": null
  },
  {
    "input": "look\uFFFAout.net",
    "output": null
  },
  {
    "input": "look\u2FF0out.net",
    "output": null
  },
  {
    "input": "look\u0341out.net",
    "output": "xn--looout-kp7b.net"
  },
  {
    "input": "look\u202Eout.net",
    "output": null
  },
  {
    "input": "look\u206Bout.net",
    "output": "lookout.net"
  },
  {
    "input": "look\uDB40\uDC01out.net",
    "output": null
  },
  {
    "input": "look\uDB40\uDC20out.net",
    "output": null
  },
  {
    "input": "look\u05BEout.net",
    "output": null
  },
  {
    "input": "B\u00FCcher.de",
    "output": "xn--bcher-kva.de"
  },
  {
    "input": "\u2665.net",
    "output": "xn--g6h.net"
  },
  {
    "input": "\u0378.net",
    "output": null
  },
  {
    "input": "\u04C0.com",
    "output": "xn--s5a.com"
  },
  {
    "input": "\uD87E\uDC68.com",
    "output": "xn--snl.com"
  },
  {
    "input": "\u2183.com",
    "output": "xn--r5g.com"
  },
  {
    "input": "look\u034Fout.net",
    "output": "lookout.net"
  },
  {
    "input": "gOoGle.com",
    "output": "google.com"
  },
  {
    "input": "\u09dc.com",
    "output": "xn--15b8c.com"
  },
  {
    "input": "\u1E9E.com",
    "output": "xn--zca.com"
  },
  {
    "input": "\u1E9E.foo.com",
    "output": "xn--zca.foo.com"
  },
  {
    "input": "-foo.bar.com",
    "output": "-foo.bar.com"
  },
  {
    "input": "foo-.bar.com",
    "output": "foo-.bar.com"
  },
  {
    "input": "ab--cd.com",
    "output": "ab--cd.com"
  },
  {
    "input": "xn--0.com",
    "output": null
  },
  {
    "input": "foo\u0300.bar.com",
    "output": "xn--fo-3ja.bar.com"
  }
]