Merge commit from fork

* fix: reject malformed IPv6 literals

* fix: restore RFC 5952 zero-run compression for IPv6 hosts

The IPv6 validation rewrite stripped per-hextet leading zeros but no longer
applied "::" zero-run compression, so normalize() was not RFC 5952 canonical
and equal() returned false for the same address across expanded and compressed
forms (e.g. [0:0:0:0:0:0:0:1] did not equal [::1]). That makes equal()-based
host checks representation-sensitive.

- add compressIPv6ZeroRun (longest run, leftmost on ties, minimum length 2)
- expand "::" before recompressing so the result is canonical regardless of
  where the input placed "::"
- add IPv6 canonical/equal tests; update the all-zeros normalize expectation

---------

Co-authored-by: Ulises Gascon <ulisesgascongonzalez@gmail.com>

Author: mcollina

index.js

@@ -362,6 +362,7 @@ function parseWithStatus (uri, opts) {
   let malformedPercentEncoding = false
   let malformedSchemeSpecific = false
   let malformedHost = false
+  let malformedIPLiteral = false
 
   let isIP = false
   if (options.reference === 'suffix') {
@@ -438,9 +439,16 @@ function parseWithStatus (uri, opts) {
     if (parsed.host) {
       const ipv4result = isIPv4(parsed.host)
       if (ipv4result === false) {
+        const bracketedIPLiteral = parsed.host[0] === '[' && parsed.host[parsed.host.length - 1] === ']'
         const ipv6result = normalizeIPv6(parsed.host)
-        parsed.host = ipv6result.host.toLowerCase()
-        isIP = ipv6result.isIPV6
+        isIP = ipv6result.isIPV6 || ipv6result.isIPVFuture === true
+        malformedIPLiteral = bracketedIPLiteral && ipv6result.error === true
+        parsed.host = isIP ? ipv6result.host : ipv6result.host.toLowerCase()
+
+        if (malformedIPLiteral) {
+          parsed.error = parsed.error || 'URI host is malformed.'
+          malformedAuthorityOrPort = true
+        }
       } else {
         isIP = true
       }
@@ -471,8 +479,9 @@ function parseWithStatus (uri, opts) {
         if (parsed.scheme !== undefined) {
           parsed.scheme = unescape(parsed.scheme)
         }
-        if (parsed.host !== undefined) {
-          parsed.host = reescapeHostDelimiters(normalizePercentEncoding(parsed.host, true), isIP)
+        if (parsed.host !== undefined && !malformedIPLiteral) {
+          const host = isIP ? parsed.host : normalizePercentEncoding(parsed.host, true)
+          parsed.host = reescapeHostDelimiters(host, isIP)
         }
       }
       if (parsed.path) {

lib/utils.js

@@ -75,12 +75,14 @@ function stringArrayToHexStripped (input) {
   return acc
 }
 
-/**
- * @typedef {Object} GetIPV6Result
- * @property {boolean} error - Indicates if there was an error parsing the IPv6 address.
- * @property {string} address - The parsed IPv6 address.
- * @property {string} [zone] - The zone identifier, if present.
- */
+/** @type {(value: string) => boolean} */
+const isHextet = RegExp.prototype.test.bind(/^[\dA-Fa-f]{1,4}$/)
+
+/** @type {(value: string) => boolean} */
+const isIPvFuture = RegExp.prototype.test.bind(/^[vV][\dA-Fa-f]+\.[A-Za-z\d\-._~!$&'()*+,;=:]+$/)
+
+/** @type {(value: string) => boolean} */
+const isZoneCharacter = RegExp.prototype.test.bind(/^[A-Za-z\d\-._~]$/)
 
 /**
  * @param {string} value
@@ -89,119 +91,154 @@ function stringArrayToHexStripped (input) {
 const nonSimpleDomain = RegExp.prototype.test.bind(/[^!"$&'()*+,\-.;=_`a-z{}~]/u)
 
 /**
- * @param {Array<string>} buffer
+ * @param {string} zone
  * @returns {boolean}
  */
-function consumeIsZone (buffer) {
-  buffer.length = 0
+function isZoneIdentifier (zone) {
+  if (zone.length === 0) return false
+
+  for (let i = 0; i < zone.length; i++) {
+    if (isZoneCharacter(zone[i])) continue
+    if (zone[i] === '%' && i + 2 < zone.length && isHexPair(zone.slice(i + 1, i + 3))) {
+      i += 2
+      continue
+    }
+    return false
+  }
+
   return true
 }
 
 /**
- * @param {Array<string>} buffer
- * @param {Array<string>} address
- * @param {GetIPV6Result} output
- * @returns {boolean}
+ * Compresses the longest run of zero hextets to "::" per RFC 5952. A run of a
+ * single zero hextet is left uncompressed. On ties the leftmost run wins.
+ *
+ * @param {string[]} hextets
+ * @returns {string}
  */
-function consumeHextets (buffer, address, output) {
-  if (buffer.length) {
-    const hex = stringArrayToHexStripped(buffer)
-    if (hex !== '') {
-      address.push(hex)
+function compressIPv6ZeroRun (hextets) {
+  let bestStart = -1
+  let bestLength = 0
+  let runStart = -1
+  let runLength = 0
+  for (let i = 0; i < hextets.length; i++) {
+    if (hextets[i] === '0') {
+      if (runStart === -1) runStart = i
+      runLength++
+      if (runLength > bestLength) {
+        bestLength = runLength
+        bestStart = runStart
+      }
     } else {
-      output.error = true
-      return false
+      runStart = -1
+      runLength = 0
     }
-    buffer.length = 0
   }
-  return true
+
+  if (bestLength < 2) return hextets.join(':')
+
+  const head = hextets.slice(0, bestStart).join(':')
+  const tail = hextets.slice(bestStart + bestLength).join(':')
+  return head + '::' + tail
 }
 
 /**
+ * Validates an IPv6 address against the alternatives in RFC 3986 section
+ * 3.2.2 and returns the same address with leading hextet zeroes removed.
+ * An embedded IPv4 address counts as two hextets and is only valid at the end.
+ *
  * @param {string} input
- * @returns {GetIPV6Result}
+ * @returns {string|undefined}
  */
-function getIPV6 (input) {
-  let tokenCount = 0
-  const output = { error: false, address: '', zone: '' }
-  /** @type {Array<string>} */
-  const address = []
-  /** @type {Array<string>} */
-  const buffer = []
-  let endipv6Encountered = false
-  let endIpv6 = false
-
-  let consume = consumeHextets
+function normalizeIPv6Address (input) {
+  const compression = input.indexOf('::')
+  if (compression !== -1 && input.indexOf('::', compression + 1) !== -1) return undefined
+
+  const left = compression === -1 ? input.split(':') : input.slice(0, compression).split(':')
+  const right = compression === -1 ? [] : input.slice(compression + 2).split(':')
+  if (compression !== -1) {
+    if (left.length === 1 && left[0] === '') left.length = 0
+    if (right.length === 1 && right[0] === '') right.length = 0
+  }
 
-  for (let i = 0; i < input.length; i++) {
-    const cursor = input[i]
-    if (cursor === '[' || cursor === ']') { continue }
-    if (cursor === ':') {
-      if (endipv6Encountered === true) {
-        endIpv6 = true
-      }
-      if (!consume(buffer, address, output)) { break }
-      if (++tokenCount > 7) {
-        // not valid
-        output.error = true
-        break
-      }
-      if (i > 0 && input[i - 1] === ':') {
-        endipv6Encountered = true
-      }
-      address.push(':')
-      continue
-    } else if (cursor === '%') {
-      if (!consume(buffer, address, output)) { break }
-      // switch to zone detection
-      consume = consumeIsZone
-    } else {
-      buffer.push(cursor)
+  const parts = left.concat(right)
+  let hextetCount = 0
+  for (let i = 0; i < parts.length; i++) {
+    const part = parts[i]
+    if (part === '') return undefined
+
+    if (part.indexOf('.') !== -1) {
+      if (i !== parts.length - 1 || (compression !== -1 && right.length === 0) || !isIPv4(part)) return undefined
+      hextetCount += 2
       continue
     }
+
+    if (!isHextet(part)) return undefined
+    parts[i] = parseInt(part, 16).toString(16)
+    hextetCount++
   }
-  if (buffer.length) {
-    if (consume === consumeIsZone) {
-      output.zone = buffer.join('')
-    } else if (endIpv6) {
-      address.push(buffer.join(''))
-    } else {
-      address.push(stringArrayToHexStripped(buffer))
-    }
+
+  if (compression === -1) {
+    if (hextetCount !== 8) return undefined
+    return compressIPv6ZeroRun(parts)
   }
-  output.address = address.join('')
-  return output
+  if (hextetCount >= 8) return undefined
+
+  // expand "::" then re-compress the longest run for a canonical result
+  const expanded = parts.slice(0, left.length)
+  for (let i = hextetCount; i < 8; i++) expanded.push('0')
+  for (let i = left.length; i < parts.length; i++) expanded.push(parts[i])
+  return compressIPv6ZeroRun(expanded)
 }
 
 /**
  * @typedef {Object} NormalizeIPv6Result
  * @property {string} host - The normalized host.
  * @property {string} [escapedHost] - The escaped host.
  * @property {boolean} isIPV6 - Indicates if the host is an IPv6 address.
+ * @property {boolean} [isIPVFuture] - Indicates if the host is an IPvFuture literal.
+ * @property {boolean} [error] - Indicates if a bracketed IP literal is malformed.
  */
 
 /**
+ * Validates and normalizes a bracketed IP literal. Raw zone separators remain
+ * accepted for backwards compatibility, while encoded separators and zone
+ * contents follow RFC 6874.
+ *
  * @param {string} host
  * @returns {NormalizeIPv6Result}
  */
 function normalizeIPv6 (host) {
-  if (findToken(host, ':') < 2) { return { host, isIPV6: false } }
-  const ipv6 = getIPV6(host)
-
-  if (!ipv6.error) {
-    let newHost = ipv6.address
-    let escapedHost = ipv6.address
-    if (ipv6.zone) {
-      // RFC 6874 encodes the zone separator as "%25" in a URI. Accept both
-      // component forms used by this API ("%zone" and "%25zone") while
-      // consuming at most that one separator escape.
-      const zone = ipv6.zone.startsWith('25') ? ipv6.zone.slice(2) : ipv6.zone
-      newHost += '%' + zone
-      escapedHost += '%25' + zone
-    }
-    return { host: newHost, isIPV6: true, escapedHost }
-  } else {
-    return { host, isIPV6: false }
+  const bracketed = host[0] === '[' && host[host.length - 1] === ']'
+  const hasBracket = host[0] === '[' || host[host.length - 1] === ']'
+  if (hasBracket && !bracketed) return { host, isIPV6: false, error: true }
+
+  let input = bracketed ? host.slice(1, -1) : host
+  if (bracketed && isIPvFuture(input)) {
+    input = input.toLowerCase()
+    return { host: `[${input}]`, escapedHost: input, isIPV6: false, isIPVFuture: true }
+  }
+
+  if (findToken(input, ':') < 2) {
+    return { host, isIPV6: false, error: bracketed }
+  }
+
+  let zoneIdentifier = ''
+  const zoneSeparator = input.indexOf('%')
+  if (zoneSeparator !== -1) {
+    const separatorLength = input.slice(zoneSeparator, zoneSeparator + 3).toLowerCase() === '%25' ? 3 : 1
+    zoneIdentifier = input.slice(zoneSeparator + separatorLength)
+    if (!isZoneIdentifier(zoneIdentifier)) return { host, isIPV6: false, error: true }
+    input = input.slice(0, zoneSeparator)
+  }
+
+  const address = normalizeIPv6Address(input)
+  if (address === undefined) return { host, isIPV6: false, error: true }
+
+  return {
+    host: address + (zoneIdentifier ? '%' + zoneIdentifier : ''),
+    escapedHost: address + (zoneIdentifier ? '%25' + zoneIdentifier : ''),
+    isIPV6: true
   }
 }
 
@@ -657,12 +694,16 @@ function recomposeAuthority (component) {
   }
 
   if (component.host !== undefined) {
-    // Decode only unreserved bytes, once. In particular, keep %25 encoded so
-    // it cannot become the introducer for a second escape during recomposition.
-    let host = normalizePercentEncoding(component.host, true)
+    let host = component.host
     if (!isIPv4(host)) {
-      const ipV6res = normalizeIPv6(host)
-      if (ipV6res.isIPV6 === true) {
+      let ipV6res = normalizeIPv6(host)
+      if (ipV6res.isIPV6 !== true && ipV6res.isIPVFuture !== true) {
+        // Decode only unreserved bytes, once. In particular, keep %25 encoded
+        // so it cannot introduce a second escape during recomposition.
+        host = normalizePercentEncoding(host, true)
+        ipV6res = normalizeIPv6(host)
+      }
+      if (ipV6res.isIPV6 === true || ipV6res.isIPVFuture === true) {
         host = `[${ipV6res.escapedHost}]`
       } else {
         host = reescapeHostDelimiters(host, false)

test/fixtures/uri-js-parse.json

@@ -53,6 +53,7 @@
       "host": "[2001:dbz::1]",
       "port": 80,
       "path": "",
+      "error": "URI host is malformed.",
       "reference": "relative"
     }
   ],
@@ -242,6 +243,7 @@
     {
       "host": "[2606:2800:220:1:248:1893:25c8:1946:43209]",
       "path": "",
+      "error": "URI host is malformed.",
       "reference": "relative"
     }
   ],

test/ipv6-canonical.test.js

@@ -0,0 +1,34 @@
+'use strict'
+
+const test = require('tape')
+const fastURI = require('..')
+
+test('IPv6 hosts normalize to the RFC 5952 canonical form', (t) => {
+  const cases = [
+    ['http://[0:0:0:0:0:0:0:1]/', 'http://[::1]/'],
+    ['http://[0:0:0:0:0:0:0:0]/', 'http://[::]/'],
+    ['http://[2001:0db8:0000:0000:0000:0000:0000:0001]/', 'http://[2001:db8::1]/'],
+    ['http://[2001:0:0:0:0:0:0:1]/', 'http://[2001::1]/'],
+    ['http://[fe80:0:0:0:0:0:0:1]/', 'http://[fe80::1]/'],
+    ['http://[1:0:0:0:2:0:0:3]/', 'http://[1::2:0:0:3]/']
+  ]
+
+  for (const [uri, normalized] of cases) {
+    t.equal(fastURI.normalize(uri), normalized, `${uri} normalizes to ${normalized}`)
+  }
+  t.end()
+})
+
+test('IPv6 equal() matches the same address across compressed and expanded forms', (t) => {
+  const pairs = [
+    ['http://[::1]/', 'http://[0:0:0:0:0:0:0:1]/'],
+    ['http://[::]/', 'http://[0:0:0:0:0:0:0:0]/'],
+    ['http://[2001:db8::1]/', 'http://[2001:0db8:0000:0000:0000:0000:0000:0001]/'],
+    ['http://[1::2:0:0:3]/', 'http://[1:0:0:0:2:0:0:3]/']
+  ]
+
+  for (const [a, b] of pairs) {
+    t.equal(fastURI.equal(a, b), true, `${a} equals ${b}`)
+  }
+  t.end()
+})