implement spec changes from a while ago (nodejs#2676)

* implement spec changes from a while ago

* fixup

lib/fetch/body.js

@@ -8,18 +8,19 @@ const {
   isReadableStreamLike,
   readableStreamClose,
   createDeferredPromise,
-  fullyReadBody
+  fullyReadBody,
+  extractMimeType
 } = require('./util')
 const { FormData } = require('./formdata')
 const { kState } = require('./symbols')
 const { webidl } = require('./webidl')
 const { Blob, File: NativeFile } = require('node:buffer')
-const { kBodyUsed, kHeadersList } = require('../core/symbols')
+const { kBodyUsed } = require('../core/symbols')
 const assert = require('node:assert')
 const { isErrored } = require('../core/util')
 const { isUint8Array, isArrayBuffer } = require('util/types')
 const { File: UndiciFile } = require('./file')
-const { parseMIMEType, serializeAMimeType } = require('./dataURL')
+const { serializeAMimeType } = require('./dataURL')
 
 /** @type {globalThis['File']} */
 const File = NativeFile ?? UndiciFile
@@ -330,7 +331,7 @@ function bodyMixinMethods (instance) {
       return specConsumeBody(this, (bytes) => {
         let mimeType = bodyMimeType(this)
 
-        if (mimeType === 'failure') {
+        if (mimeType === null) {
           mimeType = ''
         } else if (mimeType) {
           mimeType = serializeAMimeType(mimeType)
@@ -369,12 +370,11 @@ function bodyMixinMethods (instance) {
 
       throwIfAborted(this[kState])
 
-      const contentType = this.headers[kHeadersList].get('content-type', true)
-
-      const mimeType = contentType !== null ? parseMIMEType(contentType) : 'failure'
+      // 1. Let mimeType be the result of get the MIME type with this.
+      const mimeType = bodyMimeType(this)
 
       // If mimeType’s essence is "multipart/form-data", then:
-      if (mimeType !== 'failure' && mimeType.essence === 'multipart/form-data') {
+      if (mimeType !== null && mimeType.essence === 'multipart/form-data') {
         const headers = {}
         for (const [key, value] of this.headers) headers[key] = value
 
@@ -432,7 +432,7 @@ function bodyMixinMethods (instance) {
         await busboyResolve
 
         return responseFormData
-      } else if (mimeType !== 'failure' && mimeType.essence === 'application/x-www-form-urlencoded') {
+      } else if (mimeType !== null && mimeType.essence === 'application/x-www-form-urlencoded') {
         // Otherwise, if mimeType’s essence is "application/x-www-form-urlencoded", then:
 
         // 1. Let entries be the result of parsing bytes.
@@ -581,17 +581,25 @@ function parseJSONFromBytes (bytes) {
 
 /**
  * @see https://fetch.spec.whatwg.org/#concept-body-mime-type
- * @param {import('./response').Response|import('./request').Request} object
+ * @param {import('./response').Response|import('./request').Request} requestOrResponse
  */
-function bodyMimeType (object) {
-  const { headersList } = object[kState]
-  const contentType = headersList.get('content-type')
-
-  if (contentType === null) {
-    return 'failure'
+function bodyMimeType (requestOrResponse) {
+  // 1. Let headers be null.
+  // 2. If requestOrResponse is a Request object, then set headers to requestOrResponse’s request’s header list.
+  // 3. Otherwise, set headers to requestOrResponse’s response’s header list.
+  /** @type {import('./headers').HeadersList} */
+  const headers = requestOrResponse[kState].headersList
+
+  // 4. Let mimeType be the result of extracting a MIME type from headers.
+  const mimeType = extractMimeType(headers)
+
+  // 5. If mimeType is failure, then return null.
+  if (mimeType === 'failure') {
+    return null
   }
 
-  return parseMIMEType(contentType)
+  // 6. Return mimeType.
+  return mimeType
 }
 
 module.exports = {

lib/fetch/dataURL.js

@@ -1,5 +1,4 @@
 const assert = require('node:assert')
-const { isomorphicDecode } = require('./util')
 
 const encoder = new TextEncoder()
 
@@ -604,18 +603,7 @@ function isHTTPWhiteSpace (char) {
  * @param {boolean} [trailing=true]
  */
 function removeHTTPWhitespace (str, leading = true, trailing = true) {
-  let lead = 0
-  let trail = str.length - 1
-
-  if (leading) {
-    while (lead < str.length && isHTTPWhiteSpace(str.charCodeAt(lead))) lead++
-  }
-
-  if (trailing) {
-    while (trail > 0 && isHTTPWhiteSpace(str.charCodeAt(trail))) trail--
-  }
-
-  return lead === 0 && trail === str.length - 1 ? str : str.slice(lead, trail + 1)
+  return removeChars(str, leading, trailing, isHTTPWhiteSpace)
 }
 
 /**
@@ -634,20 +622,110 @@ function isASCIIWhitespace (char) {
  * @param {boolean} [trailing=true]
  */
 function removeASCIIWhitespace (str, leading = true, trailing = true) {
+  return removeChars(str, leading, trailing, isASCIIWhitespace)
+}
+
+/**
+ *
+ * @param {string} str
+ * @param {boolean} leading
+ * @param {boolean} trailing
+ * @param {(charCode: number) => boolean} predicate
+ * @returns
+ */
+function removeChars (str, leading, trailing, predicate) {
   let lead = 0
   let trail = str.length - 1
 
   if (leading) {
-    while (lead < str.length && isASCIIWhitespace(str.charCodeAt(lead))) lead++
+    while (lead < str.length && predicate(str.charCodeAt(lead))) lead++
   }
 
   if (trailing) {
-    while (trail > 0 && isASCIIWhitespace(str.charCodeAt(trail))) trail--
+    while (trail > 0 && predicate(str.charCodeAt(trail))) trail--
   }
 
   return lead === 0 && trail === str.length - 1 ? str : str.slice(lead, trail + 1)
 }
 
+/**
+ * @see https://infra.spec.whatwg.org/#isomorphic-decode
+ * @param {Uint8Array} input
+ * @returns {string}
+ */
+function isomorphicDecode (input) {
+  // 1. To isomorphic decode a byte sequence input, return a string whose code point
+  //    length is equal to input’s length and whose code points have the same values
+  //    as the values of input’s bytes, in the same order.
+  const length = input.length
+  if ((2 << 15) - 1 > length) {
+    return String.fromCharCode.apply(null, input)
+  }
+  let result = ''; let i = 0
+  let addition = (2 << 15) - 1
+  while (i < length) {
+    if (i + addition > length) {
+      addition = length - i
+    }
+    result += String.fromCharCode.apply(null, input.subarray(i, i += addition))
+  }
+  return result
+}
+
+/**
+ * @see https://mimesniff.spec.whatwg.org/#minimize-a-supported-mime-type
+ * @param {Exclude<ReturnType<typeof parseMIMEType>, 'failure'>} mimeType
+ */
+function minimizeSupportedMimeType (mimeType) {
+  switch (mimeType.essence) {
+    case 'application/ecmascript':
+    case 'application/javascript':
+    case 'application/x-ecmascript':
+    case 'application/x-javascript':
+    case 'text/ecmascript':
+    case 'text/javascript':
+    case 'text/javascript1.0':
+    case 'text/javascript1.1':
+    case 'text/javascript1.2':
+    case 'text/javascript1.3':
+    case 'text/javascript1.4':
+    case 'text/javascript1.5':
+    case 'text/jscript':
+    case 'text/livescript':
+    case 'text/x-ecmascript':
+    case 'text/x-javascript':
+      // 1. If mimeType is a JavaScript MIME type, then return "text/javascript".
+      return 'text/javascript'
+    case 'application/json':
+    case 'text/json':
+      // 2. If mimeType is a JSON MIME type, then return "application/json".
+      return 'application/json'
+    case 'image/svg+xml':
+      // 3. If mimeType’s essence is "image/svg+xml", then return "image/svg+xml".
+      return 'image/svg+xml'
+    case 'text/xml':
+    case 'application/xml':
+      // 4. If mimeType is an XML MIME type, then return "application/xml".
+      return 'application/xml'
+  }
+
+  // 2. If mimeType is a JSON MIME type, then return "application/json".
+  if (mimeType.subtype.endsWith('+json')) {
+    return 'application/json'
+  }
+
+  // 4. If mimeType is an XML MIME type, then return "application/xml".
+  if (mimeType.subtype.endsWith('+xml')) {
+    return 'application/xml'
+  }
+
+  // 5. If mimeType is supported by the user agent, then return mimeType’s essence.
+  // Technically, node doesn't support any mimetypes.
+
+  // 6. Return the empty string.
+  return ''
+}
+
 module.exports = {
   dataURLProcessor,
   URLSerializer,
@@ -656,5 +734,7 @@ module.exports = {
   stringPercentDecode,
   parseMIMEType,
   collectAnHTTPQuotedString,
-  serializeAMimeType
+  serializeAMimeType,
+  removeChars,
+  minimizeSupportedMimeType
 }

lib/fetch/index.js

@@ -44,7 +44,8 @@ const {
   clampAndCoarsenConnectionTimingInfo,
   simpleRangeHeaderValue,
   buildContentRange,
-  createInflate
+  createInflate,
+  extractMimeType
 } = require('./util')
 const { kState } = require('./symbols')
 const assert = require('node:assert')
@@ -59,7 +60,7 @@ const {
 const EE = require('node:events')
 const { Readable, pipeline } = require('node:stream')
 const { addAbortListener, isErrored, isReadable, nodeMajor, nodeMinor, bufferToLowerCasedHeaderName } = require('../core/util')
-const { dataURLProcessor, serializeAMimeType, parseMIMEType } = require('./dataURL')
+const { dataURLProcessor, serializeAMimeType, minimizeSupportedMimeType } = require('./dataURL')
 const { getGlobalDispatcher } = require('../global')
 const { webidl } = require('./webidl')
 const { STATUS_CODES } = require('node:http')
@@ -1027,11 +1028,11 @@ function fetchFinale (fetchParams, response) {
         responseStatus = response.status
 
         // 2. Let mimeType be the result of extracting a MIME type from response’s header list.
-        const mimeType = parseMIMEType(response.headersList.get('content-type', true)) // TODO: fix
+        const mimeType = extractMimeType(response.headersList)
 
         // 3. If mimeType is not failure, then set bodyInfo’s content type to the result of minimizing a supported MIME type given mimeType.
         if (mimeType !== 'failure') {
-          // TODO
+          bodyInfo.contentType = minimizeSupportedMimeType(mimeType)
         }
       }