feat!: resolve extension conflicts with mime-score, close #116 (#119)

* feat!: resolve extension conflicts with mime-score, close #116

* feat!: resolve extension conflicts with mime-score, close #116

* chore: fix lint issues

* wip: lint, logic, and test fixup

* test: tweak log output

* fix: update history.md

---------

Co-authored-by: Wes Todd <[email protected]>

Author: broofa

HISTORY.md

@@ -1,17 +1,18 @@
-2.1.35 / 2022-03-12
+unreleased
 ===================
 
-  * deps: [email protected]
-    - Add extensions from IANA for more `image/*` types
-    - Add extension `.asc` to `application/pgp-keys`
-    - Add extensions to various XML types
-    - Add new upstream MIME types
-
-2.1.34 / 2021-11-08
-===================
-
-  * deps: [email protected]
-    - Add new upstream MIME types
+* resolve extension conflicts with mime-score (#119)
+  * asc -> application/pgp-signature is now application/pgp-keys
+  * mpp -> application/vnd.ms-project is now application/dash-patch+xml
+  * ac -> application/vnd.nokia.n-gage.ac+xml is now application/pkix-attr-cert
+  * bdoc -> application/x-bdoc is now application/bdoc
+  * wmz -> application/x-msmetafile is now application/x-ms-wmz
+  * xsl -> application/xslt+xml is now application/xml
+  * wav -> audio/wave is now audio/wav
+  * rtf -> text/rtf is now application/rtf
+  * xml -> text/xml is now application/xml
+  * mp4 -> video/mp4 is now application/mp4
+  * mpg4 -> video/mp4 is now application/mp4
 
 2.1.33 / 2021-10-01
 ===================

index.js

@@ -14,6 +14,7 @@
 
 var db = require('mime-db')
 var extname = require('path').extname
+var mimeScore = require('./mimeScore')
 
 /**
  * Module variables.
@@ -35,6 +36,7 @@ exports.extension = extension
 exports.extensions = Object.create(null)
 exports.lookup = lookup
 exports.types = Object.create(null)
+exports._extensionConflicts = []
 
 // Populate the extensions/types maps
 populateMaps(exports.extensions, exports.types)
@@ -80,9 +82,7 @@ function contentType (str) {
     return false
   }
 
-  var mime = str.indexOf('/') === -1
-    ? exports.lookup(str)
-    : str
+  var mime = str.indexOf('/') === -1 ? exports.lookup(str) : str
 
   if (!mime) {
     return false
@@ -152,9 +152,6 @@ function lookup (path) {
  */
 
 function populateMaps (extensions, types) {
-  // source preference (least -> most)
-  var preference = ['nginx', 'apache', undefined, 'iana']
-
   Object.keys(db).forEach(function forEachMimeType (type) {
     var mime = db[type]
     var exts = mime.extensions
@@ -169,20 +166,46 @@ function populateMaps (extensions, types) {
     // extension -> mime
     for (var i = 0; i < exts.length; i++) {
       var extension = exts[i]
-
-      if (types[extension]) {
-        var from = preference.indexOf(db[types[extension]].source)
-        var to = preference.indexOf(mime.source)
-
-        if (types[extension] !== 'application/octet-stream' &&
-          (from > to || (from === to && types[extension].slice(0, 12) === 'application/'))) {
-          // skip the remapping
-          continue
-        }
+      types[extension] = _preferredType(extension, types[extension], type)
+
+      // DELETE (eventually): Capture extension->type maps that change as a
+      // result of switching to mime-score.  This is just to help make reviewing
+      // PR #119 easier, and can be removed once that PR is approved.
+      const legacyType = _preferredTypeLegacy(
+        extension,
+        types[extension],
+        type
+      )
+      if (legacyType !== types[extension]) {
+        exports._extensionConflicts.push([extension, legacyType, types[extension]])
       }
-
-      // set the extension -> mime
-      types[extension] = type
     }
   })
 }
+
+// Resolve type conflict using mime-score
+function _preferredType (ext, type0, type1) {
+  var score0 = type0 ? mimeScore(type0, db[type0].source) : 0
+  var score1 = type1 ? mimeScore(type1, db[type1].source) : 0
+
+  return score0 > score1 ? type0 : type1
+}
+
+// Resolve type conflict using pre-mime-score logic
+function _preferredTypeLegacy (ext, type0, type1) {
+  var SOURCE_RANK = ['nginx', 'apache', undefined, 'iana']
+
+  var score0 = type0 ? SOURCE_RANK.indexOf(db[type0].source) : 0
+  var score1 = type1 ? SOURCE_RANK.indexOf(db[type1].source) : 0
+
+  if (
+    exports.types[extension] !== 'application/octet-stream' &&
+    (score0 > score1 ||
+      (score0 === score1 &&
+        exports.types[extension]?.slice(0, 12) === 'application/'))
+  ) {
+    return type0
+  }
+
+  return score0 > score1 ? type0 : type1
+}

mimeScore.js

@@ -0,0 +1,52 @@
+// 'mime-score' back-ported to CommonJS
+
+// Score RFC facets (see https://tools.ietf.org/html/rfc6838#section-3)
+var FACET_SCORES = {
+  'prs.': 100,
+  'x-': 200,
+  'x.': 300,
+  'vnd.': 400,
+  default: 900
+}
+
+// Score mime source (Logic originally from `jshttp/mime-types` module)
+var SOURCE_SCORES = {
+  nginx: 10,
+  apache: 20,
+  iana: 40,
+  default: 30 // definitions added by `jshttp/mime-db` project?
+}
+
+var TYPE_SCORES = {
+  // prefer application/xml over text/xml
+  // prefer application/rtf over text/rtf
+  application: 1,
+
+  // prefer font/woff over application/font-woff
+  font: 2,
+
+  default: 0
+}
+
+/**
+ * Get each component of the score for a mime type.  The sum of these is the
+ * total score.  The higher the score, the more "official" the type.
+ */
+module.exports = function mimeScore (mimeType, source = 'default') {
+  if (mimeType === 'application/octet-stream') {
+    return 0
+  }
+
+  const [type, subtype] = mimeType.split('/')
+
+  const facet = subtype.replace(/(\.|x-).*/, '$1')
+
+  const facetScore = FACET_SCORES[facet] || FACET_SCORES.default
+  const sourceScore = SOURCE_SCORES[source] || SOURCE_SCORES.default
+  const typeScore = TYPE_SCORES[type] || TYPE_SCORES.default
+
+  // All else being equal prefer shorter types
+  const lengthScore = 1 - mimeType.length / 100
+
+  return facetScore + sourceScore + typeScore + lengthScore
+}

test/test.js

@@ -1,4 +1,3 @@
-
 var assert = require('assert')
 var mimeTypes = require('..')
 
@@ -220,8 +219,19 @@ describe('mimeTypes', function () {
       })
 
       it('should return mime type when there is extension, but no path', function () {
-        assert.strictEqual(mimeTypes.lookup('.config.json'), 'application/json')
+        assert.strictEqual(
+          mimeTypes.lookup('.config.json'),
+          'application/json'
+        )
       })
     })
   })
+
+  // Note the changes in extension->type mapping that result from using mime-score
+  describe('extension conflicts', function () {
+    console.warn('Mime-score logic changes extension->type mappings for the following:')
+    for (var [extension, legacy, current] of mimeTypes._extensionConflicts) {
+      console.warn(`* ${extension} -> ${legacy} is now ${current}`)
+    }
+  })
 })