Remove obsolete MIME types from IANA data

jshttp · May 14, 2022 · bbca7ce · bbca7ce
1 parent c8e4e30
commit bbca7ce
Show file tree

Hide file tree

Showing 4 changed files with 35 additions and 183 deletions.
diff --git a/HISTORY.md b/HISTORY.md
@@ -2,6 +2,7 @@ unreleased
 ==========
 
   * Add new upstream MIME types
+  * Remove obsolete MIME types from IANA data
 
 1.52.0 / 2022-02-21
 ===================

diff --git a/db.json b/db.json
@@ -451,7 +451,7 @@
     "source": "iana"
   },
   "application/ecmascript": {
-    "source": "iana",
+    "source": "apache",
     "compressible": true,
     "extensions": ["ecma"]
   },
@@ -732,7 +732,7 @@
     "extensions": ["class"]
   },
   "application/javascript": {
-    "source": "iana",
+    "source": "apache",
     "charset": "UTF-8",
     "compressible": true,
     "extensions": ["js","mjs"]
@@ -1581,7 +1581,7 @@
     "source": "iana"
   },
   "application/smil": {
-    "source": "iana"
+    "source": "apache"
   },
   "application/smil+xml": {
     "source": "iana",
@@ -1916,10 +1916,6 @@
     "source": "iana",
     "compressible": true
   },
-  "application/vnd.3gpp.mcvideo-affiliation-info+xml": {
-    "source": "iana",
-    "compressible": true
-  },
   "application/vnd.3gpp.mcvideo-info+xml": {
     "source": "iana",
     "compressible": true
@@ -2105,7 +2101,7 @@
     "extensions": ["age"]
   },
   "application/vnd.ah-barcode": {
-    "source": "iana"
+    "source": "apache"
   },
   "application/vnd.ahead.space": {
     "source": "iana",
@@ -2218,7 +2214,7 @@
     "extensions": ["pkpass"]
   },
   "application/vnd.arastra.swi": {
-    "source": "iana"
+    "source": "apache"
   },
   "application/vnd.aristanetworks.swi": {
     "source": "iana",
@@ -2976,11 +2972,11 @@
     "extensions": ["fm","frame","maker","book"]
   },
   "application/vnd.frogans.fnc": {
-    "source": "iana",
+    "source": "apache",
     "extensions": ["fnc"]
   },
   "application/vnd.frogans.ltf": {
-    "source": "iana",
+    "source": "apache",
     "extensions": ["ltf"]
   },
   "application/vnd.fsc.weblaunch": {
@@ -3070,11 +3066,11 @@
     "compressible": true
   },
   "application/vnd.geo+json": {
-    "source": "iana",
+    "source": "apache",
     "compressible": true
   },
   "application/vnd.geocube+xml": {
-    "source": "iana",
+    "source": "apache",
     "compressible": true
   },
   "application/vnd.geogebra.file": {
@@ -3277,7 +3273,7 @@
     "source": "iana"
   },
   "application/vnd.ibm.afplinedata": {
-    "source": "iana"
+    "source": "apache"
   },
   "application/vnd.ibm.electronic-media": {
     "source": "iana"
@@ -3287,7 +3283,7 @@
     "extensions": ["mpy"]
   },
   "application/vnd.ibm.modcap": {
-    "source": "iana",
+    "source": "apache",
     "extensions": ["afp","listafp","list3820"]
   },
   "application/vnd.ibm.rights-management": {
@@ -3363,7 +3359,7 @@
     "compressible": true
   },
   "application/vnd.informix-visionary": {
-    "source": "iana"
+    "source": "apache"
   },
   "application/vnd.infotech.project": {
     "source": "iana"
@@ -4072,7 +4068,7 @@
     "extensions": ["ngdat"]
   },
   "application/vnd.nokia.n-gage.symbian.install": {
-    "source": "iana",
+    "source": "apache",
     "extensions": ["n-gage"]
   },
   "application/vnd.nokia.ncd": {
@@ -4266,7 +4262,7 @@
     "compressible": true
   },
   "application/vnd.oma.bcast.drm-trigger+xml": {
-    "source": "iana",
+    "source": "apache",
     "compressible": true
   },
   "application/vnd.oma.bcast.imd+xml": {
@@ -4297,7 +4293,7 @@
     "source": "iana"
   },
   "application/vnd.oma.bcast.smartcard-trigger+xml": {
-    "source": "iana",
+    "source": "apache",
     "compressible": true
   },
   "application/vnd.oma.bcast.sprov+xml": {
@@ -5639,9 +5635,6 @@
     "source": "iana",
     "extensions": ["cmp"]
   },
-  "application/vnd.youtube.yt": {
-    "source": "iana"
-  },
   "application/vnd.zul": {
     "source": "iana",
     "extensions": ["zir","zirz"]
@@ -6935,7 +6928,7 @@
     "source": "iana"
   },
   "audio/vnd.qcelp": {
-    "source": "iana"
+    "source": "apache"
   },
   "audio/vnd.rhetorex.32kadpcm": {
     "source": "iana"
@@ -7533,7 +7526,7 @@
     "compressible": true
   },
   "message/news": {
-    "source": "iana"
+    "source": "apache"
   },
   "message/partial": {
     "source": "iana",
@@ -7545,7 +7538,7 @@
     "extensions": ["eml","mime"]
   },
   "message/s-http": {
-    "source": "iana"
+    "source": "apache"
   },
   "message/sip": {
     "source": "iana"
@@ -7557,7 +7550,7 @@
     "source": "iana"
   },
   "message/vnd.si.simp": {
-    "source": "iana"
+    "source": "apache"
   },
   "message/vnd.wfa.wsc": {
     "source": "iana",
@@ -7829,7 +7822,7 @@
     "source": "iana"
   },
   "text/ecmascript": {
-    "source": "iana"
+    "source": "apache"
   },
   "text/encaprtp": {
     "source": "iana"
@@ -8106,7 +8099,7 @@
     "source": "iana"
   },
   "text/vnd.si.uricatalogue": {
-    "source": "iana"
+    "source": "apache"
   },
   "text/vnd.sosi": {
     "source": "iana"
@@ -8500,7 +8493,7 @@
     "source": "iana"
   },
   "video/vnd.radgamettools.smacker": {
-    "source": "iana"
+    "source": "apache"
   },
   "video/vnd.sealed.mpeg1": {
     "source": "iana"

diff --git a/scripts/fetch-iana.js b/scripts/fetch-iana.js
@@ -19,7 +19,6 @@ var toArray = require('stream-to-array')
 var writedb = require('./lib/write-db')
 
 var extensionsQuotedRegExp = /^\s*(?:\d\.\s+)?File extension(?:\(s\)|s|)\s?:(?:[^'"\r\n]+)(?:"\.?([0-9a-z_-]+)"|'\.?([0-9a-z_-]+)')/im
-var intendedUsageRegExp = /^\s*(?:(?:\d{1,2}\.|o)\s+)?Intended\s+Usage\s*:\s*(\S+)/im
 var leadingSpacesRegExp = /^\s+/
 var listColonRegExp = /:(?:\s|$)/m
 var nameWithNotesRegExp = /^(\S+)(?: - (.*)$| \((.*)\)$|)/
@@ -33,6 +32,7 @@ var urlReferenceRegExp = /\[(https?:\/\/[^\]]+)]/gi
 
 var CHARSET_DEFAULT_REGEXP = /(?:\bcharset\b[^.]*(?:\.\s+default\s+(?:value\s+)?is|\bdefault[^.]*(?:of|is)|\bmust\s+have\s+the\s+value|\bvalue\s+must\s+be)\s+|\bcharset\s*\(?defaults\s+to\s+|\bdefault\b[^.]*?\bchar(?:set|act[eo]r\s+set)\b[^.]*?(?:of|is)\s+|\bcharset\s+(?:must|is)\s+always\s+(?:be\s+)?)["']?([a-z0-9]+-[a-z0-9-]+)/im
 var EXTENSIONS_REGEXP = /(?:^\s*(?:\d\.\s+)?|\s+[23]\.\s+)[Ff]ile [Ee]xtension(?:\(s\)|s|)\s?:\s+(?:\*\.|\.|)([0-9a-z_-]+|[0-9A-Z_-]+)(?:\s+or\s+(?:\*\.|\.|)([0-9a-z_-]+|[0-9A-Z_-]+)\s*)?(?:\s*[34]\.\s+|\s+[A-Z(]|\s*$)/m
+var INTENDED_USAGE_REGEXP = /^\s*(?:(?:\d{1,2}\.|o)\s+)?Intended\s+Usage\s*:\s*([0-9a-z]+)/im
 var MIME_SUBTYPE_LINE_REGEXP = /^[^:\s-]*\s*(?:MIME )?(?:[Mm]edia )?(?:[Ss]ub ?type|SUB ?TYPE)(?: (?:[Nn]ame|NAME))?\s*:\s+(?:[A-Za-z]+ [Tt]ree\s+(?:- ?)?|(?:[a-z]+ )+- )?([0-9A-Za-z][0-9A-Za-z_.+-]*)(?:\s|$)/m
 var MIME_TYPE_HAS_CHARSET_PARAMETER_REGEXP = /parameters\s*:[^.]*\bcharset\b/im
 
@@ -78,6 +78,11 @@ co(function * () {
       throw new Error('duplicate entry for ' + mime)
     }
 
+    // skip obsoleted mimes
+    if (result.usage === 'obsolete') {
+      return
+    }
+
     json[mime] = {
       charset: result.charset,
       notes: result.notes,
@@ -145,10 +150,12 @@ function addTemplateData (data, options) {
       // use extracted charset
       data.charset = extractTemplateCharset(body)
 
+      // use extracted usage
+      data.usage = extractIntendedUsage(body)
+
       // use extracted extensions
-      var useExt = opts.extensions &&
-        (opts.extensions === true || opts.extensions.test(data.mime))
-      if (useExt && extractIntendedUsage(body) === 'common') {
+      if (data.usage === 'common' && opts.extensions &&
+        (opts.extensions === true || opts.extensions.test(data.mime))) {
         data.extensions = extractTemplateExtensions(body)
       }
     }
@@ -158,7 +165,7 @@ function addTemplateData (data, options) {
 }
 
 function extractIntendedUsage (body) {
-  var match = intendedUsageRegExp.exec(body)
+  var match = INTENDED_USAGE_REGEXP.exec(body)
 
   return match
     ? match[1].toLocaleLowerCase()