Add extensions from IANA for more MIME types

Alejandrocontrera · May 14, 2022 · c1b21ca · c1b21ca
1 parent bbca7ce
commit c1b21ca
Show file tree

Hide file tree

Showing 5 changed files with 37 additions and 11 deletions.
diff --git a/HISTORY.md b/HISTORY.md
@@ -1,6 +1,7 @@
 unreleased
 ==========
 
+  * Add extensions from IANA for more MIME types
   * Add new upstream MIME types
   * Remove obsolete MIME types from IANA data
 

diff --git a/db.json b/db.json
@@ -1191,7 +1191,7 @@
   },
   "application/pgp-signature": {
     "source": "iana",
-    "extensions": ["asc","sig"]
+    "extensions": ["sig","asc"]
   },
   "application/pics-rules": {
     "source": "apache",
@@ -1307,7 +1307,8 @@
   },
   "application/prs.xsf+xml": {
     "source": "iana",
-    "compressible": true
+    "compressible": true,
+    "extensions": ["xsf"]
   },
   "application/pskc+xml": {
     "source": "iana",
@@ -4879,7 +4880,8 @@
   },
   "application/vnd.pwg-xhtml-print+xml": {
     "source": "iana",
-    "compressible": true
+    "compressible": true,
+    "extensions": ["xhtm"]
   },
   "application/vnd.qualcomm.brew-app-res": {
     "source": "iana"
@@ -5387,7 +5389,7 @@
   "application/vnd.uoml+xml": {
     "source": "iana",
     "compressible": true,
-    "extensions": ["uoml"]
+    "extensions": ["uoml","uo"]
   },
   "application/vnd.uplanet.alert": {
     "source": "iana"
@@ -7106,7 +7108,7 @@
   "image/bmp": {
     "source": "iana",
     "compressible": true,
-    "extensions": ["bmp"]
+    "extensions": ["bmp","dib"]
   },
   "image/cgm": {
     "source": "iana",
@@ -7242,7 +7244,7 @@
   },
   "image/prs.btif": {
     "source": "iana",
-    "extensions": ["btif"]
+    "extensions": ["btif","btf"]
   },
   "image/prs.pti": {
     "source": "iana",
@@ -7337,7 +7339,8 @@
     "source": "iana"
   },
   "image/vnd.mozilla.apng": {
-    "source": "iana"
+    "source": "iana",
+    "extensions": ["apng"]
   },
   "image/vnd.ms-dds": {
     "compressible": true,
@@ -7668,7 +7671,8 @@
     "extensions": ["x_t"]
   },
   "model/vnd.pytha.pyox": {
-    "source": "iana"
+    "source": "iana",
+    "extensions": ["pyo","pyox"]
   },
   "model/vnd.rosette.annotated-data-model": {
     "source": "iana"

diff --git a/scripts/fetch-iana.js b/scripts/fetch-iana.js
@@ -16,6 +16,7 @@ var getRawBody = require('raw-body')
 var cogent = require('cogent')
 var parser = require('csv-parse')
 var toArray = require('stream-to-array')
+var typer = require('media-typer')
 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
@@ -31,7 +32,7 @@ var trimQuotesRegExp = /^"|"$/gm
 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 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*,)\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
@@ -90,7 +91,10 @@ co(function * () {
     }
 
     // keep unambigious extensions
-    var extensions = (result.extensions || []).filter(function (ext) { return exts[ext] === 1 })
+    var extensions = (result.extensions || []).filter(function (ext) {
+      return exts[ext] === 1 || typer.parse(mime).subtype === ext
+    })
+
     if (extensions.length !== 0) {
       json[mime].extensions = extensions
     }

diff --git a/src/custom-types.json b/src/custom-types.json
@@ -111,6 +111,12 @@
   "application/pgp-encrypted": {
     "compressible": false
   },
+  "application/pgp-keys": {
+    "extensions": ["asc"],
+    "sources": [
+      "http://www.iana.org/assignments/media-types/application/pgp-keys"
+    ]
+  },
   "application/postscript": {
     "compressible": true
   },

diff --git a/src/iana-types.json b/src/iana-types.json
@@ -1652,13 +1652,13 @@
     ]
   },
   "application/pgp-keys": {
-    "extensions": ["asc"],
     "sources": [
       "https://tools.ietf.org/rfc/rfc3156.txt",
       "https://www.iana.org/assignments/media-types/application/pgp-keys"
     ]
   },
   "application/pgp-signature": {
+    "extensions": ["sig"],
     "sources": [
       "https://tools.ietf.org/rfc/rfc3156.txt",
       "https://www.iana.org/assignments/media-types/application/pgp-signature"
@@ -1825,6 +1825,7 @@
     ]
   },
   "application/prs.xsf+xml": {
+    "extensions": ["xsf"],
     "sources": [
       "https://www.iana.org/assignments/media-types/application/prs.xsf+xml"
     ]
@@ -3535,6 +3536,7 @@
     ]
   },
   "application/vnd.dece.ttml+xml": {
+    "extensions": ["uvt","uvvt"],
     "sources": [
       "https://www.iana.org/assignments/media-types/application/vnd.dece.ttml+xml"
     ]
@@ -3855,6 +3857,7 @@
     ]
   },
   "application/vnd.eszigno3+xml": {
+    "extensions": ["es3","et3"],
     "sources": [
       "https://www.iana.org/assignments/media-types/application/vnd.eszigno3+xml"
     ]
@@ -5808,6 +5811,7 @@
     ]
   },
   "application/vnd.oma.dd2+xml": {
+    "extensions": ["dd2"],
     "sources": [
       "https://www.iana.org/assignments/media-types/application/vnd.oma.dd2+xml"
     ]
@@ -6524,6 +6528,7 @@
     ]
   },
   "application/vnd.pwg-xhtml-print+xml": {
+    "extensions": ["xhtm"],
     "sources": [
       "https://www.iana.org/assignments/media-types/application/vnd.pwg-xhtml-print+xml"
     ]
@@ -7126,6 +7131,7 @@
     ]
   },
   "application/vnd.uoml+xml": {
+    "extensions": ["uoml","uo"],
     "sources": [
       "https://www.iana.org/assignments/media-types/application/vnd.uoml+xml"
     ]
@@ -8761,6 +8767,7 @@
     ]
   },
   "image/bmp": {
+    "extensions": ["bmp","dib"],
     "sources": [
       "https://tools.ietf.org/rfc/rfc7903.txt",
       "https://www.iana.org/assignments/media-types/image/bmp"
@@ -8948,6 +8955,7 @@
     ]
   },
   "image/prs.btif": {
+    "extensions": ["btif","btf"],
     "sources": [
       "https://www.iana.org/assignments/media-types/image/prs.btif"
     ]
@@ -9014,6 +9022,7 @@
     ]
   },
   "image/vnd.djvu": {
+    "extensions": ["djvu","djv"],
     "sources": [
       "https://www.iana.org/assignments/media-types/image/vnd.djvu"
     ]
@@ -9081,6 +9090,7 @@
     ]
   },
   "image/vnd.mozilla.apng": {
+    "extensions": ["apng"],
     "sources": [
       "https://www.iana.org/assignments/media-types/image/vnd.mozilla.apng"
     ]
@@ -9422,6 +9432,7 @@
     ]
   },
   "model/vnd.pytha.pyox": {
+    "extensions": ["pyo","pyox"],
     "sources": [
       "https://www.iana.org/assignments/media-types/model/vnd.pytha.pyox"
     ]