fix: remove duplicated mime types and prefer iana source (#7590)

* fix: remove duplicated mime types and prefer iana source

* refactor: use raw string literals just to be safe

* code review: prefer specific types over octet stream

libtransmission/mime-types.js

@@ -1,7 +1,8 @@
 #!/usr/bin/env node
+import fs from 'node:fs';
 
 const copyright =
-`// This file was generated with libtransmission/mime-types.js
+  `// This file was generated with libtransmission/mime-types.js
 // DO NOT EDIT MANUALLY
 
 // This file Copyright © Mnemosyne LLC.
@@ -9,9 +10,6 @@ const copyright =
 // or any future license endorsed by Mnemosyne LLC.
 // License text can be found in the licenses/ folder.`;
 
-const fs = require('fs');
-const https = require('https');
-
 // https://github.com/jshttp/mime-db
 // > If you're crazy enough to use this in the browser, you can just grab
 // > the JSON file using jsDelivr. It is recommended to replace master with
@@ -21,24 +19,49 @@ const https = require('https');
 // If the format changes, we'll just update this script.
 const url = 'https://cdn.jsdelivr.net/gh/jshttp/mime-db@master/db.json';
 
-https.get(url, (res) => {
+async function main() {
-  res.setEncoding('utf8');
+  const response = await fetch(url);
-  const chunks = [];
+  const mime_types = await response.json()
-  res.on('data', (chunk) => chunks.push(chunk));
+
-  res.on('end', () => {
+  const extensions = Object.entries(mime_types)
-    try {
+    .reduce((acc, [mime_type, info]) => {
-      const suffixes = [];
+      const { extensions, ...rest } = info;
-      const mime_types = JSON.parse(chunks.join(''));
+      for (const extension of extensions || [])
-      for (const [mime_type, info] of Object.entries(mime_types)) {
+        acc.push({ mime_type, extension, info: rest });
-        for (const suffix of info?.extensions || []) {
+      return acc;
-          suffixes.push([ suffix, mime_type ]);
+    }, [])
-        }
+    .sort((lhs, rhs) => {
-      }
+      // Sort by extension
+      const extension_order = lhs.extension.localeCompare(rhs.extension);
+      if (extension_order !== 0)
+        return extension_order;
+
+      // Prefer entries that has a trusted source
+      const lhs_has_source = lhs.info?.source ? 0 : 1;
+      const rhs_has_source = rhs.info?.source ? 0 : 1;
+      const has_source = lhs_has_source - rhs_has_source;
+      if (has_source !== 0)
+        return has_source;
+
+      // Prefer a specific type over the generic octet stream
+      const lhs_is_octet = lhs.mime_type === 'application/octet-stream' ? 1 : 0;
+      const rhs_is_octet = rhs.mime_type === 'application/octet-stream' ? 1 : 0;
+      const is_octet = lhs_is_octet - rhs_is_octet;
+      if (is_octet !== 0)
+        return is_octet;
+
+      // Prefer iana source
+      const lhs_is_iana = lhs.info?.source === 'iana' ? 0 : 1;
+      const rhs_is_iana = rhs.info?.source === 'iana' ? 0 : 1;
+      return lhs_is_iana - rhs_is_iana;
+    })
+    .filter(({ extension }, pos, arr) => pos === 0 || extension !== arr[pos - 1].extension);
+
+  const mime_type_lines = extensions
+    .map(({ extension, mime_type }) => `      { R"(${extension})", R"(${mime_type})" }`)
+    .join(',\n')
+    .trim();
 
-      const mime_type_lines = suffixes
-        .map(([suffix, mime_type]) => `      { "${suffix}", "${mime_type}" }`)
-        .sort()
-        .join(',\n');
   fs.writeFileSync('mime-types.h', `${copyright}
 
 #pragma once
@@ -52,13 +75,10 @@ struct mime_type_suffix
     std::string_view mime_type;
 };
 
-inline auto constexpr MimeTypeSuffixes = std::array<mime_type_suffix, ${suffixes.length}>{
+inline auto constexpr MimeTypeSuffixes = std::array<mime_type_suffix, ${extensions.length}>{
-    { ${mime_type_lines.trim()} }
+    { ${mime_type_lines} }
 };
 `);
-    } catch (e) {
+}
-      console.error(e.message);
-    }
-  });
-});
 
+main().catch(console.error);