refactor: copy/past mime-helpers (#1246)

followup of #1243

make implementation a copy-past of existing art:
<https://github.com/CycloneDX/cyclonedx-node-yarn/blob/main/src/_helpers.ts>

this should make it easier for
CycloneDX/cyclonedx-javascript-library#1162

---------

Signed-off-by: Jan Kowalleck <[email protected]>

Author: jkowalleck

src/_helpers.ts

@@ -18,7 +18,7 @@ Copyright (c) OWASP Foundation. All Rights Reserved.
 */
 
 import { readFileSync, writeSync } from 'fs'
-import { parse } from 'path'
+import { extname, parse } from 'path'
 
 export function loadJsonFile (path: string): any {
   return JSON.parse(readFileSync(path, 'utf8'))
@@ -58,24 +58,45 @@ export function tryRemoveSecretsFromUrl (url: string): string {
   }
 }
 
-export const LICENSE_FILENAME_PATTERN = /^(?:UN)?LICEN[CS]E|.\.LICEN[CS]E$|^NOTICE$/i
+// region MIME
+
+export type MimeType = string
+
+const MIME_TEXT_PLAIN: MimeType = 'text/plain'
+
+const MAP_TEXT_EXTENSION_MIME: Readonly<Record<string, MimeType>> = {
+  '': MIME_TEXT_PLAIN,
+  // https://www.iana.org/assignments/media-types/media-types.xhtml
+  '.csv': 'text/csv',
+  '.htm': 'text/html',
+  '.html': 'text/html',
+  '.md': 'text/markdown',
+  '.txt': MIME_TEXT_PLAIN,
+  '.rst': 'text/prs.fallenstein.rst',
+  '.xml': 'text/xml', // not `application/xml` -- our scope is text!
+  // add more mime types above this line. pull-requests welcome!
+  // license-specific files
+  '.license': MIME_TEXT_PLAIN,
+  '.licence': MIME_TEXT_PLAIN
+} as const
+
+export function getMimeForTextFile (filename: string): MimeType | undefined {
+  return MAP_TEXT_EXTENSION_MIME[extname(filename).toLowerCase()]
+}
+
 const LICENSE_FILENAME_BASE = new Set(['licence', 'license'])
-const LICENSE_FILENAME_EXT = new Set(['.apache', '.bsd', '.gpl', '.mit'])
-const MAP_TEXT_EXTENSION_MIME = new Map([
-  ['', 'text/plain'],
-  ['.htm', 'text/html'],
-  ['.html', 'text/html'],
-  ['.md', 'text/markdown'],
-  ['.txt', 'text/plain'],
-  ['.rst', 'text/prs.fallenstein.rst'],
-  ['.xml', 'text/xml'],
-  ['.license', 'text/plain'],
-  ['.licence', 'text/plain']
+const LICENSE_FILENAME_EXT = new Set([
+  '.apache',
+  '.bsd',
+  '.gpl',
+  '.mit'
 ])
 
-export function getMimeForLicenseFile (filename: string): string | undefined {
+export function getMimeForLicenseFile (filename: string): MimeType | undefined {
   const { name, ext } = parse(filename.toLowerCase())
   return LICENSE_FILENAME_BASE.has(name) && LICENSE_FILENAME_EXT.has(ext)
-    ? 'text/plain'
-    : MAP_TEXT_EXTENSION_MIME.get(ext)
+    ? MIME_TEXT_PLAIN
+    : MAP_TEXT_EXTENSION_MIME[ext]
 }
+
+// endregion MIME

src/builders.ts

@@ -23,12 +23,7 @@ import * as normalizePackageData from 'normalize-package-data'
 import * as path from 'path'
 import { join } from 'path'
 
-import {
-  getMimeForLicenseFile,
-  isString,
-  LICENSE_FILENAME_PATTERN,
-  loadJsonFile, tryRemoveSecretsFromUrl
-} from './_helpers'
+import { getMimeForLicenseFile, isString, loadJsonFile, tryRemoveSecretsFromUrl } from './_helpers'
 import { makeNpmRunner, type runFunc } from './npmRunner'
 import { PropertyNames, PropertyValueBool } from './properties'
 import { versionCompare } from './versionCompare'
@@ -637,10 +632,12 @@ export class BomBuilder {
     }
   }
 
+  readonly #LICENSE_FILENAME_PATTERN = /^(?:UN)?LICEN[CS]E|.\.LICEN[CS]E$|^NOTICE$/i
+
   private * fetchLicenseEvidence (path: string): Generator<Models.License | null, void, void> {
     const files = readdirSync(path)
     for (const file of files) {
-      if (!LICENSE_FILENAME_PATTERN.test(file)) {
+      if (!this.#LICENSE_FILENAME_PATTERN.test(file)) {
         continue
       }
 

tests/_data/dummy_projects/with-prepared/.gitattributes

@@ -0,0 +1,3 @@
+
+*     linguist-vendored -text
+**/*  linguist-vendored -text

tests/_data/dummy_projects/with-prepared/.gitignore

@@ -0,0 +1,13 @@
+*
+!/.gitignore
+!/.gitattributes
+
+!/node_modules
+!/node_modules/*
+!/node_modules/**/*
+!/package.json
+!/package-lock.json
+
+!/setup.sh
+!/README.md
+!/LICENCE.mit

tests/_data/dummy_projects/with-prepared/LICENCE.mit

@@ -0,0 +1 @@
+this is a dummy license file.

tests/_data/dummy_projects/with-prepared/README.md

@@ -0,0 +1,10 @@
+# setup with prepared data and structure
+
+this setup actually vendors some 3rd-party data in the `node_modules` dir:
+- package manifests
+- license files
+
+this dir is already set up as expected.
+no need to run `npm install`.
+
+if this setup is to be changed or recreated, run `./setup.sh`.