Merge pull request #1014 from TriliumNext/refactor_import-mime

refactor: refactor and add tests for `services/import/mime`

src/services/import/mime.spec.ts

@@ -0,0 +1,151 @@
+import { describe, it, expect } from "vitest";
+import mimeService from "./mime.js";
+
+type TestCase<T extends (...args: any) => any, W> = [desc: string, fnParams: Parameters<T>, expected: W];
+
+describe("#getMime", () => {
+    // prettier-ignore
+    const testCases: TestCase<typeof mimeService.getMime, string | false>[] = [
+        [
+            "Dockerfile should be handled correctly",
+            ["Dockerfile"], "text/x-dockerfile"
+        ],
+
+        [
+            "File extension that is defined in EXTENSION_TO_MIME",
+            ["test.py"], "text/x-python"
+        ],
+
+        [
+            "File extension with inconsisten capitalization that is defined in EXTENSION_TO_MIME",
+            ["test.gRoOvY"], "text/x-groovy"
+        ],
+
+        [
+            "File extension that is not defined in EXTENSION_TO_MIME should use mimeTypes.lookup",
+            ["test.zip"], "application/zip"
+        ],
+
+        [
+            "unknown MIME type not recognized by mimeTypes.lookup",
+            ["test.fake"], false
+        ],
+    ];
+
+    testCases.forEach((testCase) => {
+        const [testDesc, fnParams, expected] = testCase;
+        it(`${testDesc}: '${fnParams} should return '${expected}'`, () => {
+            const actual = mimeService.getMime(...fnParams);
+            expect(actual).toEqual(expected);
+        });
+    });
+});
+
+describe("#getType", () => {
+    // prettier-ignore
+    const testCases: TestCase<typeof mimeService.getType, string>[] = [
+        [
+            "w/ no import options set and mime type empty – it should return 'file'",
+            [{}, ""], "file"
+        ],
+
+        [
+            "w/ no import options set and non-text or non-code mime type – it should return 'file'",
+            [{}, "application/zip"], "file"
+        ],
+
+        [
+            "w/ import options set and an image mime type – it should return 'image'", 
+            [{}, "image/jpeg"], "image"
+        ],
+
+        [
+            "w/ image mime type and codeImportedAsCode: true – it should still return 'image'",
+            [{codeImportedAsCode: true}, "image/jpeg"], "image"
+        ],
+
+        [
+            "w/ image mime type and textImportedAsText: true – it should still return 'image'",
+            [{textImportedAsText: true}, "image/jpeg"], "image"
+        ],
+
+        [
+            "w/ codeImportedAsCode: true and a mime type that is in CODE_MIME_TYPES – it should return 'code'", 
+            [{codeImportedAsCode: true}, "text/css"], "code"
+        ],
+
+        [
+            "w/ codeImportedAsCode: false and a mime type that is in CODE_MIME_TYPES – it should return 'file' not 'code'", 
+            [{codeImportedAsCode: false}, "text/css"], "file"
+        ],
+
+        [
+            "w/ textImportedAsText: true and 'text/html' mime type – it should return 'text'", 
+            [{textImportedAsText: true}, "text/html"], "text"
+        ],
+
+        [
+            "w/ textImportedAsText: true and 'text/markdown' mime type – it should return 'text'", 
+            [{textImportedAsText: true}, "text/markdown"], "text"
+        ],
+
+        [
+            "w/ textImportedAsText: true and 'text/x-markdown' mime type – it should return 'text'",
+            [{textImportedAsText: true}, "text/x-markdown"], "text"
+        ],
+
+        [
+            "w/ textImportedAsText: false and 'text/x-markdown' mime type – it should return 'file'",
+            [{textImportedAsText: false}, "text/x-markdown"], "file"
+        ],
+
+        [
+            "w/ textImportedAsText: false and 'text/html' mime type – it should return 'file'",
+            [{textImportedAsText: false}, "text/html"], "file"
+        ],
+
+    ]
+
+    testCases.forEach((testCase) => {
+        const [desc, fnParams, expected] = testCase;
+        it(desc, () => {
+            const actual = mimeService.getType(...fnParams);
+            expect(actual).toEqual(expected);
+        });
+    });
+});
+
+describe("#normalizeMimeType", () => {
+    // prettier-ignore
+    const testCases: TestCase<typeof mimeService.normalizeMimeType, string | undefined>[] = [
+
+        [
+            "empty mime should return undefined", 
+            [""], undefined
+        ],
+        [
+            "a mime that's defined in CODE_MIME_TYPES should return the same mime", 
+            ["text/x-python"], "text/x-python"
+        ],
+        [
+            "a mime (with capitalization inconsistencies) that's defined in CODE_MIME_TYPES should return the same mime in lowercase", 
+            ["text/X-pYthOn"], "text/x-python"
+        ],
+        [
+            "a mime that's non defined in CODE_MIME_TYPES should return undefined", 
+            ["application/zip"], undefined
+        ],
+        [
+            "a mime that's defined in CODE_MIME_TYPES with a 'rewrite rule' should return the rewritten mime", 
+            ["text/markdown"], "text/x-markdown"
+        ]
+    ];
+
+    testCases.forEach((testCase) => {
+        const [desc, fnParams, expected] = testCase;
+        it(desc, () => {
+            const actual = mimeService.normalizeMimeType(...fnParams);
+            expect(actual).toEqual(expected);
+        });
+    });
+});

src/services/import/mime.ts

@@ -4,107 +4,109 @@ import mimeTypes from "mime-types";
 import path from "path";
 import type { TaskData } from "../task_context_interface.js";
 
-const CODE_MIME_TYPES: Record<string, boolean | string> = {
+const CODE_MIME_TYPES = new Set([
-    "text/plain": true,
+    "application/json",
-    "text/x-csrc": true,
+    "message/http",
-    "text/x-c++src": true,
+    "text/css",
-    "text/x-csharp": true,
+    "text/html",
-    "text/x-clojure": true,
+    "text/plain",
-    "text/css": true,
+    "text/x-clojure",
-    "text/x-dockerfile": true,
+    "text/x-csharp",
-    "text/x-erlang": true,
+    "text/x-c++src",
-    "text/x-feature": true,
+    "text/x-csrc",
-    "text/x-go": true,
+    "text/x-dockerfile",
-    "text/x-groovy": true,
+    "text/x-erlang",
-    "text/x-haskell": true,
+    "text/x-feature",
-    "text/html": true,
+    "text/x-go",
-    "message/http": true,
+    "text/x-groovy",
-    "text/x-java": true,
+    "text/x-haskell",
-    "application/javascript": "application/javascript;env=frontend",
+    "text/x-java",
-    "application/x-javascript": "application/javascript;env=frontend",
+    "text/x-kotlin",
-    "application/json": true,
+    "text/x-lua",
-    "text/x-kotlin": true,
+    "text/x-markdown",
-    "text/x-stex": true,
+    "text/xml",
-    "text/x-lua": true,
+    "text/x-objectivec",
+    "text/x-pascal",
+    "text/x-perl",
+    "text/x-php",
+    "text/x-python",
+    "text/x-ruby",
+    "text/x-rustsrc",
+    "text/x-scala",
+    "text/x-sh",
+    "text/x-sql",
+    "text/x-stex",
+    "text/x-swift",
+    "text/x-yaml"
+]);
+
+const CODE_MIME_TYPES_OVERRIDE = new Map<string, string>([
+    ["application/javascript", "application/javascript;env=frontend"],
+    ["application/x-javascript", "application/javascript;env=frontend"],
     // possibly later migrate to text/markdown as primary MIME
-    "text/markdown": "text/x-markdown",
+    ["text/markdown", "text/x-markdown"]
-    "text/x-markdown": true,
+]);
-    "text/x-objectivec": true,
-    "text/x-pascal": true,
-    "text/x-perl": true,
-    "text/x-php": true,
-    "text/x-python": true,
-    "text/x-ruby": true,
-    "text/x-rustsrc": true,
-    "text/x-scala": true,
-    "text/x-sh": true,
-    "text/x-sql": true,
-    "text/x-swift": true,
-    "text/xml": true,
-    "text/x-yaml": true
-};
 
 // extensions missing in mime-db
-const EXTENSION_TO_MIME: Record<string, string> = {
+const EXTENSION_TO_MIME = new Map<string, string>([
-    ".c": "text/x-csrc",
+    [".c", "text/x-csrc"],
-    ".cs": "text/x-csharp",
+    [".cs", "text/x-csharp"],
-    ".clj": "text/x-clojure",
+    [".clj", "text/x-clojure"],
-    ".erl": "text/x-erlang",
+    [".erl", "text/x-erlang"],
-    ".hrl": "text/x-erlang",
+    [".hrl", "text/x-erlang"],
-    ".feature": "text/x-feature",
+    [".feature", "text/x-feature"],
-    ".go": "text/x-go",
+    [".go", "text/x-go"],
-    ".groovy": "text/x-groovy",
+    [".groovy", "text/x-groovy"],
-    ".hs": "text/x-haskell",
+    [".hs", "text/x-haskell"],
-    ".lhs": "text/x-haskell",
+    [".lhs", "text/x-haskell"],
-    ".http": "message/http",
+    [".http", "message/http"],
-    ".kt": "text/x-kotlin",
+    [".kt", "text/x-kotlin"],
-    ".m": "text/x-objectivec",
+    [".m", "text/x-objectivec"],
-    ".py": "text/x-python",
+    [".py", "text/x-python"],
-    ".rb": "text/x-ruby",
+    [".rb", "text/x-ruby"],
-    ".scala": "text/x-scala",
+    [".scala", "text/x-scala"],
-    ".swift": "text/x-swift"
+    [".swift", "text/x-swift"]
-};
+]);
 
 /** @returns false if MIME is not detected */
 function getMime(fileName: string) {
-    if (fileName.toLowerCase() === "dockerfile") {
+    const fileNameLc = fileName?.toLowerCase();
+
+    if (fileNameLc === "dockerfile") {
         return "text/x-dockerfile";
     }
 
-    const ext = path.extname(fileName).toLowerCase();
+    const ext = path.extname(fileNameLc);
+    const mimeFromExt = EXTENSION_TO_MIME.get(ext);
 
-    if (ext in EXTENSION_TO_MIME) {
+    return mimeFromExt || mimeTypes.lookup(fileNameLc);
-        return EXTENSION_TO_MIME[ext];
-    }
-
-    return mimeTypes.lookup(fileName);
 }
 
 function getType(options: TaskData, mime: string) {
-    mime = mime ? mime.toLowerCase() : "";
+    const mimeLc = mime?.toLowerCase();
 
-    if (options.textImportedAsText && (mime === "text/html" || ["text/markdown", "text/x-markdown"].includes(mime))) {
+    switch (true) {
+        case options.textImportedAsText && ["text/html", "text/markdown", "text/x-markdown"].includes(mimeLc):
             return "text";
-    } else if (options.codeImportedAsCode && mime in CODE_MIME_TYPES) {
+
+        case options.codeImportedAsCode && CODE_MIME_TYPES.has(mimeLc):
             return "code";
-    } else if (mime.startsWith("image/")) {
+
+        case mime.startsWith("image/"):
             return "image";
-    } else {
+
+        default:
             return "file";
     }
 }
 
 function normalizeMimeType(mime: string) {
-    mime = mime ? mime.toLowerCase() : "";
+    const mimeLc = mime.toLowerCase();
-    const mappedMime = CODE_MIME_TYPES[mime];
 
-    if (mappedMime === true) {
+    //prettier-ignore
-        return mime;
+    return CODE_MIME_TYPES.has(mimeLc)
-    } else if (typeof mappedMime === "string") {
+        ? mimeLc
-        return mappedMime;
+        : CODE_MIME_TYPES_OVERRIDE.get(mimeLc);
-    }
-
-    return undefined;
 }
 
 export default {