From 2524fb3272171a166f777bbd0486aef6ab0ca85e Mon Sep 17 00:00:00 2001
From: anatawa12 <anatawa12@icloud.com>
Date: Thu, 9 Jan 2025 13:38:59 +0900
Subject: [PATCH] refactor: move `splitSegments` to another file

---
 packages/backend/src/core/MfmService.ts       | 45 +----------------
 packages/backend/src/misc/split-segments.ts   | 48 +++++++++++++++++++
 .../backend/test/unit/misc/split-segments.ts  | 20 ++++++++
 3 files changed, 69 insertions(+), 44 deletions(-)
 create mode 100644 packages/backend/src/misc/split-segments.ts
 create mode 100644 packages/backend/test/unit/misc/split-segments.ts

diff --git a/packages/backend/src/core/MfmService.ts b/packages/backend/src/core/MfmService.ts
index befe443d98..90a657b484 100644
--- a/packages/backend/src/core/MfmService.ts
+++ b/packages/backend/src/core/MfmService.ts
@@ -14,6 +14,7 @@ import { intersperse } from '@/misc/prelude/array.js';
 import { normalizeForSearch } from '@/misc/normalize-for-search.js';
 import type { IMentionedRemoteUsers } from '@/models/Note.js';
 import { bindThis } from '@/decorators.js';
+import { splitSegments } from '@/misc/split-segments.js';
 import type { DefaultTreeAdapterMap } from 'parse5';
 import type * as mfm from 'mfm-js';
 
@@ -250,50 +251,6 @@ export class MfmService {
 			}
 		}
 
-		/**
-		 * textをregexesで分割するが、分割するときに regexes にマッチした部分も含める。
-		 */
-		function splitSegments(text: string, regexes: RegExp[]): [regexIdx: number, text: string][] {
-			const result: [regexIdx: number, text: string][] = [];
-
-			let rest = text;
-			for (;;) {
-				let matchRegex: [number, RegExpExecArray] | null = null;
-
-				for (let i = 0; i < regexes.length; i++) {
-					const regex = regexes[i];
-					regex.lastIndex = 0;
-					const matchCurrent = regex.exec(rest);
-					if (matchCurrent) {
-						if (matchRegex != null) {
-							if (matchCurrent.index < matchRegex[1].index) {
-								matchRegex = [i, matchCurrent];
-							}
-						} else {
-							matchRegex = [i, matchCurrent];
-						}
-					}
-				}
-
-				if (matchRegex != null) {
-					const [i, match] = matchRegex;
-
-					const head = rest.slice(0, match.index);
-					const segment = match[0];
-					const tail = rest.slice(match.index + segment.length);
-
-					result.push([-1, head]);
-					result.push([i, segment]);
-					rest = tail;
-				} else {
-					result.push([-1, rest]);
-					break;
-				}
-			}
-
-			return result;
-		}
-
 		const emojiCodeRegex = /(?<![a-z0-9]):[a-z0-9_]+:(?![a-z0-9])/i;
 
 		return splitSegments(text, [twemojiRegex.default, emojiCodeRegex])
diff --git a/packages/backend/src/misc/split-segments.ts b/packages/backend/src/misc/split-segments.ts
new file mode 100644
index 0000000000..07ee5fa7c1
--- /dev/null
+++ b/packages/backend/src/misc/split-segments.ts
@@ -0,0 +1,48 @@
+/*
+ * SPDX-FileCopyrightText: syuilo and misskey-project
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+/**
+ * textをregexesで分割するが、分割するときに regexes にマッチした部分も含める。
+ */
+export function splitSegments(text: string, regexes: RegExp[]): [regexIdx: number, text: string][] {
+	const result: [regexIdx: number, text: string][] = [];
+
+	let rest = text;
+	for (;;) {
+		let matchRegex: [number, RegExpExecArray] | null = null;
+
+		for (let i = 0; i < regexes.length; i++) {
+			const regex = regexes[i];
+			regex.lastIndex = 0;
+			const matchCurrent = regex.exec(rest);
+			if (matchCurrent) {
+				if (matchRegex != null) {
+					if (matchCurrent.index < matchRegex[1].index) {
+						matchRegex = [i, matchCurrent];
+					}
+				} else {
+					matchRegex = [i, matchCurrent];
+				}
+			}
+		}
+
+		if (matchRegex != null) {
+			const [i, match] = matchRegex;
+
+			const head = rest.slice(0, match.index);
+			const segment = match[0];
+			const tail = rest.slice(match.index + segment.length);
+
+			result.push([-1, head]);
+			result.push([i, segment]);
+			rest = tail;
+		} else {
+			result.push([-1, rest]);
+			break;
+		}
+	}
+
+	return result;
+}
diff --git a/packages/backend/test/unit/misc/split-segments.ts b/packages/backend/test/unit/misc/split-segments.ts
new file mode 100644
index 0000000000..41df9b6c61
--- /dev/null
+++ b/packages/backend/test/unit/misc/split-segments.ts
@@ -0,0 +1,20 @@
+/*
+ * SPDX-FileCopyrightText: syuilo and misskey-project
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+import { describe, expect, test } from '@jest/globals';
+import { splitSegments } from '@/misc/split-segments.js';
+
+describe('misc:split-segments', () => {
+	test('simple', () => {
+		expect(splitSegments('abcdefghijklmn', [/c/])).toStrictEqual([[-1, 'ab'], [0, 'c'], [-1, 'defghijklmn']]);
+		expect(splitSegments('abcdefgabcdefg', [/c/])).toStrictEqual([[-1, 'ab'], [0, 'c'], [-1, 'defgab'], [0, 'c'], [-1, 'defg']]);
+
+		expect(splitSegments('abcdefghijklmn', [/c/, /x/])).toStrictEqual([[-1, 'ab'], [0, 'c'], [-1, 'defghijklmn']]);
+		expect(splitSegments('abcdefghijklmn', [/x/, /c/])).toStrictEqual([[-1, 'ab'], [1, 'c'], [-1, 'defghijklmn']]);
+	});
+	test('match multiple regex', () => {
+		expect(splitSegments('abcdefgabcdefg', [/c/, /f/])).toStrictEqual([[-1, 'ab'], [0, 'c'], [-1, 'de'], [1, 'f'], [-1, 'gab'], [0, 'c'], [-1, 'de'], [1, 'f'], [-1, 'g']]);
+	});
+});