From b7a987208f22fc9bcc1ad003f08dac85e7d7d66c Mon Sep 17 00:00:00 2001
From: Vjacheslav Trushkin <cyberalien@gmail.com>
Date: Sat, 10 Dec 2022 13:17:26 +0200
Subject: [PATCH] fix: support various formats when parsing emoji string

---
 packages/utils/src/emoji/cleanup.ts        |  6 +++++-
 packages/utils/src/emoji/regex/create.ts   | 11 +++++++++++
 packages/utils/src/emoji/regex/parse.ts    |  0
 packages/utils/tests/emoji-cleanup-test.ts | 13 +++++++++++++
 4 files changed, 29 insertions(+), 1 deletion(-)
 create mode 100644 packages/utils/src/emoji/regex/parse.ts

diff --git a/packages/utils/src/emoji/cleanup.ts b/packages/utils/src/emoji/cleanup.ts
index 4f16aa2..1da68f8 100644
--- a/packages/utils/src/emoji/cleanup.ts
+++ b/packages/utils/src/emoji/cleanup.ts
@@ -5,7 +5,11 @@ import { emojiTones, joinerEmoji, vs16Emoji } from './data';
  * Get emoji sequence from string
  */
 export function getEmojiSequenceFromString(value: string): number[] {
-	return value.trim().split(/[\s-]/).map(getEmojiCodePoint);
+	return value
+		.trim()
+		.split(/[^0-9A-F]+/i)
+		.filter((item) => item.length > 0)
+		.map(getEmojiCodePoint);
 }
 
 /**
diff --git a/packages/utils/src/emoji/regex/create.ts b/packages/utils/src/emoji/regex/create.ts
index a09d103..203b337 100644
--- a/packages/utils/src/emoji/regex/create.ts
+++ b/packages/utils/src/emoji/regex/create.ts
@@ -24,6 +24,17 @@ export function createOptimisedRegexForEmojiSequences(
 
 /**
  * Create optimised regex for emojis
+ *
+ * First parameter is array of emojis, entry can be either list of
+ * code points or emoji sequence as a string
+ *
+ * Examples of acceptable strings (case insensitive):
+ * 	'1F636 200D 1F32B FE0F' - space separated UTF32 sequence
+ * 	'1f636-200d-1f32b-fe0f' - dash separated UTF32 sequence
+ *  'd83d-de36-200d-d83c-df2b-fe0f' - dash separated UTF16 sequence
+ *  '\\uD83D\\uDE36\\u200D\\uD83C\\uDF2B\\uFE0F' - UTF16 sequence escaped with '\\u'
+ *
+ * All examples above refer to the same emoji and will generate the same regex result
  */
 export function createOptimisedRegex(
 	emojis: (string | number[])[],
diff --git a/packages/utils/src/emoji/regex/parse.ts b/packages/utils/src/emoji/regex/parse.ts
new file mode 100644
index 0000000..e69de29
diff --git a/packages/utils/tests/emoji-cleanup-test.ts b/packages/utils/tests/emoji-cleanup-test.ts
index fad4c5c..d1f6f18 100644
--- a/packages/utils/tests/emoji-cleanup-test.ts
+++ b/packages/utils/tests/emoji-cleanup-test.ts
@@ -1,4 +1,5 @@
 /* eslint-disable @typescript-eslint/no-non-null-assertion */
+import { convertEmojiSequenceToUTF32 } from '../lib';
 import {
 	getEmojiSequenceFromString,
 	joinEmojiSequences,
@@ -16,6 +17,18 @@ describe('Testing formatting emoji cleanup', () => {
 		);
 		expect(sequence).toEqual([0x1f441, 0xfe0f, 0x200d, 0x1f5e8, 0xfe0f]);
 
+		// Various representations of the same sequence
+		expect(
+			getEmojiSequenceFromString('1f441-fe0f-200d-1f5e8-fe0f')
+		).toEqual(sequence);
+		expect(
+			convertEmojiSequenceToUTF32(
+				getEmojiSequenceFromString(
+					'\\uD83D\\uDC41\\uFE0F\\u200D\\uD83D\\uDDE8\\uFE0F'
+				)
+			)
+		).toEqual(sequence);
+
 		// Split
 		const split = splitEmojiSequences(sequence);
 		expect(split).toEqual([