From 21ed7633875b71881ad2b8b403291c3ee3ea2644 Mon Sep 17 00:00:00 2001
From: Aleksandr Hovhannisyan <aleksandrhovhannisyan@gmail.com>
Date: Sat, 7 Sep 2024 12:07:33 -0500
Subject: [PATCH] Add support for WebVTT (#658)

---
 core.d.ts                         |  2 ++
 core.js                           | 12 ++++++++++++
 fixture/fixture-vtt-eof.vtt       |  1 +
 fixture/fixture-vtt-linebreak.vtt |  4 ++++
 fixture/fixture-vtt-space.vtt     |  2 ++
 fixture/fixture-vtt-tab.vtt       |  2 ++
 package.json                      |  3 ++-
 readme.md                         |  1 +
 supported.js                      |  2 ++
 test.js                           |  6 ++++++
 10 files changed, 34 insertions(+), 1 deletion(-)
 create mode 100644 fixture/fixture-vtt-eof.vtt
 create mode 100644 fixture/fixture-vtt-linebreak.vtt
 create mode 100644 fixture/fixture-vtt-space.vtt
 create mode 100644 fixture/fixture-vtt-tab.vtt

diff --git a/core.d.ts b/core.d.ts
index 16179ec9..1f1ad94f 100644
--- a/core.d.ts
+++ b/core.d.ts
@@ -165,6 +165,7 @@ export type FileExtension =
 	| 'icc'
 	| 'fbx'
 	| 'vsdx'
+	| 'vtt'
 	; // eslint-disable-line semi-style
 
 export type MimeType =
@@ -270,6 +271,7 @@ export type MimeType =
 	| 'audio/x-musepack'
 	| 'text/calendar'
 	| 'text/vcard'
+	| 'text/vtt'
 	| 'model/gltf-binary'
 	| 'application/vnd.tcpdump.pcap'
 	| 'audio/x-dsf' // Non-standard
diff --git a/core.js b/core.js
index 9444b66a..81787c3d 100644
--- a/core.js
+++ b/core.js
@@ -1139,6 +1139,18 @@ export class FileTypeParser {
 			}
 		}
 
+		if (
+			this.checkString('WEBVTT')
+			&&	(
+				// One of LF, CR, tab, space, or end of file must follow "WEBVTT" per the spec (see `fixture/fixture-vtt-*.vtt` for examples). Note that `\0` is technically the null character (there is no such thing as an EOF character). However, checking for `\0` gives us the same result as checking for the end of the stream.
+				(['\n', '\r', '\t', ' ', '\0'].some(char7 => this.checkString(char7, {offset: 6}))))
+		) {
+			return {
+				ext: 'vtt',
+				mime: 'text/vtt',
+			};
+		}
+
 		// -- 8-byte signatures --
 
 		if (this.check([0x89, 0x50, 0x4E, 0x47, 0x0D, 0x0A, 0x1A, 0x0A])) {
diff --git a/fixture/fixture-vtt-eof.vtt b/fixture/fixture-vtt-eof.vtt
new file mode 100644
index 00000000..af1827dd
--- /dev/null
+++ b/fixture/fixture-vtt-eof.vtt
@@ -0,0 +1 @@
+WEBVTT
\ No newline at end of file
diff --git a/fixture/fixture-vtt-linebreak.vtt b/fixture/fixture-vtt-linebreak.vtt
new file mode 100644
index 00000000..ed33e96d
--- /dev/null
+++ b/fixture/fixture-vtt-linebreak.vtt
@@ -0,0 +1,4 @@
+WEBVTT
+
+00:11.000 --> 00:13.000
+<v Speaker Name>Test WEBVTT prefix followed by a line break
\ No newline at end of file
diff --git a/fixture/fixture-vtt-space.vtt b/fixture/fixture-vtt-space.vtt
new file mode 100644
index 00000000..e3f02d79
--- /dev/null
+++ b/fixture/fixture-vtt-space.vtt
@@ -0,0 +1,2 @@
+WEBVTT 00:11.000 --> 00:13.000
+<v Speaker Name>Test WEBVTT prefix followed by a space
\ No newline at end of file
diff --git a/fixture/fixture-vtt-tab.vtt b/fixture/fixture-vtt-tab.vtt
new file mode 100644
index 00000000..5349f53d
--- /dev/null
+++ b/fixture/fixture-vtt-tab.vtt
@@ -0,0 +1,2 @@
+WEBVTT	00:11.000 --> 00:13.000
+<v Speaker Name>Test WEBVTT prefix followed by a tab
\ No newline at end of file
diff --git a/package.json b/package.json
index 08f1ca3b..f86f5232 100644
--- a/package.json
+++ b/package.json
@@ -215,7 +215,8 @@
 		"avro",
 		"icc",
 		"fbx",
-		"vsdx"
+		"vsdx",
+		"vtt"
 	],
 	"dependencies": {
 		"get-stream": "^9.0.1",
diff --git a/readme.md b/readme.md
index 21e540ae..5ef754cf 100644
--- a/readme.md
+++ b/readme.md
@@ -507,6 +507,7 @@ console.log(fileType);
 - [`vcf`](https://en.wikipedia.org/wiki/VCard) - vCard
 - [`voc`](https://wiki.multimedia.cx/index.php/Creative_Voice) - Creative Voice File
 - [`vsdx`](https://en.wikipedia.org/wiki/Microsoft_Visio) - Microsoft Visio File
+- [`vtt`](https://en.wikipedia.org/wiki/WebVTT) - WebVTT File (for video captions)
 - [`wasm`](https://en.wikipedia.org/wiki/WebAssembly) - WebAssembly intermediate compiled format
 - [`wav`](https://en.wikipedia.org/wiki/WAV) - Waveform Audio file
 - [`webm`](https://en.wikipedia.org/wiki/WebM) - Web video file
diff --git a/supported.js b/supported.js
index 24feb8fe..b64f5b59 100644
--- a/supported.js
+++ b/supported.js
@@ -152,6 +152,7 @@ export const extensions = [
 	'icc',
 	'fbx',
 	'vsdx',
+	'vtt',
 ];
 
 export const mimeTypes = [
@@ -257,6 +258,7 @@ export const mimeTypes = [
 	'audio/x-musepack',
 	'text/calendar',
 	'text/vcard',
+	'text/vtt',
 	'model/gltf-binary',
 	'application/vnd.tcpdump.pcap',
 	'audio/x-dsf', // Non-standard
diff --git a/test.js b/test.js
index 3d4761a9..3b937533 100644
--- a/test.js
+++ b/test.js
@@ -261,6 +261,12 @@ const names = {
 		'fixture-vsdx',
 		'fixture-vstx',
 	],
+	vtt: [
+		'fixture-vtt-linebreak',
+		'fixture-vtt-space',
+		'fixture-vtt-tab',
+		'fixture-vtt-eof',
+	],
 };
 
 // Define an entry here only if the file type has potential