From 2056e056a4bd2850d5b1b90ea3b8c0e5a9cfacb7 Mon Sep 17 00:00:00 2001
From: rlaphoenix <rlaphoenix@pm.me>
Date: Thu, 18 Jan 2024 16:23:31 +0000
Subject: [PATCH] Unescape HTML Entities in Subtitles after Downloading

This fixes some Subtitles having e.g., `&amp;` instead of just `&`, but especially for special entities like `&rlm;` which enables Right-to-Left mode on Hebrew and Arabic Subtitles.
---
 devine/commands/dl.py         | 1 +
 devine/core/manifests/dash.py | 2 ++
 devine/core/manifests/hls.py  | 3 +++
 3 files changed, 6 insertions(+)

diff --git a/devine/commands/dl.py b/devine/commands/dl.py
index 216bfd5..6654297 100644
--- a/devine/commands/dl.py
+++ b/devine/commands/dl.py
@@ -913,6 +913,7 @@ class dl:
                         if isinstance(track, Subtitle):
                             track_data = track.path.read_bytes()
                             track_data = try_ensure_utf8(track_data)
+                            track_data = html.unescape(track_data.decode("utf8")).encode("utf8")
                             track.path.write_bytes(track_data)
 
                         progress(downloaded="Downloaded")
diff --git a/devine/core/manifests/dash.py b/devine/core/manifests/dash.py
index adc2f01..718f93e 100644
--- a/devine/core/manifests/dash.py
+++ b/devine/core/manifests/dash.py
@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import base64
+import html
 import logging
 import math
 import re
@@ -473,6 +474,7 @@ class DASH:
                         track.codec not in (Subtitle.Codec.fVTT, Subtitle.Codec.fTTML)
                     ):
                         segment_data = try_ensure_utf8(segment_data)
+                        segment_data = html.unescape(segment_data.decode("utf8")).encode("utf8")
                     f.write(segment_data)
                     segment_file.unlink()
 
diff --git a/devine/core/manifests/hls.py b/devine/core/manifests/hls.py
index 6ba9b2c..bf23ca1 100644
--- a/devine/core/manifests/hls.py
+++ b/devine/core/manifests/hls.py
@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import html
 import logging
 import re
 import sys
@@ -314,6 +315,8 @@ class HLS:
                 segment_data = segment_file.read_bytes()
                 if isinstance(track, Subtitle):
                     segment_data = try_ensure_utf8(segment_data)
+                    if track.codec not in (Subtitle.Codec.fVTT, Subtitle.Codec.fTTML):
+                        segment_data = html.unescape(segment_data.decode("utf8")).encode("utf8")
                 f.write(segment_data)
                 segment_file.unlink()