From b23b6b716293ab758e7ff6c8b503764d397d043e Mon Sep 17 00:00:00 2001
From: Barodius1 <jean.astor1500@gmail.com>
Date: Sun, 2 Aug 2026 00:37:12 -0700
Subject: [PATCH] test: cover UTF-16 description command offsets

Add regression coverage for emoji before and inside commands, BMP characters, and HTML escaping.
---
 spec/invidious/videos/description_spec.cr | 90 +++++++++++++++++++++++
 1 file changed, 90 insertions(+)
 create mode 100644 spec/invidious/videos/description_spec.cr

diff --git a/spec/invidious/videos/description_spec.cr b/spec/invidious/videos/description_spec.cr
new file mode 100644
index 000000000..18885b350
--- /dev/null
+++ b/spec/invidious/videos/description_spec.cr
@@ -0,0 +1,90 @@
+require "../../spec_helper"
+require "../../../src/invidious/videos/description"
+
+Spectator.describe "Video description parser" do
+  describe "#parse_description" do
+    it "uses UTF-16 offsets when an emoji precedes a command" do
+      description = JSON.parse(<<-JSON)
+        {
+          "content": "🚀 Visit example now",
+          "commandRuns": [{
+            "startIndex": 3,
+            "length": 5,
+            "onTap": {
+              "innertubeCommand": {
+                "urlEndpoint": {"url": "https://example.com"}
+              }
+            }
+          }]
+        }
+        JSON
+
+      expect(parse_description(description, "video-id")).to eq(
+        %(🚀 <a href="https://example.com">Visit</a> example now)
+      )
+    end
+
+    it "uses UTF-16 lengths when an emoji is inside a command" do
+      description = JSON.parse(<<-JSON)
+        {
+          "content": "Open 😀 docs",
+          "commandRuns": [{
+            "startIndex": 0,
+            "length": 7,
+            "onTap": {
+              "innertubeCommand": {
+                "urlEndpoint": {"url": "https://example.com"}
+              }
+            }
+          }]
+        }
+        JSON
+
+      expect(parse_description(description, "video-id")).to eq(
+        %(<a href="https://example.com">Open 😀</a> docs)
+      )
+    end
+
+    it "continues to count basic multilingual plane characters once" do
+      description = JSON.parse(<<-JSON)
+        {
+          "content": "Café link",
+          "commandRuns": [{
+            "startIndex": 5,
+            "length": 4,
+            "onTap": {
+              "innertubeCommand": {
+                "urlEndpoint": {"url": "https://example.com"}
+              }
+            }
+          }]
+        }
+        JSON
+
+      expect(parse_description(description, "video-id")).to eq(
+        %(Café <a href="https://example.com">link</a>)
+      )
+    end
+
+    it "escapes HTML around and inside commands" do
+      description = JSON.parse(<<-JSON)
+        {
+          "content": "🚀 <click> & end",
+          "commandRuns": [{
+            "startIndex": 3,
+            "length": 7,
+            "onTap": {
+              "innertubeCommand": {
+                "urlEndpoint": {"url": "https://example.com"}
+              }
+            }
+          }]
+        }
+        JSON
+
+      expect(parse_description(description, "video-id")).to eq(
+        %(🚀 <a href="https://example.com">&lt;click&gt;</a> &amp; end)
+      )
+    end
+  end
+end