channels: handle null topic metadata safely

spec/invidious/channels/about_spec.cr

@@ -254,6 +254,23 @@ Spectator.describe "extract_auto_generated_channel_header" do
     expect(header[:is_family_friendly]).to be_false
   end
 
+  it "defaults an explicit familySafe: null to true" do
+    initdata = JSON.parse(<<-JSON).as_h
+      {
+        "header": {
+          "pageHeaderRenderer": {"pageTitle": "Gaming"}
+        },
+        "microformat": {
+          "microformatDataRenderer": {"familySafe": null}
+        }
+      }
+      JSON
+
+    header = extract_auto_generated_channel_header(initdata, "UCNullFamilySafe")
+
+    expect(header[:is_family_friendly]).to be_true
+  end
+
   it "raises when the header shape is unknown" do
     initdata = JSON.parse(<<-JSON).as_h
       {
@@ -298,6 +315,40 @@ Spectator.describe "extract_topic_channel_details" do
     expect(extract_topic_channel_details(initdata)).to be_nil
   end
 
+  it "returns nil when carousel contents are explicitly null" do
+    initdata = JSON.parse(<<-JSON).as_h
+      {
+        "header": {
+          "carouselHeaderRenderer": {
+            "contents": null
+          }
+        }
+      }
+      JSON
+
+    expect(extract_topic_channel_details(initdata)).to be_nil
+  end
+
+  it "skips an explicitly null topic renderer before a valid one" do
+    initdata = JSON.parse(<<-JSON).as_h
+      {
+        "header": {
+          "carouselHeaderRenderer": {
+            "contents": [
+              {"topicChannelDetailsRenderer": null},
+              {"topicChannelDetailsRenderer": {"title": {"simpleText": "Sports"}}}
+            ]
+          }
+        }
+      }
+      JSON
+
+    details = extract_topic_channel_details(initdata)
+
+    expect(details).not_to be_nil
+    expect(details.not_nil!.dig("title", "simpleText").as_s).to eq("Sports")
+  end
+
   it "exposes the subscriber count carried by the subtitle" do
     # `subscriberCountText` is part of the renderer but comes back null, so the
     # count is only available as free text in the subtitle.
@@ -322,11 +373,6 @@ Spectator.describe "extract_topic_channel_details" do
       }
       JSON
 
-    details = extract_topic_channel_details(initdata)
+    expect(extract_topic_channel_subscriber_count(initdata)).to eq(74_300_000)
-
-    expect(details).not_to be_nil
-    sub_text = details.not_nil!.dig("subtitle", "simpleText").as_s
-    expect(sub_text).to eq("74.3M subscribers")
-    expect(short_text_to_number(sub_text.split(" ")[0])).to eq(74_300_000_i64)
   end
 end

src/invidious/channels/about.cr

@@ -24,12 +24,29 @@ record AboutChannel,
 # rather than by index.
 # ex: https://www.youtube.com/channel/UCEgdi0XIXXZ-qJOFPf4JSKw
 def extract_topic_channel_details(initdata : Hash(String, JSON::Any)) : JSON::Any?
-  contents = initdata.dig?("header", "carouselHeaderRenderer", "contents")
+  contents = initdata.dig?("header", "carouselHeaderRenderer", "contents").try &.as_a?
-  return nil if contents.nil?
+  return nil unless contents
 
-  contents.as_a
+  contents.each do |content|
-    .find { |content| !content.dig?("topicChannelDetailsRenderer").nil? }
+    details = content.dig?("topicChannelDetailsRenderer")
-    .try &.dig?("topicChannelDetailsRenderer")
+    return details if details.try &.as_h?
+  end
+
+  nil
+end
+
+def extract_topic_channel_subscriber_count(initdata : Hash(String, JSON::Any)) : Int32
+  details = extract_topic_channel_details(initdata)
+  return 0 unless details
+
+  sub_text = details.dig?("subscriberCountText", "simpleText").try &.as_s?
+  unless sub_text.try &.includes?("subscriber")
+    sub_text = details.dig?("subtitle", "simpleText").try &.as_s?
+  end
+
+  return 0 unless sub_text.try &.includes?("subscriber")
+
+  short_text_to_number(sub_text.split(" ")[0]).to_i32
 end
 
 # Auto-generated channels come with one of three header shapes. This is only
@@ -96,7 +113,7 @@ def extract_auto_generated_channel_header(initdata : Hash(String, JSON::Any), uc
 
   # `microformat` is absent from these payloads, so a missing flag defaults to
   # safe. An explicit `false` is still preserved.
-  family_safe = initdata.dig?("microformat", "microformatDataRenderer", "familySafe").try(&.as_bool)
+  family_safe = initdata.dig?("microformat", "microformatDataRenderer", "familySafe").try &.as_bool?
 
   {
     author:             author,
@@ -265,18 +282,8 @@ def get_about_info(ucid) : AboutChannel
 
       break if sub_count != 0 && !pronouns.nil?
     end
-  elsif (topic_details = extract_topic_channel_details(initdata))
+  elsif initdata.dig?("header", "carouselHeaderRenderer")
-    # Topic channels carry the subscriber count as free text in the subtitle,
+    sub_count = extract_topic_channel_subscriber_count(initdata)
-    # ex: "74.3M subscribers". `subscriberCountText` is part of the same
-    # renderer but comes back null, so it is only used as a first choice.
-    sub_text = topic_details.dig?("subscriberCountText", "simpleText").try &.as_s
-    unless sub_text.try &.includes?("subscriber")
-      sub_text = topic_details.dig?("subtitle", "simpleText").try &.as_s
-    end
-
-    if sub_text && sub_text.includes?("subscriber")
-      sub_count = short_text_to_number(sub_text.split(" ")[0]).to_i32
-    end
   end
 
   AboutChannel.new(