From 873c4e617d7e031c1e04352dfb9896bd77675f64 Mon Sep 17 00:00:00 2001
From: llama <100429699+iamllama@users.noreply.github.com>
Date: Sat, 25 Jan 2025 15:16:24 +0800
Subject: [PATCH] Recognise and check for media referenced in <source> tags
 (#3763)

* recognise and check <source> tags for media

* add test
---
 pylib/anki/media.py      |  4 ++--
 rslib/src/media/check.rs | 23 +++++++++++++++++++++++
 rslib/src/text.rs        |  4 ++--
 3 files changed, 27 insertions(+), 4 deletions(-)

diff --git a/pylib/anki/media.py b/pylib/anki/media.py
index b76d7116c..5d8653a97 100644
--- a/pylib/anki/media.py
+++ b/pylib/anki/media.py
@@ -33,9 +33,9 @@ class MediaManager(DeprecatedNamesMixin):
     sound_regexps = [r"(?i)(\[sound:(?P<fname>[^]]+)\])"]
     html_media_regexps = [
         # src element quoted case
-        r"(?i)(<(?:img|audio)\b[^>]* src=(?P<str>[\"'])(?P<fname>[^>]+?)(?P=str)[^>]*>)",
+        r"(?i)(<(?:img|audio|source)\b[^>]* src=(?P<str>[\"'])(?P<fname>[^>]+?)(?P=str)[^>]*>)",
         # unquoted case
-        r"(?i)(<(?:img|audio)\b[^>]* src=(?!['\"])(?P<fname>[^ >]+)[^>]*?>)",
+        r"(?i)(<(?:img|audio|source)\b[^>]* src=(?!['\"])(?P<fname>[^ >]+)[^>]*?>)",
         # src element quoted case
         r"(?i)(<object\b[^>]* data=(?P<str>[\"'])(?P<fname>[^>]+?)(?P=str)[^>]*>)",
         # unquoted case
diff --git a/rslib/src/media/check.rs b/rslib/src/media/check.rs
index bc85b177b..20d684826 100644
--- a/rslib/src/media/check.rs
+++ b/rslib/src/media/check.rs
@@ -846,4 +846,27 @@ Unused: unused.jpg
 
         Ok(())
     }
+
+    #[test]
+    fn source_tags() -> Result<()> {
+        let (_dir, _mgr, mut col) = common_setup()?;
+        let mut checker = col.media_checker()?;
+
+        let field = "<audio controls><source src='foo-ss.mp3' /><source type='audio/ogg' src='bar-ss.ogg' /></audio>";
+        let seen = normalize_and_maybe_rename_files_helper(&mut checker, field);
+        assert!(seen.contains("foo-ss.mp3"));
+        assert!(seen.contains("bar-ss.ogg"));
+
+        let field = r#"
+            <picture>
+                <source src="foo-dd.webp" media="(orientation: portrait)" />
+                <img src="bar-dd.gif" alt="fancy jif" />
+            </picture>
+        "#;
+        let seen = normalize_and_maybe_rename_files_helper(&mut checker, field);
+        assert!(seen.contains("foo-dd.webp"));
+        assert!(seen.contains("bar-dd.gif"));
+
+        Ok(())
+    }
 }
diff --git a/rslib/src/text.rs b/rslib/src/text.rs
index aebdb3caf..f83332ff8 100644
--- a/rslib/src/text.rs
+++ b/rslib/src/text.rs
@@ -107,8 +107,8 @@ static HTML_LINEBREAK_TAGS: LazyLock<Regex> = LazyLock::new(|| {
 pub static HTML_MEDIA_TAGS: LazyLock<Regex> = LazyLock::new(|| {
     Regex::new(
         r#"(?xsi)
-            # the start of the image, audio, or object tag
-            <\b(?:img|audio|video|object)\b
+            # the start of the image, audio, object, or source tag
+            <\b(?:img|audio|video|object|source)\b
 
             # any non-`>`, except inside `"` or `'`
             (?: