handle quoted html chars in media check

https://forums.ankiweb.net/t/unable-to-play-longer-audio-on-cards/1313/30

rslib/src/media/check.rs

@@ -455,7 +455,7 @@ fn normalize_and_maybe_rename_files<'a>(
         }
 
         // normalize fname into NFC
-        let mut fname = normalize_to_nfc(media_ref.fname);
+        let mut fname = normalize_to_nfc(&media_ref.fname_decoded);
         // and look it up to see if it's been renamed
         if let Some(new_name) = renamed.get(fname.as_ref()) {
             fname = new_name.to_owned().into();
@@ -486,7 +486,13 @@ fn normalize_and_maybe_rename_files<'a>(
 }
 
 fn rename_media_ref_in_field(field: &str, media_ref: &MediaRef, new_name: &str) -> String {
-    let updated_tag = media_ref.full_ref.replace(media_ref.fname, new_name);
+    let new_name = if matches!(media_ref.fname_decoded, Cow::Owned(_)) {
+        // filename had quoted characters like &amp; - need to re-encode
+        htmlescape::encode_minimal(new_name)
+    } else {
+        new_name.into()
+    };
+    let updated_tag = media_ref.full_ref.replace(media_ref.fname, &new_name);
     field.replace(media_ref.full_ref, &updated_tag)
 }
 
@@ -522,6 +528,7 @@ pub(crate) mod test {
     pub(crate) const MEDIACHECK_ANKI2: &[u8] =
         include_bytes!("../../tests/support/mediacheck.anki2");
 
+    use super::normalize_and_maybe_rename_files;
     use crate::collection::{open_collection, Collection};
     use crate::err::Result;
     use crate::i18n::I18n;
@@ -530,7 +537,7 @@ pub(crate) mod test {
     use crate::media::files::trash_folder;
     use crate::media::MediaManager;
     use std::path::Path;
-    use std::{fs, io};
+    use std::{collections::HashMap, fs, io};
     use tempfile::{tempdir, TempDir};
 
     fn common_setup() -> Result<(TempDir, MediaManager, Collection)> {
@@ -730,4 +737,12 @@ Unused: unused.jpg
 
         Ok(())
     }
+
+    #[test]
+    fn html_encoding() {
+        let field = "[sound:a &amp; b.mp3]";
+        let mut seen = Default::default();
+        normalize_and_maybe_rename_files(field, &HashMap::new(), &mut seen, Path::new("/tmp"));
+        assert!(seen.contains("a & b.mp3"));
+    }
 }

rslib/src/text.rs

@@ -142,28 +142,36 @@ pub fn extract_av_tags<'a>(text: &'a str, question_side: bool) -> (Cow<'a, str>,
 pub(crate) struct MediaRef<'a> {
     pub full_ref: &'a str,
     pub fname: &'a str,
+    /// audio files may have things like &amp; that need decoding
+    pub fname_decoded: Cow<'a, str>,
 }
 
 pub(crate) fn extract_media_refs(text: &str) -> Vec<MediaRef> {
     let mut out = vec![];
 
     for caps in IMG_TAG.captures_iter(text) {
+        let fname = caps
+            .get(1)
+            .or_else(|| caps.get(2))
+            .or_else(|| caps.get(3))
+            .unwrap()
+            .as_str();
+        let fname_decoded = fname.into();
         out.push(MediaRef {
             full_ref: caps.get(0).unwrap().as_str(),
-            fname: caps
+            fname,
-                .get(1)
+            fname_decoded,
-                .or_else(|| caps.get(2))
-                .or_else(|| caps.get(3))
-                .unwrap()
-                .as_str(),
         });
     }
 
     for caps in AV_TAGS.captures_iter(text) {
         if let Some(m) = caps.get(1) {
+            let fname = m.as_str();
+            let fname_decoded = decode_entities(fname);
             out.push(MediaRef {
                 full_ref: caps.get(0).unwrap().as_str(),
-                fname: m.as_str(),
+                fname,
+                fname_decoded,
             });
         }
     }