fix(core): match a Takeout sidecar by prefix only where Google cut its name

The prefix rule for sidecars whose names Google cut at 51 characters also
matched names that were never cut, so IMG_1.jpg's sidecar gave its title
and date to an IMG_12.jpg that had none. It now applies only to a sidecar
name at the cap. Edited copies, which have no sidecar of their own, take
their original's date by a rule of their own and keep their own name
(B225).

crates/pdfs-core/src/takeout.rs

@@ -364,9 +364,13 @@ fn assemble(entries: Vec<RawEntry>, sidecars: &HashMap<String, Vec<u8>>) -> Take
             continue;
         };
 
-        let sidecar = per_folder_sidecars
+        let found = per_folder_sidecars
             .get(&folder)
-            .and_then(|in_folder| find_sidecar(&file_name, in_folder))
+            .and_then(|in_folder| find_sidecar(&file_name, in_folder));
+        // An edited copy keeps its own name: the title is its original's.
+        let title_is_ours = found.is_some_and(|((rule, _), _)| rule != EDITED_COPY);
+        let sidecar = found
+            .map(|(_, body)| body)
             .and_then(|body| serde_json::from_slice::<PhotoSidecar>(body).ok())
             .unwrap_or_default();
 
@@ -388,6 +392,7 @@ fn assemble(entries: Vec<RawEntry>, sidecars: &HashMap<String, Vec<u8>>) -> Take
         let name = sidecar
             .title
             .as_deref()
+            .filter(|_| title_is_ours)
             .map(str::trim)
             .filter(|title| !title.is_empty() && same_extension(title, &file_name))
             .unwrap_or(&file_name)
@@ -627,15 +632,25 @@ fn swap_duplicate_marker(file_name: &str) -> Option<String> {
     Some(format!("{}{}{}", &file_name[..marker_start], tail, marker))
 }
 
-/// Find the sidecar belonging to `file_name` among the JSON files of its folder.
+/// How long Google lets a sidecar's file name get, `.json` included. A longer
+/// one is cut, into its suffix first and then into the photo name.
+const SIDECAR_NAME_CAP: usize = 51;
+
+/// The rank [`sidecar_rank`] gives an edited copy that shares its original's
+/// sidecar. Its date is the original's, but its name is its own.
+const EDITED_COPY: u8 = 4;
+
+/// Find the sidecar belonging to `file_name` among the JSON files of its folder,
+/// with the rank it matched at.
 ///
 /// Tried in order of confidence: the exact name, the `(N)`-swapped name, an
-/// unknown trailing suffix, then a prefix match for the case where truncation
-/// cut into the photo name itself. The best-ranked candidate in the folder wins.
+/// unknown trailing suffix, a prefix match for the case where truncation cut
+/// into the photo name itself, and last an edited copy's original. The
+/// best-ranked candidate in the folder wins.
 fn find_sidecar<'a>(
     file_name: &str,
     in_folder: &BTreeMap<String, &'a Vec<u8>>,
-) -> Option<&'a Vec<u8>> {
+) -> Option<((u8, usize), &'a Vec<u8>)> {
     let mut best: Option<((u8, usize), &'a Vec<u8>)> = None;
     for (sidecar_name, body) in in_folder {
         let Some(rank) = sidecar_rank(strip_json_suffix(sidecar_name), file_name) else {
@@ -645,7 +660,7 @@ fn find_sidecar<'a>(
             best = Some((rank, body));
         }
     }
-    best.map(|(_, body)| body)
+    best
 }
 
 /// How well a sidecar's `.json`-stripped `stem` matches `file_name`, as a sort
@@ -683,19 +698,44 @@ fn sidecar_rank(stem: &str, file_name: &str) -> Option<(u8, usize)> {
     }
     // 3 — truncation ate into the photo name itself, leaving only a prefix of
     // it. Every suffix boundary in the stem is tried, so this holds for a
-    // localized suffix too; the longest surviving prefix wins, because a folder
-    // with `IMG_1.jpg` and `IMG_12.jpg` must not let the shorter stem claim the
-    // longer photo.
-    let mut longest: Option<usize> = None;
-    for head in suffix_boundaries(stem) {
-        if !head.is_empty()
-            && file_name.starts_with(head)
-            && longest.is_none_or(|len| head.len() > len)
-        {
-            longest = Some(head.len());
+    // localized suffix too; the longest surviving prefix wins.
+    //
+    // Only for a sidecar whose name is as long as Google lets one get: a name
+    // that short of the cap was not cut, so a prefix of it is the name of a
+    // different photo. Without this, `IMG_1.jpg`'s sidecar gave its name and
+    // date to an `IMG_12.jpg` that had none (`docs/BUGS.md` B225).
+    if stem.chars().count() + ".json".len() >= SIDECAR_NAME_CAP - 4 {
+        let mut longest: Option<usize> = None;
+        for head in suffix_boundaries(stem) {
+            if !head.is_empty()
+                && file_name.starts_with(head)
+                && longest.is_none_or(|len| head.len() > len)
+            {
+                longest = Some(head.len());
+            }
+        }
+        if let Some(len) = longest {
+            return Some((3, usize::MAX - len));
         }
     }
-    longest.map(|len| (3, usize::MAX - len))
+    // 4 — an edited copy, `IMG_1-edited.jpg` (the word is localized), which
+    // Google exports without a sidecar of its own next to `IMG_1.jpg`'s.
+    suffix_boundaries(stem)
+        .filter_map(|head| head.rsplit_once('.'))
+        .any(|(original, extension)| is_edited_copy_of(file_name, original, extension))
+        .then_some((EDITED_COPY, 0))
+}
+
+/// Whether `file_name` is `<original>-<word>.<extension>`, the name Google
+/// gives an edited copy of `<original>.<extension>`.
+fn is_edited_copy_of(file_name: &str, original: &str, extension: &str) -> bool {
+    !original.is_empty()
+        && file_name
+            .strip_prefix(original)
+            .and_then(|rest| rest.strip_prefix('-'))
+            .and_then(|rest| rest.strip_suffix(extension))
+            .and_then(|rest| rest.strip_suffix('.'))
+            .is_some_and(|word| !word.is_empty() && word.chars().all(char::is_alphabetic))
 }
 
 /// The stem plus each of its prefixes ending before a `.`, longest first — the
@@ -994,21 +1034,25 @@ mod tests {
 
     #[test]
     fn longest_prefix_wins_when_the_photo_name_itself_was_truncated() {
+        // Both sidecar names are at Google's cap, so both were cut into the
+        // photo name; the longer surviving prefix is the better claim.
+        let short = "Takeout/Google Photos/Photos from 2019/PXL_20190101_120000000.PORTRAIT_ORIGINAL_E.json";
+        let long = "Takeout/Google Photos/Photos from 2019/PXL_20190101_120000000.PORTRAIT_ORIGINAL_EXPO.json";
         let mut sidecars = HashMap::new();
         sidecar(
             &mut sidecars,
-            "Takeout/Google Photos/Photos from 2019/IMG_1.json",
+            short,
             r#"{"photoTakenTime":{"timestamp":"1"}}"#,
         );
         sidecar(
             &mut sidecars,
-            "Takeout/Google Photos/Photos from 2019/IMG_12.json",
+            long,
             r#"{"photoTakenTime":{"timestamp":"2"}}"#,
         );
         let scan = assemble(
             vec![entry(
                 0,
-                "Takeout/Google Photos/Photos from 2019/IMG_12345.jpg",
+                "Takeout/Google Photos/Photos from 2019/PXL_20190101_120000000.PORTRAIT_ORIGINAL_EXPOSURE.jpg",
                 10,
             )],
             &sidecars,
@@ -1017,6 +1061,53 @@ mod tests {
         assert_eq!(scan.photos[0].capture_time, Some(2));
     }
 
+    /// A short sidecar name was not cut, so a prefix of it names another photo:
+    /// `IMG_1.jpg`'s sidecar gave its title and date to `IMG_12.jpg`
+    /// (`docs/BUGS.md` B225).
+    #[test]
+    fn a_photo_without_a_sidecar_does_not_take_a_shorter_named_siblings() {
+        let mut sidecars = HashMap::new();
+        sidecar(
+            &mut sidecars,
+            "Takeout/Google Photos/Photos from 2019/IMG_1.jpg.supplemental-metadata.json",
+            r#"{"title":"IMG_1.jpg","photoTakenTime":{"timestamp":"1"}}"#,
+        );
+        let scan = assemble(
+            vec![
+                entry(0, "Takeout/Google Photos/Photos from 2019/IMG_1.jpg", 10),
+                entry(0, "Takeout/Google Photos/Photos from 2019/IMG_12.jpg", 10),
+            ],
+            &sidecars,
+        );
+
+        assert_eq!(scan.photos[0].capture_time, Some(1));
+        assert_eq!(scan.photos[1].name, "IMG_12.jpg");
+        assert_eq!(scan.photos[1].capture_time, None);
+    }
+
+    /// An edited copy has no sidecar of its own. It takes its original's date,
+    /// but keeps its own name rather than the original's title.
+    #[test]
+    fn an_edited_copy_takes_its_originals_date_and_keeps_its_name() {
+        let mut sidecars = HashMap::new();
+        sidecar(
+            &mut sidecars,
+            "Takeout/Google Fotos/Urlaub/IMG_2.jpg.ergänzende-Metadaten.json",
+            r#"{"title":"Holiday.jpg","photoTakenTime":{"timestamp":"7"}}"#,
+        );
+        let scan = assemble(
+            vec![
+                entry(0, "Takeout/Google Fotos/Urlaub/IMG_2.jpg", 10),
+                entry(0, "Takeout/Google Fotos/Urlaub/IMG_2-bearbeitet.jpg", 10),
+            ],
+            &sidecars,
+        );
+
+        assert_eq!(scan.photos[0].name, "Holiday.jpg");
+        assert_eq!(scan.photos[1].name, "IMG_2-bearbeitet.jpg");
+        assert_eq!(scan.photos[1].capture_time, Some(7));
+    }
+
     #[test]
     fn sidecar_title_replaces_a_mangled_on_disk_name() {
         let mut sidecars = HashMap::new();