feat: improve subtitle matching logic

Author: arabcoders

backend/app/subtitles.py

@@ -196,9 +196,6 @@ def _collect_matching_subtitles(subtitle_root: Path, target_stem: str) -> dict[s
             continue
 
         candidate_stem = normalize_subtitle_stem(candidate.stem)
-        if not candidate_stem.startswith(target_stem):
-            continue
-
         resolved_candidate = _resolve_within_root(candidate, subtitle_root)
         if resolved_candidate is None:
             continue
@@ -207,6 +204,9 @@ def _collect_matching_subtitles(subtitle_root: Path, target_stem: str) -> dict[s
             exact_matches_by_format[source_format].append(resolved_candidate)
             continue
 
+        if not _contains_subtitle_stem(candidate_stem, target_stem):
+            continue
+
         prefix_matches_by_format[source_format].append(resolved_candidate)
 
     for source_format in SUPPORTED_SUBTITLE_SOURCE_FORMATS:
@@ -224,3 +224,16 @@ def _resolve_within_root(candidate: Path, subtitle_root: Path) -> Path | None:
             return resolved_candidate
 
     return None
+
+
+def _contains_subtitle_stem(candidate_stem: str, target_stem: str) -> bool:
+    match_start = candidate_stem.find(target_stem)
+    while match_start != -1:
+        match_end = match_start + len(target_stem)
+        has_left_boundary = match_start == 0 or not candidate_stem[match_start - 1].isalnum()
+        has_right_boundary = match_end == len(candidate_stem) or not candidate_stem[match_end].isalnum()
+        if has_left_boundary and has_right_boundary:
+            return True
+        match_start = candidate_stem.find(target_stem, match_start + 1)
+
+    return False

backend/tests/test_subtitles.py

@@ -157,6 +157,37 @@ async def test_list_file_subtitles_supports_prefix_language_suffixes(client, tes
         )
 
 
+@pytest.mark.asyncio
+async def test_list_file_subtitles_supports_duplicate_leading_prefixes(client, test_run_dir):
+    subtitle_root = test_run_dir / "subtitles-duplicate-prefix"
+    _write_subtitle_file(
+        subtitle_root, "260503 260503 Rajira! Sunday 川﨑桜、梅澤美波 [REQUEST].ass", "[Script Info]\nTitle: Duplicate Prefix\n"
+    )
+
+    with (
+        patch("backend.app.config.settings.subtitle_path", str(subtitle_root)),
+        patch("backend.app.security.settings.allow_public_downloads", True),
+    ):
+        token_data = await create_token(client, max_uploads=1)
+        upload_token = token_data["token"]
+        download_token = token_data["download_token"]
+
+        upload_data = await initiate_upload(client, upload_token, "260503 Rajira! Sunday 川﨑桜、梅澤美波.mp4", 5, "video/mp4")
+        upload_id = upload_data["upload_id"]
+        status_code = await upload_file_via_tus(client, upload_id, b"hello", upload_token)
+
+        assert status_code == status.HTTP_200_OK, "Upload should complete successfully"
+
+        response = await client.get(
+            app.url_path_for("get_file_subtitle", download_token=download_token, upload_id=upload_id, source_format="ass")
+        )
+
+        assert response.status_code == status.HTTP_200_OK, "Repeated leading prefixes should still allow subtitle matching"
+        assert "Title: Duplicate Prefix" in response.text, (
+            "Subtitle matching should tolerate duplicate leading date prefixes before the shared filename stem"
+        )
+
+
 @pytest.mark.asyncio
 async def test_list_file_subtitles_prefers_exact_stem_over_prefix_matches(client, test_run_dir):
     subtitle_root = test_run_dir / "subtitles-exact-first"
@@ -185,6 +216,31 @@ async def test_list_file_subtitles_prefers_exact_stem_over_prefix_matches(client
         assert "Title: Exact" in response.text, "Exact stem matches should win over language-suffixed prefix matches"
 
 
+@pytest.mark.asyncio
+async def test_list_file_subtitles_ignores_midword_contains_matches(client, test_run_dir):
+    subtitle_root = test_run_dir / "subtitles-midword"
+    _write_subtitle_file(subtitle_root, "myepisode.ass", "[Script Info]\nTitle: Midword\n")
+
+    with (
+        patch("backend.app.config.settings.subtitle_path", str(subtitle_root)),
+        patch("backend.app.security.settings.allow_public_downloads", True),
+    ):
+        token_data = await create_token(client, max_uploads=1)
+        upload_token = token_data["token"]
+        download_token = token_data["download_token"]
+
+        upload_data = await initiate_upload(client, upload_token, "episode.mkv", 5, "video/mp4")
+        upload_id = upload_data["upload_id"]
+        status_code = await upload_file_via_tus(client, upload_id, b"hello", upload_token)
+
+        assert status_code == status.HTTP_200_OK, "Upload should complete successfully"
+
+        response = await client.get(app.url_path_for("list_file_subtitles", download_token=download_token, upload_id=upload_id))
+
+        assert response.status_code == status.HTTP_200_OK, "Subtitle manifest should still be returned"
+        assert response.json()["subtitles"] == [], "Contains matching should not match unrelated mid-word stems"
+
+
 @pytest.mark.asyncio
 async def test_list_file_subtitles_skips_ambiguous_extension_matches(client, test_run_dir):
     subtitle_root = test_run_dir / "subtitles-ambiguous"