Expose yt-dlp media metadata for snapshot cards

archivebox/core/models.py

@@ -4628,12 +4628,26 @@ class ArchiveResult(ModelWithDeleteAfter, ModelWithOutputDir, ModelWithNotes):
                 if Path(candidate).name.lower() == preferred_name:
                     return candidate
 
-        ext_groups = (
-            (".html", ".htm", ".mhtml", ".mht", ".pdf"),
-            (".png", ".jpg", ".jpeg", ".gif", ".webp", ".svg", ".ico"),
-            (".json", ".jsonl", ".txt", ".md", ".csv", ".tsv"),
-            (".mp4", ".webm", ".mp3", ".opus", ".ogg", ".wav"),
-        )
+        plugin_lower = (plugin_name or "").lower()
+        if plugin_lower in ("ytdlp", "yt-dlp", "youtube-dl"):
+            # yt-dlp commonly emits a thumbnail plus several media formats.
+            # Prefer something browsers can play directly; otherwise cards can
+            # select a large MKV or thumbnail that the plugin player cannot use.
+            ext_groups = (
+                (".mp4", ".webm", ".m4v", ".ogv"),
+                (".mp3", ".m4a", ".aac", ".opus", ".ogg", ".wav", ".flac"),
+                (".mkv", ".mov", ".avi", ".flv", ".wmv", ".mpg", ".mpeg", ".ts", ".m2ts", ".mts", ".3gp", ".3g2"),
+                (".html", ".htm", ".mhtml", ".mht", ".pdf"),
+                (".png", ".jpg", ".jpeg", ".gif", ".webp", ".svg", ".ico"),
+                (".json", ".jsonl", ".txt", ".md", ".csv", ".tsv", ".srt", ".vtt"),
+            )
+        else:
+            ext_groups = (
+                (".html", ".htm", ".mhtml", ".mht", ".pdf"),
+                (".png", ".jpg", ".jpeg", ".gif", ".webp", ".svg", ".ico"),
+                (".json", ".jsonl", ".txt", ".md", ".csv", ".tsv"),
+                (".mp4", ".webm", ".mp3", ".opus", ".ogg", ".wav"),
+            )
         for ext_group in ext_groups:
             group_candidates = [candidate for candidate in candidates if Path(candidate).suffix.lower() in ext_group]
             if group_candidates:

archivebox/core/templatetags/core_tags.py

@@ -32,7 +32,7 @@ _TEXT_PREVIEW_EXTS = (".json", ".jsonl", ".txt", ".csv", ".tsv", ".xml", ".yml",
 _IMAGE_PREVIEW_EXTS = (".png", ".jpg", ".jpeg", ".gif", ".webp", ".bmp", ".ico", ".avif")
 _STATIC_URL_SAFE = "/@-._~!$&'()*+,;="
 
-_MEDIA_FILE_EXTS = {
+_VIDEO_FILE_EXTS = {
     ".mp4",
     ".webm",
     ".mkv",
@@ -49,6 +49,9 @@ _MEDIA_FILE_EXTS = {
     ".3gp",
     ".3g2",
     ".ogv",
+}
+
+_AUDIO_FILE_EXTS = {
     ".mp3",
     ".m4a",
     ".aac",
@@ -66,6 +69,10 @@ _MEDIA_FILE_EXTS = {
     ".dts",
 }
 
+_MEDIA_FILE_EXTS = _VIDEO_FILE_EXTS | _AUDIO_FILE_EXTS
+_BROWSER_VIDEO_FILE_EXTS = {".mp4", ".webm", ".m4v", ".ogv"}
+_BROWSER_AUDIO_FILE_EXTS = {".mp3", ".m4a", ".aac", ".ogg", ".oga", ".opus", ".wav", ".flac"}
+
 
 def _normalize_output_files(output_files: Any) -> dict[str, dict[str, Any]]:
     if isinstance(output_files, dict):
@@ -159,15 +166,27 @@ def _list_media_files(result) -> list[dict]:
 
     for rel_path, size in candidates:
         href = str(Path(result.plugin) / rel_path)
+        suffix = rel_path.suffix.lower()
+        media_type = "video" if suffix in _VIDEO_FILE_EXTS else "audio"
         media_files.append(
             {
                 "name": rel_path.name,
                 "path": href,
                 "size": size,
+                "media_type": media_type,
+                "is_video": media_type == "video",
+                "is_audio": media_type == "audio",
+                "is_browser_playable": suffix in _BROWSER_VIDEO_FILE_EXTS or suffix in _BROWSER_AUDIO_FILE_EXTS,
             },
         )
 
-    media_files.sort(key=lambda item: item["name"].lower())
+    media_files.sort(
+        key=lambda item: (
+            0 if item["is_video"] and item["is_browser_playable"] else 1 if item["is_audio"] and item["is_browser_playable"] else 2,
+            -int(item.get("size") or 0),
+            item["name"].lower(),
+        ),
+    )
     return media_files
 
 

archivebox/tests/test_ui_admin_snapshot.py

@@ -485,10 +485,46 @@ class TestSnapshotProgressStats:
 
         assert _count_media_files(result) == 2
         assert _list_media_files(result) == [
-            {"name": "audio.mp3", "path": "ytdlp/audio.mp3", "size": 222},
-            {"name": "video.mp4", "path": "ytdlp/video.mp4", "size": 111},
+            {
+                "name": "video.mp4",
+                "path": "ytdlp/video.mp4",
+                "size": 111,
+                "media_type": "video",
+                "is_video": True,
+                "is_audio": False,
+                "is_browser_playable": True,
+            },
+            {
+                "name": "audio.mp3",
+                "path": "ytdlp/audio.mp3",
+                "size": 222,
+                "media_type": "audio",
+                "is_video": False,
+                "is_audio": True,
+                "is_browser_playable": True,
+            },
         ]
 
+    def test_ytdlp_discover_outputs_prefers_browser_playable_video(self, snapshot):
+        from archivebox.core.models import ArchiveResult
+
+        ArchiveResult.objects.create(
+            snapshot=snapshot,
+            plugin="ytdlp",
+            status="succeeded",
+            output_files={
+                "thumbnail.jpg": {"size": 20_000, "mimetype": "image/jpeg", "extension": "jpg"},
+                "large.mkv": {"size": 10_000, "mimetype": "video/x-matroska", "extension": "mkv"},
+                "small.mp4": {"size": 111, "mimetype": "video/mp4", "extension": "mp4"},
+            },
+            output_size=30_111,
+        )
+
+        outputs = snapshot.discover_outputs(include_filesystem_fallback=False)
+        ytdlp_output = next(output for output in outputs if output["name"] == "ytdlp")
+
+        assert ytdlp_output["path"] == "ytdlp/small.mp4"
+
     def test_discover_outputs_falls_back_to_hashes_index_without_filesystem_walk(
         self,
         snapshot,