added mhtml support and fixed some bugs in the process

tool/ytdlp.py

@@ -613,6 +613,18 @@ def is_browseable_format(fmt: Any) -> bool:
     
     if format_id.lower().startswith("sb"):
         return False
+
+    protocol = str(fmt.get("protocol") or "").strip().lower()
+    size_bytes = fmt.get("filesize") or fmt.get("filesize_approx")
+    if (
+        protocol in {"m3u8", "m3u8_native"}
+        and re.fullmatch(r"\d+-\d+", format_id)
+        and not size_bytes
+    ):
+        vcodec = str(fmt.get("vcodec", "none"))
+        acodec = str(fmt.get("acodec", "none"))
+        if vcodec != "none" and acodec != "none":
+            return False
     
     # Filter out formats with no audio and no video
     vcodec = str(fmt.get("vcodec", "none"))
@@ -620,6 +632,80 @@ def is_browseable_format(fmt: Any) -> bool:
     return not (vcodec == "none" and acodec == "none")
 
 
+def get_selection_format_id(
+    fmt: Dict[str, Any],
+    *,
+    video_audio_suffix: str = "ba",
+) -> str:
+    format_id = str(fmt.get("format_id") or "").strip()
+    if not format_id:
+        return ""
+
+    vcodec = str(fmt.get("vcodec", "none"))
+    acodec = str(fmt.get("acodec", "none"))
+    selector_id = format_id
+
+    match = re.fullmatch(r"(?P<base>\d+)-[A-Za-z0-9]+", format_id)
+    if match and vcodec == "none" and acodec != "none":
+        selector_id = match.group("base")
+
+    if selector_id and vcodec != "none" and acodec == "none" and video_audio_suffix:
+        selector_id = f"{selector_id}+{video_audio_suffix}"
+
+    return selector_id
+
+
+def get_display_format_id(fmt: Dict[str, Any]) -> str:
+    format_id = str(fmt.get("format_id") or "").strip()
+    if not format_id:
+        return ""
+    selector_id = get_selection_format_id(fmt, video_audio_suffix="")
+    return selector_id or format_id
+
+
+def _picker_format_score(fmt: Dict[str, Any]) -> tuple[int, int, float]:
+    note = str(fmt.get("format_note") or fmt.get("format") or "").strip().lower()
+    format_id = str(fmt.get("format_id") or "").strip().lower()
+    prefers_original = 1 if ("original" in note or "default" in note) else 0
+    avoids_drc = 0 if ("-drc" in format_id or "drc" in note) else 1
+    magnitude = 0.0
+    for key in ("filesize", "filesize_approx", "abr", "tbr"):
+        value = fmt.get(key)
+        if isinstance(value, (int, float)):
+            magnitude = float(value)
+            break
+        if isinstance(value, str):
+            try:
+                magnitude = float(value.strip())
+                break
+            except Exception:
+                pass
+    return (prefers_original, avoids_drc, magnitude)
+
+
+def collapse_picker_formats(
+    formats: Sequence[Dict[str, Any]],
+    *,
+    video_audio_suffix: str = "ba",
+) -> List[Dict[str, Any]]:
+    collapsed: Dict[str, Dict[str, Any]] = {}
+    order: List[str] = []
+    for fmt in formats:
+        if not isinstance(fmt, dict) or not is_browseable_format(fmt):
+            continue
+        selector_id = get_selection_format_id(fmt, video_audio_suffix=video_audio_suffix)
+        if not selector_id:
+            continue
+        current = collapsed.get(selector_id)
+        if current is None:
+            collapsed[selector_id] = fmt
+            order.append(selector_id)
+            continue
+        if _picker_format_score(fmt) > _picker_format_score(current):
+            collapsed[selector_id] = fmt
+    return [collapsed[key] for key in order if key in collapsed]
+
+
 def format_for_table_selection(
     fmt: Dict[str, Any],
     url: str,
@@ -648,6 +734,7 @@ def format_for_table_selection(
         results = [format_for_table_selection(f, url, i+1) for i, f in enumerate(browseable)]
     """
     format_id = fmt.get("format_id", "")
+    display_format_id = get_display_format_id(fmt)
     resolution = fmt.get("resolution", "")
     ext = fmt.get("ext", "")
     vcodec = fmt.get("vcodec", "none")
@@ -657,10 +744,10 @@ def format_for_table_selection(
     
     # If not provided, compute selection format ID (add +ba for video-only)
     if selection_format_id is None:
-        selection_format_id = format_id
+        selection_format_id = get_selection_format_id(fmt, video_audio_suffix="ba")
         try:
-            if vcodec != "none" and acodec == "none" and format_id:
-                selection_format_id = f"{format_id}+ba"
+            if not selection_format_id and format_id:
+                selection_format_id = format_id
         except Exception:
             from SYS.logger import logger
             logger.exception("Failed to compute selection_format_id for format: %s", fmt)
@@ -694,14 +781,14 @@ def format_for_table_selection(
     # Build table row
     return {
         "table": "download-file",
-        "title": f"Format {format_id}",
+        "title": f"Format {display_format_id or format_id}",
         "url": url,
         "target": url,
         "detail": format_desc,
         "annotations": [ext, resolution] if resolution else [ext],
         "media_kind": "format",
         "columns": [
-            ("ID", format_id),
+            ("ID", display_format_id or format_id),
             ("Resolution", resolution or "N/A"),
             ("Ext", ext),
             ("Size", size_str or ""),