WIP, BROKEN

Better support for "enhanced broadcast" streams

issue #154

CHANGELOG.md

@@ -3,6 +3,12 @@ twitch-dl changelog
 
 <!-- Do not edit. This file is automatically generated from changelog.yaml.-->
 
+### [2.3.0 (2024-04-27)](https://github.com/ihabunek/twitch-dl/releases/tag/2.3.0)
+
+* Show more playlist data when choosing quality
+* Improve detection of 'source' quality for Twitch Enhanced Broadcast Streams
+  (#154)
+
 ### [2.2.4 (2024-04-25)](https://github.com/ihabunek/twitch-dl/releases/tag/2.2.4)
 
 * Add m dot url support to video and clip regexes (thanks @localnerve)

Makefile

@@ -7,7 +7,7 @@ dist:
 
 clean :
 	find . -name "*pyc" | xargs rm -rf $1
-	rm -rf build dist bundle MANIFEST htmlcov deb_dist twitch-dl.*.pyz twitch-dl.1.man twitch_dl.egg-info
+	rm -rf build dist book bundle MANIFEST htmlcov deb_dist twitch-dl.*.pyz twitch-dl.1.man twitch_dl.egg-info
 
 bundle:
 	mkdir bundle

changelog.yaml

@@ -1,3 +1,9 @@
+2.3.0:
+  date: 2024-04-27
+  changes:
+    - "Show more playlist data when choosing quality"
+    - "Improve detection of 'source' quality for Twitch Enhanced Broadcast Streams (#154)"
+
 2.2.4:
   date: 2024-04-25
   changes:

docs/changelog.md

@@ -3,6 +3,12 @@ twitch-dl changelog
 
 <!-- Do not edit. This file is automatically generated from changelog.yaml.-->
 
+### [2.3.0 (2024-04-27)](https://github.com/ihabunek/twitch-dl/releases/tag/2.3.0)
+
+* Show more playlist data when choosing quality
+* Improve detection of 'source' quality for Twitch Enhanced Broadcast Streams
+  (#154)
+
 ### [2.2.4 (2024-04-25)](https://github.com/ihabunek/twitch-dl/releases/tag/2.2.4)
 
 * Add m dot url support to video and clip regexes (thanks @localnerve)

pyproject.toml

@@ -22,7 +22,7 @@ classifiers = [
 dependencies = [
     "click>=8.0.0,<9.0.0",
     "httpx>=0.17.0,<1.0.0",
-    "m3u8>=1.0.0,<4.0.0",
+    "m3u8>=3.0.0,<5.0.0",
 ]
 
 [tool.setuptools]

tests/test_progress.py

@@ -23,26 +23,31 @@ def test_downloaded():
     assert progress.progress_perc == 0
 
     progress.advance(1, 100)
+    progress._recalculate()
     assert progress.downloaded == 100
     assert progress.progress_bytes == 100
     assert progress.progress_perc == 11
 
     progress.advance(2, 200)
+    progress._recalculate()
     assert progress.downloaded == 300
     assert progress.progress_bytes == 300
     assert progress.progress_perc == 33
 
     progress.advance(3, 150)
+    progress._recalculate()
     assert progress.downloaded == 450
     assert progress.progress_bytes == 450
     assert progress.progress_perc == 50
 
     progress.advance(1, 50)
+    progress._recalculate()
     assert progress.downloaded == 500
     assert progress.progress_bytes == 500
     assert progress.progress_perc == 55
 
     progress.abort(2)
+    progress._recalculate()
     assert progress.downloaded == 500
     assert progress.progress_bytes == 300
     assert progress.progress_perc == 33
@@ -52,6 +57,7 @@ def test_downloaded():
     progress.advance(1, 150)
     progress.advance(2, 300)
     progress.advance(3, 150)
+    progress._recalculate()
 
     assert progress.downloaded == 1100
     assert progress.progress_bytes == 900
@@ -71,12 +77,15 @@ def test_estimated_total():
     assert progress.estimated_total is None
 
     progress.start(1, 12000)
+    progress._recalculate()
     assert progress.estimated_total == 12000 * 3
 
     progress.start(2, 11000)
+    progress._recalculate()
     assert progress.estimated_total == 11500 * 3
 
     progress.start(3, 10000)
+    progress._recalculate()
     assert progress.estimated_total == 11000 * 3
 
 

twitchdl/commands/download.py

@@ -7,7 +7,7 @@ import subprocess
 import tempfile
 from os import path
 from pathlib import Path
-from typing import Dict, List
+from typing import Dict, List, Optional
 from urllib.parse import urlencode, urlparse
 
 import click
@@ -30,6 +30,10 @@ from twitchdl.twitch import Chapter, Clip, ClipAccessToken, Video
 
 
 def download(ids: List[str], args: DownloadOptions):
+    if not ids:
+        print_log("No IDs to downlad given")
+        return
+
     for video_id in ids:
         download_one(video_id, args)
 
@@ -156,7 +160,7 @@ def _crete_temp_dir(base_uri: str) -> str:
     return str(temp_dir)
 
 
-def _get_clip_url(access_token: ClipAccessToken, quality: str) -> str:
+def _get_clip_url(access_token: ClipAccessToken, quality: Optional[str]) -> str:
     qualities = access_token["videoQualities"]
 
     # Quality given as an argument
@@ -184,7 +188,7 @@ def _get_clip_url(access_token: ClipAccessToken, quality: str) -> str:
     return selected_quality["sourceURL"]
 
 
-def get_clip_authenticated_url(slug: str, quality: str):
+def get_clip_authenticated_url(slug: str, quality: Optional[str]):
     print_log("Fetching access token...")
     access_token = twitch.get_clip_access_token(slug)
 
@@ -278,7 +282,8 @@ def _download_video(video_id: str, args: DownloadOptions) -> None:
         return
 
     base_uri = re.sub("/[^/]+$", "/", playlist.url)
-    target_dir = _crete_temp_dir(base_uri)
+    target_dir = f".twitch_dl_{video_id}_{playlist.group_id}"
+    os.makedirs(target_dir, exist_ok=True)
 
     # Save playlists for debugging purposes
     with open(path.join(target_dir, "playlists.m3u8"), "w") as f:
@@ -292,7 +297,7 @@ def _download_video(video_id: str, args: DownloadOptions) -> None:
     targets = [os.path.join(target_dir, f"{vod.index:05d}.ts") for vod in vods]
     asyncio.run(download_all(sources, targets, args.max_workers, rate_limit=args.rate_limit))
 
-    join_playlist = make_join_playlist(vods_m3u8, vods, targets)
+    join_playlist = make_join_playlist(vods, targets)
     join_playlist_path = path.join(target_dir, "playlist_downloaded.m3u8")
     join_playlist.dump(join_playlist_path)  # type: ignore
     click.echo()

twitchdl/output.py

@@ -25,15 +25,24 @@ def print_log(message: Any):
     click.secho(message, err=True, dim=True)
 
 
+def visual_len(text: str):
+    return len(click.unstyle(text))
+
+
+def ljust(text: str, width: int):
+    diff = width - visual_len(text)
+    return text + (" " * diff) if diff > 0 else text
+
+
 def print_table(headers: List[str], data: List[List[str]]):
-    widths = [[len(cell) for cell in row] for row in data + [headers]]
+    widths = [[visual_len(cell) for cell in row] for row in data + [headers]]
     widths = [max(width) for width in zip(*widths)]
     underlines = ["-" * width for width in widths]
 
     def print_row(row: List[str]):
         for idx, cell in enumerate(row):
             width = widths[idx]
-            click.echo(cell.ljust(width), nl=False)
+            click.echo(ljust(cell, width), nl=False)
             click.echo("  ", nl=False)
         click.echo()
 

twitchdl/playlists.py

@@ -3,20 +3,23 @@ Parse and manipulate m3u8 playlists.
 """
 
 from dataclasses import dataclass
-from typing import Generator, List, Optional, OrderedDict
+from os.path import basename
+from typing import Generator, List, Optional
 
 import click
 import m3u8
 
 from twitchdl import utils
-from twitchdl.output import bold, dim
+from twitchdl.output import bold, dim, print_table
 
 
 @dataclass
 class Playlist:
     name: str
+    group_id: str
     resolution: Optional[str]
     url: str
+    is_source: bool
 
 
 @dataclass
@@ -34,17 +37,17 @@ def parse_playlists(playlists_m3u8: str) -> List[Playlist]:
         document = load_m3u8(source)
 
         for p in document.playlists:
-            if p.stream_info.resolution:
-                name = p.media[0].name
-                resolution = "x".join(str(r) for r in p.stream_info.resolution)
-            else:
-                name = p.media[0].group_id
-                resolution = None
+            resolution = (
+                "x".join(str(r) for r in p.stream_info.resolution)
+                if p.stream_info.resolution
+                else None
+            )
 
-            yield Playlist(name, resolution, p.uri)
+            media = p.media[0]
+            is_source = media.group_id == "chunked"
+            yield Playlist(media.name, media.group_id, resolution, p.uri, is_source)
 
-    # Move audio to bottom, it has no resolution
-    return sorted(_parse(playlists_m3u8), key=lambda p: p.resolution is None)
+    return list(_parse(playlists_m3u8))
 
 
 def load_m3u8(playlist_m3u8: str) -> m3u8.M3U8:
@@ -76,23 +79,15 @@ def enumerate_vods(
     return vods
 
 
-def make_join_playlist(
-    playlist: m3u8.M3U8,
-    vods: List[Vod],
-    targets: List[str],
-) -> m3u8.Playlist:
+def make_join_playlist(vods: List[Vod], targets: List[str]) -> m3u8.Playlist:
     """
     Make a modified playlist which references downloaded VODs
     Keep only the downloaded segments and skip the rest
     """
-    org_segments = playlist.segments.copy()
+    playlist = m3u8.M3U8()
 
-    path_map = OrderedDict(zip([v.path for v in vods], targets))
-    playlist.segments.clear()
-    for segment in org_segments:
-        if segment.uri in path_map:
-            segment.uri = path_map[segment.uri]
-            playlist.segments.append(segment)
+    for vod, target in zip(vods, targets):
+        playlist.add_segment(m3u8.Segment(uri=basename(target), duration=vod.duration))
 
     return playlist
 
@@ -107,10 +102,13 @@ def select_playlist(playlists: List[Playlist], quality: Optional[str]) -> Playli
 
 def select_playlist_by_name(playlists: List[Playlist], quality: str) -> Playlist:
     if quality == "source":
-        return playlists[0]
+        for playlist in playlists:
+            if playlist.is_source:
+                return playlist
+        raise click.ClickException("Source quality not found, please report an issue on github.")
 
     for playlist in playlists:
-        if playlist.name == quality:
+        if playlist.name == quality or playlist.group_id == quality:
             return playlist
 
     available = ", ".join([p.name for p in playlists])
@@ -119,13 +117,47 @@ def select_playlist_by_name(playlists: List[Playlist], quality: str) -> Playlist
 
 
 def select_playlist_interactive(playlists: List[Playlist]) -> Playlist:
-    click.echo("\nAvailable qualities:")
-    for n, playlist in enumerate(playlists):
-        if playlist.resolution:
-            click.echo(f"{n + 1}) {bold(playlist.name)} {dim(f'({playlist.resolution})')}")
-        else:
-            click.echo(f"{n + 1}) {bold(playlist.name)}")
+    playlists = sorted(playlists, key=_playlist_key)
+    headers = ["#", "Name", "Group ID", "Resolution"]
 
-    no = utils.read_int("Choose quality", min=1, max=len(playlists) + 1, default=1)
+    rows = [
+        [
+            f"{n + 1})",
+            bold(playlist.name),
+            dim(playlist.group_id),
+            dim(playlist.resolution or ""),
+        ]
+        for n, playlist in enumerate(playlists)
+    ]
+
+    click.echo()
+    print_table(headers, rows)
+
+    default = 1
+    for index, playlist in enumerate(playlists):
+        if playlist.is_source:
+            default = index + 1
+
+    no = utils.read_int("\nChoose quality", min=1, max=len(playlists) + 1, default=default)
     playlist = playlists[no - 1]
     return playlist
+
+
+MAX = 1_000_000
+
+
+def _playlist_key(playlist: Playlist) -> int:
+    """Attempt to sort playlists so that source quality is on top, audio only
+    is on bottom and others are sorted descending by resolution."""
+    if playlist.is_source:
+        return 0
+
+    if playlist.group_id == "audio_only":
+        return MAX
+
+    try:
+        return MAX - int(playlist.name.split("p")[0])
+    except Exception:
+        pass
+
+    return MAX

twitchdl/progress.py

@@ -1,7 +1,7 @@
 import logging
 import time
 from collections import deque
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from statistics import mean
 from typing import Deque, Dict, NamedTuple, Optional
 
@@ -31,28 +31,25 @@ class Sample(NamedTuple):
     timestamp: float
 
 
-@dataclass
 class Progress:
-    vod_count: int
-    downloaded: int = 0
-    estimated_total: Optional[int] = None
-    last_printed: float = field(default_factory=time.time)
-    progress_bytes: int = 0
-    progress_perc: int = 0
-    remaining_time: Optional[int] = None
-    speed: Optional[float] = None
-    start_time: float = field(default_factory=time.time)
-    tasks: Dict[TaskId, Task] = field(default_factory=dict)
-    vod_downloaded_count: int = 0
-    samples: Deque[Sample] = field(default_factory=lambda: deque(maxlen=100))
+    def __init__(self, vod_count: int):
+        self.downloaded: int = 0
+        self.estimated_total: Optional[int] = None
+        self.last_printed: Optional[float] = None
+        self.progress_bytes: int = 0
+        self.progress_perc: int = 0
+        self.remaining_time: Optional[int] = None
+        self.samples: Deque[Sample] = deque(maxlen=1000)
+        self.speed: Optional[float] = None
+        self.tasks: Dict[TaskId, Task] = {}
+        self.vod_count = vod_count
+        self.vod_downloaded_count: int = 0
 
     def start(self, task_id: int, size: int):
         if task_id in self.tasks:
             raise ValueError(f"Task {task_id}: cannot start, already started")
 
         self.tasks[task_id] = Task(task_id, size)
-        self._calculate_total()
-        self._calculate_progress()
         self.print()
 
     def advance(self, task_id: int, size: int):
@@ -63,7 +60,6 @@ class Progress:
         self.progress_bytes += size
         self.tasks[task_id].advance(size)
         self.samples.append(Sample(self.downloaded, time.time()))
-        self._calculate_progress()
         self.print()
 
     def already_downloaded(self, task_id: int, size: int):
@@ -73,8 +69,6 @@ class Progress:
         self.tasks[task_id] = Task(task_id, size)
         self.progress_bytes += size
         self.vod_downloaded_count += 1
-        self._calculate_total()
-        self._calculate_progress()
         self.print()
 
     def abort(self, task_id: int):
@@ -83,9 +77,6 @@ class Progress:
 
         del self.tasks[task_id]
         self.progress_bytes = sum(t.downloaded for t in self.tasks.values())
-
-        self._calculate_total()
-        self._calculate_progress()
         self.print()
 
     def end(self, task_id: int):
@@ -101,12 +92,10 @@ class Progress:
         self.vod_downloaded_count += 1
         self.print()
 
-    def _calculate_total(self):
+    def _recalculate(self):
         self.estimated_total = (
             int(mean(t.size for t in self.tasks.values()) * self.vod_count) if self.tasks else None
         )
-
-    def _calculate_progress(self):
         self.speed = self._calculate_speed()
         self.progress_perc = (
             int(100 * self.progress_bytes / self.estimated_total) if self.estimated_total else 0
@@ -133,9 +122,11 @@ class Progress:
         now = time.time()
 
         # Don't print more often than 10 times per second
-        if now - self.last_printed < 0.1:
+        if self.last_printed and now - self.last_printed < 0.1:
             return
 
+        self._recalculate()
+
         click.echo(f"\rDownloaded {self.vod_downloaded_count}/{self.vod_count} VODs", nl=False)
         click.secho(f" {self.progress_perc}%", fg="blue", nl=False)