Нашел +- быстрый способ скачивания файла с помощью yt-dlp

app/utils/youtube.py

@@ -1,6 +1,9 @@
-from typing import Optional
+import subprocess
+from typing import Literal
 from dataclasses import dataclass
 
+from app.utils.uploader import ChunkUploadBackend
+
 
 @dataclass
 class MediaContent:
@@ -9,9 +12,10 @@ class MediaContent:
     chunk_size: int
 
 
-class YtDlpInfo:
-    def __init__(self, url: str):
+class YtDlpManager:
+    def __init__(self, url: str, backend: ChunkUploadBackend):
         self.url = url
+        self.backend = backend
         self.info = None
         self._extract_info()
 
@@ -32,56 +36,35 @@ class YtDlpInfo:
         """Возвращает title видео"""
         return self.info.get("title", "unknown")
 
-    def get_video_url(self, resolution: Optional[str] = None) -> Optional[MediaContent]:
-        """
-        Возвращает ссылку на видеопоток с указанным разрешением
-        resolution: например '1080p', '720p', '480p'
-        """
-        formats = self.info.get("formats", [])
-        video_formats = [
-            f
-            for f in formats
-            if f.get("vcodec") != "none"  # есть видео
-            and f.get("acodec") == "none"  # без аудио
+    def download_video(self, size: Literal[240, 360, 480, 720, 1080, 1440, 2160]):
+        command = [
+            "yt-dlp",
+            "-f",
+            f"bestvideo[height<={size}]",
+            "--no-part",
+            "--quiet",
+            "--no-warnings",
+            "-o",
+            "-",
+            self.url
         ]
-        if resolution:
-            # ищем точное соответствие разрешению
-            for f in video_formats:
-                if f.get("format_note") == resolution:
-                    return MediaContent(
-                        f.get("url", ""),
-                        f.get("http_headers", {}),
-                        f.get("downloader_options", {}).get("http_chunk_size", 1024**2),
-                    )
-        # если разрешение не указано или не найдено — берем лучший
-        if video_formats:
-            # сортируем по height
-            video_formats.sort(key=lambda x: x.get("height") or 0, reverse=True)
-            return MediaContent(
-                video_formats[0].get("url", ""),
-                video_formats[0].get("http_headers", {}),
-                video_formats[0]
-                .get("downloader_options", {})
-                .get("http_chunk_size", 1024**2),
-            )
-        return None
-
-    def get_audio_url(self) -> Optional[MediaContent]:
-        """Возвращает ссылку на аудиопоток"""
-        formats = self.info.get("formats", [])
-        audio_formats = [
-            f
-            for f in formats
-            if f.get("vcodec") == "none" and f.get("acodec") != "none"
-        ]
-        if audio_formats:
-            # берем наилучшее качество
-            audio_formats.sort(key=lambda x: x.get("abr") or 0, reverse=True)
-            return MediaContent(
-                audio_formats[0].get("url", ""),
-                audio_formats[0].get("http_headers", {}),
-                audio_formats[0]
-                .get("downloader_options", {})
-                .get("http_chunk_size", 1024**2),
-            )
-        return None
+        print("Start processing")
+        process = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.DEVNULL, bufsize=0)
+        print("Write filename to upload backend")
+        self.backend.start(self.title + ".mp4")
+        print("Start write chunk to upload backend")
+        chunk_size = 1024 ** 2
+        length = 0
+        while True:
+            chunk = process.stdout.read(chunk_size)
+            if not chunk:
+                break
+            length += chunk_size
+            print("Write chunk to backend", length)
+            self.backend.upload_chunk(chunk)
+        print("End writing to backend")
+        ret = process.wait()
+        print("Check ret status")
+        if ret != 0:
+            self.backend.abort()
+            raise RuntimeError(f"yt-dlp failed, status code: {ret}")

test.py

@@ -1,19 +1,21 @@
 from app.utils.downloader import HttpStreamingDownloader
 from app.utils.uploader import DiskChunkUploadBackend
-from app.utils.youtube import YtDlpInfo
+from app.utils.youtube import YtDlpManager
 
 
 def download(url: str):
     upload_backend = DiskChunkUploadBackend("trash_holder")
-    downloader = HttpStreamingDownloader(upload_backend)
-    youtube = YtDlpInfo(url)
-    video = youtube.get_video_url("480p")
-    video_name = youtube.title + ".mp4"
-    audio = youtube.get_audio_url()
-    audio_name = youtube.title + ".m4a"
+    youtube = YtDlpManager(url, upload_backend)
+    youtube.download_video(360)
+    # downloader = HttpStreamingDownloader(upload_backend)
+    # youtube = YtDlpInfo(url)
+    # video = youtube.get_video_url("480p")
+    # video_name = youtube.title + ".mp4"
+    # audio = youtube.get_audio_url()
+    # audio_name = youtube.title + ".m4a"
 
-    downloader.download(video.url, video_name, video.headers, video.chunk_size)
-    downloader.download(audio.url, audio_name, audio.headers, audio.chunk_size)
+    # downloader.download(video.url, video_name, video.headers, video.chunk_size)
+    # downloader.download(audio.url, audio_name, audio.headers, audio.chunk_size)
 
 
 def main():