direct audio conversion

ytdlbot/downloader.py

@@ -184,8 +184,9 @@ def can_convert_mp4(video_path, uid):
         return True
 
 
-def ytdl_download(url, tempdir, bm) -> dict:
+def ytdl_download(url, tempdir, bm, **kwargs) -> dict:
     chat_id = bm.chat.id
+    hijack = kwargs.get("hijack")
     response = {"status": True, "error": "", "filepath": []}
     output = pathlib.Path(tempdir, "%(title).70s.%(ext)s").as_posix()
     ydl_opts = {
@@ -193,14 +194,14 @@ def ytdl_download(url, tempdir, bm) -> dict:
         'outtmpl': output,
         'restrictfilenames': False,
         'quiet': True,
-        # "proxy": "socks5://127.0.0.1:1080"
+        "proxy": os.getenv("YTDL_PROXY")
     }
     formats = [
         "bestvideo[ext=mp4]+bestaudio[ext=m4a]/bestvideo+bestaudio",
         "bestvideo[vcodec^=avc]+bestaudio[acodec^=mp4a]/best[vcodec^=avc]/best",
         None
     ]
-    adjust_formats(chat_id, url, formats)
+    adjust_formats(chat_id, url, formats, hijack)
     add_instagram_cookies(url, ydl_opts)
 
     address = ["::", "0.0.0.0"] if IPv6 else [None]
@@ -249,7 +250,7 @@ def ytdl_download(url, tempdir, bm) -> dict:
     if settings[2] == "video" or isinstance(settings[2], MagicMock):
         # only convert if send type is video
         convert_to_mp4(response, bm)
-    if settings[2] == "audio":
+    if settings[2] == "audio" or hijack == "bestaudio[ext=m4a]":
         convert_audio_format(response, bm)
     # disable it for now
     # split_large_video(response)

ytdlbot/limit.py

@@ -107,17 +107,18 @@ class VIP(Redis, MySQL):
     @staticmethod
     def extract_canonical_link(url):
         # canonic link works for many websites. It will strip out unnecessary stuff
+        proxy = {"https": os.getenv("YTDL_PROXY")} if os.getenv("YTDL_PROXY") else None
         props = ["canonical", "alternate", "shortlinkUrl"]
         headers = {
             "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.163 Safari/537.36"}
         # send head request first
-        r = requests.head(url, headers=headers)
+        r = requests.head(url, headers=headers, proxies=proxy)
         if r.status_code != http.HTTPStatus.METHOD_NOT_ALLOWED and "text/html" not in r.headers.get("content-type"):
             # get content-type, if it's not text/html, there's no need to issue a GET request
             logging.warning("%s Content-type is not text/html, no need to GET for extract_canonical_link", url)
             return url
 
-        html_doc = requests.get(url, headers=headers, timeout=5).text
+        html_doc = requests.get(url, headers=headers, timeout=5, proxies=proxy).text
         soup = BeautifulSoup(html_doc, "html.parser")
         for prop in props:
             element = soup.find("link", rel=prop)

ytdlbot/tasks.py

@@ -211,28 +211,17 @@ def direct_normal_download(bot_msg, client, url):
 
 def normal_audio(bot_msg, client):
     chat_id = bot_msg.chat.id
-    fn = getattr(bot_msg.video, "file_name", None) or getattr(bot_msg.document, "file_name", None)
+    # fn = getattr(bot_msg.video, "file_name", None) or getattr(bot_msg.document, "file_name", None)
     status_msg = bot_msg.reply_text("Converting to audio...please wait patiently", quote=True)
+    orig_url: "str" = re.findall(r"http[s]://.*", bot_msg.caption)[0]
     with tempfile.TemporaryDirectory(prefix="ytdl-") as tmp:
-        logging.info("downloading to %s", tmp)
-        base_path = pathlib.Path(tmp)
-        video_path = base_path.joinpath(fn)
-        audio = base_path.joinpath(fn).with_suffix(f".{AUDIO_FORMAT}")
-        client.send_chat_action(chat_id, 'record_video_note')
-        status_msg.edit_text("Preparing your conversion....")
-        client.download_media(bot_msg, video_path)
-        logging.info("downloading complete %s", video_path)
-        # execute ffmpeg
         client.send_chat_action(chat_id, 'record_audio')
-        try:
-            run_ffmpeg(["ffmpeg", "-y", "-i", video_path, "-vn", "-acodec", "copy", audio], status_msg)
-        except subprocess.CalledProcessError:
-            # CPU consuming if re-encoding.
-            run_ffmpeg(["ffmpeg", "-y", "-i", video_path, audio], status_msg)
-
+        # just try to download the audio using yt-dlp
+        resp = ytdl_download(orig_url, tmp, status_msg, hijack="bestaudio[ext=m4a]")
         status_msg.edit_text("Sending audio now...")
         client.send_chat_action(chat_id, 'upload_audio')
-        client.send_audio(chat_id, audio)
+        for f in resp["filepath"]:
+            client.send_audio(chat_id, f)
         status_msg.edit_text("✅ Conversion complete.")
         Redis().update_metrics("audio_success")
 

ytdlbot/utils.py

@@ -76,10 +76,14 @@ def is_youtube(url: "str"):
         return True
 
 
-def adjust_formats(user_id: "str", url: "str", formats: "list"):
+def adjust_formats(user_id: "str", url: "str", formats: "list", hijack=None):
     # high: best quality, 720P, 1080P, 2K, 4K, 8K
     # medium: 480P
     # low: 360P+240P
+    if hijack:
+        formats.insert(0, hijack)
+        return
+
     mapping = {"high": [], "medium": [480], "low": [240, 360]}
     settings = get_user_settings(user_id)
     if settings and is_youtube(url):