stream-api/backend/stream_providers.py

import youtube_dl
import streamlink
import requests
import asyncio
import html.parser
import re

ytimg_pattern = re.compile(r'(https:\/\/[a-z0-9.]+ytimg\.com\/.+\/)[a-z0-9]+(\.[a-z0-9]+)')

class DummyLogger():
    def debug(self, msg):
        pass
    def warning(self, msg):
        pass
    def error(self, msg):
        pass

class MetaParser(html.parser.HTMLParser):
    def __init__(self):
        self.meta_data = {}
        self.accepted_attrs = []
        self.accepted_attrs.append("og:title")
        self.accepted_attrs.append("og:description")
        self.accepted_attrs.append("og:image")
        super().__init__()
    def handle_starttag(self, tag, attrs):
        if tag == "meta":
            name = None
            for attr in (attrs + attrs):
                if len(attr) == 2:
                    if isinstance(name, str):
                        if attr[0] == "content":
                            self.meta_data[name] = attr[1]
                            return
                    elif attr[0] == "property" and attr[1] in self.accepted_attrs:
                        name = attr[1]

class StreamData():
    def __init__(self, upstream, thumbnail, title, description, override):
        self.values = {}
        self.values["upstream"] = upstream
        self.values["thumbnail"] = thumbnail
        self.values["title"] = title
        self.values["description"] = description
        self.override = override
    def update(self, key, value, override):
        missing = not isinstance(self.values.get(key), str)
        override = override and isinstance(value, str)
        if missing or override:
            self.values[key] = value
    def upstream(self):
        return self.values.get("upstream")
    def thumbnail(self):
        return self.values.get("thumbnail")
    def title(self):
        return self.values.get("title")
    def description(self):
        return self.values.get("description")
    def meta(self):
        data = []
        if isinstance(self.values.get("thumbnail"), str):
            data.append(("og:image", self.values.get("thumbnail")))
        if isinstance(self.values.get("title"), str):
            data.append(("og:title", self.values.get("title")))
        if isinstance(self.values.get("description"), str):
            data.append(("og:description", self.values.get("description")))
        return data

class StreamProvider():
    def __init__(self, upstream, proxy, logger):
        self.upstream = upstream
        self.proxy = None
        self.logger = logger
        proxy = str(proxy)
        if len(proxy) > 5:
            self.proxy = "socks5://" + proxy

class StreamlinkRunner(StreamProvider):
    def stream(self):
        try:
            session = streamlink.Streamlink()
            if self.proxy is not None:
                session.set_option("https-proxy", self.proxy)
                session.set_option("http-proxy", self.proxy)
            streams = session.streams(self.upstream)
            if streams is not None:
                for key in reversed(streams):
                    stream = streams.get(key)
                    if hasattr(stream, "url"):
                        return StreamData(stream.url, None, None, None, False)
        except Exception as e:
            self.logger.info(e)
        return StreamData(None, None, None, None, False)
    async def run(self):
        return await asyncio.to_thread(self.stream)

class YoutubeRunner(StreamProvider):
    def stream(self):
        best_url = None
        thumbnail = None
        title = None
        description = None
        try:
            opts = {}
            opts["logger"] = DummyLogger()
            if isinstance(self.proxy, str):
                opts["proxy"] = self.proxy
            with youtube_dl.YoutubeDL(opts) as ydl:
                info = ydl.extract_info(self.upstream, download=False)
                vformats = info.get("formats")
                thumbnail = info.get("thumbnail")
                description = info.get("channel")
                title = info.get("title")
                best_format = {}
                best_format["width"] = 10
                best_format["height"] = 10
                if isinstance(vformats, list):
                    for vformat in vformats:
                        acodec = vformat.get("acodec")
                        vcodec = vformat.get("vcodec")
                        current_width = vformat.get("height")
                        current_height = vformat.get("width")
                        best_width = best_format.get("width")
                        best_height = best_format.get("height")
                        new_url = vformat.get("url")
                        if (isinstance(best_width, int) and 
                                isinstance(best_height, int) and
                                isinstance(current_width, int) and
                                isinstance(current_height, int) and
                                isinstance(new_url, str) and
                                current_width > best_width and
                                current_height > best_height and
                                acodec != "none" and
                                vcodec != "none"):
                            best_format = vformat
                            best_url = new_url
        except Exception as e:
            self.logger.info(e)
        return StreamData(best_url, thumbnail, title, description, True)
    async def run(self):
        return await asyncio.to_thread(self.stream)

class MetaRunner(StreamProvider):
    def stream(self):
        data = {}
        try:
            resp = requests.get(self.upstream)
            parser = MetaParser()
            parser.feed(resp.text)
            data = parser.meta_data
        except Exception as e:
            self.logger.info(e)
        return StreamData(None, data.get("og:image"), data.get("og:title"), data.get("og:description"), False)
    async def run(self):
        return await asyncio.to_thread(self.stream)

async def get_ytdl(upstream, proxy, logger):
    result = None
    try:
        runner = YoutubeRunner(upstream, proxy, logger)
        result_temp = await runner.run()
    except Exception as e:
        logger.info(e)
    else:
        result = result_temp
    return result

async def get_streamlink(upstream, proxy, logger):
    result = None
    try:
        runner = StreamlinkRunner(upstream, proxy, logger)
        result_temp = await runner.run()
    except Exception as e:
        logger.info(e)
    else:
        result = result_temp
    return result

async def get_meta(upstream, proxy, logger):
    result = None
    try:
        runner = MetaRunner(upstream, proxy, logger)
        result_temp = await runner.run()
    except Exception as e:
        logger.info(e)
    else:
        result = result_temp
    return result

async def get_any(upstream, proxy, logger):
    tasks = []
    tasks.append(asyncio.create_task(get_streamlink(upstream, proxy, logger)))
    tasks.append(asyncio.create_task(get_ytdl(upstream, proxy, logger)))
    tasks.append(asyncio.create_task(get_meta(upstream, proxy, logger)))
    result = StreamData(None, None, None, None, False)
    for task in asyncio.as_completed(tasks, timeout=5.0):
        temp_result = None
        try:
            temp_result = await task
        except Exception as e:
            logger.info(e)
        if isinstance(temp_result, StreamData):
            result.update("upstream", temp_result.upstream(), temp_result.override)
            result.update("thumbnail", temp_result.thumbnail(), temp_result.override)
            result.update("title", temp_result.title(), temp_result.override)
            result.update("description", temp_result.description(), temp_result.override)
    for task in tasks:
        if not task.done():
            task.cancel()
    return result