reduce number of requests

backend/stream.py

@@ -12,7 +12,6 @@ import tornado.web
 import tornado.routing
 import aiohttp
 import aiohttp_socks
-import html.parser
 import stream_providers
 
 logging.basicConfig(format='[%(filename)s:%(lineno)d] %(message)s', stream=sys.stdout, level=logging.INFO)
@@ -112,26 +111,6 @@ for key in providers:
         for proxy in current:
             proxies[key].append(ProxyElem(proxy))
 
-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 UpstreamHandler():
     def __init__(self):
         self.provider = None
@@ -192,76 +171,6 @@ class UpstreamHandler():
                 for session in sessions:
                     await session.close()
 
-    async def meta(self):
-        data = {}
-        try:
-            embed_url = f'https://noembed.com/embed?url={self.upstream_safe}'
-            async with self.proxy.session() as session:
-                resp_embed_future = session.get(embed_url)
-                resp_upstream_future = session.get(self.upstream)
-                try:
-                    resp_embed = await resp_embed_future
-                except Exception as e:
-                    logger.info(e)
-                    resp_embed = None
-                try:
-                    resp_upstream = await resp_upstream_future
-                except Exception as e:
-                    logger.info(e)
-                    resp_upstream = None
-                text_embed_future = resp_embed.text()
-                text_upstream_future = resp_upstream.text()
-                try:
-                    text_embed = await text_embed_future
-                except Exception as e:
-                    logger.info(e)
-                    text_embed = None
-                try:
-                    text_upstream = await text_upstream_future
-                except Exception as e:
-                    logger.info(e)
-                    text_upstream = None
-                parser = MetaParser()
-                parser.feed(text_upstream)
-                data_raw = json.loads(text_embed)
-                if isinstance(data_raw, dict):
-                    data_new = {}
-                    data_valid = True
-                    data_new["og:title"] = data_raw.get("title")
-                    data_new["og:description"] = data_raw.get("author_name")
-                    data_new["og:image"] = data_raw.get("thumbnail_url")
-                    data_filtered = {}
-                    for key in data_new:
-                        value = data_new.get(key)
-                        if isinstance(value, str):
-                            data_filtered[key] = value
-                    data_filtered.update(parser.meta_data)
-                    data = data_filtered
-                    image = data.get("og:image")
-                    if isinstance(image, str):
-                        if self.provider == "youtube":
-                            full_image = re.sub(r'\/[a-zA-Z0-9]+\.([a-zA-Z0-9]+)$', r'/maxresdefault.\1', image)
-                            standard_image = re.sub(r'\/[a-zA-Z0-9]+\.([a-zA-Z0-9]+)$', r'/sddefault.\1', image)
-                            image_status_full_future = session.head(full_image)
-                            image_status_standard_future = session.head(standard_image)
-                            try:
-                                image_status_full = await image_status_full_future
-                            except Exception as e:
-                                logger.info(e)
-                                image_status_full = None
-                            try:
-                                image_status_standard = await image_status_standard_future
-                            except Exception as e:
-                                logger.info(e)
-                                image_status_standard = None
-                            if hasattr(image_status_full, "status") and (image_status_full.status < 400):
-                                data["og:image"] = full_image
-                            elif hasattr(image_status_standard, "status") and (image_status_standard.status < 400):
-                                data["og:image"] = standard_image
-        except Exception as e:
-            logger.info(e)
-        return data
-
 if icecast_server is not None and stream_server is not None:
     try:
         with open("/app/sources.json", "r") as f:
@@ -376,32 +285,29 @@ class MainHandler(tornado.web.RequestHandler):
     
     async def handle_render(self, handler):
         if script_file is not None and template_html is not None:
-            meta = await handler.meta()
-            meta_list = list(meta.items())
-            title = meta.get("og:title")
+            provider_data = await stream_providers.get_any(handler.upstream, handler.proxy, logger)
             data["script"] = script_file
             data["videojs_version"] = videojs_version
             data["chromecast_version"] = chromecast_version
             data["font_awesome_version"] = font_awesome_version
             data["custom_style"] = custom_style
-            rendered_html = template_html.generate(data=data, meta=meta_list, title=title)
+            rendered_html = template_html.generate(data=data, meta=provider_data.meta(), title=provider_data.title)
             self.write(rendered_html)
         else:
             self.set_status(404)
             self.write("HTML template missing.")
 
     async def handle_stream(self, handler, redir):
+        upstream = None
         if handler.provider == "nextcloud":
             upstream = handler.upstream + "/download"
         else:
-            if not redir:
-                meta = await handler.meta()
-                image = meta.get("og:image")
-                if isinstance(image, str):
-                    image = await handler.proxy.proxy_url(image, None)
+            provider_data = await stream_providers.get_any(handler.upstream, handler.proxy, logger)
+            upstream = provider_data.upstream()
+            if isinstance(provider_data.thumbnail(), str):
+                image = await handler.proxy.proxy_url(provider_data.thumbnail(), None)
                 if isinstance(image, str):
                     self.set_header("Custom-Poster", image)
-            upstream = await stream_providers.get_any(handler.upstream, handler.proxy, logger)
         if upstream is None:
             logger.info(f'invalid upstream ({handler.provider})')
             self.set_status(404)

backend/stream_providers.py

@@ -1,7 +1,11 @@
-#!/usr/bin/env python3
 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):
@@ -11,71 +15,149 @@ class DummyLogger():
     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):
+    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):
-        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 stream.url
-        return None
+        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):
-        opts = {}
-        opts["logger"] = DummyLogger()
-        if isinstance(self.proxy, str):
-            opts["proxy"] = self.proxy
         best_url = None
-        with youtube_dl.YoutubeDL(opts) as ydl:
-            info = ydl.extract_info(self.upstream, download=False)
-            vformats = info.get("formats")
-            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
-        return best_url
+        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)
+        runner = YoutubeRunner(upstream, proxy, logger)
         result_temp = await runner.run()
     except Exception as e:
         logger.info(e)
@@ -86,7 +168,18 @@ async def get_ytdl(upstream, proxy, logger):
 async def get_streamlink(upstream, proxy, logger):
     result = None
     try:
-        runner = StreamlinkRunner(upstream, proxy)
+        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)
@@ -98,16 +191,19 @@ 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)))
-    result = None
+    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, str):
-            result = temp_result
-            break
+        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()