Merge tag 'network' into develop

v0.2.5

pytchat/__init__.py

@@ -2,7 +2,7 @@
 pytchat is a lightweight python library to browse youtube livechat without Selenium or BeautifulSoup.
 """
 __copyright__    = 'Copyright (C) 2019 taizan-hokuto'
-__version__      = '0.2.4'
+__version__      = '0.2.5'
 __license__      = 'MIT'
 __author__       = 'taizan-hokuto'
 __author_email__ = '55448286+taizan-hokuto@users.noreply.github.com'

pytchat/cli/__init__.py

@@ -5,9 +5,10 @@ import signal
 import time
 from json.decoder import JSONDecodeError
 from pathlib import Path
+from httpcore import ReadTimeout as HCReadTimeout, NetworkError as HCNetworkError
 from .arguments import Arguments
 from .progressbar import ProgressBar
-from .. exceptions import InvalidVideoIdException, NoContents, PatternUnmatchError
+from .. exceptions import InvalidVideoIdException, NoContents, PatternUnmatchError, UnknownConnectionError
 from .. processors.html_archiver import HTMLArchiver
 from .. tool.extract.extractor import Extractor
 from .. tool.videoinfo import VideoInfo
@@ -50,6 +51,9 @@ def main():
     for counter, video_id in enumerate(Arguments().video_ids):
         if '[' in video_id:
             video_id = video_id.replace('[', '').replace(']', '')
+        if len(Arguments().video_ids) > 1:
+            print(f"\n{'-' * 10} video:{counter + 1} of {len(Arguments().video_ids)} {'-' * 10}")
+
         try:
             video_id = extract_video_id(video_id)
             if os.path.exists(Arguments().output):
@@ -71,8 +75,6 @@ def main():
                     util.save(err.doc, "ERR", ".dat")
                 continue
 
-            if len(Arguments().video_ids) > 1:
-                print(f"\n{'-' * 10} video:{counter + 1} of {len(Arguments().video_ids)} {'-' * 10}")
             print(f"\n"
                   f" video_id: {video_id}\n"
                   f" channel:  {info.get_channel_name()}\n"
@@ -112,6 +114,12 @@ def main():
             print("JSONDecodeError.:{}".format(video_id))
             if Arguments().save_error_data:
                 util.save(e.doc, "ERR_JSON_DECODE", ".dat")
+        except (UnknownConnectionError, HCNetworkError, HCReadTimeout) as e:
+            print(f"An unknown network error occurred during the processing of [{video_id}]. : " + str(e))
+        except PatternUnmatchError:
+            print(f"PatternUnmatchError [{video_id}]. ")
+        except Exception as e:
+            print(type(e), str(e))
 
     return
 

pytchat/processors/html_archiver.py

@@ -1,9 +1,12 @@
+import httpx
 import os
 import re
-import httpx
+import time
 from base64 import standard_b64encode
+from httpx import NetworkError, ReadTimeout
 from .chat_processor import ChatProcessor
 from .default.processor import DefaultProcessor
+from ..exceptions import UnknownConnectionError
 
 
 PATTERN = re.compile(r"(.*)\(([0-9]+)\)$")
@@ -112,7 +115,17 @@ class HTMLArchiver(ChatProcessor):
                        for item in message_items)
 
     def _encode_img(self, url):
-        resp = httpx.get(url)
+        err = None
+        for _ in range(3):
+            try:
+                resp = httpx.get(url)
+                break
+            except (NetworkError, ReadTimeout) as e:
+                err = e
+                time.sleep(3)
+        else:
+            raise UnknownConnectionError(str(err))
+
         return standard_b64encode(resp.content).decode()
 
     def _set_emoji_table(self, item: dict):

pytchat/tool/extract/asyncdl.py

@@ -8,9 +8,11 @@ from ... import config
 from ... paramgen import arcparam
 from ... exceptions import UnknownConnectionError
 from concurrent.futures import CancelledError
+from httpx import NetworkError, ReadTimeout
 from json import JSONDecodeError
 from urllib.parse import quote
 
+
 headers = config.headers
 REPLAY_URL = "https://www.youtube.com/live_chat_replay/" \
              "get_live_chat_replay?continuation="
@@ -66,6 +68,7 @@ def ready_blocks(video_id, duration, div, callback):
     async def _create_block(session, video_id, seektime, callback):
         continuation = arcparam.getparam(video_id, seektime=seektime)
         url = f"{REPLAY_URL}{quote(continuation)}&pbj=1"
+        err = None
         for _ in range(MAX_RETRY_COUNT):
             try:
                 if continuation in param_set:
@@ -77,9 +80,12 @@ def ready_blocks(video_id, duration, div, callback):
                 break
             except JSONDecodeError:
                 await asyncio.sleep(3)
+            except (NetworkError, ReadTimeout) as e:
+                err = e
+                await asyncio.sleep(3)
         else:
             cancel()
-            raise UnknownConnectionError("Abort: Unknown connection error.")
+            raise UnknownConnectionError("Abort:" + str(err))
 
         if actions:
             first = parser.get_offset(actions[0])
@@ -118,6 +124,7 @@ def fetch_patch(callback, blocks, video_id):
 
     async def _fetch(continuation, session) -> Patch:
         url = f"{REPLAY_URL}{quote(continuation)}&pbj=1"
+        err = None
         for _ in range(MAX_RETRY_COUNT):
             try:
                 if continuation in param_set:
@@ -129,9 +136,12 @@ def fetch_patch(callback, blocks, video_id):
                 break
             except JSONDecodeError:
                 await asyncio.sleep(3)
+            except (NetworkError, ReadTimeout) as e:
+                err = e
+                await asyncio.sleep(3)
         else:
             cancel()
-            raise UnknownConnectionError("Abort: Unknown connection error.")
+            raise UnknownConnectionError("Abort:" + str(err))
 
         if actions:
             last = parser.get_offset(actions[-1])

pytchat/tool/extract/extractor.py

@@ -93,4 +93,5 @@ class Extractor:
         return ret
 
     def cancel(self):
+        print("cancel")
         asyncdl.cancel()

pytchat/tool/videoinfo.py

@@ -1,10 +1,13 @@
+import httpx
 import json
 import re
-import httpx
+import time
+from httpx import ConnectError, NetworkError
 from .. import config
-from ..exceptions import InvalidVideoIdException, PatternUnmatchError
+from ..exceptions import InvalidVideoIdException, PatternUnmatchError, UnknownConnectionError
 from ..util.extract_video_id import extract_video_id
 
+
 headers = config.headers
 
 pattern = re.compile(r"'PLAYER_CONFIG': ({.*}}})")
@@ -85,8 +88,18 @@ class VideoInfo:
 
     def _get_page_text(self, video_id):
         url = f"https://www.youtube.com/embed/{video_id}"
-        resp = httpx.get(url, headers=headers)
+        err = None
-        resp.raise_for_status()
+        for _ in range(3):
+            try:
+                resp = httpx.get(url, headers=headers)
+                resp.raise_for_status()
+                break
+            except (ConnectError, NetworkError) as e:
+                err = e
+                time.sleep(3)
+        else:
+            raise UnknownConnectionError(str(err))
+
         return resp.text
 
     def _parse(self, text):