warp-beacon 2.3.15__py3-none-any.whl → 2.3.17__py3-none-any.whl

warp_beacon/__version__.py

@@ -1,2 +1,2 @@
-__version__ = "2.3.15"
+__version__ = "2.3.17"
 

warp_beacon/jobs/abstract.py

@@ -44,6 +44,7 @@ class JobSettings(TypedDict):
 	job_postponed_until: int
 	message_leftover: str
 	replay: bool
+	short_text: bool
 
 class AbstractJob(ABC):
 	job_id: uuid.UUID = None
@@ -80,6 +81,7 @@ class AbstractJob(ABC):
 	job_postponed_until: int = -1
 	message_leftover: str = ""
 	replay: bool = False
+	short_text: bool = False
 
 	def __init__(self, **kwargs: Unpack[JobSettings]) -> None:
 		if kwargs:

warp_beacon/scheduler/scheduler.py

@@ -1,6 +1,7 @@
 import os
 import time
 from random import randrange
+import datetime
 import threading
 import json
 
@@ -81,6 +82,12 @@ class IGScheduler(object):
 		self.load_state()
 		while self.running:
 			try:
+				now = datetime.datetime.now()
+				if 4 <= now.hour < 7:
+					logging.info("Scheduler is paused due to night hours (4:00 - 7:00)")
+					self.event.wait(timeout=10800)
+					continue
+
 				if self.state["remaining"] <= 0:
 					self.state["remaining"] = randrange(8400, 26200)
 					logging.info("Next scheduler activity in '%s' seconds", self.state["remaining"])

warp_beacon/scraper/account_selector.py

@@ -56,8 +56,15 @@ class AccountSelector(object):
 						logging.info("Account proxy matched '%s'", proxy)
 						matched_proxy.append(proxy)
 				if matched_proxy:
-					random.seed(random.seed(time.time_ns() ^ int.from_bytes(os.urandom(len(matched_proxy)), "big")))
+					if len(matched_proxy) > 1:
+						random.seed(random.seed(time.time_ns() ^ int.from_bytes(os.urandom(len(matched_proxy)), "big")))
+						# ensure new proxy in case if previous account required captcha
+						last_proxy = self.accounts.get("last_proxy", None)
+						if last_proxy and last_proxy in matched_proxy:
+							matched_proxy.remove(last_proxy)
 					prox_choice = random.choice(matched_proxy)
+					# saving chosen proxy for history
+					self.accounts["last_proxy"] = prox_choice
 					logging.info("Chosen proxy: '%s'", prox_choice)
 					return prox_choice
 			except Exception as e:
@@ -103,8 +110,8 @@ class AccountSelector(object):
 	def bump_acc_fail(self, key: str, amount: int = 1) -> int:
 		try:
 			idx = self.account_index[self.current_module_name].value
-			self.accounts_meta_data[idx][key] += amount
-			return self.accounts_meta_data[idx][key]
+			self.accounts_meta_data[self.current_module_name][idx][key] += amount
+			return self.accounts_meta_data[self.current_module_name][idx][key]
 		except Exception as e:
 			logging.warning("Failed to record fail stats")
 			logging.exception(e)

warp_beacon/storage/__init__.py

@@ -52,12 +52,15 @@ class Storage(object):
 		path = urlparse(url).path.strip('/')
 		return path
 	
-	def db_find(self, uniq_id: str) -> list[dict]:
+	def db_find(self, uniq_id: str, origin: str = "") -> list[dict]:
 		document = None
 		ret = []
 		try:
 			logging.debug("uniq_id to search is '%s'", uniq_id)
-			cursor = self.db.find({"uniq_id": uniq_id})
+			find_opts = {"uniq_id": uniq_id}
+			if origin:
+				find_opts["origin"] = origin
+			cursor = self.db.find(find_opts)
 			for document in cursor:
 				ret.append(
 				{

warp_beacon/telegram/bot.py

@@ -21,6 +21,7 @@ from warp_beacon.uploader import AsyncUploader
 from warp_beacon.jobs.upload_job import UploadJob
 from warp_beacon.jobs.types import JobType
 from warp_beacon.telegram.utils import Utils
+from warp_beacon.telegram.caption_shortener import CaptionShortner
 from warp_beacon.scheduler.scheduler import IGScheduler
 
 import logging
@@ -90,7 +91,9 @@ class Bot(object):
 		self.client.add_handler(MessageHandler(self.handlers.help, filters.command("help")))
 		self.client.add_handler(MessageHandler(self.handlers.random, filters.command("random")))
 		self.client.add_handler(MessageHandler(self.handlers.handler))
-		self.client.add_handler(CallbackQueryHandler(self.handlers.simple_button_handler))
+		self.client.add_handler(self.handlers.simple_button_handler, filters=filters.create(lambda _, q: not q.data.startswith("readmore:")))
+		self.client.add_handler(CallbackQueryHandler(self.handlers.read_more_handler, filters=filters.create(lambda _, q: q.data.startswith("readmore:"))))
+
 
 		self.placeholder = PlaceholderMessage(self)
 
@@ -158,9 +161,14 @@ class Bot(object):
 
 	def build_signature_caption(self, job: UploadJob) -> str:
 		caption = ""
+		is_group = job.chat_type in (ChatType.GROUP, ChatType.SUPERGROUP)
 		if job.canonical_name:
-			caption = f"<b>{html.escape(job.canonical_name)}</b>"
-		if job.chat_type in (ChatType.GROUP, ChatType.SUPERGROUP):
+			if is_group and CaptionShortner.need_short(job.canonical_name):
+				caption = f"{html.escape(CaptionShortner.smart_truncate_html(job.canonical_name))} ..."
+				job.short_text = True
+			else:
+				caption = f"<b>{html.escape(job.canonical_name)}</b>"
+		if is_group:
 			if job.canonical_name:
 				caption += "\n—\n"
 			if job.message_leftover:
@@ -336,8 +344,16 @@ class Bot(object):
 			args["disable_notification"] = True
 			args["reply_to_message_id"] = job.message_id
 
-		if os.environ.get("ENABLE_DONATES", None) == "true" and job.media_type is not JobType.COLLECTION:
-			args["reply_markup"] = InlineKeyboardMarkup([[InlineKeyboardButton("❤ Donate", url=os.environ.get("DONATE_LINK", "https://pay.cryptocloud.plus/pos/W5BMtNQt5bJFoW2E"))]])
+		if job.media_type is not JobType.COLLECTION:
+			render_donates = os.environ.get("ENABLE_DONATES", None) == "true"
+			keyboard_buttons = [[]]
+			if job.short_text:
+				keyboard_buttons[0].append(InlineKeyboardButton("📖 Read more", callback_data=f"read_more:{job.job_origin}:{job.uniq_id}"))
+			if render_donates:
+				keyboard_buttons[0].append(InlineKeyboardButton("❤ Donate", url=os.environ.get("DONATE_LINK", "https://pay.cryptocloud.plus/pos/W5BMtNQt5bJFoW2E")))
+
+			if keyboard_buttons[0]:  #job.short_text or render_donates:
+				args["reply_markup"] = InlineKeyboardMarkup(keyboard_buttons)
 
 		return args
 

warp_beacon/telegram/caption_shortener.py

@@ -0,0 +1,66 @@
+import logging
+from typing import Union
+from bs4 import NavigableString, Tag, Comment
+from bs4 import BeautifulSoup
+
+CAPTION_LENGTH_LIMIT = 85
+
+class CaptionShortner(object):
+	@staticmethod
+	def strip_html(text: str) -> str:
+		try:
+			soup = BeautifulSoup(text, "html.parser")
+			return soup.get_text()
+		except Exception as e:
+			logging.warning("Failed to stript HTML tags!")
+			logging.exception(e)
+
+		return text
+
+	@staticmethod
+	def need_short(text: str) -> bool:
+		wo_html = CaptionShortner.strip_html(text)
+		if len(wo_html) > CAPTION_LENGTH_LIMIT:
+			return True
+		return False
+
+	@staticmethod
+	def smart_truncate_html(html: str, limit: int = CAPTION_LENGTH_LIMIT) -> str:
+		result = ""
+		try:
+			soup = BeautifulSoup(html, "html.parser")
+			length = 0
+
+			def walk(node: Union[NavigableString, Tag, Comment]) -> None:
+				nonlocal result, length
+				if length >= limit:
+					return
+
+				if isinstance(node, str):
+					words = node.split()
+					for word in words:
+						if length + len(word) + 1 > limit:
+							return
+						if result and not result.endswith(" "):
+							result += " "
+							length += 1
+						result += word
+						length += len(word)
+				elif isinstance(node, Tag):
+					if node.name == '[document]':
+						for child in node.children:
+							walk(child)
+						return
+
+					for child in node.children:
+						walk(child)
+						if length >= limit:
+							break
+
+					result += f"</{node.name}>"
+
+			walk(soup)
+		except Exception as e:
+			logging.warning("Fail in smart_truncate_html!")
+			logging.exception(e)
+		return result

warp_beacon/telegram/handlers.py

@@ -213,6 +213,7 @@ class Handlers(object):
 		if chat.type not in (ChatType.GROUP, ChatType.SUPERGROUP) and not urls:
 			await self.bot.send_text(text=reply_text, reply_id=effective_message_id, chat_id=chat.id)
 
+	#TODO refactor to callback router
 	async def simple_button_handler(self, client: Client, query: CallbackQuery) -> None:
 		await client.answer_callback_query(
 			callback_query_id=query.id,
@@ -220,4 +221,28 @@ class Handlers(object):
 			show_alert=True
 		)
 		self.bot.downloader.auth_event.set()
-		self.bot.downloader.auth_event.clear()
+		self.bot.downloader.auth_event.clear()
+
+	async def read_more_handler(self, client: Client, query: CallbackQuery) -> None:
+		origin, uniq_id = '', ''
+		#read_more:{job.job_origin}:{job.uniq_id}
+		if query.data:
+			parts = query.data.split(':')
+			if len(parts) == 3:
+				_, origin, uniq_id = parts
+		db_results = []
+		if uniq_id and origin:
+			db_results = self.storage.db_find(uniq_id=uniq_id.strip(), origin=origin.strip())
+		first_entity = {}
+		if db_results:
+			first_entity = db_results[0]
+
+		try:
+			await client.answer_callback_query(
+				callback_query_id=query.id,
+				show_alert=True,
+				text=first_entity.get("canonical_name", "Failed to fetch data.")
+			)
+		except Exception as e:
+			logging.warning("read_more_handler: Failed for uniq_id='%s', origin='%s", uniq_id, origin)
+			logging.exception(e)

{warp_beacon-2.3.15.dist-info → warp_beacon-2.3.17.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: warp_beacon
-Version: 2.3.15
+Version: 2.3.17
 Summary: Telegram bot for expanding external media links
 Home-page: https://github.com/sb0y/warp_beacon
 Author: Andrey Bagrintsev
@@ -236,6 +236,7 @@ Requires-Dist: urlextract
 Requires-Dist: pillow
 Requires-Dist: pymongo
 Requires-Dist: instagrapi==2.0.0
+Requires-Dist: bs4
 Dynamic: author
 Dynamic: home-page
 

{warp_beacon-2.3.15.dist-info → warp_beacon-2.3.17.dist-info}/RECORD

@@ -4,12 +4,12 @@ var/warp_beacon/accounts.json,sha256=OsXdncs6h88xrF_AP6_WDCK1waGBn9SR-uYdIeK37GM
 var/warp_beacon/placeholder.gif,sha256=cE5CGJVaop4Sx21zx6j4AyoHU0ncmvQuS2o6hJfEH88,6064
 var/warp_beacon/proxies.json,sha256=VnjlQDXumOEq72ZFjbh6IqHS1TEHqn8HPYAZqWCeSIA,95
 warp_beacon/__init__.py,sha256=_rThNODmz0nDp_n4mWo_HKaNFE5jk1_7cRhHyYaencI,163
-warp_beacon/__version__.py,sha256=JVaWB6HomdBMOznRLcZrplBgUWb-fYoP0lJ0CYpYETM,24
+warp_beacon/__version__.py,sha256=7kEUywY6HMmkf0hezoDB_dwDQMiBqA4wHs6q4S35P6Y,24
 warp_beacon/warp_beacon.py,sha256=7KEtZDj-pdhtl6m-zFLsSojs1ZR4o7L0xbqtdmYPvfE,342
 warp_beacon/compress/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 warp_beacon/compress/video.py,sha256=_PDMVYCyzLYxHv1uZmmzGcG_8rjaZr7BTXsXTTy_oS4,2846
 warp_beacon/jobs/__init__.py,sha256=ED8_tPle4iL4kqNW0apAVkgNQtRRTnYfAJwBjO1g0JY,180
-warp_beacon/jobs/abstract.py,sha256=LyM4PzOXXlQxll1_tjpD-zjKpkJHDXl1pC4ziEdrLTM,3043
+warp_beacon/jobs/abstract.py,sha256=lYZplMq4Uo9vUB6ziMsC_FPer2NsvhdmFJ_I54L7v7Y,3087
 warp_beacon/jobs/download_job.py,sha256=5HiPcnJppFMhO14___3eSkoMygM3y-vhpGkMAuNhK7s,854
 warp_beacon/jobs/types.py,sha256=Ae8zINgbs7cOcYkYoOCOACA7duyhnIGMQAJ_SJB1QRQ,176
 warp_beacon/jobs/upload_job.py,sha256=_ul4psPej1jLEs-BMcMR80GbXDSmm38jE9yoZtecclY,741
@@ -19,10 +19,10 @@ warp_beacon/mediainfo/audio.py,sha256=ous88kwQj4bDIChN5wnGil5LqTs0IQHH0d-nyrL0-Z
 warp_beacon/mediainfo/silencer.py,sha256=qxMuViOoVwUYb60uCVvqHiGrqByR1_4_rqMT-XdMkwc,1813
 warp_beacon/mediainfo/video.py,sha256=UBZrhTN5IDI-aYu6tsJEILo9nFkjHhkldGVFmvV7tEI,2480
 warp_beacon/scheduler/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-warp_beacon/scheduler/scheduler.py,sha256=F9jF6myIHOI0TmC0-rBt2Un8gVvuhBW9zL-nEAEj0Bc,2585
+warp_beacon/scheduler/scheduler.py,sha256=gcV-e-9MjL8JEfu7T3Gy1imUrXz266ub9Ub4O27W5Hw,2786
 warp_beacon/scraper/__init__.py,sha256=V_C5SmAmRtjFfVBtTotOHCW3ILMQ8m_ulrBF6ry59_A,16944
 warp_beacon/scraper/abstract.py,sha256=6A6KuBUHZhu8VAyBwLgmnxMPHJcLpgwLapmULy8hpoA,2726
-warp_beacon/scraper/account_selector.py,sha256=zYFn9qd5x_BOox5g1iG7QjqaLe_5ezKD_2xFy06fPrk,4731
+warp_beacon/scraper/account_selector.py,sha256=zv5ci_Y7W-tzKeFklffpLts19a1wFg69fqefK-_i_pk,5122
 warp_beacon/scraper/exceptions.py,sha256=Qkz76yo-X5kucEZIP9tWaK-oYO-kvsPEl8Y0W63oDhU,1300
 warp_beacon/scraper/fail_handler.py,sha256=_blvckfTZ4xWVancQKVRXH5ClKGwfrBxMwvXIFZh1qA,975
 warp_beacon/scraper/link_resolver.py,sha256=Rc9ZuMyOo3iPywDHwjngy-WRQ2SXhJwxcg-5ripx7tM,2447
@@ -33,17 +33,18 @@ warp_beacon/scraper/youtube/abstract.py,sha256=cBtExei2Cb3o4YBtHqi8If_FdmE6NyJqN
 warp_beacon/scraper/youtube/music.py,sha256=qbijpSv54fsrIYHeY-nfmw4vo6oBmedQHsVG8pXNfrc,1380
 warp_beacon/scraper/youtube/shorts.py,sha256=ujGEV7ILXHqBRa99SyITsnR7ulAHJDtumAh51kVX880,1231
 warp_beacon/scraper/youtube/youtube.py,sha256=fGrbjBngvvNdpzhb1yZVedNW0_tCrLc31VSYMSHzcQY,2135
-warp_beacon/storage/__init__.py,sha256=2uvyIR0APIW6gOxwJRvCji7wS2q6I7dghvLyWsRqRxo,3312
+warp_beacon/storage/__init__.py,sha256=xEzexwWZTMtjVesQ73BLhS0VkWoqRXSUmuXS4h5HOvY,3402
 warp_beacon/storage/mongo.py,sha256=qC4ZiO8XXvPnP0rJwz4CJx42pqFsyAjCiW10W5QdT6E,527
 warp_beacon/telegram/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-warp_beacon/telegram/bot.py,sha256=WDaUphlF7p6pQkVajIM_F07LvswZXrzOUV5oiAzbb-Y,15625
-warp_beacon/telegram/handlers.py,sha256=7gldKrX-CTDQNfD1NQ08cC31uPR-95mhN59OujAEqZg,7704
+warp_beacon/telegram/bot.py,sha256=YgHTCZuPTC04Pkl18A9b6eA7pZBbDeQn_iEBtnRoVxs,16438
+warp_beacon/telegram/caption_shortener.py,sha256=6J4Dp35mOmJEyctiuN8WXM39bkQ13m78HoghsYX3fbU,1548
+warp_beacon/telegram/handlers.py,sha256=P9eR8DzFBSEQVaU0Fp0ABu8qeqFowu0GSU2SdZDfEdc,8511
 warp_beacon/telegram/placeholder_message.py,sha256=wN9-BRiyrtHG-EvXtZkGJHt2CX71munQ57ITttjt0mw,6400
 warp_beacon/telegram/utils.py,sha256=9uebX53G16mV7ER7WgfdWBLFHHw14S8HBt9URrIskg0,4440
 warp_beacon/uploader/__init__.py,sha256=5KRWsxPRGuQ56YhCEnJsXnb-yQp8dpvWEsPDf0dD-fw,4964
-warp_beacon-2.3.15.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-warp_beacon-2.3.15.dist-info/METADATA,sha256=jpCB5tW5_C8ybCWMY4Z1CgZi_FLS-bNtAMCB7HzTZGs,21704
-warp_beacon-2.3.15.dist-info/WHEEL,sha256=52BFRY2Up02UkjOa29eZOS2VxUrpPORXg1pkohGGUS8,91
-warp_beacon-2.3.15.dist-info/entry_points.txt,sha256=eSB61Rb89d56WY0O-vEIQwkn18J-4CMrJcLA_R_8h3g,119
-warp_beacon-2.3.15.dist-info/top_level.txt,sha256=XusFB1mGUStI2gqbtRuJzQrfrAXmdMaT3QTyfvdLEXI,1064
-warp_beacon-2.3.15.dist-info/RECORD,,
+warp_beacon-2.3.17.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+warp_beacon-2.3.17.dist-info/METADATA,sha256=NeHzbOuWLvTkhwo8d4UPZ7j3J4HyEKSYSn_HL5qD3KE,21723
+warp_beacon-2.3.17.dist-info/WHEEL,sha256=52BFRY2Up02UkjOa29eZOS2VxUrpPORXg1pkohGGUS8,91
+warp_beacon-2.3.17.dist-info/entry_points.txt,sha256=eSB61Rb89d56WY0O-vEIQwkn18J-4CMrJcLA_R_8h3g,119
+warp_beacon-2.3.17.dist-info/top_level.txt,sha256=2iKFlYwJ-meO9sCX4OGEP1hhQN17t2KFksQ5dXMhXUA,1103
+warp_beacon-2.3.17.dist-info/RECORD,,

{warp_beacon-2.3.15.dist-info → warp_beacon-2.3.17.dist-info}/top_level.txt

@@ -31,6 +31,7 @@ warp_beacon/storage
 warp_beacon/storage/mongo
 warp_beacon/telegram
 warp_beacon/telegram/bot
+warp_beacon/telegram/caption_shortener
 warp_beacon/telegram/handlers
 warp_beacon/telegram/placeholder_message
 warp_beacon/telegram/utils