warp-beacon 2.3.15__tar.gz → 2.3.16__tar.gz

{warp_beacon-2.3.15/warp_beacon.egg-info → warp_beacon-2.3.16}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: warp_beacon
-Version: 2.3.15
+Version: 2.3.16
 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 → warp_beacon-2.3.16}/pyproject.toml

@@ -24,7 +24,8 @@ dependencies = [
 	"urlextract",
 	"pillow",
 	"pymongo",
-	"instagrapi == 2.0.0"
+	"instagrapi == 2.0.0",
+	"bs4"
 ]
 keywords = ["bot", "telegram", "instagram", "export"]
 classifiers = [

{warp_beacon-2.3.15 → warp_beacon-2.3.16}/setup.py

@@ -64,6 +64,7 @@ setup(
 		"warp_beacon/telegram/placeholder_message",
 		"warp_beacon/telegram/handlers",
 		"warp_beacon/telegram/utils",
+		"warp_beacon/telegram/caption_shortener",
 		"warp_beacon/jobs/abstract",
 		"warp_beacon/jobs/download_job",
 		"warp_beacon/jobs/upload_job",

warp_beacon-2.3.16/warp_beacon/__version__.py

@@ -0,0 +1,2 @@
+__version__ = "2.3.16"
+

{warp_beacon-2.3.15 → warp_beacon-2.3.16}/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-2.3.15 → warp_beacon-2.3.16}/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-2.3.15 → warp_beacon-2.3.16}/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-2.3.15 → warp_beacon-2.3.16}/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-2.3.15 → warp_beacon-2.3.16}/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
@@ -91,6 +92,8 @@ class Bot(object):
 		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(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-2.3.16/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-2.3.15 → warp_beacon-2.3.16}/warp_beacon/telegram/handlers.py

@@ -220,4 +220,28 @@ 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 → warp_beacon-2.3.16/warp_beacon.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: warp_beacon
-Version: 2.3.15
+Version: 2.3.16
 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 → warp_beacon-2.3.16}/warp_beacon.egg-info/SOURCES.txt

@@ -50,6 +50,7 @@ warp_beacon/storage/__init__.py
 warp_beacon/storage/mongo.py
 warp_beacon/telegram/__init__.py
 warp_beacon/telegram/bot.py
+warp_beacon/telegram/caption_shortener.py
 warp_beacon/telegram/handlers.py
 warp_beacon/telegram/placeholder_message.py
 warp_beacon/telegram/utils.py

{warp_beacon-2.3.15 → warp_beacon-2.3.16}/warp_beacon.egg-info/requires.txt

@@ -10,3 +10,4 @@ urlextract
 pillow
 pymongo
 instagrapi==2.0.0
+bs4

{warp_beacon-2.3.15 → warp_beacon-2.3.16}/warp_beacon.egg-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

warp_beacon-2.3.15/warp_beacon/__version__.py

@@ -1,2 +0,0 @@
-__version__ = "2.3.15"
-