warp-beacon 2.8.5__tar.gz → 2.8.7__tar.gz

{warp_beacon-2.8.5/warp_beacon.egg-info → warp_beacon-2.8.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: warp_beacon
-Version: 2.8.5
+Version: 2.8.7
 Summary: Telegram bot for expanding external media links
 Home-page: https://github.com/sb0y/warp_beacon
 Author: Andrey Bagrintsev

warp_beacon-2.8.7/warp_beacon/__version__.py

@@ -0,0 +1,2 @@
+__version__ = "2.8.7"
+

{warp_beacon-2.8.5 → warp_beacon-2.8.7}/warp_beacon/scraper/X/X.py

@@ -5,7 +5,7 @@ from mimetypes import guess_extension
 from urllib.parse import urlparse
 import requests
 import yt_dlp
-from playwright.sync_api import sync_playwright
+from playwright.sync_api import sync_playwright, Page
 
 from warp_beacon.telegram.utils import Utils
 from warp_beacon.scraper.utils import ScraperUtils
@@ -100,7 +100,7 @@ class XScraper(XAbstract):
 				chunk = []
 				for media in media_chunk:
 					chunk.append({
-						"local_media_path": media,
+						"local_media_path": self.rename_local_file(media),
 						"canonical_name": post_text,
 						"media_type": JobType.IMAGE
 					})
@@ -112,7 +112,7 @@ class XScraper(XAbstract):
 		else:
 			if local_file:
 				res.append({
-					"local_media_path": local_file,
+					"local_media_path": self.rename_local_file(local_file),
 					"performer": media_info.get("uploader", "Unknown"),
 					"canonical_name": post_text,
 					"media_type": job_type
@@ -200,7 +200,34 @@ class XScraper(XAbstract):
 				downloaded_imgs.append(filepath)
 
 		return downloaded_imgs, post_text
-	
+
+	def extract_post_text(self, page: Page) -> str:
+		try:
+			tweet_texts = []
+			# collecting text blocks from post
+			containers = page.query_selector_all('div[data-testid="tweetText"]')
+			for container in containers:
+				try:
+					spans = container.query_selector_all("span")
+					if spans:
+						for span in spans:
+							text = span.inner_text().strip()
+							if text:
+								tweet_texts.append(text)
+					else:
+						# to span's try container itself
+						text = container.inner_text().strip()
+						if text:
+							tweet_texts.append(text)
+				except Exception:
+					continue
+
+			return " ".join(tweet_texts).strip()
+		except Exception as e:
+			logging.warning("Failed to extract tweet text.", exc_info=e)
+
+		return ""
+
 	def extract_image_urls_from_x_post(self, url: str, timeout: int = 60) -> tuple[list[str], str]:
 		img_urls, post_text = [], ''
 
@@ -224,9 +251,7 @@ class XScraper(XAbstract):
 
 					#page.wait_for_timeout(3000)
 					page.wait_for_selector("img[src*='pbs.twimg.com/media']", timeout=(timeout*1000))
-					text_element = page.query_selector('[data-testid="tweetText"]')
-					if text_element:
-						post_text = str(text_element.inner_text())
+					post_text = self.extract_post_text(page)
 
 					image_elements = page.query_selector_all("img")
 					image_urls = []

{warp_beacon-2.8.5 → warp_beacon-2.8.7}/warp_beacon/telegram/bot.py

@@ -371,15 +371,25 @@ class Bot(object):
 								height=j.media_info["height"],
 								duration=round(j.media_info["duration"]),
 								thumb=j.media_info["thumb"],
-								caption=self.build_signature_caption(job)
+								caption=self.build_signature_caption(j)
 							)
 							tg_chunk.append(vid)
 						elif j.media_type == JobType.IMAGE:
 							photo = InputMediaPhoto(
 								media=j.local_media_path,
-								caption=self.build_signature_caption(job)
+								caption=self.build_signature_caption(j)
 							)
 							tg_chunk.append(photo)
+						elif j.media_type == JobType.ANIMATION:
+							anim = InputMediaAnimation(
+								media=j.local_media_path,
+								thumb=j.media_info["thumb"],
+								caption=self.build_signature_caption(j),
+								width=j.media_info["width"],
+								height=j.media_info["height"],
+								duration=round(j.media_info["duration"])
+							)
+							tg_chunk.append(anim)
 					mediafs.append(tg_chunk)
 				args["media"] = mediafs
 

{warp_beacon-2.8.5 → warp_beacon-2.8.7/warp_beacon.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: warp_beacon
-Version: 2.8.5
+Version: 2.8.7
 Summary: Telegram bot for expanding external media links
 Home-page: https://github.com/sb0y/warp_beacon
 Author: Andrey Bagrintsev

warp_beacon-2.8.5/warp_beacon/__version__.py

@@ -1,2 +0,0 @@
-__version__ = "2.8.5"
-