warp-beacon 2.8.4__py3-none-any.whl → 2.8.6__py3-none-any.whl

warp_beacon/__version__.py

@@ -1,2 +1,2 @@
-__version__ = "2.8.4"
+__version__ = "2.8.6"
 

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
@@ -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:
-						text = text_element.inner_text()
+					post_text = self.extract_post_text(page)
 
 					image_elements = page.query_selector_all("img")
 					image_urls = []
@@ -237,5 +262,4 @@ class XScraper(XAbstract):
 							image_urls.append(src)
 					
 					img_urls = list(set(image_urls))
-					post_text = str(text)
 		return img_urls, post_text

{warp_beacon-2.8.4.dist-info → warp_beacon-2.8.6.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: warp_beacon
-Version: 2.8.4
+Version: 2.8.6
 Summary: Telegram bot for expanding external media links
 Home-page: https://github.com/sb0y/warp_beacon
 Author: Andrey Bagrintsev

{warp_beacon-2.8.4.dist-info → warp_beacon-2.8.6.dist-info}/RECORD

@@ -4,7 +4,7 @@ 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=DLLNgrgTvXrqxrEVoOSS4LUYpuTs6_9DUJqLwjFw3Dg,23
+warp_beacon/__version__.py,sha256=Hm_e9QMByvoNNxgb1GLg8_JIDnCEz-Y05F0A2SJvVNM,23
 warp_beacon/warp_beacon.py,sha256=ADCR30uGXIsDrt9WoiI9Ghu2QtWs0qZIK6x3pQKM_B4,1109
 warp_beacon/yt_auth.py,sha256=GUTKqYr_tzDC-07Lx_ahWXSag8EyLxXBUnQbDBIkEmk,6022
 warp_beacon/compress/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -29,7 +29,7 @@ warp_beacon/scraper/exceptions.py,sha256=EKwoF0oH2xZWbNU-v8DOaWK5skKwa3s1yTIBdlc
 warp_beacon/scraper/fail_handler.py,sha256=5ODu4b8ndZWAcHIXrcUufsWFihetzNUoAi8IgAkreyQ,998
 warp_beacon/scraper/link_resolver.py,sha256=Rc9ZuMyOo3iPywDHwjngy-WRQ2SXhJwxcg-5ripx7tM,2447
 warp_beacon/scraper/utils.py,sha256=AOZmDki2Pbr84IG-j_wN2UghKCiWFVDYdx6HJl0JTBs,1258
-warp_beacon/scraper/X/X.py,sha256=UasMEYGsbNddKbvrPHUVq_hZ6tuYIskVZCUdvvgrhlM,7284
+warp_beacon/scraper/X/X.py,sha256=N9-nbqBv2aNNEUd-QtZa-sBsCog-wek85Zdw1i0Q_lM,7907
 warp_beacon/scraper/X/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 warp_beacon/scraper/X/abstract.py,sha256=pCzZPTCtn8pRbBx2SeuBUpMkEHqnOLtwLBAHYceL12Q,5475
 warp_beacon/scraper/X/types.py,sha256=i36Nu2cHpHCkvoeobBQC3B13Ke_N40tgCCApcm_FBFY,76
@@ -56,9 +56,9 @@ warp_beacon/telegram/progress_file_reader.py,sha256=e3equyNKlKs764AD-iE9QRsh3YDH
 warp_beacon/telegram/types.py,sha256=Kvdng6uCF1HRoqQgGW1ZYYPJoVuYkFb-LDvMBbW5Hjk,89
 warp_beacon/telegram/utils.py,sha256=zTF8VQfAWetBSjAPbmNe_Zi_LN5fAcWptJKjLaFNHaE,5073
 warp_beacon/uploader/__init__.py,sha256=1enK6qMWaTZEaK456JwaKOfvCvznHA8cjgceOsrF6Po,5732
-warp_beacon-2.8.4.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-warp_beacon-2.8.4.dist-info/METADATA,sha256=sKXHXCqoOzXc4nYB5n9jgkOSESkgVgsTxz4ie7b0jcU,23214
-warp_beacon-2.8.4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-warp_beacon-2.8.4.dist-info/entry_points.txt,sha256=eSB61Rb89d56WY0O-vEIQwkn18J-4CMrJcLA_R_8h3g,119
-warp_beacon-2.8.4.dist-info/top_level.txt,sha256=RraB0PWGvRK2zPYkuICKNgStLG1C5s7rPHHJEHJbkgA,1510
-warp_beacon-2.8.4.dist-info/RECORD,,
+warp_beacon-2.8.6.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+warp_beacon-2.8.6.dist-info/METADATA,sha256=8MtpB3yYJK75N8q1spUukoa6qbeBImLIPhDsxalL8YQ,23214
+warp_beacon-2.8.6.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+warp_beacon-2.8.6.dist-info/entry_points.txt,sha256=eSB61Rb89d56WY0O-vEIQwkn18J-4CMrJcLA_R_8h3g,119
+warp_beacon-2.8.6.dist-info/top_level.txt,sha256=RraB0PWGvRK2zPYkuICKNgStLG1C5s7rPHHJEHJbkgA,1510
+warp_beacon-2.8.6.dist-info/RECORD,,