fix: retain line breaks

This commit is contained in:
Ernest Poletaev
2024-10-25 22:33:24 +07:00
parent 96e52767ad
commit 1f8791953e

View File

@@ -69,5 +69,6 @@ class ScrapeWebsiteTool(BaseTool):
parsed = BeautifulSoup(page.text, "html.parser")
text = parsed.get_text(" ")
text = re.sub('\s+', ' ', text)
text = re.sub('[ \t]+', ' ', text)
text = re.sub('\\s+\n\\s+', '\n', text)
return text