Spaces:
Sleeping
Sleeping
Pycrolis
commited on
Commit
·
9e2fac8
1
Parent(s):
8576a1a
feat(web_page_extractor_tool): wikipedia tables are moved to the beginning of the return text
Browse files
tools/web_page_information_extractor.py
CHANGED
@@ -86,7 +86,7 @@ def _get_text_from_url(url: str) -> str:
|
|
86 |
|
87 |
text = markdown_converter.handle(html)
|
88 |
if tables_text:
|
89 |
-
text
|
90 |
|
91 |
return text
|
92 |
|
@@ -106,4 +106,4 @@ if __name__ == "__main__":
|
|
106 |
# "request": "What is the surname of the equine veterinarian mentioned"})
|
107 |
# print(result)
|
108 |
|
109 |
-
print(
|
|
|
86 |
|
87 |
text = markdown_converter.handle(html)
|
88 |
if tables_text:
|
89 |
+
text = f'Tables:\n{tables_text}\n\nContent\n{text}'
|
90 |
|
91 |
return text
|
92 |
|
|
|
106 |
# "request": "What is the surname of the equine veterinarian mentioned"})
|
107 |
# print(result)
|
108 |
|
109 |
+
print(_get_text_from_url("https://en.wikipedia.org/wiki/Malko_Competition"))
|