kennethreitz/requests-html
1
0
Fork 0
mirror of https://github.com/kennethreitz/requests-html.git synced 2026-06-05 23:00:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix things

Browse Source 
Signed-off-by: Kenneth Reitz <me@kennethreitz.org>
This commit is contained in:
Kenneth Reitz
2018-02-27 19:32:53 -05:00
parent 6340ad919c
commit fcfdaf2503
2 changed files with 18 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
requests_html.py
+7 -1
View File
@@ -214,8 +214,14 @@ class HTML(BaseParser):
"""An HTML document, ready for parsing."""
def __init__(self, *, url=DEFAULT_URL, html, default_encoding=DEFAULT_ENCODING) -> None:
# Convert incoming unicode HTML into bytes.
if isinstance(html, str):
html = html.encode(DEFAULT_ENCODING)
super(HTML, self).__init__(
element=PyQuery(html)('html'),
# Convert unicode HTML to bytes.
element=PyQuery(html)('html') or PyQuery('<html>{}</html>'.format(html))('html'),
html=html,
url=url,
default_encoding=default_encoding
tests/test_requests_html.py
+11 -3
View File
@@ -1,6 +1,6 @@
import os
from requests_html import HTMLSession
from requests_html import HTMLSession, HTML
from requests_file import FileAdapter
session = HTMLSession()
@@ -59,6 +59,14 @@ def test_xpath():
html = r.html.xpath('/html', first=True)
assert 'no-js' in html.attrs['class']
if __name__ == '__main__':
test_css_selector()
def test_html_loading():
doc = """<a href='https://httpbin.org'>"""
html = HTML(html=doc)
assert 'https://httpbin.org' in html.links
if __name__ == '__main__':
test_html_loading()