backup/pywb
1
0
Fork 0
mirror of https://github.com/webrecorder/pywb.git synced 2025-03-24 06:59:52 +01:00
Code Issues Projects Releases Activity

Updated RewriteInfo._resolve_text_type to recognize the fr_ rewrite modifier (indicates that the content is from a frameset's frame) (#438)

Browse Source 
Added a test, test_rewrite_frameset_frame_content, to test_content_rewriter.py for these changes
This commit is contained in:
John Berlin 2019-02-05 18:11:21 -05:00 committed by Ilya Kreymer
parent 529a587cdc
commit 777cc30e82
2 changed files with 19 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
pywb/rewrite/content_rewriter.py
View File

@ -434,7 +434,7 @@ class RewriteInfo(object):
# if html or no-content type, allow resolving on js, css, # if html or no-content type, allow resolving on js, css,
# or other templates # or other templates
if text_type in ('guess-text', 'guess-html'): if text_type in ('guess-text', 'guess-html'):
if not is_js_or_css and mod not in ('if_', 'mp_', 'bn_', ''): if not is_js_or_css and mod not in ('fr_', 'if_', 'mp_', 'bn_', ''):
return None return None
# if application/octet-stream binary, only resolve if in js/css content # if application/octet-stream binary, only resolve if in js/css content

18
pywb/rewrite/test/test_content_rewriter.py
View File

@ -524,6 +524,24 @@ class TestContentRewriter(object):
assert b''.join(gen).decode('utf-8') == '{"ssid":"5678"}' assert b''.join(gen).decode('utf-8') == '{"ssid":"5678"}'
def test_rewrite_frameset_frame_content(self):
"""Determines if the content rewriter correctly determines that HTML loaded via a frameset's frame,
frame's src url is rewritten with the **fr_** rewrite modifier, is content to be rewritten
"""
headers = {'Content-Type': 'text/html; charset=UTF-8'}
prefix = 'http://localhost:8080/live/'
dt = '20190205180554%s'
content = '<!DOCTYPE html><head><link rel="icon" href="http://r-u-ins.tumblr.com/img/favicon/72.png" ' \
'type="image/x-icon"></head>'
rw_headers, gen, is_rw = self.rewrite_record(headers, content, ts=dt % 'fr_',
prefix=prefix,
url='http://r-u-ins.tumblr.com/',
is_live='1')
# is_rw should be true indicating the content was rewritten
assert is_rw
assert b''.join(gen).decode('utf-8') == content.replace('href="', 'href="%s%s' % (prefix, dt % 'oe_/'))
assert rw_headers.headers == [('Content-Type', 'text/html; charset=UTF-8')]
def test_custom_live_only(self): def test_custom_live_only(self):
headers = {'Content-Type': 'application/json'} headers = {'Content-Type': 'application/json'}
content = '{"foo":"bar", "dash": {"on": "true"}, "some": ["list"]' content = '{"foo":"bar", "dash": {"on": "true"}, "some": ["list"]'