Cleanup page decode
Signed-off-by: Nathaniel Clark <Nathaniel.Clark@misrule.us>
This commit is contained in:
parent
76b4675d1b
commit
09b145cb85
|
@ -233,15 +233,8 @@ def process_url(url):
|
||||||
file=sys.stderr,
|
file=sys.stderr,
|
||||||
)
|
)
|
||||||
return []
|
return []
|
||||||
data = rxscript.sub(
|
data = rxscript.sub("", rxcomment.sub("", rxblank.sub(" ", resp.read().decode())))
|
||||||
"",
|
|
||||||
rxcomment.sub(
|
|
||||||
"",
|
|
||||||
rxblank.sub(
|
|
||||||
" ", "".join(map(lambda x: x.decode("utf-8"), resp.readlines()))
|
|
||||||
),
|
|
||||||
),
|
|
||||||
)
|
|
||||||
dom = html.fromstring(data)
|
dom = html.fromstring(data)
|
||||||
|
|
||||||
return dom.xpath('//table[contains(@class, "simpleTable")]')
|
return dom.xpath('//table[contains(@class, "simpleTable")]')
|
||||||
|
|
Loading…
Reference in New Issue