diff options
author | zsloan | 2020-10-29 14:35:09 -0500 |
---|---|---|
committer | zsloan | 2020-10-29 14:35:09 -0500 |
commit | 7c1c9e2a519ba662e9f293eea73eb7922b2160e4 (patch) | |
tree | f6d4db465d338c1433bbb126e911062a6c31748b /test/requests/link_checker.py | |
parent | 5a1f69aa85809768577069ae63d92c9ef6aecc02 (diff) | |
parent | 6e6911b466c2727b16a190d8b714f55d7842d7e2 (diff) | |
download | genenetwork2-7c1c9e2a519ba662e9f293eea73eb7922b2160e4.tar.gz |
Merge branch 'testing' of github.com:genenetwork/genenetwork2 into reaper_bootstrap_fix
Diffstat (limited to 'test/requests/link_checker.py')
-rw-r--r-- | test/requests/link_checker.py | 16 |
1 files changed, 6 insertions, 10 deletions
diff --git a/test/requests/link_checker.py b/test/requests/link_checker.py index d040ba54..6ac26ba7 100644 --- a/test/requests/link_checker.py +++ b/test/requests/link_checker.py @@ -1,4 +1,3 @@ -from __future__ import print_function import re import requests from lxml.html import parse @@ -23,12 +22,9 @@ def is_in_page_link(link): return pattern.match(link) def get_links(doc): - return filter( - lambda x: not ( + return [x for x in [y.get("href") for y in doc.cssselect("a")] if not ( is_root_link(x) - or is_mailto_link(x)) - , map(lambda y: y.get("href") - , doc.cssselect("a"))) + or is_mailto_link(x))] def verify_link(link): if link[0] == "#": @@ -58,7 +54,7 @@ def verify_static_file(link): try: result = requests.get(link, timeout=20, verify=False) if (result.status_code == 200 and - result.content.find("Error: 404 Not Found") <= 0): + result.content.find(bytes("Error: 404 Not Found", "utf-8")) <= 0): print(link+" ==> OK") else: print("ERROR: link {}".format(link)) @@ -72,9 +68,9 @@ def check_page(host, start_url): print("Checking links host "+host+" in page `"+start_url+"`") doc = parse(start_url).getroot() links = get_links(doc) - in_page_links = filter(is_in_page_link, links) - internal_links = filter(is_internal_link, links) - external_links = filter(lambda x: not (is_internal_link(x) or is_in_page_link(x)), links) + in_page_links = list(filter(is_in_page_link, links)) + internal_links = list(filter(is_internal_link, links)) + external_links = [x for x in links if not (is_internal_link(x) or is_in_page_link(x))] for link in internal_links: verify_link(host+link) |