mirror of
https://gitlab.archlinux.org/archlinux/infrastructure.git
synced 2026-09-19 16:19:09 +02:00
Christian Heusel
588f3fa4d1
For some reason it seems like the crawlers really like to use the `returnto=` parameter, which is why we should penalize this more with anubis. Signed-off-by: Christian Heusel <christian@heusel.eu>
54 lines
2.3 KiB
YAML
54 lines
2.3 KiB
YAML
- name: archwiki
|
|
action: ALLOW
|
|
expression:
|
|
any:
|
|
- path == "/load.php"
|
|
- path.startsWith("/skins/") || path.startsWith("/resources/")
|
|
- query.size() == 0
|
|
- '"Cookie" in headers && headers["Cookie"].matches("(^| )(archwiki_session|archwikiToken)=")'
|
|
- path == "/api.php" && "action" in query && query["action"] == "opensearch"
|
|
- path.matches("^/rest\\.php/v1/search(/title|$)")
|
|
|
|
# Inspiration taken from Extension:CrawlerProtection
|
|
- name: archwiki-expensive
|
|
action: WEIGH
|
|
weight:
|
|
adjust: 10
|
|
expression:
|
|
any:
|
|
- path.startsWith("/title/Special:Contributions/")
|
|
- path == "/index.php" && "title" in query && query["title"].startsWith("Special:Contributions")
|
|
- path.startsWith("/title/Special:Diff/")
|
|
- path == "/index.php" && "title" in query && query["title"].startsWith("Special:Diff")
|
|
- path.startsWith("/title/Special:ListUsers/")
|
|
- path == "/index.php" && "title" in query && query["title"].startsWith("Special:ListUsers")
|
|
- path.startsWith("/title/Special:Log/")
|
|
- path == "/index.php" && "title" in query && query["title"].startsWith("Special:Log")
|
|
- path.startsWith("/title/Special:RecentChanges/")
|
|
- path == "/index.php" && "title" in query && query["title"].startsWith("Special:RecentChanges")
|
|
- path.startsWith("/title/Special:RecentChangesLinked/")
|
|
- path == "/index.php" && "title" in query && query["title"].startsWith("Special:RecentChangesLinked")
|
|
- path.startsWith("/title/Special:Search/")
|
|
- path == "/index.php" && "title" in query && query["title"].startsWith("Special:Search")
|
|
- path.startsWith("/title/Special:WhatLinksHere/")
|
|
- path == "/index.php" && "title" in query && query["title"].startsWith("Special:WhatLinksHere")
|
|
- path == "/index.php" && "oldid" in query
|
|
- path == "/index.php" && "diff" in query
|
|
|
|
# This should only affect "real" users on login
|
|
- name: archwiki-crawler-suspicious
|
|
action: WEIGH
|
|
weight:
|
|
adjust: 20
|
|
expression:
|
|
any:
|
|
- path == "/index.php" && "returnto" in query
|
|
|
|
# The crawlers use very innocent user agents, the challenge for WEIGH=10 should be fast for regular users
|
|
- name: archwiki-suspicios-user-agents
|
|
user_agent_regex: >-
|
|
Windows NT|Macintosh
|
|
action: WEIGH
|
|
weight:
|
|
adjust: 10
|