From d83fe3653a6e4cd00a99b009f882ee9cbbbab0e9 Mon Sep 17 00:00:00 2001
From: WeebDataHoarder <weebdatahoarder@noreply.gammaspectra.live>
Date: Wed, 23 Apr 2025 07:25:06 +0200
Subject: [PATCH] examples: update bot matches, allow badges to be fetched

---
 examples/forgejo.yml | 6 ++----
 examples/generic.yml | 2 +-
 2 files changed, 3 insertions(+), 5 deletions(-)

diff --git a/examples/forgejo.yml b/examples/forgejo.yml
index 0c78257..eb2d46b 100644
--- a/examples/forgejo.yml
+++ b/examples/forgejo.yml
@@ -193,7 +193,7 @@ conditions:
     - 'path.matches("^/[^/]+/[^/]+/(git-upload-pack|git-receive-pack|HEAD|info/refs|info/lfs|objects)")'
 
   is-generic-robot-ua:
-    - 'userAgent.contains("compatible;") && !userAgent.contains("Trident/")'
+    - 'userAgent.matches("compatible[;)]") && !userAgent.contains("Trident/")'
     - 'userAgent.matches("\\+https?://")'
     - 'userAgent.contains("@")'
     - 'userAgent.matches("[bB]ot/[0-9]")'
@@ -383,7 +383,7 @@ rules:
       # generic /*/*/ match gave too many options for scrapers to trigger random endpoints
       # this is a negative match of endpoints that Forgejo holds as reserved as users or orgs
       # see https://codeberg.org/forgejo/forgejo/src/branch/forgejo/models/user/user.go#L582
-      - '(path.matches("^/[^/]+/[^/]+/?$") || path.matches("^/[^/]+/[^/]+/(issues|pulls)/[0-9]+$") || (path.matches("^/[^/]+/?$") && size(query) == 0)) && !path.matches("(?i)^/(api|metrics|v2|assets|attachments|avatar|avatars|repo-avatars|captcha|login|org|repo|user|admin|devtest|explore|issues|pulls|milestones|notifications|ghost)(/|$)")'
+      - '(path.matches("^/[^/]+/[^/]+/?$") || path.matches("^/[^/]+/[^/]+/badges/") || path.matches("^/[^/]+/[^/]+/(issues|pulls)/[0-9]+$") || (path.matches("^/[^/]+/?$") && size(query) == 0)) && !path.matches("(?i)^/(api|metrics|v2|assets|attachments|avatar|avatars|repo-avatars|captcha|login|org|repo|user|admin|devtest|explore|issues|pulls|milestones|notifications|ghost)(/|$)")'
     action: pass
 
   - name: desired-crawlers
@@ -409,8 +409,6 @@ rules:
         action: check
         settings:
           challenges: [ self-resource-load, js-pow-sha256, http-cookie-check ]
-    settings:
-      challenges: [self-preload-link, self-header-refresh, js-pow-sha256, http-cookie-check]
 
   - name: standard-bots
     action: check
diff --git a/examples/generic.yml b/examples/generic.yml
index 5523bab..cf3a5c0 100644
--- a/examples/generic.yml
+++ b/examples/generic.yml
@@ -112,7 +112,7 @@ conditions:
 
 
   is-generic-robot-ua:
-    - 'userAgent.contains("compatible;") && !userAgent.contains("Trident/")'
+    - 'userAgent.matches("compatible[;)]") && !userAgent.contains("Trident/")'
     - 'userAgent.matches("\\+https?://")'
     - 'userAgent.contains("@")'
     - 'userAgent.matches("[bB]ot/[0-9]")'