From 6dab0a4ebd9ef6538ca0a256f5eb858eb0a5008d Mon Sep 17 00:00:00 2001
From: "Sotnikov, Dmitri" <dmitri.sotnikov@gmail.com>
Date: Mon, 23 Apr 2018 17:47:41 -0400
Subject: [PATCH] added support for filtering out content

---
 README.md         | 5 ++++-
 mastodon-bot.cljs | 9 ++++++++-
 2 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 85b45b5..4311799 100644
--- a/README.md
+++ b/README.md
@@ -37,7 +37,10 @@ the bot will post the timeline from the specified Twitter/Tumblr accounts and RS
             :api_url "https://botsin.space/api/v1/"
             :max-post-length 300
             ;; optional signature for posts
-            :signature "#newsbot"}}
+            :signature "#newsbot"
+            ;; optional content filter regexes
+            ;; any posts matching the regexes will be filtered out
+            :content-filters [".*bannedsite.*"]}}
 ```
 
 * the bot looks for `config.edn` at its relative path by default, an alternative location can be specified either using the `MASTODON_BOT_CONFIG` environment variable or passing the path to config as an argument
diff --git a/mastodon-bot.cljs b/mastodon-bot.cljs
index 790e750..5afdd54 100755
--- a/mastodon-bot.cljs
+++ b/mastodon-bot.cljs
@@ -20,6 +20,11 @@
 
 (def config (-> (find-config) fs/readFileSync str edn/read-string))
 
+(def content-filter-regexes (mapv re-pattern (-> config :mastodon :content-filters)))
+
+(defn blocked-content? [text]
+  (boolean (some #(re-matches % text) content-filter-regexes)))
+
 (def max-post-length (or (-> config :mastodon :max-post-length) 300))
 
 (def mastodon-client (or (some-> config :mastodon clj->js mastodon.)
@@ -72,7 +77,9 @@
   (.then (.get mastodon-client "timelines/home" #js {}) #(-> % .-data js->edn callback)))
 
 (defn post-items [last-post-time items]
-  (doseq [{:keys [text media-links]} (filter #(> (:created-at %) last-post-time) items)]
+  (doseq [{:keys [text media-links]} (->> items
+                                          (remove #(blocked-content? (:text %)))
+                                          (filter #(> (:created-at %) last-post-time)))]
     (if media-links
       (post-status-with-images text media-links)
       (post-status text))))