From b30df3f8d4d638602c10b60336dc563a095c4664 Mon Sep 17 00:00:00 2001
From: Dragory <2606411+Dragory@users.noreply.github.com>
Date: Sat, 11 Apr 2020 16:56:55 +0300
Subject: [PATCH] Automod: add include_words/exclude_words and
 include_regex/exclude_regex to match_links trigger

---
 backend/src/plugins/Automod/Automod.ts | 51 ++++++++++++++++++++++----
 backend/src/plugins/Automod/types.ts   |  4 ++
 backend/src/utils.ts                   | 15 +++++---
 3 files changed, 57 insertions(+), 13 deletions(-)

diff --git a/backend/src/plugins/Automod/Automod.ts b/backend/src/plugins/Automod/Automod.ts
index d9707af4..17b69153 100644
--- a/backend/src/plugins/Automod/Automod.ts
+++ b/backend/src/plugins/Automod/Automod.ts
@@ -458,14 +458,39 @@ export class AutomodPlugin extends ZeppelinPlugin<TConfigSchema, ICustomOverride
     for (const link of links) {
       const normalizedHostname = link.hostname.toLowerCase();
 
-      if (trigger.include_domains) {
-        for (const domain of trigger.include_domains) {
-          const normalizedDomain = domain.toLowerCase();
-          if (normalizedDomain === normalizedHostname) {
-            return domain;
+      // Exclude > Include
+      // In order of specificity, regex > word > domain
+
+      if (trigger.exclude_regex) {
+        for (const pattern of trigger.exclude_regex) {
+          if (pattern.test(link.input)) {
+            return null;
           }
-          if (trigger.include_subdomains && normalizedHostname.endsWith(`.${domain}`)) {
-            return domain;
+        }
+      }
+
+      if (trigger.include_regex) {
+        for (const pattern of trigger.include_regex) {
+          if (pattern.test(link.input)) {
+            return link.input;
+          }
+        }
+      }
+
+      if (trigger.exclude_words) {
+        for (const word of trigger.exclude_words) {
+          const regex = new RegExp(escapeStringRegexp(word), "i");
+          if (regex.test(link.input)) {
+            return null;
+          }
+        }
+      }
+
+      if (trigger.include_words) {
+        for (const word of trigger.include_words) {
+          const regex = new RegExp(escapeStringRegexp(word), "i");
+          if (regex.test(link.input)) {
+            return link.input;
           }
         }
       }
@@ -483,6 +508,18 @@ export class AutomodPlugin extends ZeppelinPlugin<TConfigSchema, ICustomOverride
 
         return link.toString();
       }
+
+      if (trigger.include_domains) {
+        for (const domain of trigger.include_domains) {
+          const normalizedDomain = domain.toLowerCase();
+          if (normalizedDomain === normalizedHostname) {
+            return domain;
+          }
+          if (trigger.include_subdomains && normalizedHostname.endsWith(`.${domain}`)) {
+            return domain;
+          }
+        }
+      }
     }
 
     return null;
diff --git a/backend/src/plugins/Automod/types.ts b/backend/src/plugins/Automod/types.ts
index e590a605..271ce33b 100644
--- a/backend/src/plugins/Automod/types.ts
+++ b/backend/src/plugins/Automod/types.ts
@@ -170,6 +170,10 @@ export const MatchLinksTrigger = t.type({
   include_domains: tNullable(t.array(t.string)),
   exclude_domains: tNullable(t.array(t.string)),
   include_subdomains: t.boolean,
+  include_words: tNullable(t.array(t.string)),
+  exclude_words: tNullable(t.array(t.string)),
+  include_regex: tNullable(t.array(TSafeRegex)),
+  exclude_regex: tNullable(t.array(TSafeRegex)),
   match_messages: t.boolean,
   match_embeds: t.boolean,
   match_visible_names: t.boolean,
diff --git a/backend/src/utils.ts b/backend/src/utils.ts
index b44032c8..c1fe584c 100644
--- a/backend/src/utils.ts
+++ b/backend/src/utils.ts
@@ -390,18 +390,21 @@ export async function findRelevantAuditLogEntry(
 const urlRegex = /(\S+\.\S+)/g;
 const protocolRegex = /^[a-z]+:\/\//;
 
-export function getUrlsInString(str: string, unique = false): url.URL[] {
+interface MatchedURL extends url.URL {
+  input: string;
+}
+
+export function getUrlsInString(str: string, unique = false): MatchedURL[] {
   let matches = str.match(urlRegex) || [];
   if (unique) matches = Array.from(new Set(matches));
 
   return matches.reduce((urls, match) => {
-    if (!protocolRegex.test(match)) {
-      match = `https://${match}`;
-    }
+    const withProtocol = protocolRegex.test(match) ? match : `https://${match}`;
 
-    let matchUrl: url.URL;
+    let matchUrl: MatchedURL;
     try {
-      matchUrl = new url.URL(match);
+      matchUrl = new url.URL(withProtocol) as MatchedURL;
+      matchUrl.input = match;
     } catch (e) {
       return urls;
     }