feat: sanity check video IDs with regex (#6037)

* feat: sanity check video IDs with regex

* Return 400 instead

* fix wrong function

* add coderabbit suggestion about PCRE2 regex and length of the string with a new line

* Use validate_video_id on more places

* Add video id validation to /watch route.

src/invidious/exceptions.cr

@@ -38,3 +38,15 @@ end
 # some important informations, and that the query should be sent again.
 class RetryOnceException < Exception
 end
+
+# Exception for invalid video IDs.
+class InvalidVideoID < InfoException
+  getter id : String?
+
+  def initialize(@id)
+  end
+
+  def message
+    return "Invalid video ID '#{id}'"
+  end
+end

src/invidious/helpers/utils.cr

@@ -404,3 +404,10 @@ def invidious_companion_encrypt(data)
   encrypted_data = encrypt_ecb_without_salt("#{timestamp}|#{data}", CONFIG.invidious_companion_key)
   return Base64.urlsafe_encode(encrypted_data)
 end
+
+def validate_video_id(id : String) : Bool
+  # This is the video ID regex. May be need to be changed
+  # if Youtube ever decides to add more characters to their
+  # video IDs.
+  /\A[a-zA-Z0-9_-]{11}\z/.matches?(id)
+end

src/invidious/routes/api/v1/authenticated.cr

@@ -81,9 +81,10 @@ module Invidious::Routes::API::V1::Authenticated
       return error_json(409, "Watch history is disabled in preferences.")
     end
 
+    # Sanity checks
     id = env.params.url["id"]
-    if !id.match(/^[a-zA-Z0-9_-]{11}$/)
-      return error_json(400, "Invalid video id.")
+    unless validate_video_id(id)
+      return error_json(400, InvalidVideoID.new(id))
     end
 
     Invidious::Database::Users.mark_watched(user, id)
@@ -97,12 +98,12 @@ module Invidious::Routes::API::V1::Authenticated
       return error_json(409, "Watch history is disabled in preferences.")
     end
 
-    id = env.params.url["id"]
-    if !id.match(/^[a-zA-Z0-9_-]{11}$/)
-      return error_json(400, "Invalid video id.")
+    video_id = env.params.url["id"]
+    unless video_id && validate_video_id(video_id)
+      return error_json(400, InvalidVideoID.new(video_id))
     end
 
-    Invidious::Database::Users.mark_unwatched(user, id)
+    Invidious::Database::Users.mark_unwatched(user, video_id)
     env.response.status_code = 204
   end
 
@@ -309,8 +310,9 @@ module Invidious::Routes::API::V1::Authenticated
     end
 
     video_id = env.params.json["videoId"].try &.as(String)
-    if !video_id
-      return error_json(403, "Invalid videoId")
+    # Sanity checks
+    unless video_id && validate_video_id(video_id)
+      return error_json(400, InvalidVideoID.new(video_id))
     end
 
     begin

src/invidious/routes/api/v1/videos.cr

@@ -32,8 +32,9 @@ module Invidious::Routes::API::V1::Videos
     id = env.params.url["id"]
     region = env.params.query["region"]? || env.params.body["region"]?
 
-    if id.nil? || id.size != 11 || !id.matches?(/^[\w-]+$/)
-      return error_json(400, "Invalid video ID")
+    # Sanity checks
+    unless validate_video_id(id)
+      return error_json(400, InvalidVideoID.new(id))
     end
 
     # See https://github.com/ytdl-org/youtube-dl/blob/6ab30ff50bf6bd0585927cb73c7421bef184f87a/youtube_dl/extractor/youtube.py#L1354
@@ -256,11 +257,11 @@ module Invidious::Routes::API::V1::Videos
   def self.annotations(env)
     env.response.content_type = "text/xml"
 
-    id = env.params.url["id"]
+    video_id = env.params.url["id"]
     source = env.params.query["source"]?
     source ||= "archive"
 
-    if !id.match(/[a-zA-Z0-9_-]{11}/)
+    unless video_id && validate_video_id(video_id)
       haltf env, 400
     end
 
@@ -268,21 +269,21 @@ module Invidious::Routes::API::V1::Videos
 
     case source
     when "archive"
-      if CONFIG.cache_annotations && (cached_annotation = Invidious::Database::Annotations.select(id))
+      if CONFIG.cache_annotations && (cached_annotation = Invidious::Database::Annotations.select(video_id))
         annotations = cached_annotation.annotations
       else
-        index = CHARS_SAFE.index!(id[0]).to_s.rjust(2, '0')
+        index = CHARS_SAFE.index!(video_id[0]).to_s.rjust(2, '0')
 
         # IA doesn't handle leading hyphens,
         # so we use https://archive.org/details/youtubeannotations_64
         if index == "62"
           index = "64"
-          id = id.sub(/^-/, 'A')
+          video_id = video_id.sub(/^-/, 'A')
         end
 
-        file = URI.encode_www_form("#{id[0, 3]}/#{id}.xml")
+        file = URI.encode_www_form("#{video_id[0, 3]}/#{video_id}.xml")
 
-        location = make_client(INTERNET_ARCHIVE_URL, &.get("/download/youtubeannotations_#{index}/#{id[0, 2]}.tar/#{file}"))
+        location = make_client(INTERNET_ARCHIVE_URL, &.get("/download/youtubeannotations_#{index}/#{video_id[0, 2]}.tar/#{file}"))
 
         if !location.headers["Location"]?
           env.response.status_code = location.status_code
@@ -300,10 +301,10 @@ module Invidious::Routes::API::V1::Videos
 
         annotations = response.body
 
-        Helpers.cache_annotation(id, annotations)
+        Helpers.cache_annotation(video_id, annotations)
       end
     else # "youtube"
-      response = YT_POOL.client &.get("/annotations_invideo?video_id=#{id}")
+      response = YT_POOL.client &.get("/annotations_invideo?video_id=#{video_id}")
 
       if response.status_code != 200
         haltf env, response.status_code

src/invidious/routes/video_playback.cr

@@ -268,8 +268,8 @@ module Invidious::Routes::VideoPlayback
     itag = env.params.query["itag"]?.try &.to_i?
 
     # Sanity checks
-    if id.nil? || id.size != 11 || !id.matches?(/^[\w-]+$/)
-      return error_template(400, "Invalid video ID")
+    unless id && validate_video_id(id)
+      return error_template(400, InvalidVideoID.new(id))
     end
 
     if !itag.nil? && (itag <= 0 || itag >= 1000)

src/invidious/routes/watch.cr

@@ -18,14 +18,8 @@ module Invidious::Routes::Watch
         return error_template(400, "Invalid parameters.")
       end
 
-      if id.size > 11
-        url = "/watch?v=#{id[0, 11]}"
-        env.params.query.delete_all("v")
-        if env.params.query.size > 0
-          url += "&#{env.params.query}"
-        end
-
-        return env.redirect url
+      unless validate_video_id(id)
+        return error_template(400, InvalidVideoID.new(id))
       end
     else
       return env.redirect "/"
@@ -235,7 +229,7 @@ module Invidious::Routes::Watch
     token = env.params.body["csrf_token"]?
 
     id = env.params.query["id"]?
-    if !id
+    unless id && validate_video_id(id)
       env.response.status_code = 400
       return
     end