From 05c220619fc2e36f3731512f281d09611ad42619 Mon Sep 17 00:00:00 2001
From: rhymes <rhymes@hey.com>
Date: Fri, 16 Apr 2021 20:20:11 +0200
Subject: [PATCH] [15 min fix] Enforce an upper limit for
 articles.body_markdown (#13418)

* Enforce a length limit for articles.body_markdown

* Fix spec assertion

* Add bytesize validation to body_markdown
---
 app/models/article.rb                |  4 +++-
 app/validators/bytesize_validator.rb | 35 ++++++++++++++++++++++++++++
 spec/models/article_spec.rb          | 19 +++++++++++++--
 3 files changed, 55 insertions(+), 3 deletions(-)
 create mode 100644 app/validators/bytesize_validator.rb

diff --git a/app/models/article.rb b/app/models/article.rb
index c156e90bc..565772369 100644
--- a/app/models/article.rb
+++ b/app/models/article.rb
@@ -54,7 +54,9 @@ class Article < ApplicationRecord
            inverse_of: :commentable,
            class_name: "Comment"
 
-  validates :body_markdown, length: { minimum: 0, allow_nil: false }, uniqueness: { scope: %i[user_id title] }
+  validates :body_markdown, bytesize: { maximum: 800.kilobytes, too_long: "is too long." }
+  validates :body_markdown, length: { minimum: 0, allow_nil: false }
+  validates :body_markdown, uniqueness: { scope: %i[user_id title] }
   validates :boost_states, presence: true
   validates :cached_tag_list, length: { maximum: 126 }
   validates :canonical_url, uniqueness: { allow_nil: true }
diff --git a/app/validators/bytesize_validator.rb b/app/validators/bytesize_validator.rb
new file mode 100644
index 000000000..533ac83ff
--- /dev/null
+++ b/app/validators/bytesize_validator.rb
@@ -0,0 +1,35 @@
+# Adapted from https://github.com/rails/rails/issues/19570#issuecomment-348366536
+# and https://github.com/rails/rails/blob/v6.1.3.1/activemodel/lib/active_model/validations/length.rb
+
+# Currently only supports `:maximum`
+class BytesizeValidator < ActiveModel::EachValidator
+  MESSAGES = { maximum: :too_long }.freeze
+  CHECKS = { maximum: :<= }.freeze
+  RESERVED_OPTIONS = %i[maximum too_long].freeze
+
+  ERROR_MISSING_OPTIONS_MESSAGE = "Specify the :maximum option.".freeze
+  ERROR_NON_NEGATIVE_MESSAGE = ":maximum must be a non-negative Integer".freeze
+
+  def check_validity!
+    raise ArgumentError, ERROR_MESSAGE unless options.key?(:maximum)
+
+    maximum = options[:maximum]
+    raise ArgumentError, ERROR_NON_NEGATIVE_MESSAGE unless maximum.is_a?(Integer) && maximum >= 0
+  end
+
+  def validate_each(record, attribute, value)
+    key = :maximum
+    value_bytesize = value.respond_to?(:bytesize) ? value.bytesize : value.to_s.bytesize
+    errors_options = options.except(*RESERVED_OPTIONS)
+
+    check_value = options[key]
+    return if value_bytesize.public_send(CHECKS[key], check_value)
+
+    errors_options[:count] = check_value
+
+    default_message = options[MESSAGES[key]]
+    errors_options[:message] ||= default_message if default_message
+
+    record.errors.add(attribute, MESSAGES[key], **errors_options)
+  end
+end
diff --git a/spec/models/article_spec.rb b/spec/models/article_spec.rb
index 9f0d4e301..95f04a804 100644
--- a/spec/models/article_spec.rb
+++ b/spec/models/article_spec.rb
@@ -32,6 +32,7 @@ RSpec.describe Article, type: :model do
     it { is_expected.to have_many(:tags) }
     it { is_expected.to have_many(:user_subscriptions).dependent(:nullify) }
 
+    it { is_expected.to validate_length_of(:body_markdown).is_at_least(0) }
     it { is_expected.to validate_length_of(:cached_tag_list).is_at_most(126) }
     it { is_expected.to validate_length_of(:title).is_at_most(128) }
 
@@ -81,11 +82,25 @@ RSpec.describe Article, type: :model do
     end
 
     describe "#body_markdown" do
-      it "is unique scoped for user_id and title" do
+      it "is unique scoped for user_id and title", :aggregate_failures do
         art2 = build(:article, body_markdown: article.body_markdown, user: article.user, title: article.title)
 
         expect(art2).not_to be_valid
-        expect(art2.errors.full_messages.to_sentence).to match("markdown has already been taken")
+        expect(art2.errors_as_sentence).to match("markdown has already been taken")
+      end
+
+      # using https://unicode-table.com/en/11A15/ multibyte char
+      it "is valid if its bytesize is less than 800 kilobytes" do
+        article.body_markdown = "𑨕" * 204_800 # 4 bytes x 204800 = 800 kilobytes
+
+        expect(article).to be_valid
+      end
+
+      it "is not valid if its bytesize exceeds 800 kilobytes" do
+        article.body_markdown = "𑨕" * 204_801
+
+        expect(article).not_to be_valid
+        expect(article.errors_as_sentence).to match("too long")
       end
     end