Add latest changes from gitlab-org/security/gitlab@15-3-stable-ee

6 files changed, 63 insertions, 25 deletions

diff --git a/lib/banzai/filter/commit_trailers_filter.rb b/lib/banzai/filter/commit_trailers_filter.rb
index a615abc1989..817bea42757 100644
--- a/lib/banzai/filter/commit_trailers_filter.rb
+++ b/lib/banzai/filter/commit_trailers_filter.rb
@@ -17,21 +17,10 @@ module Banzai
       include ActionView::Helpers::TagHelper
       include AvatarsHelper
 
-      TRAILER_REGEXP = /(?<label>[[:alpha:]-]+-by:)/i.freeze
-      AUTHOR_REGEXP = /(?<author_name>.+)/.freeze
-      # Devise.email_regexp wouldn't work here since its designed to match
-      # against strings that only contains email addresses; the \A and \z
-      # around the expression will only match if the string being matched
-      # contains just the email nothing else.
-      MAIL_REGEXP = /&lt;(?<author_email>[^@\s]+@[^@\s]+)&gt;/.freeze
-      FILTER_REGEXP = /(?<trailer>^\s*#{TRAILER_REGEXP}\s*#{AUTHOR_REGEXP}\s+#{MAIL_REGEXP}$)/mi.freeze
-
       def call
         doc.xpath('descendant-or-self::text()').each do |node|
           content = node.to_html
 
-          next unless content.match(FILTER_REGEXP)
-
           html = trailer_filter(content)
 
           next if html == content
@@ -52,11 +41,24 @@ module Banzai
       # Returns a String with all trailer lines replaced with links to GitLab
       # users and mailto links to non GitLab users. All links have `data-trailer`
       # and `data-user` attributes attached.
+      #
+      # The code intentionally avoids using Regex for security and performance
+      # reasons: https://gitlab.com/gitlab-org/gitlab/-/issues/363734
       def trailer_filter(text)
-        text.gsub(FILTER_REGEXP) do |author_match|
-          label = $~[:label]
-          "#{label} #{parse_user($~[:author_name], $~[:author_email], label)}"
-        end
+        text.lines.map! do |line|
+          trailer, rest = line.split(':', 2)
+
+          next line unless trailer.downcase.end_with?('-by') && rest.present?
+
+          chunks = rest.split
+          author_email = chunks.pop.delete_prefix('&lt;').delete_suffix('&gt;')
+          next line unless Devise.email_regexp.match(author_email)
+
+          author_name = chunks.join(' ').strip
+          trailer = "#{trailer.strip}:"
+
+          "#{trailer} #{link_to_user_or_email(author_name, author_email, trailer)}\n"
+        end.join
       end
 
       # Find a GitLab user using the supplied email and generate
@@ -67,7 +69,7 @@ module Banzai
       # trailer - String trailer used in the commit message
       #
       # Returns a String with a link to the user.
-      def parse_user(name, email, trailer)
+      def link_to_user_or_email(name, email, trailer)
         link_to_user User.find_by_any_email(email),
           name: name,
           email: email,
diff --git a/lib/banzai/filter/pathological_markdown_filter.rb b/lib/banzai/filter/pathological_markdown_filter.rb
new file mode 100644
index 00000000000..0f94150c7a1
--- /dev/null
+++ b/lib/banzai/filter/pathological_markdown_filter.rb
@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Banzai
+  module Filter
+    class PathologicalMarkdownFilter < HTML::Pipeline::TextFilter
+      # It's not necessary for this to be precise - we just need to detect
+      # when there are a non-trivial number of unclosed image links.
+      # So we don't really care about code blocks, etc.
+      # See https://gitlab.com/gitlab-org/gitlab/-/issues/370428
+      REGEX = /!\[(?:[^\]])+?!\[/.freeze
+      DETECTION_MAX = 10
+
+      def call
+        count = 0
+
+        @text.scan(REGEX) do |_match|
+          count += 1
+          break if count > DETECTION_MAX
+        end
+
+        return @text if count <= DETECTION_MAX
+
+        "_Unable to render markdown - too many unclosed markdown image links detected._"
+      end
+    end
+  end
+end
diff --git a/lib/banzai/pipeline/plain_markdown_pipeline.rb b/lib/banzai/pipeline/plain_markdown_pipeline.rb
index 1da0f72996b..fb6f6e9077d 100644
--- a/lib/banzai/pipeline/plain_markdown_pipeline.rb
+++ b/lib/banzai/pipeline/plain_markdown_pipeline.rb
@@ -5,6 +5,7 @@ module Banzai
     class PlainMarkdownPipeline < BasePipeline
       def self.filters
         FilterArray[
+          Filter::PathologicalMarkdownFilter,
           Filter::MarkdownPreEscapeFilter,
           Filter::MarkdownFilter,
           Filter::MarkdownPostEscapeFilter
diff --git a/lib/gitlab/git/rugged_impl/tree.rb b/lib/gitlab/git/rugged_impl/tree.rb
index bc0af12d7e3..66cfc02130b 100644
--- a/lib/gitlab/git/rugged_impl/tree.rb
+++ b/lib/gitlab/git/rugged_impl/tree.rb
@@ -16,9 +16,10 @@ module Gitlab
           TREE_SORT_ORDER = { tree: 0, blob: 1, commit: 2 }.freeze
 
           override :tree_entries
-          def tree_entries(repository, sha, path, recursive, pagination_params = nil)
+          def tree_entries(repository, sha, path, recursive, skip_flat_paths, pagination_params = nil)
             if use_rugged?(repository, :rugged_tree_entries)
-              entries = execute_rugged_call(:tree_entries_with_flat_path_from_rugged, repository, sha, path, recursive)
+              entries = execute_rugged_call(
+                :tree_entries_with_flat_path_from_rugged, repository, sha, path, recursive, skip_flat_paths)
 
               if pagination_params
                 paginated_response(entries, pagination_params[:limit], pagination_params[:page_token].to_s)
@@ -60,11 +61,11 @@ module Gitlab
             [result, cursor]
           end
 
-          def tree_entries_with_flat_path_from_rugged(repository, sha, path, recursive)
+          def tree_entries_with_flat_path_from_rugged(repository, sha, path, recursive, skip_flat_paths)
             tree_entries_from_rugged(repository, sha, path, recursive).tap do |entries|
               # This was an optimization to reduce N+1 queries for Gitaly
               # (https://gitlab.com/gitlab-org/gitaly/issues/530).
-              rugged_populate_flat_path(repository, sha, path, entries)
+              rugged_populate_flat_path(repository, sha, path, entries) unless skip_flat_paths
             end
           end
 
diff --git a/lib/gitlab/git/tree.rb b/lib/gitlab/git/tree.rb
index eb008507397..f0eef619e13 100644
--- a/lib/gitlab/git/tree.rb
+++ b/lib/gitlab/git/tree.rb
@@ -15,15 +15,16 @@ module Gitlab
         # Uses rugged for raw objects
         #
         # Gitaly migration: https://gitlab.com/gitlab-org/gitaly/issues/320
-        def where(repository, sha, path = nil, recursive = false, pagination_params = nil)
+        def where(repository, sha, path = nil, recursive = false, skip_flat_paths = true, pagination_params = nil)
           path = nil if path == '' || path == '/'
 
-          tree_entries(repository, sha, path, recursive, pagination_params)
+          tree_entries(repository, sha, path, recursive, skip_flat_paths, pagination_params)
         end
 
-        def tree_entries(repository, sha, path, recursive, pagination_params = nil)
+        def tree_entries(repository, sha, path, recursive, skip_flat_paths, pagination_params = nil)
           wrapped_gitaly_errors do
-            repository.gitaly_commit_client.tree_entries(repository, sha, path, recursive, pagination_params)
+            repository.gitaly_commit_client.tree_entries(
+              repository, sha, path, recursive, skip_flat_paths, pagination_params)
           end
         end
 
diff --git a/lib/gitlab/gitaly_client/commit_service.rb b/lib/gitlab/gitaly_client/commit_service.rb
index 9fb34f74c82..0f306a9825d 100644
--- a/lib/gitlab/gitaly_client/commit_service.rb
+++ b/lib/gitlab/gitaly_client/commit_service.rb
@@ -5,6 +5,8 @@ module Gitlab
     class CommitService
       include Gitlab::EncodingHelper
 
+      TREE_ENTRIES_DEFAULT_LIMIT = 100_000
+
       def initialize(repository)
         @gitaly_repo = repository.gitaly_repository
         @repository = repository
@@ -111,12 +113,16 @@ module Gitlab
         nil
       end
 
-      def tree_entries(repository, revision, path, recursive, pagination_params)
+      def tree_entries(repository, revision, path, recursive, skip_flat_paths, pagination_params)
+        pagination_params ||= {}
+        pagination_params[:limit] ||= TREE_ENTRIES_DEFAULT_LIMIT
+
         request = Gitaly::GetTreeEntriesRequest.new(
           repository: @gitaly_repo,
           revision: encode_binary(revision),
           path: path.present? ? encode_binary(path) : '.',
           recursive: recursive,
+          skip_flat_paths: skip_flat_paths,
           pagination_params: pagination_params
         )
         request.sort = Gitaly::GetTreeEntriesRequest::SortBy::TREES_FIRST if pagination_params