Merge branch 'georgekoltsov/update-import-archive-file-validator' into 'master'

Update Import archive size validator

See merge request gitlab-org/gitlab!52893

lib/gitlab/import_export/decompressed_archive_size_validator.rb

 # frozen_string_literal: true
 
-require 'zlib'
-
 module Gitlab
   module ImportExport
     class DecompressedArchiveSizeValidator
       include Gitlab::Utils::StrongMemoize
 
       DEFAULT_MAX_BYTES = 10.gigabytes.freeze
-      CHUNK_SIZE = 4096.freeze
-
-      attr_reader :error
+      TIMEOUT_LIMIT = 60.seconds
 
       def initialize(archive_path:, max_bytes: self.class.max_bytes)
         @archive_path = archive_path
         @max_bytes = max_bytes
-        @bytes_read = 0
-        @total_reads = 0
-        @denominator = 5
-        @error = nil
       end
 
       def valid?
@@ -31,59 +23,62 @@ module Gitlab
         DEFAULT_MAX_BYTES
       end
 
-      def archive_file
-        @archive_file ||= File.open(@archive_path)
-      end
-
       private
 
       def validate
-        until archive_file.eof?
-          compressed_chunk = archive_file.read(CHUNK_SIZE)
+        pgrp = nil
+        valid_archive = true
 
-          inflate_stream.inflate(compressed_chunk) do |chunk|
-            @bytes_read += chunk.size
-            @total_reads += 1
-          end
+        Timeout.timeout(TIMEOUT_LIMIT) do
+          stdin, stdout, stderr, wait_thr = Open3.popen3(command, pgroup: true)
+          stdin.close
+          pgrp = Process.getpgid(wait_thr[:pid])
+          status = wait_thr.value
 
-          # Start garbage collection every 5 reads in order
-          # to prevent memory bloat during archive decompression
-          GC.start if gc_start?
+          if status.success?
+            result = stdout.readline
 
-          if @bytes_read > @max_bytes
-            @error = error_message
+            if result.to_i > @max_bytes
+              valid_archive = false
 
-            return false
+              log_error('Decompressed archive size limit reached')
             end
+          else
+            valid_archive = false
+
+            log_error(stderr.readline)
           end
 
-        true
-      rescue => e
-        @error = error_message
+        ensure
+          stdout.close
+          stderr.close
+        end
 
-        Gitlab::ErrorTracking.track_exception(e)
+        valid_archive
+      rescue Timeout::Error
+        log_error('Timeout reached during archive decompression')
 
-        Gitlab::Import::Logger.info(
-          message: @error,
-          error: e.message
-        )
+        Process.kill(-1, pgrp) if pgrp
 
         false
-      ensure
-        inflate_stream.close
-        archive_file.close
-      end
+      rescue => e
+        log_error(e.message)
 
-      def inflate_stream
-        @inflate_stream ||= Zlib::Inflate.new(Zlib::MAX_WBITS + 32)
+        Process.kill(-1, pgrp) if pgrp
+
+        false
       end
 
-      def gc_start?
-        @total_reads % @denominator == 0
+      def command
+        "gzip -dc #{@archive_path} | wc -c"
       end
 
-      def error_message
-        _('Decompressed archive size validation failed.')
+      def log_error(error)
+        Gitlab::Import::Logger.info(
+          message: error,
+          import_upload_archive_path: @archive_path,
+          import_upload_archive_size: File.size(@archive_path)
+        )
       end
     end
   end

lib/gitlab/import_export/file_importer.rb

@@ -87,7 +87,7 @@ module Gitlab
       end
 
       def validate_decompressed_archive_size
-        raise ImporterError.new(size_validator.error) unless size_validator.valid?
+        raise ImporterError.new(_('Decompressed archive size validation failed.')) unless size_validator.valid?
       end
 
       def size_validator

spec/lib/gitlab/import_export/decompressed_archive_size_validator_spec.rb

@@ -27,25 +27,55 @@ RSpec.describe Gitlab::ImportExport::DecompressedArchiveSizeValidator do
     end
 
     context 'when file exceeds allowed decompressed size' do
-      it 'returns false' do
+      it 'logs error message returns false' do
+        expect(Gitlab::Import::Logger)
+          .to receive(:info)
+          .with(
+            import_upload_archive_path: filepath,
+            import_upload_archive_size: File.size(filepath),
+            message: 'Decompressed archive size limit reached'
+          )
         expect(subject.valid?).to eq(false)
       end
     end
 
-    context 'when something goes wrong during decompression' do
+    context 'when exception occurs during decompression' do
+      shared_examples 'logs raised exception and terminates validator process group' do
+        let(:std) { double(:std, close: nil, value: nil) }
+        let(:wait_thr) { double }
+
         before do
-        allow(subject.archive_file).to receive(:eof?).and_raise(StandardError)
+          allow(Process).to receive(:getpgid).and_return(2)
+          allow(Open3).to receive(:popen3).and_return([std, std, std, wait_thr])
+          allow(wait_thr).to receive(:[]).with(:pid).and_return(1)
+          allow(wait_thr).to receive(:value).and_raise(exception)
         end
 
-      it 'logs and tracks raised exception' do
-        expect(Gitlab::ErrorTracking).to receive(:track_exception).with(instance_of(StandardError))
-        expect(Gitlab::Import::Logger).to receive(:info).with(hash_including(message: 'Decompressed archive size validation failed.'))
+        it 'logs raised exception and terminates validator process group' do
+          expect(Gitlab::Import::Logger)
+            .to receive(:info)
+            .with(
+              import_upload_archive_path: filepath,
+              import_upload_archive_size: File.size(filepath),
+              message: error_message
+            )
+          expect(Process).to receive(:kill).with(-1, 2)
+          expect(subject.valid?).to eq(false)
+        end
+      end
 
-        subject.valid?
+      context 'when timeout occurs' do
+        let(:error_message) { 'Timeout reached during archive decompression' }
+        let(:exception) { Timeout::Error }
+
+        include_examples 'logs raised exception and terminates validator process group'
       end
 
-      it 'returns false' do
-        expect(subject.valid?).to eq(false)
+      context 'when exception occurs' do
+        let(:error_message) { 'Error!' }
+        let(:exception) { StandardError.new(error_message) }
+
+        include_examples 'logs raised exception and terminates validator process group'
       end
     end
   end