debian-mirror-gitlab/app/services/bulk_imports/file_download_service.rb

157 lines
4.6 KiB
Ruby
Raw Normal View History

2021-09-04 01:27:46 +05:30
# frozen_string_literal: true
2021-09-30 23:02:18 +05:30
# Downloads a remote file. If no filename is given, it'll use the remote filename
2021-09-04 01:27:46 +05:30
module BulkImports
class FileDownloadService
ServiceError = Class.new(StandardError)
2021-09-30 23:02:18 +05:30
REMOTE_FILENAME_PATTERN = %r{filename="(?<filename>[^"]+)"}.freeze
FILENAME_SIZE_LIMIT = 255 # chars before the extension
2021-12-11 22:18:48 +05:30
DEFAULT_FILE_SIZE_LIMIT = 5.gigabytes
DEFAULT_ALLOWED_CONTENT_TYPES = %w(application/gzip application/octet-stream).freeze
def initialize(
configuration:,
relative_url:,
dir:,
file_size_limit: DEFAULT_FILE_SIZE_LIMIT,
allowed_content_types: DEFAULT_ALLOWED_CONTENT_TYPES,
filename: nil)
2021-09-04 01:27:46 +05:30
@configuration = configuration
@relative_url = relative_url
@filename = filename
@dir = dir
2021-09-30 23:02:18 +05:30
@file_size_limit = file_size_limit
@allowed_content_types = allowed_content_types
2021-09-04 01:27:46 +05:30
end
def execute
validate_dir
validate_url
validate_content_type
validate_content_length
download_file
validate_symlink
filepath
end
private
2021-09-30 23:02:18 +05:30
attr_reader :configuration, :relative_url, :dir, :file_size_limit, :allowed_content_types
2021-09-04 01:27:46 +05:30
def download_file
File.open(filepath, 'wb') do |file|
bytes_downloaded = 0
http_client.stream(relative_url) do |chunk|
bytes_downloaded += chunk.size
2021-09-30 23:02:18 +05:30
validate_size!(bytes_downloaded)
2021-09-04 01:27:46 +05:30
raise(ServiceError, "File download error #{chunk.code}") unless chunk.code == 200
file.write(chunk)
end
end
rescue StandardError => e
File.delete(filepath) if File.exist?(filepath)
raise e
end
def http_client
@http_client ||= BulkImports::Clients::HTTP.new(
2021-09-30 23:02:18 +05:30
url: configuration.url,
2021-09-04 01:27:46 +05:30
token: configuration.access_token
)
end
def allow_local_requests?
::Gitlab::CurrentSettings.allow_local_requests_from_web_hooks_and_services?
end
def headers
@headers ||= http_client.head(relative_url).headers
end
def validate_dir
raise(ServiceError, 'Invalid target directory') unless dir.start_with?(Dir.tmpdir)
end
def validate_symlink
if File.lstat(filepath).symlink?
File.delete(filepath)
raise(ServiceError, 'Invalid downloaded file')
end
end
def validate_url
::Gitlab::UrlBlocker.validate!(
http_client.resource_url(relative_url),
allow_localhost: allow_local_requests?,
allow_local_network: allow_local_requests?,
schemes: %w(http https)
)
end
def validate_content_length
2021-09-30 23:02:18 +05:30
validate_size!(headers['content-length'])
end
2021-09-04 01:27:46 +05:30
2021-09-30 23:02:18 +05:30
def validate_size!(size)
if size.blank?
raise ServiceError, 'Missing content-length header'
elsif size.to_i > file_size_limit
raise ServiceError, "File size %{size} exceeds limit of %{limit}" % {
size: ActiveSupport::NumberHelper.number_to_human_size(size),
limit: ActiveSupport::NumberHelper.number_to_human_size(file_size_limit)
}
end
2021-09-04 01:27:46 +05:30
end
def validate_content_type
content_type = headers['content-type']
2021-09-30 23:02:18 +05:30
raise(ServiceError, 'Invalid content type') if content_type.blank? || allowed_content_types.exclude?(content_type)
end
def filepath
@filepath ||= File.join(@dir, filename)
end
def filename
@filename.presence || remote_filename
end
# Fetch the remote filename information from the request content-disposition header
# - Raises if the filename does not exist
# - If the filename is longer then 255 chars truncate it
# to be a total of 255 chars (with the extension)
def remote_filename
@remote_filename ||=
headers['content-disposition'].to_s
.match(REMOTE_FILENAME_PATTERN) # matches the filename pattern
.then { |match| match&.named_captures || {} } # ensures the match is a hash
.fetch('filename') # fetches the 'filename' key or raise KeyError
.then(&File.method(:basename)) # Ensures to remove path from the filename (../ for instance)
.then(&method(:ensure_filename_size)) # Ensures the filename is within the FILENAME_SIZE_LIMIT
rescue KeyError
raise ServiceError, 'Remote filename not provided in content-disposition header'
end
def ensure_filename_size(filename)
if filename.length <= FILENAME_SIZE_LIMIT
filename
else
extname = File.extname(filename)
basename = File.basename(filename, extname)[0, FILENAME_SIZE_LIMIT]
"#{basename}#{extname}"
end
2021-09-04 01:27:46 +05:30
end
end
end