Sha256: e06af2fba408e65fd4512c54a7c68bd041955b0694349adefc95a088933ffd62
Contents?: true
Size: 904 Bytes
Versions: 27
Compression:
Stored size: 904 Bytes
Contents
require 'marc'
require 'json'
require 'zlib'
# Read newline-delimited JSON file, where each line is a marc-in-json string.
# UTF-8 encoding is required.
class Traject::NDJReader
include Enumerable
def initialize(input_stream, settings)
@settings = settings
@input_stream = input_stream
if @settings['command_line.filename'] =~ /\.gz$/
@input_stream = Zlib::GzipReader.new(@input_stream, :external_encoding => "UTF-8")
end
end
def logger
@logger ||= (@settings[:logger] || Yell.new(STDERR, :level => "gt.fatal")) # null logger)
end
def each
unless block_given?
return enum_for(:each)
end
@input_stream.each_with_index do |json, i|
begin
yield MARC::Record.new_from_hash(JSON.parse(json))
rescue Exception => e
self.logger.error("Problem with JSON record on line #{i}: #{e.message}")
end
end
end
end
Version data entries
27 entries across 27 versions & 1 rubygems