38778a1e-2dd3-4ec3-b4d6-a6c4d301682a

Committed 07 Dec 2023 05:02PM UTC coverage: 87.536% (+0.6%) from 86.964%

Build # 38778a1e-2dd3-4ec3-b4d6-a6c4d301682a

Build Type

push

circleci

Committed by

web-flow

Commit Message

Merge pull request #79 from pulibrary/i77-jrgriffiniii-date-display

Updates Rubocop, refactors DataSpace Solr Document generation, and ensures that invalid embargo dates trigger log warnings

Run Details

190 of 196 new or added lines in 5 files covered. (96.94%)

12 existing lines in 1 file now uncovered.

604 of 690 relevant lines covered (87.54%)

18.31 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

69.05

/lib/orangetheses/harvester.rb

# frozen_string_literal: true

require 'oai'
require 'tmpdir'

module Orangetheses
  class Harvester
    def self.default_server
      return 'https://dataspace-staging.princeton.edu/oai/request' if test?

      'https://dataspace.princeton.edu/oai/request'
    end

    def self.default_metadata_prefix
      'oai_dc'
    end

    def self.default_verb
      'ListRecords'
    end

    def self.default_set
      'com_88435_dsp019c67wm88m'
    end

    # @param [Hash] opts  options to pass to the client
    # @option opts [String] :dir  Directory in which to save files. A temporary
    #   directory will be created if this option is not included.
    # @option opts [String] :server ('http://dataspace.princeton.edu/oai/')
    # @option opts [String] :metadata_prefix ('oai_dc')
    # @option opts [String] :verb ('ListRecords')
    # @option opts [String] :set ('hdl_88435_dsp019c67wm88m')
    def initialize(dir: Dir.mktmpdir,
                   server: nil,
                   metadata_prefix: nil,
                   verb: nil,
                   set: nil)

      @dir = dir
      @server = server || self.class.default_server
      @metadata_prefix = metadata_prefix || self.class.default_metadata_prefix
      @verb = verb || self.class.default_verb
      @set = set || self.class.default_set
    end

    # @return [Array<String>] A list of directories containing metadata records
    def harvest_all
      dirs = []
      dir = nil
      client.list_records(headers).full.each_with_index do |record, i|
        if (i % 1000).zero?
          dir = Dir.mktmpdir(nil, @dir)
          dirs << dir
        end
        File.open(File.join(dir, "#{i}.xml"), 'w') do |f|
          f.write(record.metadata)
        end
      end
      dirs
    end

    # @return [Array<String>]
    def index_all(indexer)
      client.list_records(headers).full.each_with_index do |record, _i|
        indexer.index(record.metadata)
      end
    end

    def index_item(indexer:, identifier:)
      record_response = client.get_record(identifier:)
      record = record_response.record

      indexer.index(record.metadata)
    rescue StandardError => e
      Rails.logger.warn("Error indexing the OAI-PMH record using #{identifier}: #{e}")
    end

    private

    def headers
      {
        metadataPrefix: @metadata_prefix,
        set: @set
      }
    end

    def client
      @client ||= OAI::Client.new(@server)
    end
  end
end

1	# frozen_string_literal: true
2
3	require 'oai'	1✔
4	require 'tmpdir'	1✔
5
6	module Orangetheses	1✔
7	class Harvester	1✔
8	def self.default_server	1✔
9	return 'https://dataspace-staging.princeton.edu/oai/request' if test?	2✔
10
11	'https://dataspace.princeton.edu/oai/request'	2✔
12	end
13
14	def self.default_metadata_prefix	1✔
15	'oai_dc'	3✔
16	end
17
18	def self.default_verb	1✔
19	'ListRecords'	3✔
20	end
21
22	def self.default_set	1✔
23	'com_88435_dsp019c67wm88m'	3✔
24	end
25
26	# @param [Hash] opts options to pass to the client
27	# @option opts [String] :dir Directory in which to save files. A temporary
28	# directory will be created if this option is not included.
29	# @option opts [String] :server ('http://dataspace.princeton.edu/oai/')
30	# @option opts [String] :metadata_prefix ('oai_dc')
31	# @option opts [String] :verb ('ListRecords')
32	# @option opts [String] :set ('hdl_88435_dsp019c67wm88m')
33	def initialize(dir: Dir.mktmpdir,	1✔
34	server: nil,
35	metadata_prefix: nil,
36	verb: nil,
37	set: nil)
38
39	@dir = dir	3✔
40	@server = server \|\| self.class.default_server	3✔
41	@metadata_prefix = metadata_prefix \|\| self.class.default_metadata_prefix	3✔
42	@verb = verb \|\| self.class.default_verb	3✔
43	@set = set \|\| self.class.default_set	3✔
44	end
45
46	# @return [Array<String>] A list of directories containing metadata records
47	def harvest_all	1✔
48	dirs = []	×
49	dir = nil	×
50	client.list_records(headers).full.each_with_index do \|record, i\|	×
NEW 51	if (i % 1000).zero?	×
52	dir = Dir.mktmpdir(nil, @dir)	×
53	dirs << dir	×
54	end
55	File.open(File.join(dir, "#{i}.xml"), 'w') do \|f\|	×
56	f.write(record.metadata)	×
57	end
58	end
59	dirs	×
60	end
61
62	# @return [Array<String>]
63	def index_all(indexer)	1✔
64	client.list_records(headers).full.each_with_index do \|record, _i\|	×
65	indexer.index(record.metadata)	×
66	end
67	end
68
69	def index_item(indexer:, identifier:)	1✔
70	record_response = client.get_record(identifier:)	1✔
71	record = record_response.record	1✔
72
73	indexer.index(record.metadata)	1✔
74	rescue StandardError => e
75	Rails.logger.warn("Error indexing the OAI-PMH record using #{identifier}: #{e}")	×
76	end
77
78	private	1✔
79
80	def headers	1✔
81	{
82	metadataPrefix: @metadata_prefix,	×
83	set: @set
84	}
85	end
86
87	def client	1✔
88	@client \|\|= OAI::Client.new(@server)	3✔
89	end
90	end
91	end

pulibrary / orangetheses / 38778a1e-2dd3-4ec3-b4d6-a6c4d301682a

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous