traject 2.0.0-java
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +18 -0
- data/.travis.yml +27 -0
- data/.yardopts +3 -0
- data/Gemfile +12 -0
- data/LICENSE.txt +20 -0
- data/README.md +461 -0
- data/Rakefile +21 -0
- data/bench/bench.rb +30 -0
- data/bin/traject +16 -0
- data/doc/batch_execution.md +243 -0
- data/doc/extending.md +190 -0
- data/doc/indexing_rules.md +265 -0
- data/doc/other_commands.md +47 -0
- data/doc/settings.md +101 -0
- data/lib/tasks/load_maps.rake +48 -0
- data/lib/traject.rb +11 -0
- data/lib/traject/command_line.rb +301 -0
- data/lib/traject/csv_writer.rb +34 -0
- data/lib/traject/debug_writer.rb +47 -0
- data/lib/traject/delimited_writer.rb +110 -0
- data/lib/traject/indexer.rb +613 -0
- data/lib/traject/indexer/settings.rb +110 -0
- data/lib/traject/json_writer.rb +51 -0
- data/lib/traject/line_writer.rb +63 -0
- data/lib/traject/macros/basic.rb +9 -0
- data/lib/traject/macros/marc21.rb +223 -0
- data/lib/traject/macros/marc21_semantics.rb +584 -0
- data/lib/traject/macros/marc_format_classifier.rb +197 -0
- data/lib/traject/marc_extractor.rb +410 -0
- data/lib/traject/marc_reader.rb +89 -0
- data/lib/traject/mock_reader.rb +97 -0
- data/lib/traject/ndj_reader.rb +40 -0
- data/lib/traject/null_writer.rb +22 -0
- data/lib/traject/qualified_const_get.rb +40 -0
- data/lib/traject/solr_json_writer.rb +277 -0
- data/lib/traject/thread_pool.rb +161 -0
- data/lib/traject/translation_map.rb +267 -0
- data/lib/traject/util.rb +52 -0
- data/lib/traject/version.rb +3 -0
- data/lib/traject/yaml_writer.rb +9 -0
- data/lib/translation_maps/lcc_top_level.yaml +26 -0
- data/lib/translation_maps/marc_genre_007.yaml +9 -0
- data/lib/translation_maps/marc_genre_leader.yaml +22 -0
- data/lib/translation_maps/marc_geographic.yaml +589 -0
- data/lib/translation_maps/marc_instruments.yaml +102 -0
- data/lib/translation_maps/marc_languages.yaml +490 -0
- data/test/debug_writer_test.rb +38 -0
- data/test/delimited_writer_test.rb +104 -0
- data/test/indexer/each_record_test.rb +59 -0
- data/test/indexer/macros_marc21_semantics_test.rb +391 -0
- data/test/indexer/macros_marc21_test.rb +190 -0
- data/test/indexer/macros_test.rb +40 -0
- data/test/indexer/map_record_test.rb +209 -0
- data/test/indexer/read_write_test.rb +101 -0
- data/test/indexer/settings_test.rb +152 -0
- data/test/indexer/to_field_test.rb +77 -0
- data/test/marc_extractor_test.rb +412 -0
- data/test/marc_format_classifier_test.rb +98 -0
- data/test/marc_reader_test.rb +110 -0
- data/test/solr_json_writer_test.rb +248 -0
- data/test/test_helper.rb +90 -0
- data/test/test_support/245_no_ab.marc +1 -0
- data/test/test_support/880_with_no_6.utf8.marc +1 -0
- data/test/test_support/bad_subfield_code.marc +1 -0
- data/test/test_support/bad_utf_byte.utf8.marc +1 -0
- data/test/test_support/date_resort_to_260.marc +1 -0
- data/test/test_support/date_type_r_missing_date2.marc +1 -0
- data/test/test_support/date_with_u.marc +1 -0
- data/test/test_support/demo_config.rb +155 -0
- data/test/test_support/emptyish_record.marc +1 -0
- data/test/test_support/escaped_character_reference.marc8.marc +1 -0
- data/test/test_support/george_eliot.marc +1 -0
- data/test/test_support/hebrew880s.marc +1 -0
- data/test/test_support/louis_armstrong.marc +1 -0
- data/test/test_support/manufacturing_consent.marc +1 -0
- data/test/test_support/manuscript_online_thesis.marc +1 -0
- data/test/test_support/microform_online_conference.marc +1 -0
- data/test/test_support/multi_era.marc +1 -0
- data/test/test_support/multi_geo.marc +1 -0
- data/test/test_support/musical_cage.marc +1 -0
- data/test/test_support/nature.marc +1 -0
- data/test/test_support/one-marc8.mrc +1 -0
- data/test/test_support/online_only.marc +1 -0
- data/test/test_support/packed_041a_lang.marc +1 -0
- data/test/test_support/test_data.utf8.json +30 -0
- data/test/test_support/test_data.utf8.marc.xml +2609 -0
- data/test/test_support/test_data.utf8.mrc +1 -0
- data/test/test_support/test_data.utf8.mrc.gz +0 -0
- data/test/test_support/the_business_ren.marc +1 -0
- data/test/translation_map_test.rb +225 -0
- data/test/translation_maps/bad_ruby.rb +8 -0
- data/test/translation_maps/bad_yaml.yaml +1 -0
- data/test/translation_maps/both_map.rb +1 -0
- data/test/translation_maps/both_map.yaml +1 -0
- data/test/translation_maps/default_literal.rb +10 -0
- data/test/translation_maps/default_passthrough.rb +10 -0
- data/test/translation_maps/marc_040a_translate_test.yaml +1 -0
- data/test/translation_maps/properties_map.properties +5 -0
- data/test/translation_maps/ruby_map.rb +10 -0
- data/test/translation_maps/translate_array_test.yaml +8 -0
- data/test/translation_maps/yaml_map.yaml +7 -0
- data/traject.gemspec +47 -0
- metadata +382 -0
@@ -0,0 +1,152 @@
|
|
1
|
+
require 'test_helper'
|
2
|
+
|
3
|
+
describe "Traject::Indexer#settings" do
|
4
|
+
before do
|
5
|
+
@indexer = Traject::Indexer.new
|
6
|
+
end
|
7
|
+
|
8
|
+
it "starts out a Hash, that can fill in it's defaults" do
|
9
|
+
assert_kind_of Hash, @indexer.settings
|
10
|
+
|
11
|
+
Traject::Indexer::Settings.defaults.each_pair do |key, value|
|
12
|
+
assert_equal value, @indexer.settings[key]
|
13
|
+
end
|
14
|
+
end
|
15
|
+
|
16
|
+
it "can fill_in_defaults!" do
|
17
|
+
@indexer.settings.fill_in_defaults!
|
18
|
+
|
19
|
+
assert_equal Traject::Indexer::Settings.defaults, @indexer.settings
|
20
|
+
end
|
21
|
+
|
22
|
+
it "doesn't overwrite with fill_in_defaults!" do
|
23
|
+
key = Traject::Indexer::Settings.defaults.keys.first
|
24
|
+
@indexer.settings[ key ] = "MINE KEEP IT"
|
25
|
+
|
26
|
+
@indexer.settings.fill_in_defaults!
|
27
|
+
|
28
|
+
assert_equal "MINE KEEP IT", @indexer.settings[key]
|
29
|
+
end
|
30
|
+
|
31
|
+
it "can take argument to set" do
|
32
|
+
@indexer.settings("foo" => "foo", "bar" => "bar")
|
33
|
+
|
34
|
+
assert_equal "foo", @indexer.settings["foo"]
|
35
|
+
assert_equal "bar", @indexer.settings["bar"]
|
36
|
+
end
|
37
|
+
|
38
|
+
it "has settings DSL to set" do
|
39
|
+
@indexer.instance_eval do
|
40
|
+
settings do
|
41
|
+
store "foo", "foo"
|
42
|
+
end
|
43
|
+
end
|
44
|
+
|
45
|
+
assert_equal "foo", @indexer.settings["foo"]
|
46
|
+
end
|
47
|
+
|
48
|
+
it "merges new values, not completely replaces" do
|
49
|
+
@indexer.settings("one" => "original", "two" => "original", "three" => "original", "four" => "original")
|
50
|
+
|
51
|
+
@indexer.settings do
|
52
|
+
store "two", "second"
|
53
|
+
store "three", "second"
|
54
|
+
end
|
55
|
+
|
56
|
+
@indexer.settings do
|
57
|
+
store "three", "third"
|
58
|
+
end
|
59
|
+
|
60
|
+
@indexer.settings("four" => "fourth")
|
61
|
+
|
62
|
+
{"one" => "original", "two" => "second", "three" => "third", "four" => "fourth"}.each_pair do |key, value|
|
63
|
+
assert_equal value, @indexer.settings[key]
|
64
|
+
end
|
65
|
+
end
|
66
|
+
|
67
|
+
it "is indifferent between string and symbol" do
|
68
|
+
@indexer.settings[:foo] = "foo 1"
|
69
|
+
@indexer.settings["foo"] = "foo 2"
|
70
|
+
|
71
|
+
assert_equal "foo 2", @indexer.settings[:foo]
|
72
|
+
|
73
|
+
@indexer.settings do
|
74
|
+
store "foo", "foo 3"
|
75
|
+
store :foo, "foo 4"
|
76
|
+
end
|
77
|
+
|
78
|
+
assert_equal "foo 4", @indexer.settings["foo"]
|
79
|
+
end
|
80
|
+
|
81
|
+
it "implements #provide as cautious setter" do
|
82
|
+
@indexer.settings[:a] = "original"
|
83
|
+
|
84
|
+
@indexer.settings do
|
85
|
+
provide :a, "new"
|
86
|
+
provide :b, "new"
|
87
|
+
end
|
88
|
+
|
89
|
+
assert_equal "original", @indexer.settings[:a]
|
90
|
+
assert_equal "new", @indexer.settings[:b]
|
91
|
+
end
|
92
|
+
|
93
|
+
it "has reverse_merge" do
|
94
|
+
settings = Traject::Indexer::Settings.new("a" => "original", "b" => "original")
|
95
|
+
|
96
|
+
new_settings = settings.reverse_merge(:a => "new", :c => "new")
|
97
|
+
|
98
|
+
assert_kind_of Traject::Indexer::Settings, new_settings
|
99
|
+
|
100
|
+
assert_equal "original", new_settings["a"]
|
101
|
+
assert_equal "original", new_settings["b"]
|
102
|
+
assert_equal "new", new_settings["c"]
|
103
|
+
end
|
104
|
+
|
105
|
+
it "has reverse_merge!" do
|
106
|
+
settings = Traject::Indexer::Settings.new("a" => "original", "b" => "original")
|
107
|
+
|
108
|
+
settings.reverse_merge!(:a => "new", :c => "new")
|
109
|
+
|
110
|
+
assert_kind_of Traject::Indexer::Settings, settings
|
111
|
+
|
112
|
+
assert_equal "original", settings["a"]
|
113
|
+
assert_equal "original", settings["b"]
|
114
|
+
assert_equal "new", settings["c"]
|
115
|
+
end
|
116
|
+
|
117
|
+
describe "inspect" do
|
118
|
+
it "keeps keys ending in 'password' out of inspect" do
|
119
|
+
settings = Traject::Indexer::Settings.new("a" => "a",
|
120
|
+
"password" => "password", "some_password" => "password",
|
121
|
+
"some.password" => "password")
|
122
|
+
|
123
|
+
parsed = eval( settings.inspect )
|
124
|
+
assert_equal( {"a" => "a", "password" => "[hidden]", "some_password" => "[hidden]", "some.password" => "[hidden]"}, parsed)
|
125
|
+
end
|
126
|
+
end
|
127
|
+
|
128
|
+
describe "JRuby / MRI" do
|
129
|
+
before do
|
130
|
+
@indexer = Traject::Indexer.new
|
131
|
+
end
|
132
|
+
|
133
|
+
it "has the right indexer name" do
|
134
|
+
if defined? JRUBY_VERSION
|
135
|
+
assert_equal "Traject::Marc4JReader", @indexer.settings['reader_class_name']
|
136
|
+
else
|
137
|
+
assert_equal "Traject::MarcReader", @indexer.settings['reader_class_name']
|
138
|
+
end
|
139
|
+
end
|
140
|
+
|
141
|
+
# This next one has the added effect of making sure the correct class
|
142
|
+
# has actually been loaded -- otherwise the constant wouldn't be available
|
143
|
+
it "has the correct default indexer class based on platform" do
|
144
|
+
if defined? JRUBY_VERSION
|
145
|
+
assert_equal Traject::Marc4JReader, @indexer.reader_class
|
146
|
+
else
|
147
|
+
assert_equal Traject::MarcReader, @indexer.reader_class
|
148
|
+
end
|
149
|
+
end
|
150
|
+
end
|
151
|
+
|
152
|
+
end
|
@@ -0,0 +1,77 @@
|
|
1
|
+
require 'test_helper'
|
2
|
+
|
3
|
+
describe "Traject::Indexer.to_field" do
|
4
|
+
before do
|
5
|
+
@indexer = Traject::Indexer.new
|
6
|
+
end
|
7
|
+
describe "checks it's arguments" do
|
8
|
+
it "rejects nil first arg" do
|
9
|
+
assert_raises(Traject::Indexer::NamingError) { @indexer.to_field(nil) }
|
10
|
+
end
|
11
|
+
it "rejects empty string first arg" do
|
12
|
+
assert_raises(Traject::Indexer::NamingError) {@indexer.to_field("")}
|
13
|
+
end
|
14
|
+
it "rejects non-string first arg" do
|
15
|
+
assert_raises(Traject::Indexer::NamingError) {@indexer.to_field(:symbol)}
|
16
|
+
end
|
17
|
+
|
18
|
+
it "rejects one-arg lambda" do
|
19
|
+
assert_raises(Traject::Indexer::ArityError) do
|
20
|
+
@indexer.to_field("foo") do |one_arg|
|
21
|
+
end
|
22
|
+
end
|
23
|
+
end
|
24
|
+
it "rejects four-arg lambda" do
|
25
|
+
assert_raises(Traject::Indexer::ArityError) do
|
26
|
+
@indexer.to_field("foo") do |one_arg, two_arg, three_arg, four_arg|
|
27
|
+
end
|
28
|
+
end
|
29
|
+
end
|
30
|
+
it "accepts two arg lambda" do
|
31
|
+
@indexer.to_field("foo") do |one, two|
|
32
|
+
end
|
33
|
+
end
|
34
|
+
it "accepts three arg lambda" do
|
35
|
+
@indexer.to_field("foo") {|one, two, three| one }
|
36
|
+
end
|
37
|
+
it "accepts variable lambda" do
|
38
|
+
@indexer.to_field("foo") do |*variable|
|
39
|
+
end
|
40
|
+
end
|
41
|
+
end
|
42
|
+
|
43
|
+
it "outputs error with source location" do
|
44
|
+
begin
|
45
|
+
@indexer.to_field('foo') {|one, two| }
|
46
|
+
@indexer.to_field('') {|one, two| } # bad field name
|
47
|
+
flunk("Should have rejected empty field name")
|
48
|
+
rescue Traject::Indexer::NamingError => e
|
49
|
+
assert_match(/at .*\/.*:\d+/, e.message)
|
50
|
+
rescue
|
51
|
+
flunk("Should only fail with a NamingError")
|
52
|
+
end
|
53
|
+
end
|
54
|
+
|
55
|
+
# Just verifying this is how it works
|
56
|
+
it "doesn't allow you to just wholesale assignment to the accumulator" do
|
57
|
+
@indexer.to_field('foo') do |rec, acc|
|
58
|
+
acc = ['hello']
|
59
|
+
end
|
60
|
+
output = @indexer.map_record('never looked at')
|
61
|
+
assert_equal nil, output['foo']
|
62
|
+
end
|
63
|
+
|
64
|
+
it "allows use of accumulator.replace" do
|
65
|
+
@indexer.to_field('foo') do |rec, acc|
|
66
|
+
acc.replace ['hello']
|
67
|
+
end
|
68
|
+
output = @indexer.map_record('never looked at')
|
69
|
+
assert_equal ['hello'], output['foo']
|
70
|
+
end
|
71
|
+
|
72
|
+
|
73
|
+
end
|
74
|
+
|
75
|
+
|
76
|
+
|
77
|
+
|
@@ -0,0 +1,412 @@
|
|
1
|
+
# encoding: UTF-8
|
2
|
+
|
3
|
+
require 'test_helper'
|
4
|
+
require 'traject/marc_extractor'
|
5
|
+
|
6
|
+
require 'marc'
|
7
|
+
|
8
|
+
describe "Traject::MarcExtractor" do
|
9
|
+
it "is frozen read-only" do
|
10
|
+
extractor = Traject::MarcExtractor.new("100abcde", :seperator => ";")
|
11
|
+
assert extractor.frozen?
|
12
|
+
assert extractor.spec_hash.frozen?
|
13
|
+
assert extractor.options.frozen?
|
14
|
+
end
|
15
|
+
|
16
|
+
|
17
|
+
describe "#parse_marc_spec" do
|
18
|
+
it "parses single spec with all elements" do
|
19
|
+
parsed = Traject::MarcExtractor.parse_string_spec("245|1*|abcg")
|
20
|
+
|
21
|
+
assert_kind_of Hash, parsed
|
22
|
+
assert_equal 1, parsed.keys.length
|
23
|
+
spec = parsed['245'].first
|
24
|
+
assert_kind_of Traject::MarcExtractor::Spec, spec
|
25
|
+
|
26
|
+
assert_equal "1", spec.indicator1
|
27
|
+
assert_nil spec.indicator2
|
28
|
+
|
29
|
+
assert_kind_of Array, spec.subfields
|
30
|
+
end
|
31
|
+
|
32
|
+
it "parses a mixed bag" do
|
33
|
+
parsed = Traject::MarcExtractor.parse_string_spec("245abcde:810:700|*4|bcd")
|
34
|
+
spec245 = parsed['245'].first
|
35
|
+
spec810 = parsed['810'].first
|
36
|
+
spec700 = parsed['700'].first
|
37
|
+
|
38
|
+
assert_length 3, parsed
|
39
|
+
|
40
|
+
#245abcde
|
41
|
+
assert spec245
|
42
|
+
assert_nil spec245.indicator1
|
43
|
+
assert_nil spec245.indicator2
|
44
|
+
assert_equal %w{a b c d e}, spec245.subfields
|
45
|
+
|
46
|
+
#810
|
47
|
+
assert spec810
|
48
|
+
assert_nil spec810.indicator1
|
49
|
+
assert_nil spec810.indicator2
|
50
|
+
assert_nil spec810.subfields, "No subfields"
|
51
|
+
|
52
|
+
#700-*4bcd
|
53
|
+
assert spec700
|
54
|
+
assert_nil spec700.indicator1
|
55
|
+
assert_equal "4", spec700.indicator2
|
56
|
+
assert_equal %w{b c d}, spec700.subfields
|
57
|
+
end
|
58
|
+
|
59
|
+
it "parses fixed field byte offsets" do
|
60
|
+
parsed = Traject::MarcExtractor.parse_string_spec("005[5]:008[7-10]")
|
61
|
+
|
62
|
+
assert_equal 5, parsed["005"].first.bytes
|
63
|
+
assert_equal 7..10, parsed["008"].first.bytes
|
64
|
+
end
|
65
|
+
|
66
|
+
it "allows arrays of specs" do
|
67
|
+
parsed = Traject::MarcExtractor.parse_string_spec %w(
|
68
|
+
245abcde
|
69
|
+
810
|
70
|
+
700|*4|bcd
|
71
|
+
)
|
72
|
+
assert_length 3, parsed
|
73
|
+
end
|
74
|
+
|
75
|
+
it "allows mixture of array and colon-delimited specs" do
|
76
|
+
parsed = Traject::MarcExtractor.parse_string_spec %w(
|
77
|
+
245abcde
|
78
|
+
100:110:111
|
79
|
+
810
|
80
|
+
700|*4|bcd
|
81
|
+
)
|
82
|
+
assert_length 6, parsed
|
83
|
+
end
|
84
|
+
|
85
|
+
|
86
|
+
end
|
87
|
+
|
88
|
+
# Mostly an internal method, not neccesarily API, but
|
89
|
+
# an important one, so we unit test some parts of it.
|
90
|
+
describe "#specs_covering_field" do
|
91
|
+
describe "for alternate script tags" do
|
92
|
+
before do
|
93
|
+
@record = MARC::Reader.new(support_file_path "hebrew880s.marc").to_a.first
|
94
|
+
@extractor = Traject::MarcExtractor.new("245")
|
95
|
+
|
96
|
+
@a245 = @record.fields.find {|f| f.tag == "245"}
|
97
|
+
assert ! @a245.nil?, "Found a 245 to test"
|
98
|
+
|
99
|
+
@a880_245 = @record.fields.find do |field|
|
100
|
+
(field.tag == "880") && field['6'] &&
|
101
|
+
"245" == field['6'].slice(0,3)
|
102
|
+
end
|
103
|
+
assert ! @a880_245.nil?, "Found an 880-245 to test"
|
104
|
+
|
105
|
+
@a880_100 = @record.fields.find do |field|
|
106
|
+
(field.tag == "880") && field['6'] &&
|
107
|
+
"100" == field['6'].slice(0,3)
|
108
|
+
end
|
109
|
+
|
110
|
+
assert ! @a880_100.nil?, "Found an 880-100 to test"
|
111
|
+
end
|
112
|
+
it "finds spec for relevant 880" do
|
113
|
+
assert_equal( [Traject::MarcExtractor::Spec.new(:tag => "245")], @extractor.specs_covering_field(@a880_245) )
|
114
|
+
assert_equal [], @extractor.specs_covering_field(@a880_100)
|
115
|
+
end
|
116
|
+
it "does not find spec for 880 if disabled" do
|
117
|
+
@extractor = Traject::MarcExtractor.new("245", :alternate_script => false)
|
118
|
+
assert_equal [], @extractor.specs_covering_field(@a880_245)
|
119
|
+
end
|
120
|
+
it "finds only 880 if so configured" do
|
121
|
+
@extractor = Traject::MarcExtractor.new("245", :alternate_script => :only)
|
122
|
+
assert_equal [], @extractor.specs_covering_field(@a245)
|
123
|
+
assert_equal([Traject::MarcExtractor::Spec.new(:tag => "245")], @extractor.specs_covering_field(@a880_245))
|
124
|
+
end
|
125
|
+
end
|
126
|
+
end
|
127
|
+
|
128
|
+
describe "#extract_by_spec" do
|
129
|
+
before do
|
130
|
+
@record = MARC::Reader.new(support_file_path "manufacturing_consent.marc").to_a.first
|
131
|
+
end
|
132
|
+
|
133
|
+
describe "extracts a basic case" do
|
134
|
+
before do
|
135
|
+
parsed_spec = Traject::MarcExtractor.parse_string_spec("700abcdef:856|*2|:505|1*|:245ba")
|
136
|
+
@values = Traject::MarcExtractor.new(parsed_spec).extract(@record)
|
137
|
+
end
|
138
|
+
|
139
|
+
it "returns an array" do
|
140
|
+
assert_kind_of Array, @values
|
141
|
+
end
|
142
|
+
|
143
|
+
it "handles no subfields given" do
|
144
|
+
a856s = @record.find_all {|f| f.tag == "856"}
|
145
|
+
assert a856s, "Record must have 856 fields for this test to work"
|
146
|
+
|
147
|
+
a856s.each do |field|
|
148
|
+
assert @values.include?( field.subfields.collect(&:value).join(" "))
|
149
|
+
end
|
150
|
+
end
|
151
|
+
|
152
|
+
it "does not have 505, due to non-matching indicators" do
|
153
|
+
assert ! @values.find {|s| s.include? "propaganda model"}
|
154
|
+
end
|
155
|
+
|
156
|
+
|
157
|
+
|
158
|
+
it "respects original record order, for both fields and subfields" do
|
159
|
+
expected = ["Manufacturing consent : the political economy of the mass media /",
|
160
|
+
"Chomsky, Noam.",
|
161
|
+
"Contributor biographical information http://www.loc.gov/catdir/bios/random051/2001050014.html",
|
162
|
+
"Publisher description http://www.loc.gov/catdir/description/random044/2001050014.html"]
|
163
|
+
assert_equal expected, @values
|
164
|
+
end
|
165
|
+
end
|
166
|
+
|
167
|
+
describe "extracts fixed fields" do
|
168
|
+
it ", complete" do
|
169
|
+
parsed_spec = Traject::MarcExtractor.parse_string_spec("001")
|
170
|
+
values = Traject::MarcExtractor.new(parsed_spec).extract(@record)
|
171
|
+
|
172
|
+
assert_equal ["2710183"], values
|
173
|
+
end
|
174
|
+
it ", single byte offset" do
|
175
|
+
parsed_spec = Traject::MarcExtractor.parse_string_spec("008[5]")
|
176
|
+
values = Traject::MarcExtractor.new(parsed_spec).extract(@record)
|
177
|
+
|
178
|
+
assert_equal ["1"], values
|
179
|
+
end
|
180
|
+
it ", byte range" do
|
181
|
+
parsed_spec = Traject::MarcExtractor.parse_string_spec("008[7-10]")
|
182
|
+
values = Traject::MarcExtractor.new(parsed_spec).extract(@record)
|
183
|
+
|
184
|
+
assert_equal ["2002"], values
|
185
|
+
end
|
186
|
+
end
|
187
|
+
|
188
|
+
describe "separator argument" do
|
189
|
+
it "causes non-join when nil" do
|
190
|
+
parsed_spec = Traject::MarcExtractor.parse_string_spec("245")
|
191
|
+
values = Traject::MarcExtractor.new(parsed_spec, :separator => nil).extract(@record)
|
192
|
+
|
193
|
+
assert_length 3, values
|
194
|
+
end
|
195
|
+
|
196
|
+
it "can be non-default" do
|
197
|
+
parsed_spec = Traject::MarcExtractor.parse_string_spec("245")
|
198
|
+
values = Traject::MarcExtractor.new(parsed_spec, :separator => "!! ").extract(@record)
|
199
|
+
|
200
|
+
assert_length 1, values
|
201
|
+
assert_equal "Manufacturing consent :!! the political economy of the mass media /!! Edward S. Herman and Noam Chomsky ; with a new introduction by the authors.", values.first
|
202
|
+
end
|
203
|
+
end
|
204
|
+
|
205
|
+
describe "extracts alternate script" do
|
206
|
+
before do
|
207
|
+
@record = MARC::Reader.new(support_file_path "hebrew880s.marc").to_a.first
|
208
|
+
@parsed_spec = Traject::MarcExtractor.parse_string_spec("245b")
|
209
|
+
end
|
210
|
+
it "from default :include" do
|
211
|
+
|
212
|
+
values = Traject::MarcExtractor.new(@parsed_spec).extract(@record)
|
213
|
+
|
214
|
+
assert_length 2, values # both the original and the 880
|
215
|
+
assert_equal ["ben Marṭin Buber le-Aharon Daṿid Gordon /", "בין מרטין בובר לאהרן דוד גורדון /"], values
|
216
|
+
end
|
217
|
+
it "with :only" do
|
218
|
+
values = Traject::MarcExtractor.new(@parsed_spec, :alternate_script => :only).extract(@record)
|
219
|
+
|
220
|
+
assert_length 1, values
|
221
|
+
assert_equal ["בין מרטין בובר לאהרן דוד גורדון /"], values
|
222
|
+
end
|
223
|
+
it "with false" do
|
224
|
+
values = Traject::MarcExtractor.new(@parsed_spec, :alternate_script => false).extract(@record)
|
225
|
+
|
226
|
+
assert_length 1, values
|
227
|
+
assert_equal ["ben Marṭin Buber le-Aharon Daṿid Gordon /"], values
|
228
|
+
end
|
229
|
+
end
|
230
|
+
|
231
|
+
it "works with string second arg too" do
|
232
|
+
values = Traject::MarcExtractor.new("245abc").extract(@record)
|
233
|
+
|
234
|
+
assert_length 1, values
|
235
|
+
assert values.first.include?("Manufacturing consent"), "Extracted value includes title"
|
236
|
+
end
|
237
|
+
|
238
|
+
it "returns empty array if no matching tags" do
|
239
|
+
values = Traject::MarcExtractor.new("999abc").extract(@record)
|
240
|
+
assert_equal [], values
|
241
|
+
|
242
|
+
values = Traject::MarcExtractor.new("999").extract(@record)
|
243
|
+
assert_equal [], values
|
244
|
+
end
|
245
|
+
|
246
|
+
it "returns empty array if matching tag but no subfield" do
|
247
|
+
values = Traject::MarcExtractor.new("245xyz").extract(@record)
|
248
|
+
assert_equal [], values
|
249
|
+
end
|
250
|
+
|
251
|
+
end
|
252
|
+
|
253
|
+
describe "with bad data" do
|
254
|
+
it "can ignore an 880 with no $6" do
|
255
|
+
@record = MARC::Reader.new(support_file_path "880_with_no_6.utf8.marc").to_a.first
|
256
|
+
values = Traject::MarcExtractor.new("001").extract(@record)
|
257
|
+
assert_equal ["3468569"], values
|
258
|
+
end
|
259
|
+
end
|
260
|
+
|
261
|
+
describe "#each_matching_line" do
|
262
|
+
before do
|
263
|
+
@record = MARC::Reader.new(support_file_path "manufacturing_consent.marc").to_a.first
|
264
|
+
@extractor = Traject::MarcExtractor.new("245abc")
|
265
|
+
end
|
266
|
+
it "yields two args" do
|
267
|
+
called = false
|
268
|
+
@extractor.each_matching_line(@record) do |field, spec|
|
269
|
+
called = true
|
270
|
+
assert_kind_of MARC::DataField, field
|
271
|
+
assert_kind_of Traject::MarcExtractor::Spec, spec
|
272
|
+
end
|
273
|
+
assert called, "calls block"
|
274
|
+
end
|
275
|
+
it "yields three args" do
|
276
|
+
called = false
|
277
|
+
@extractor.each_matching_line(@record) do |field, spec, extractor|
|
278
|
+
called = true
|
279
|
+
assert_kind_of MARC::DataField, field
|
280
|
+
assert_kind_of Traject::MarcExtractor::Spec, spec
|
281
|
+
assert_kind_of Traject::MarcExtractor, extractor
|
282
|
+
assert_same @extractor, extractor
|
283
|
+
end
|
284
|
+
assert called, "calls block"
|
285
|
+
end
|
286
|
+
end
|
287
|
+
|
288
|
+
describe "#collect_matching_lines" do
|
289
|
+
before do
|
290
|
+
@record = MARC::Reader.new(support_file_path "manufacturing_consent.marc").to_a.first
|
291
|
+
@extractor = Traject::MarcExtractor.new("245abc")
|
292
|
+
end
|
293
|
+
it "collects with custom block" do
|
294
|
+
results = @extractor.collect_matching_lines(@record) do |field, spec, extractor|
|
295
|
+
extractor.collect_subfields(field, spec)
|
296
|
+
end
|
297
|
+
assert_equal ["Manufacturing consent : the political economy of the mass media / Edward S. Herman and Noam Chomsky ; with a new introduction by the authors."], results
|
298
|
+
end
|
299
|
+
end
|
300
|
+
|
301
|
+
describe "MarcExtractor.cached" do
|
302
|
+
it "creates" do
|
303
|
+
extractor = Traject::MarcExtractor.cached("245abc", :separator => nil)
|
304
|
+
spec_hash = extractor.spec_hash
|
305
|
+
|
306
|
+
assert extractor.options[:separator].nil?, "extractor options[:separator] is nil"
|
307
|
+
assert_equal({"245"=>[Traject::MarcExtractor::Spec.new(:tag => "245", :subfields=>["a", "b", "c"])]}, spec_hash)
|
308
|
+
end
|
309
|
+
it "caches" do
|
310
|
+
ext1 = Traject::MarcExtractor.cached("245abc", :separator => nil)
|
311
|
+
ext2 = Traject::MarcExtractor.cached("245abc", :separator => nil)
|
312
|
+
|
313
|
+
assert_same ext1, ext2
|
314
|
+
end
|
315
|
+
end
|
316
|
+
|
317
|
+
|
318
|
+
describe "Allows multiple uses of the same tag" do
|
319
|
+
before do
|
320
|
+
@record = MARC::Reader.new(support_file_path "manufacturing_consent.marc").to_a.first
|
321
|
+
end
|
322
|
+
|
323
|
+
it "allows repated tags for a variable field" do
|
324
|
+
extractor = Traject::MarcExtractor.new("245a:245b")
|
325
|
+
values = extractor.extract(@record)
|
326
|
+
assert_equal ['Manufacturing consent :', 'the political economy of the mass media /'], values
|
327
|
+
end
|
328
|
+
|
329
|
+
it "allows repeated tags with indicators specs" do
|
330
|
+
extractor = Traject::MarcExtractor.new("245|1*|a:245|2*|b")
|
331
|
+
@record.append(MARC::DataField.new('245', '2', '0', ['a', 'Subfield A Value'], ['b', 'Subfield B Value']))
|
332
|
+
results = extractor.extract(@record)
|
333
|
+
assert_equal ['Manufacturing consent :', 'Subfield B Value'], results
|
334
|
+
end
|
335
|
+
|
336
|
+
|
337
|
+
|
338
|
+
|
339
|
+
it "provides multiple values for repeated subfields with single specified subfield" do
|
340
|
+
ex = Traject::MarcExtractor.new("245a")
|
341
|
+
f = @record.fields('245').first
|
342
|
+
title_a = f['a']
|
343
|
+
f.append(MARC::Subfield.new('a', title_a))
|
344
|
+
results = ex.extract(@record)
|
345
|
+
assert_equal [title_a, title_a], results
|
346
|
+
end
|
347
|
+
|
348
|
+
it "concats single subfield spec when given as eg 245aa" do
|
349
|
+
ex = Traject::MarcExtractor.new("245aa")
|
350
|
+
f = @record.fields('245').first
|
351
|
+
title_a = f['a']
|
352
|
+
f.append(MARC::Subfield.new('a', title_a))
|
353
|
+
results = ex.extract(@record)
|
354
|
+
assert_equal ["#{title_a} #{title_a}"], results
|
355
|
+
end
|
356
|
+
|
357
|
+
it "provides single value for repeated subfields with multiple specified subfields" do
|
358
|
+
ex = Traject::MarcExtractor.new("245ab")
|
359
|
+
f = @record.fields('245').first
|
360
|
+
title_a = f['a']
|
361
|
+
title_b = f['b']
|
362
|
+
f.append(MARC::Subfield.new('a', title_a))
|
363
|
+
results = ex.extract(@record)
|
364
|
+
assert_equal ["#{title_a} #{title_b} #{title_a}"], results
|
365
|
+
|
366
|
+
end
|
367
|
+
|
368
|
+
it "provides single value for repeated subfields with no specified subfield" do
|
369
|
+
ex = Traject::MarcExtractor.new("245")
|
370
|
+
f = @record.fields('245').first
|
371
|
+
title_a = f['a']
|
372
|
+
f.append(MARC::Subfield.new('a', title_a))
|
373
|
+
results = ex.extract(@record)
|
374
|
+
assert_equal 1, results.size
|
375
|
+
end
|
376
|
+
|
377
|
+
|
378
|
+
|
379
|
+
|
380
|
+
it "allows repeated tags for a control field" do
|
381
|
+
extractor = Traject::MarcExtractor.new("001[0-1]:001[0-3]")
|
382
|
+
values = extractor.extract(@record)
|
383
|
+
assert_equal ["27", "2710"], values
|
384
|
+
end
|
385
|
+
|
386
|
+
it "associates indicators properly with repeated tags" do
|
387
|
+
@record = MARC::Record.new
|
388
|
+
@record.append MARC::DataField.new("100", '1', ' ', ['a', '100a first indicator 1'], ['b', 'should not include 100|1|b'])
|
389
|
+
@record.append MARC::DataField.new("100", '2', ' ', ['b', '100b first indicator 2'], ['a', 'should not include 100|2|a'])
|
390
|
+
|
391
|
+
extractor = Traject::MarcExtractor.new("100|1*|a:100|2*|b")
|
392
|
+
|
393
|
+
values = extractor.extract(@record)
|
394
|
+
|
395
|
+
assert_equal ['100a first indicator 1', '100b first indicator 2'], values
|
396
|
+
end
|
397
|
+
|
398
|
+
end
|
399
|
+
|
400
|
+
describe "MarcExtractor::Spec" do
|
401
|
+
describe "==" do
|
402
|
+
it "equals when equal" do
|
403
|
+
assert_equal Traject::MarcExtractor::Spec.new(:subfields => %w{a b c}), Traject::MarcExtractor::Spec.new(:subfields => %w{a b c})
|
404
|
+
end
|
405
|
+
it "does not equal when not" do
|
406
|
+
refute_equal Traject::MarcExtractor::Spec.new(:subfields => %w{a b c}), Traject::MarcExtractor::Spec.new(:subfields => %w{a b c}, :indicator2 => '1')
|
407
|
+
end
|
408
|
+
end
|
409
|
+
end
|
410
|
+
|
411
|
+
|
412
|
+
end
|