berkeley_library-alma 0.0.5 → 0.0.6
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.idea/alma.iml +13 -13
- data/.idea/inspectionProfiles/Project_Default.xml +2 -0
- data/CHANGES.md +4 -0
- data/berkeley_library-alma.gemspec +1 -1
- data/lib/berkeley_library/alma/module_info.rb +1 -1
- data/lib/berkeley_library/alma/record_id.rb +2 -6
- data/lib/berkeley_library/alma/sru/sru.rb +102 -0
- data/lib/berkeley_library/alma/sru/xml_reader.rb +148 -0
- data/lib/berkeley_library/alma/sru.rb +1 -79
- data/spec/data/availability-sru-page-1.xml +522 -0
- data/spec/data/availability-sru-page-2.xml +266 -0
- data/spec/lib/berkeley_library/alma/sru/sru_spec.rb +98 -0
- data/spec/lib/berkeley_library/alma/sru/xml_reader_spec.rb +75 -0
- metadata +14 -6
- data/spec/lib/berkeley_library/alma/sru_spec.rb +0 -29
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: ab227f94abea9d1275bceb4ed361cf2d75021ae7d9cf49283e6139404e9e327f
|
4
|
+
data.tar.gz: 2370efa8b2bc2b72e6e506cf38271a74ee8a785cbebf7f42b19e49bf84b4a6da
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 438181aea5e19163ca57dd7c5c9086fd6d324ac5bb6e05fd33917cc70327ca5682e1f07a9ffc5280104207562f6fdda4443070bdc8ba3163b519ba70f96f2558
|
7
|
+
data.tar.gz: dc9cafbf0b36ee0ac1875d8cc89b0b4a10bdccd37456f67004e0c3abb0097fde7b8d5da31c946ab1eb4f300dd1f14ad5dcaf550ca697aef0de4f8b6d81521bdb
|
data/.idea/alma.iml
CHANGED
@@ -12,15 +12,15 @@
|
|
12
12
|
</content>
|
13
13
|
<orderEntry type="jdk" jdkName="RVM: ruby-2.7.5" jdkType="RUBY_SDK" />
|
14
14
|
<orderEntry type="sourceFolder" forTests="false" />
|
15
|
-
<orderEntry type="library" scope="PROVIDED" name="actionpack (
|
16
|
-
<orderEntry type="library" scope="PROVIDED" name="actionview (
|
17
|
-
<orderEntry type="library" scope="PROVIDED" name="activesupport (
|
15
|
+
<orderEntry type="library" scope="PROVIDED" name="actionpack (v7.0.2.4, RVM: ruby-2.7.5) [gem]" level="application" />
|
16
|
+
<orderEntry type="library" scope="PROVIDED" name="actionview (v7.0.2.4, RVM: ruby-2.7.5) [gem]" level="application" />
|
17
|
+
<orderEntry type="library" scope="PROVIDED" name="activesupport (v7.0.2.4, RVM: ruby-2.7.5) [gem]" level="application" />
|
18
18
|
<orderEntry type="library" scope="PROVIDED" name="addressable (v2.8.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
19
19
|
<orderEntry type="library" scope="PROVIDED" name="amazing_print (v1.4.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
20
20
|
<orderEntry type="library" scope="PROVIDED" name="ast (v2.4.2, RVM: ruby-2.7.5) [gem]" level="application" />
|
21
|
-
<orderEntry type="library" scope="PROVIDED" name="berkeley_library-logging (v0.2.
|
21
|
+
<orderEntry type="library" scope="PROVIDED" name="berkeley_library-logging (v0.2.6, RVM: ruby-2.7.5) [gem]" level="application" />
|
22
22
|
<orderEntry type="library" scope="PROVIDED" name="berkeley_library-marc (v0.3.1, RVM: ruby-2.7.5) [gem]" level="application" />
|
23
|
-
<orderEntry type="library" scope="PROVIDED" name="berkeley_library-util (v0.1.
|
23
|
+
<orderEntry type="library" scope="PROVIDED" name="berkeley_library-util (v0.1.2, RVM: ruby-2.7.5) [gem]" level="application" />
|
24
24
|
<orderEntry type="library" scope="PROVIDED" name="builder (v3.2.4, RVM: ruby-2.7.5) [gem]" level="application" />
|
25
25
|
<orderEntry type="library" scope="PROVIDED" name="bundle-audit (v0.1.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
26
26
|
<orderEntry type="library" scope="PROVIDED" name="bundler (v2.1.4, RVM: ruby-2.7.5) [gem]" level="application" />
|
@@ -28,7 +28,7 @@
|
|
28
28
|
<orderEntry type="library" scope="PROVIDED" name="ci_reporter (v2.0.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
29
29
|
<orderEntry type="library" scope="PROVIDED" name="ci_reporter_rspec (v1.0.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
30
30
|
<orderEntry type="library" scope="PROVIDED" name="colorize (v0.8.1, RVM: ruby-2.7.5) [gem]" level="application" />
|
31
|
-
<orderEntry type="library" scope="PROVIDED" name="concurrent-ruby (v1.1.
|
31
|
+
<orderEntry type="library" scope="PROVIDED" name="concurrent-ruby (v1.1.10, RVM: ruby-2.7.5) [gem]" level="application" />
|
32
32
|
<orderEntry type="library" scope="PROVIDED" name="crack (v0.4.5, RVM: ruby-2.7.5) [gem]" level="application" />
|
33
33
|
<orderEntry type="library" scope="PROVIDED" name="crass (v1.0.6, RVM: ruby-2.7.5) [gem]" level="application" />
|
34
34
|
<orderEntry type="library" scope="PROVIDED" name="diff-lcs (v1.5.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
@@ -39,16 +39,16 @@
|
|
39
39
|
<orderEntry type="library" scope="PROVIDED" name="http-accept (v1.7.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
40
40
|
<orderEntry type="library" scope="PROVIDED" name="http-cookie (v1.0.4, RVM: ruby-2.7.5) [gem]" level="application" />
|
41
41
|
<orderEntry type="library" scope="PROVIDED" name="i18n (v1.10.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
42
|
-
<orderEntry type="library" scope="PROVIDED" name="lograge (v0.
|
43
|
-
<orderEntry type="library" scope="PROVIDED" name="loofah (v2.
|
42
|
+
<orderEntry type="library" scope="PROVIDED" name="lograge (v0.12.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
43
|
+
<orderEntry type="library" scope="PROVIDED" name="loofah (v2.17.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
44
44
|
<orderEntry type="library" scope="PROVIDED" name="marc (v1.1.1, RVM: ruby-2.7.5) [gem]" level="application" />
|
45
45
|
<orderEntry type="library" scope="PROVIDED" name="method_source (v1.0.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
46
46
|
<orderEntry type="library" scope="PROVIDED" name="mime-types (v3.4.1, RVM: ruby-2.7.5) [gem]" level="application" />
|
47
47
|
<orderEntry type="library" scope="PROVIDED" name="mime-types-data (v3.2022.0105, RVM: ruby-2.7.5) [gem]" level="application" />
|
48
|
-
<orderEntry type="library" scope="PROVIDED" name="mini_portile2 (v2.
|
48
|
+
<orderEntry type="library" scope="PROVIDED" name="mini_portile2 (v2.8.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
49
49
|
<orderEntry type="library" scope="PROVIDED" name="minitest (v5.15.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
50
50
|
<orderEntry type="library" scope="PROVIDED" name="netrc (v0.11.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
51
|
-
<orderEntry type="library" scope="PROVIDED" name="nokogiri (v1.13.
|
51
|
+
<orderEntry type="library" scope="PROVIDED" name="nokogiri (v1.13.5, RVM: ruby-2.7.5) [gem]" level="application" />
|
52
52
|
<orderEntry type="library" scope="PROVIDED" name="oj (v3.13.11, RVM: ruby-2.7.5) [gem]" level="application" />
|
53
53
|
<orderEntry type="library" scope="PROVIDED" name="ougai (v1.9.1, RVM: ruby-2.7.5) [gem]" level="application" />
|
54
54
|
<orderEntry type="library" scope="PROVIDED" name="parallel (v1.21.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
@@ -60,7 +60,7 @@
|
|
60
60
|
<orderEntry type="library" scope="PROVIDED" name="rack-test (v1.1.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
61
61
|
<orderEntry type="library" scope="PROVIDED" name="rails-dom-testing (v2.0.3, RVM: ruby-2.7.5) [gem]" level="application" />
|
62
62
|
<orderEntry type="library" scope="PROVIDED" name="rails-html-sanitizer (v1.4.2, RVM: ruby-2.7.5) [gem]" level="application" />
|
63
|
-
<orderEntry type="library" scope="PROVIDED" name="railties (
|
63
|
+
<orderEntry type="library" scope="PROVIDED" name="railties (v7.0.2.4, RVM: ruby-2.7.5) [gem]" level="application" />
|
64
64
|
<orderEntry type="library" scope="PROVIDED" name="rainbow (v3.1.1, RVM: ruby-2.7.5) [gem]" level="application" />
|
65
65
|
<orderEntry type="library" scope="PROVIDED" name="rake (v13.0.6, RVM: ruby-2.7.5) [gem]" level="application" />
|
66
66
|
<orderEntry type="library" scope="PROVIDED" name="regexp_parser (v2.2.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
@@ -87,7 +87,7 @@
|
|
87
87
|
<orderEntry type="library" scope="PROVIDED" name="typesafe_enum (v0.3.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
88
88
|
<orderEntry type="library" scope="PROVIDED" name="tzinfo (v2.0.4, RVM: ruby-2.7.5) [gem]" level="application" />
|
89
89
|
<orderEntry type="library" scope="PROVIDED" name="unf (v0.1.4, RVM: ruby-2.7.5) [gem]" level="application" />
|
90
|
-
<orderEntry type="library" scope="PROVIDED" name="unf_ext (v0.0.8, RVM: ruby-2.7.5) [gem]" level="application" />
|
90
|
+
<orderEntry type="library" scope="PROVIDED" name="unf_ext (v0.0.8.1, RVM: ruby-2.7.5) [gem]" level="application" />
|
91
91
|
<orderEntry type="library" scope="PROVIDED" name="unicode-display_width (v2.1.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
92
92
|
<orderEntry type="library" scope="PROVIDED" name="webmock (v3.14.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
93
93
|
<orderEntry type="library" scope="PROVIDED" name="webrick (v1.7.0, RVM: ruby-2.7.5) [gem]" level="application" />
|
@@ -109,7 +109,7 @@
|
|
109
109
|
</RakeTaskImpl>
|
110
110
|
<RakeTaskImpl description="Run all specs in spec directory, with coverage" fullCommand="coverage" id="coverage" />
|
111
111
|
<RakeTaskImpl description="Run tests, check test coverage, check code style, check for vulnerabilities, build gem" fullCommand="default" id="default" />
|
112
|
-
<RakeTaskImpl description="Build berkeley_library-alma.gemspec as berkeley_library-alma-0.0.
|
112
|
+
<RakeTaskImpl description="Build berkeley_library-alma.gemspec as berkeley_library-alma-0.0.5.gem" fullCommand="gem" id="gem" />
|
113
113
|
<RakeTaskImpl description="Run RuboCop with auto-correct, and output results to console" fullCommand="ra" id="ra" />
|
114
114
|
<RakeTaskImpl description="Run rubocop with HTML output" fullCommand="rubocop" id="rubocop" />
|
115
115
|
<RakeTaskImpl id="rubocop">
|
@@ -7,6 +7,7 @@
|
|
7
7
|
<inspection_tool class="GoUnusedGlobalVariable" enabled="false" level="WEAK WARNING" enabled_by_default="false" />
|
8
8
|
<inspection_tool class="GoUnusedVariable" enabled="true" level="WARNING" enabled_by_default="true" />
|
9
9
|
<inspection_tool class="GrazieInspection" enabled="false" level="TYPO" enabled_by_default="false" />
|
10
|
+
<inspection_tool class="HttpUrlsUsage" enabled="false" level="WEAK WARNING" enabled_by_default="false" />
|
10
11
|
<inspection_tool class="LanguageDetectionInspection" enabled="false" level="WARNING" enabled_by_default="false" />
|
11
12
|
<inspection_tool class="Rubocop" enabled="false" level="WARNING" enabled_by_default="false" />
|
12
13
|
<inspection_tool class="RubyCaseWithoutElseBlockInspection" enabled="false" level="WARNING" enabled_by_default="false" />
|
@@ -15,6 +16,7 @@
|
|
15
16
|
</inspection_tool>
|
16
17
|
<inspection_tool class="RubyNilAnalysis" enabled="true" level="WEAK WARNING" enabled_by_default="true" />
|
17
18
|
<inspection_tool class="RubyStringKeysInHashInspection" enabled="true" level="INFORMATION" enabled_by_default="true" />
|
19
|
+
<inspection_tool class="RubyTooManyInstanceVariablesInspection" enabled="false" level="WARNING" enabled_by_default="false" />
|
18
20
|
<inspection_tool class="RubyUnnecessaryReturnStatement" enabled="true" level="WEAK WARNING" enabled_by_default="true" />
|
19
21
|
<inspection_tool class="SpellCheckingInspection" enabled="false" level="TYPO" enabled_by_default="false">
|
20
22
|
<option name="processCode" value="true" />
|
data/CHANGES.md
CHANGED
@@ -24,7 +24,7 @@ Gem::Specification.new do |spec|
|
|
24
24
|
|
25
25
|
spec.add_dependency 'berkeley_library-logging', '~> 0.2'
|
26
26
|
spec.add_dependency 'berkeley_library-marc', '~> 0.3.1'
|
27
|
-
spec.add_dependency 'berkeley_library-util', '~> 0.1', '>= 0.1.
|
27
|
+
spec.add_dependency 'berkeley_library-util', '~> 0.1', '>= 0.1.2'
|
28
28
|
spec.add_dependency 'nokogiri', '~> 1.12'
|
29
29
|
|
30
30
|
spec.add_development_dependency 'bundle-audit', '~> 0.1'
|
@@ -7,7 +7,7 @@ module BerkeleyLibrary
|
|
7
7
|
SUMMARY = 'Alma/Primo utilities for the UC Berkeley Library'.freeze
|
8
8
|
DESCRIPTION = 'A gem providing Alma/Primo-related utility code for the UC Berkeley Library'.freeze
|
9
9
|
LICENSE = 'MIT'.freeze
|
10
|
-
VERSION = '0.0.
|
10
|
+
VERSION = '0.0.6'.freeze
|
11
11
|
HOMEPAGE = 'https://github.com/BerkeleyLibrary/alma'.freeze
|
12
12
|
end
|
13
13
|
end
|
@@ -60,12 +60,8 @@ module BerkeleyLibrary
|
|
60
60
|
# @return [MARC::Record, nil] the MARC record
|
61
61
|
# rubocop:disable Naming/AccessorMethodName
|
62
62
|
def get_marc_record
|
63
|
-
|
64
|
-
|
65
|
-
return nil
|
66
|
-
end
|
67
|
-
|
68
|
-
logger.warn("GET #{marc_uri} did not return a MARC record") unless (marc_record = marc_reader.first)
|
63
|
+
records = SRU.marc_records_for(sru_query_value)
|
64
|
+
logger.warn("GET #{marc_uri} did not return a MARC record") unless (marc_record = records.first)
|
69
65
|
marc_record
|
70
66
|
end
|
71
67
|
# rubocop:enable Naming/AccessorMethodName
|
@@ -0,0 +1,102 @@
|
|
1
|
+
require 'uri'
|
2
|
+
require 'berkeley_library/util/uris'
|
3
|
+
require 'berkeley_library/alma/config'
|
4
|
+
require 'berkeley_library/alma/constants'
|
5
|
+
require 'berkeley_library/alma/record_id'
|
6
|
+
require 'berkeley_library/alma/sru/xml_reader'
|
7
|
+
|
8
|
+
module BerkeleyLibrary
|
9
|
+
module Alma
|
10
|
+
module SRU
|
11
|
+
include BerkeleyLibrary::Logging
|
12
|
+
include Constants
|
13
|
+
|
14
|
+
class << self
|
15
|
+
include SRU
|
16
|
+
end
|
17
|
+
|
18
|
+
# Given a list of record IDs, returns the MARC records for each ID (if found).
|
19
|
+
# Note that the order of the records is not guaranteed.
|
20
|
+
#
|
21
|
+
# @param record_ids [Array<String, RecordId>] the record IDs to look up
|
22
|
+
# @param freeze [Boolean] whether to freeze the records
|
23
|
+
# @return [Enumerator::Lazy<MARC::Record>] the records
|
24
|
+
def get_marc_records(*record_ids, freeze: false)
|
25
|
+
# noinspection RubyMismatchedReturnType
|
26
|
+
parsed_ids = record_ids.filter_map { |id| RecordId.parse(id) }
|
27
|
+
raise ArgumentError, "Argument #{record_ids.inspect} contain no valid record IDs" if parsed_ids.empty?
|
28
|
+
|
29
|
+
sru_query_value = parsed_ids.map(&:sru_query_value).join(' or ')
|
30
|
+
SRU.marc_records_for(sru_query_value, freeze: freeze)
|
31
|
+
end
|
32
|
+
|
33
|
+
# Returns a URI for retrieving records for the specified query
|
34
|
+
# via SRU. Requires {Config#alma_sru_base_uri} to be set.
|
35
|
+
#
|
36
|
+
# @return [URI] the MARC URI
|
37
|
+
def sru_query_uri(sru_query_value)
|
38
|
+
query_string = URI.encode_www_form(
|
39
|
+
'version' => '1.2',
|
40
|
+
'operation' => 'searchRetrieve',
|
41
|
+
'query' => sru_query_value
|
42
|
+
)
|
43
|
+
|
44
|
+
Util::URIs.append(Config.alma_sru_base_uri, '?', query_string)
|
45
|
+
end
|
46
|
+
|
47
|
+
# Makes an SRU query for the specified query value and returns the query response
|
48
|
+
# as a string.
|
49
|
+
#
|
50
|
+
# @param query_value [String] the value of the SRU query parameter
|
51
|
+
# @return [String, nil] the SRU query response body, or nil in the event of an error.
|
52
|
+
def make_sru_query(query_value)
|
53
|
+
uri = sru_query_uri(query_value)
|
54
|
+
do_get(uri)
|
55
|
+
end
|
56
|
+
|
57
|
+
# Makes an SRU query for the specified query value and returns the query response
|
58
|
+
# as MARC records.
|
59
|
+
#
|
60
|
+
# @param query_value [String] the value of the SRU query parameter
|
61
|
+
# @param freeze [Boolean] whether to freeze the records
|
62
|
+
# @return [Enumerator::Lazy<MARC::Record>] the records
|
63
|
+
def marc_records_for(query_value, freeze: false)
|
64
|
+
Enumerator.new do |y|
|
65
|
+
uri = sru_query_uri(query_value)
|
66
|
+
perform_query(uri, freeze: freeze) { |rec| y << rec }
|
67
|
+
end.lazy
|
68
|
+
end
|
69
|
+
|
70
|
+
private
|
71
|
+
|
72
|
+
def perform_query(query_uri, start_record: nil, freeze: false, &block)
|
73
|
+
full_query_uri = full_query_uri_for(query_uri, start_record)
|
74
|
+
next_start_record = perform_single_query(full_query_uri, freeze, &block)
|
75
|
+
return unless next_start_record
|
76
|
+
|
77
|
+
perform_query(query_uri, start_record: next_start_record, freeze: freeze, &block)
|
78
|
+
end
|
79
|
+
|
80
|
+
def full_query_uri_for(query_uri, start_record)
|
81
|
+
return query_uri unless start_record
|
82
|
+
|
83
|
+
BerkeleyLibrary::Util::URIs.append(query_uri, "&startRecord=#{start_record}")
|
84
|
+
end
|
85
|
+
|
86
|
+
def perform_single_query(query_uri, freeze, &block)
|
87
|
+
return unless (xml = do_get(query_uri))
|
88
|
+
|
89
|
+
xml_reader = XMLReader.read(xml, freeze: freeze)
|
90
|
+
xml_reader.each(&block)
|
91
|
+
xml_reader.next_record_position
|
92
|
+
end
|
93
|
+
|
94
|
+
def do_get(uri)
|
95
|
+
Util::URIs.get(uri, headers: { user_agent: DEFAULT_USER_AGENT })
|
96
|
+
rescue RestClient::Exception => e
|
97
|
+
logger.warn("GET #{uri} failed", e)
|
98
|
+
nil
|
99
|
+
end
|
100
|
+
end
|
101
|
+
end
|
102
|
+
end
|
@@ -0,0 +1,148 @@
|
|
1
|
+
require 'nokogiri'
|
2
|
+
require 'marc/xml_parsers'
|
3
|
+
require 'berkeley_library/util/files'
|
4
|
+
|
5
|
+
module BerkeleyLibrary
|
6
|
+
module Alma
|
7
|
+
module SRU
|
8
|
+
# A customized XML reader for reading MARC records from SRU search results.
|
9
|
+
class XMLReader
|
10
|
+
include Enumerable
|
11
|
+
include ::MARC::NokogiriReader
|
12
|
+
include BerkeleyLibrary::Util::Files
|
13
|
+
|
14
|
+
# ############################################################
|
15
|
+
# Constants
|
16
|
+
|
17
|
+
NS_SRW = 'http://www.loc.gov/zing/srw/'.freeze
|
18
|
+
NS_MARC = 'http://www.loc.gov/MARC21/slim'.freeze
|
19
|
+
|
20
|
+
# ############################################################
|
21
|
+
# Attributes
|
22
|
+
|
23
|
+
# @return [Integer, nil] the record identifier of the most recently parsed record, if any
|
24
|
+
attr_reader :last_record_id
|
25
|
+
|
26
|
+
# @return [Integer, nil] the record position of the most recently parsed record, if any
|
27
|
+
attr_reader :last_record_position
|
28
|
+
|
29
|
+
# @return [Integer, nil] the next record position, if present
|
30
|
+
attr_reader :next_record_position
|
31
|
+
|
32
|
+
# Returns the total number of records, based on the `<numberOfRecords/>` tag
|
33
|
+
# returned in the SRU response.
|
34
|
+
#
|
35
|
+
# Note that the total is not guaranteed to be present, and if present,
|
36
|
+
# may not be present unless at least some records have been parsed.
|
37
|
+
#
|
38
|
+
# @return [Integer, nil] the total number of records, or `nil` if the total has not been read yet
|
39
|
+
def num_records
|
40
|
+
@num_records&.to_i
|
41
|
+
end
|
42
|
+
|
43
|
+
# Returns the number of records yielded.
|
44
|
+
#
|
45
|
+
# @return [Integer] the number of records yielded.
|
46
|
+
def records_yielded
|
47
|
+
@records_yielded ||= 0
|
48
|
+
end
|
49
|
+
|
50
|
+
# ############################################################
|
51
|
+
# Initializer
|
52
|
+
|
53
|
+
def initialize(source, freeze: false)
|
54
|
+
@handle = ensure_io(source)
|
55
|
+
@freeze = freeze
|
56
|
+
init
|
57
|
+
end
|
58
|
+
|
59
|
+
class << self
|
60
|
+
# Reads MARC records from an XML datasource given either as an XML string, a file path,
|
61
|
+
# or as an IO object.
|
62
|
+
#
|
63
|
+
# @param source [String, Pathname, IO] an XML string, the path to a file, or an IO to read from directly
|
64
|
+
# @param freeze [Boolean] whether to freeze each record after reading
|
65
|
+
def read(source, freeze: false)
|
66
|
+
new(source, freeze: freeze)
|
67
|
+
end
|
68
|
+
end
|
69
|
+
|
70
|
+
# ############################################################
|
71
|
+
# MARC::GenericPullParser overrides
|
72
|
+
|
73
|
+
def yield_record
|
74
|
+
@record[:record].tap do |record|
|
75
|
+
record.freeze if @freeze
|
76
|
+
end
|
77
|
+
|
78
|
+
super
|
79
|
+
ensure
|
80
|
+
increment_records_yielded!
|
81
|
+
end
|
82
|
+
|
83
|
+
# ############################################################
|
84
|
+
# Nokogiri::XML::SAX::Document overrides
|
85
|
+
|
86
|
+
# @see Nokogiri::XML::Sax::Document#start_element_namespace
|
87
|
+
# rubocop:disable Metrics/ParameterLists
|
88
|
+
def start_element_namespace(name, attrs = [], prefix = nil, uri = nil, ns = [])
|
89
|
+
super
|
90
|
+
|
91
|
+
@current_element_ns = uri
|
92
|
+
@current_element_name = name
|
93
|
+
end
|
94
|
+
# rubocop:enable Metrics/ParameterLists
|
95
|
+
|
96
|
+
# @see Nokogiri::XML::Sax::Document#end_element_namespace
|
97
|
+
def end_element_namespace(name, prefix = nil, uri = nil)
|
98
|
+
# Delay yielding record till we reach the end of the outer SRU <record/>
|
99
|
+
# element (not the inner MARC <record/> element), so we can record the
|
100
|
+
# values of <recordIdentifier> and <recordPosition/>
|
101
|
+
if name.downcase == 'record'
|
102
|
+
yield_record if uri == NS_SRW
|
103
|
+
elsif uri == NS_MARC
|
104
|
+
super
|
105
|
+
end
|
106
|
+
|
107
|
+
@current_element_name = nil
|
108
|
+
end
|
109
|
+
|
110
|
+
# @see Nokogiri::XML::Sax::Document#characters
|
111
|
+
# rubocop:disable Metrics/MethodLength
|
112
|
+
def characters(string)
|
113
|
+
return super unless NS_SRW == @current_element_ns
|
114
|
+
return unless (name = @current_element_name)
|
115
|
+
|
116
|
+
case name
|
117
|
+
when 'numberOfRecords'
|
118
|
+
@num_records = string
|
119
|
+
when 'recordIdentifier'
|
120
|
+
@last_record_id = string
|
121
|
+
when 'recordPosition'
|
122
|
+
@last_record_position = string.to_i
|
123
|
+
when 'nextRecordPosition'
|
124
|
+
@next_record_position = string.to_i
|
125
|
+
end
|
126
|
+
end
|
127
|
+
# rubocop:enable Metrics/MethodLength
|
128
|
+
|
129
|
+
# ############################################################
|
130
|
+
# Private
|
131
|
+
|
132
|
+
private
|
133
|
+
|
134
|
+
def ensure_io(file)
|
135
|
+
return file if reader_like?(file)
|
136
|
+
return File.new(file) if file_exists?(file)
|
137
|
+
return StringIO.new(file) if file =~ /^\s*</x
|
138
|
+
|
139
|
+
raise ArgumentError, "Don't know how to read XML from #{file.inspect}: not an IO, file path, or XML text"
|
140
|
+
end
|
141
|
+
|
142
|
+
def increment_records_yielded!
|
143
|
+
@records_yielded = records_yielded + 1
|
144
|
+
end
|
145
|
+
end
|
146
|
+
end
|
147
|
+
end
|
148
|
+
end
|
@@ -1,79 +1 @@
|
|
1
|
-
|
2
|
-
require 'berkeley_library/alma/config'
|
3
|
-
require 'berkeley_library/alma/constants'
|
4
|
-
require 'berkeley_library/alma/record_id'
|
5
|
-
require 'berkeley_library/util/uris'
|
6
|
-
|
7
|
-
module BerkeleyLibrary
|
8
|
-
module Alma
|
9
|
-
module SRU
|
10
|
-
include BerkeleyLibrary::Logging
|
11
|
-
include Constants
|
12
|
-
|
13
|
-
class << self
|
14
|
-
include SRU
|
15
|
-
end
|
16
|
-
|
17
|
-
# Given a list of record IDs, returns the MARC records for each ID (if found).
|
18
|
-
#
|
19
|
-
# @param record_ids [Array<String, RecordId>] the record IDs to look up
|
20
|
-
# @return [MARC::XMLReader, nil] a reader for the MARC records, or nil if
|
21
|
-
# the records could not be read
|
22
|
-
def get_marc_records(*record_ids)
|
23
|
-
# noinspection RubyMismatchedReturnType
|
24
|
-
parsed_ids = record_ids.filter_map { |id| RecordId.parse(id) }
|
25
|
-
raise ArgumentError, "Argument #{record_ids.inspect} contain no valid record IDs" if parsed_ids.empty?
|
26
|
-
|
27
|
-
sru_query_value = parsed_ids.map(&:sru_query_value).join(' or ')
|
28
|
-
SRU.marc_records_for(sru_query_value)
|
29
|
-
end
|
30
|
-
|
31
|
-
# Returns a URI for retrieving records for the specified query
|
32
|
-
# via SRU. Requires {Config#alma_sru_base_uri} to be set.
|
33
|
-
#
|
34
|
-
# @return [URI] the MARC URI
|
35
|
-
def sru_query_uri(sru_query_value)
|
36
|
-
query_string = URI.encode_www_form(
|
37
|
-
'version' => '1.2',
|
38
|
-
'operation' => 'searchRetrieve',
|
39
|
-
'query' => sru_query_value
|
40
|
-
)
|
41
|
-
|
42
|
-
Util::URIs.append(Config.alma_sru_base_uri, '?', query_string)
|
43
|
-
end
|
44
|
-
|
45
|
-
# Makes an SRU query for the specified query value and returns the query response
|
46
|
-
# as a string.
|
47
|
-
#
|
48
|
-
# @param query_value [String] the value of the SRU query parameter
|
49
|
-
# @return [String, nil] the SRU query response body, or nil in the event of an error.
|
50
|
-
def make_sru_query(query_value)
|
51
|
-
uri = sru_query_uri(query_value)
|
52
|
-
do_get(uri)
|
53
|
-
end
|
54
|
-
|
55
|
-
# Makes an SRU query for the specified query value and returns the query response
|
56
|
-
# as MARC records.
|
57
|
-
#
|
58
|
-
# @param query_value [String] the value of the SRU query parameter
|
59
|
-
# @return [MARC::XMLReader, nil] a reader for the MARC records, or nil if
|
60
|
-
# the records could not be read
|
61
|
-
def marc_records_for(query_value)
|
62
|
-
return unless (xml = make_sru_query(query_value))
|
63
|
-
|
64
|
-
input = StringIO.new(xml.scrub)
|
65
|
-
MARC::XMLReader.new(input, parser: 'nokogiri')
|
66
|
-
end
|
67
|
-
|
68
|
-
private
|
69
|
-
|
70
|
-
def do_get(uri)
|
71
|
-
# TODO: can we get the XML as an IO rather than as a string?
|
72
|
-
Util::URIs.get(uri, headers: { user_agent: DEFAULT_USER_AGENT })
|
73
|
-
rescue RestClient::Exception => e
|
74
|
-
logger.warn("GET #{uri} failed", e)
|
75
|
-
nil
|
76
|
-
end
|
77
|
-
end
|
78
|
-
end
|
79
|
-
end
|
1
|
+
Dir.glob(File.expand_path('sru/*.rb', __dir__)).sort.each(&method(:require))
|