puree 1.3.0 → 1.4.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +4 -0
- data/lib/puree.rb +9 -5
- data/lib/puree/extractor/project.rb +4 -1
- data/lib/puree/extractor/publication.rb +2 -0
- data/lib/puree/model/external_organisation_header.rb +26 -0
- data/lib/puree/model/project.rb +10 -0
- data/lib/puree/model/publication.rb +6 -0
- data/lib/puree/version.rb +1 -1
- data/lib/puree/xml_extractor/dataset.rb +1 -15
- data/lib/puree/xml_extractor/mixins/associated_mixin.rb +27 -0
- data/lib/puree/xml_extractor/mixins/external_organisations_mixin.rb +18 -0
- data/lib/puree/xml_extractor/project.rb +9 -0
- data/lib/puree/xml_extractor/publication.rb +2 -0
- data/lib/puree/xml_extractor/shared.rb +17 -0
- data/spec/resource/project_http_spec.rb +13 -0
- data/spec/resource/publication_http_spec.rb +40 -4
- data/spec/spec_helper.rb +9 -5
- metadata +5 -2
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 3f4b875fcc3fc9d0c03faf62d20332c2be02b8c7
|
4
|
+
data.tar.gz: eacdf05a1625cda281f653009389b16dc2d10822
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: d644dbf426c9137f442e14b385b9535d9f65e4fd90c5bc24eb1417416d169b867f6aa828d16d44ffb95e7b39e2a49462a95aaff79b322c5385d30ae493afe25c
|
7
|
+
data.tar.gz: 9f40cdd393fb7606d55431c9de9ff68db7ee2341234997e48db252ec240a6fc50a55282ffb120820485379c69aa5e3d83f7f35d0c9c50dfc48adfe875250e508
|
data/CHANGELOG.md
CHANGED
@@ -5,6 +5,10 @@ This project adheres to [Semantic Versioning](http://semver.org/).
|
|
5
5
|
## Unreleased
|
6
6
|
- Metadata: activity?, clipping?, externalPerson?
|
7
7
|
|
8
|
+
## 1.4.0 - 2017-09-01
|
9
|
+
- Project - external_organisations, funded.
|
10
|
+
- Publication - associated, external_organisations.
|
11
|
+
|
8
12
|
## 1.3.0 - 2017-04-03
|
9
13
|
### Added
|
10
14
|
- Publication type - JournalArticle, Paper, ConferencePaper.
|
data/lib/puree.rb
CHANGED
@@ -6,6 +6,14 @@ require 'puree/version'
|
|
6
6
|
require 'puree/util/date'
|
7
7
|
require 'puree/util/string'
|
8
8
|
|
9
|
+
require 'puree/xml_extractor/mixins/associated_mixin'
|
10
|
+
require 'puree/xml_extractor/mixins/doi_mixin'
|
11
|
+
require 'puree/xml_extractor/mixins/event_mixin'
|
12
|
+
require 'puree/xml_extractor/mixins/external_organisations_mixin'
|
13
|
+
require 'puree/xml_extractor/mixins/pages_mixin'
|
14
|
+
require 'puree/xml_extractor/mixins/page_range_mixin'
|
15
|
+
require 'puree/xml_extractor/mixins/peer_reviewed_mixin'
|
16
|
+
|
9
17
|
require 'puree/xml_extractor/shared'
|
10
18
|
require 'puree/xml_extractor/base'
|
11
19
|
require 'puree/xml_extractor/collection'
|
@@ -18,11 +26,6 @@ require 'puree/xml_extractor/organisation'
|
|
18
26
|
require 'puree/xml_extractor/person'
|
19
27
|
require 'puree/xml_extractor/project'
|
20
28
|
|
21
|
-
require 'puree/xml_extractor/mixins/doi_mixin'
|
22
|
-
require 'puree/xml_extractor/mixins/event_mixin'
|
23
|
-
require 'puree/xml_extractor/mixins/pages_mixin'
|
24
|
-
require 'puree/xml_extractor/mixins/page_range_mixin'
|
25
|
-
require 'puree/xml_extractor/mixins/peer_reviewed_mixin'
|
26
29
|
require 'puree/xml_extractor/publication'
|
27
30
|
require 'puree/xml_extractor/thesis'
|
28
31
|
require 'puree/xml_extractor/doctoral_thesis'
|
@@ -71,6 +74,7 @@ require 'puree/model/spatial_point'
|
|
71
74
|
require 'puree/model/address'
|
72
75
|
require 'puree/model/copyright_license'
|
73
76
|
require 'puree/model/event_header'
|
77
|
+
require 'puree/model/external_organisation_header'
|
74
78
|
require 'puree/model/file'
|
75
79
|
require 'puree/model/legal_condition'
|
76
80
|
require 'puree/model/organisation_header'
|
@@ -9,7 +9,7 @@ module Puree
|
|
9
9
|
# @option (see Puree::Extractor::Resource#initialize)
|
10
10
|
def initialize(config)
|
11
11
|
super
|
12
|
-
@latest_api = false #
|
12
|
+
@latest_api = false # current API does not return person roles
|
13
13
|
setup :project
|
14
14
|
end
|
15
15
|
|
@@ -18,7 +18,10 @@ module Puree
|
|
18
18
|
def combine_metadata
|
19
19
|
super
|
20
20
|
@model.acronym = @extractor.acronym
|
21
|
+
# @model.associated = @extractor.associated # not present in stable API
|
21
22
|
@model.description = @extractor.description
|
23
|
+
@model.external_organisations = @extractor.external_organisations
|
24
|
+
@model.funded = @extractor.funded?
|
22
25
|
@model.organisations = @extractor.organisations
|
23
26
|
@model.owner = @extractor.owner
|
24
27
|
@model.persons_internal = @extractor.persons_internal
|
@@ -17,10 +17,12 @@ module Puree
|
|
17
17
|
def combine_metadata
|
18
18
|
super
|
19
19
|
|
20
|
+
@model.associated = @extractor.associated
|
20
21
|
@model.bibliographical_note = @extractor.bibliographical_note
|
21
22
|
@model.category = @extractor.category
|
22
23
|
@model.description = @extractor.description
|
23
24
|
@model.dois = @extractor.dois
|
25
|
+
@model.external_organisations = @extractor.external_organisations
|
24
26
|
@model.files = @extractor.files
|
25
27
|
@model.keywords = @extractor.keywords
|
26
28
|
@model.language = @extractor.language
|
@@ -0,0 +1,26 @@
|
|
1
|
+
module Puree
|
2
|
+
module Model
|
3
|
+
|
4
|
+
# A minimal representation of an ExternalOrganisation.
|
5
|
+
#
|
6
|
+
class ExternalOrganisationHeader < Puree::Model::Structure
|
7
|
+
|
8
|
+
# @return [String, nil]
|
9
|
+
attr_reader :uuid
|
10
|
+
|
11
|
+
# @return [String, nil]
|
12
|
+
attr_reader :name
|
13
|
+
|
14
|
+
# @param v [String]
|
15
|
+
def uuid=(v)
|
16
|
+
@uuid = v if v && !v.empty?
|
17
|
+
end
|
18
|
+
|
19
|
+
# @param v [String]
|
20
|
+
def name=(v)
|
21
|
+
@name = v if v && !v.empty?
|
22
|
+
end
|
23
|
+
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
data/lib/puree/model/project.rb
CHANGED
@@ -8,9 +8,19 @@ module Puree
|
|
8
8
|
# @return [String, nil]
|
9
9
|
attr_accessor :acronym
|
10
10
|
|
11
|
+
# not present in stable API
|
12
|
+
# @return [Array<Puree::Model::RelatedContentHeader>]
|
13
|
+
# attr_accessor :associated
|
14
|
+
|
11
15
|
# @return [String, nil]
|
12
16
|
attr_accessor :description
|
13
17
|
|
18
|
+
# @return [Array<Puree::Model::ExternalOrganisationHeader>]
|
19
|
+
attr_accessor :external_organisations
|
20
|
+
|
21
|
+
# @return [Boolean]
|
22
|
+
attr_accessor :funded
|
23
|
+
|
14
24
|
# @return [Array<Puree::Model::OrganisationHeader>]
|
15
25
|
attr_accessor :organisations
|
16
26
|
|
@@ -5,6 +5,9 @@ module Puree
|
|
5
5
|
#
|
6
6
|
class Publication < Resource
|
7
7
|
|
8
|
+
# @return [Array<Puree::Model::RelatedContentHeader>]
|
9
|
+
attr_accessor :associated
|
10
|
+
|
8
11
|
# @return [String, nil]
|
9
12
|
attr_accessor :bibliographical_note
|
10
13
|
|
@@ -17,6 +20,9 @@ module Puree
|
|
17
20
|
# @return [Array<String>]
|
18
21
|
attr_accessor :dois
|
19
22
|
|
23
|
+
# @return [Array<Puree::Model::ExternalOrganisationHeader>]
|
24
|
+
attr_accessor :external_organisations
|
25
|
+
|
20
26
|
# @return [Array<Puree::Model::File>]
|
21
27
|
attr_accessor :files
|
22
28
|
|
data/lib/puree/version.rb
CHANGED
@@ -5,6 +5,7 @@ module Puree
|
|
5
5
|
# Dataset XML extractor.
|
6
6
|
#
|
7
7
|
class Dataset < Puree::XMLExtractor::Resource
|
8
|
+
include Puree::XMLExtractor::AssociatedMixin
|
8
9
|
|
9
10
|
def initialize(xml:)
|
10
11
|
super
|
@@ -17,21 +18,6 @@ module Puree
|
|
17
18
|
xpath_query_for_single_value '/openAccessPermission/term/localizedString'
|
18
19
|
end
|
19
20
|
|
20
|
-
# Combines projects and publications
|
21
|
-
# @return [Array<Puree::Model::RelatedContentHeader>]
|
22
|
-
def associated
|
23
|
-
xpath_result = xpath_query '/associatedContent/relatedContent'
|
24
|
-
data_arr = []
|
25
|
-
xpath_result.each { |i|
|
26
|
-
related = Puree::Model::RelatedContentHeader.new
|
27
|
-
related.type = i.xpath('typeClassification').text.strip
|
28
|
-
related.title = i.xpath('title').text.strip
|
29
|
-
related.uuid = i.attr('uuid').strip
|
30
|
-
data_arr << related
|
31
|
-
}
|
32
|
-
data_arr.uniq { |d| d.uuid }
|
33
|
-
end
|
34
|
-
|
35
21
|
# Date made available
|
36
22
|
# @return [Time, nil]
|
37
23
|
def available
|
@@ -0,0 +1,27 @@
|
|
1
|
+
module Puree
|
2
|
+
|
3
|
+
module XMLExtractor
|
4
|
+
|
5
|
+
# Associated extractor mixin.
|
6
|
+
#
|
7
|
+
module AssociatedMixin
|
8
|
+
|
9
|
+
# Combines projects and publications
|
10
|
+
# @return [Array<Puree::Model::RelatedContentHeader>]
|
11
|
+
def associated
|
12
|
+
xpath_result = xpath_query '/associatedContent/relatedContent'
|
13
|
+
data_arr = []
|
14
|
+
xpath_result.each { |i|
|
15
|
+
related = Puree::Model::RelatedContentHeader.new
|
16
|
+
related.type = i.xpath('typeClassification').text.strip
|
17
|
+
related.title = i.xpath('title').text.strip
|
18
|
+
related.uuid = i.attr('uuid').strip
|
19
|
+
data_arr << related
|
20
|
+
}
|
21
|
+
data_arr.uniq { |d| d.uuid }
|
22
|
+
end
|
23
|
+
|
24
|
+
end
|
25
|
+
|
26
|
+
end
|
27
|
+
end
|
@@ -0,0 +1,18 @@
|
|
1
|
+
module Puree
|
2
|
+
|
3
|
+
module XMLExtractor
|
4
|
+
|
5
|
+
# External organisations extractor mixin.
|
6
|
+
#
|
7
|
+
module ExternalOrganisationsMixin
|
8
|
+
|
9
|
+
# @return [Array<Puree::Model::ExternalOrganisationHeader>]
|
10
|
+
def external_organisations
|
11
|
+
xpath_result = xpath_query '/associatedExternalOrganisations/externalOrganisation'
|
12
|
+
Puree::XMLExtractor::Shared.external_organisation_multi_header xpath_result
|
13
|
+
end
|
14
|
+
|
15
|
+
end
|
16
|
+
|
17
|
+
end
|
18
|
+
end
|
@@ -5,6 +5,8 @@ module Puree
|
|
5
5
|
# Project XML extractor.
|
6
6
|
#
|
7
7
|
class Project < Puree::XMLExtractor::Resource
|
8
|
+
# include Puree::XMLExtractor::AssociatedMixin # not present in stable API
|
9
|
+
include Puree::XMLExtractor::ExternalOrganisationsMixin
|
8
10
|
|
9
11
|
def initialize(xml:)
|
10
12
|
super
|
@@ -21,6 +23,13 @@ module Puree
|
|
21
23
|
xpath_query_for_single_value '/description/localizedString'
|
22
24
|
end
|
23
25
|
|
26
|
+
# @return [Boolean]
|
27
|
+
def funded?
|
28
|
+
xpath_result = xpath_query_for_single_value '/type'
|
29
|
+
return false if xpath_result.downcase.include? 'nonfunded'
|
30
|
+
true
|
31
|
+
end
|
32
|
+
|
24
33
|
# @return [Array<Puree::Model::OrganisationHeader>]
|
25
34
|
def organisations
|
26
35
|
xpath_result = xpath_query '/organisations/association/organisation'
|
@@ -6,6 +6,23 @@ module Puree
|
|
6
6
|
#
|
7
7
|
module Shared
|
8
8
|
|
9
|
+
# @return [Puree::Model::ExternalOrganisationHeader]
|
10
|
+
def self.external_organisation_header(nokogiri_xml_element)
|
11
|
+
h = Puree::Model::ExternalOrganisationHeader.new
|
12
|
+
h.uuid = nokogiri_xml_element.xpath('@uuid').text.strip
|
13
|
+
h.name = nokogiri_xml_element.xpath('name').text.strip
|
14
|
+
h
|
15
|
+
end
|
16
|
+
|
17
|
+
# @return [Array<Puree::Model::ExternalOrganisationHeader>]
|
18
|
+
def self.external_organisation_multi_header(nokogiri_xml_nodeset)
|
19
|
+
data = []
|
20
|
+
nokogiri_xml_nodeset.each do |i|
|
21
|
+
data << external_organisation_header(i)
|
22
|
+
end
|
23
|
+
data.uniq { |d| d.uuid }
|
24
|
+
end
|
25
|
+
|
9
26
|
# @return [Puree::Model::OrganisationHeader]
|
10
27
|
def self.organisation_header(nokogiri_xml_element)
|
11
28
|
h = Puree::Model::OrganisationHeader.new
|
@@ -23,10 +23,23 @@ describe 'Project' do
|
|
23
23
|
expect(@p.acronym).to be_a String if @p.acronym
|
24
24
|
end
|
25
25
|
|
26
|
+
# not present in stable API
|
27
|
+
# it '#associated' do
|
28
|
+
# expect(@p.associated).to all( be_a Puree::Model::RelatedContentHeader )
|
29
|
+
# end
|
30
|
+
|
26
31
|
it '#description' do
|
27
32
|
expect(@p.description).to be_a String if @p.description
|
28
33
|
end
|
29
34
|
|
35
|
+
it '#external_organisations' do
|
36
|
+
expect(@p.external_organisations).to all( be_a Puree::Model::ExternalOrganisationHeader )
|
37
|
+
end
|
38
|
+
|
39
|
+
it '#funded' do
|
40
|
+
expect(@p.funded).to be(true).or be(false)
|
41
|
+
end
|
42
|
+
|
30
43
|
it '#organisations' do
|
31
44
|
expect(@p.organisations).to all( be_a Puree::Model::OrganisationHeader )
|
32
45
|
end
|
@@ -17,6 +17,14 @@ describe 'Publication' do
|
|
17
17
|
expect(@p).to be_a Puree::Model::Publication
|
18
18
|
end
|
19
19
|
|
20
|
+
it '#associated' do
|
21
|
+
expect(@p.associated).to all( be_a Puree::Model::RelatedContentHeader )
|
22
|
+
end
|
23
|
+
|
24
|
+
it '#bibliographical_note' do
|
25
|
+
expect(@p.bibliographical_note).to be_a String if @p.bibliographical_note
|
26
|
+
end
|
27
|
+
|
20
28
|
it '#category' do
|
21
29
|
expect(@p.category).to be_a String if @p.category
|
22
30
|
end
|
@@ -25,10 +33,30 @@ describe 'Publication' do
|
|
25
33
|
expect(@p.description).to be_a String if @p.description
|
26
34
|
end
|
27
35
|
|
36
|
+
it '#dois' do
|
37
|
+
expect(@p.dois).to all( be_a String )
|
38
|
+
end
|
39
|
+
|
40
|
+
it '#external_organisations' do
|
41
|
+
expect(@p.external_organisations).to all( be_a Puree::Model::ExternalOrganisationHeader )
|
42
|
+
end
|
43
|
+
|
28
44
|
it '#files' do
|
29
45
|
expect(@p.files).to all( be_a Puree::Model::File )
|
30
46
|
end
|
31
47
|
|
48
|
+
it '#keywords' do
|
49
|
+
expect(@p.keywords).to all( be_a String )
|
50
|
+
end
|
51
|
+
|
52
|
+
it '#language' do
|
53
|
+
expect(@p.language).to be_a String if @p.language
|
54
|
+
end
|
55
|
+
|
56
|
+
it '#links' do
|
57
|
+
expect(@p.links).to all( be_a String )
|
58
|
+
end
|
59
|
+
|
32
60
|
it '#organisations' do
|
33
61
|
expect(@p.organisations).to all( be_a Puree::Model::OrganisationHeader )
|
34
62
|
end
|
@@ -49,6 +77,14 @@ describe 'Publication' do
|
|
49
77
|
expect(@p.persons_other).to all( be_a Puree::Model::EndeavourPerson )
|
50
78
|
end
|
51
79
|
|
80
|
+
it '#publication_place' do
|
81
|
+
expect(@p.publication_place).to be_a String if @p.publication_place
|
82
|
+
end
|
83
|
+
|
84
|
+
it '#publisher' do
|
85
|
+
expect(@p.publisher).to be_a String if @p.publisher
|
86
|
+
end
|
87
|
+
|
52
88
|
it '#statuses' do
|
53
89
|
expect(@p.statuses).to all( be_a Puree::Model::PublicationStatus )
|
54
90
|
end
|
@@ -61,14 +97,14 @@ describe 'Publication' do
|
|
61
97
|
expect(@p.title).to be_a String if @p.title
|
62
98
|
end
|
63
99
|
|
64
|
-
it '#translated_title' do
|
65
|
-
expect(@p.translated_title).to be_a String if @p.translated_title
|
66
|
-
end
|
67
|
-
|
68
100
|
it '#translated_subtitle' do
|
69
101
|
expect(@p.translated_subtitle).to be_a String if @p.translated_subtitle
|
70
102
|
end
|
71
103
|
|
104
|
+
it '#translated_title' do
|
105
|
+
expect(@p.translated_title).to be_a String if @p.translated_title
|
106
|
+
end
|
107
|
+
|
72
108
|
it '#type' do
|
73
109
|
expect(@p.type).to be_a String if @p.type
|
74
110
|
end
|
data/spec/spec_helper.rb
CHANGED
@@ -6,6 +6,14 @@ require 'puree/version'
|
|
6
6
|
require 'puree/util/date'
|
7
7
|
require 'puree/util/string'
|
8
8
|
|
9
|
+
require 'puree/xml_extractor/mixins/associated_mixin'
|
10
|
+
require 'puree/xml_extractor/mixins/doi_mixin'
|
11
|
+
require 'puree/xml_extractor/mixins/event_mixin'
|
12
|
+
require 'puree/xml_extractor/mixins/external_organisations_mixin'
|
13
|
+
require 'puree/xml_extractor/mixins/pages_mixin'
|
14
|
+
require 'puree/xml_extractor/mixins/page_range_mixin'
|
15
|
+
require 'puree/xml_extractor/mixins/peer_reviewed_mixin'
|
16
|
+
|
9
17
|
require 'puree/xml_extractor/shared'
|
10
18
|
require 'puree/xml_extractor/base'
|
11
19
|
require 'puree/xml_extractor/collection'
|
@@ -18,11 +26,6 @@ require 'puree/xml_extractor/organisation'
|
|
18
26
|
require 'puree/xml_extractor/person'
|
19
27
|
require 'puree/xml_extractor/project'
|
20
28
|
|
21
|
-
require 'puree/xml_extractor/mixins/doi_mixin'
|
22
|
-
require 'puree/xml_extractor/mixins/event_mixin'
|
23
|
-
require 'puree/xml_extractor/mixins/pages_mixin'
|
24
|
-
require 'puree/xml_extractor/mixins/page_range_mixin'
|
25
|
-
require 'puree/xml_extractor/mixins/peer_reviewed_mixin'
|
26
29
|
require 'puree/xml_extractor/publication'
|
27
30
|
require 'puree/xml_extractor/thesis'
|
28
31
|
require 'puree/xml_extractor/doctoral_thesis'
|
@@ -71,6 +74,7 @@ require 'puree/model/spatial_point'
|
|
71
74
|
require 'puree/model/address'
|
72
75
|
require 'puree/model/copyright_license'
|
73
76
|
require 'puree/model/event_header'
|
77
|
+
require 'puree/model/external_organisation_header'
|
74
78
|
require 'puree/model/file'
|
75
79
|
require 'puree/model/legal_condition'
|
76
80
|
require 'puree/model/organisation_header'
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: puree
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 1.
|
4
|
+
version: 1.4.0
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Adrian Albin-Clark
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2017-
|
11
|
+
date: 2017-09-01 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: http
|
@@ -100,6 +100,7 @@ files:
|
|
100
100
|
- lib/puree/model/endeavour_person.rb
|
101
101
|
- lib/puree/model/event.rb
|
102
102
|
- lib/puree/model/event_header.rb
|
103
|
+
- lib/puree/model/external_organisation_header.rb
|
103
104
|
- lib/puree/model/file.rb
|
104
105
|
- lib/puree/model/helper/validation.rb
|
105
106
|
- lib/puree/model/journal.rb
|
@@ -140,8 +141,10 @@ files:
|
|
140
141
|
- lib/puree/xml_extractor/journal.rb
|
141
142
|
- lib/puree/xml_extractor/journal_article.rb
|
142
143
|
- lib/puree/xml_extractor/masters_thesis.rb
|
144
|
+
- lib/puree/xml_extractor/mixins/associated_mixin.rb
|
143
145
|
- lib/puree/xml_extractor/mixins/doi_mixin.rb
|
144
146
|
- lib/puree/xml_extractor/mixins/event_mixin.rb
|
147
|
+
- lib/puree/xml_extractor/mixins/external_organisations_mixin.rb
|
145
148
|
- lib/puree/xml_extractor/mixins/page_range_mixin.rb
|
146
149
|
- lib/puree/xml_extractor/mixins/pages_mixin.rb
|
147
150
|
- lib/puree/xml_extractor/mixins/peer_reviewed_mixin.rb
|