cirneco 0.9.13 → 0.9.14
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/Gemfile.lock +2 -2
- data/lib/cirneco/data_center.rb +5 -0
- data/lib/cirneco/doi.rb +6 -1
- data/lib/cirneco/file_utils.rb +368 -0
- data/lib/cirneco/media.rb +3 -1
- data/lib/cirneco/metadata.rb +2 -0
- data/lib/cirneco/utils.rb +5 -379
- data/lib/cirneco/version.rb +1 -1
- data/lib/cirneco/work.rb +2 -0
- data/spec/api_spec.rb +3 -3
- data/spec/doi_spec.rb +7 -7
- data/spec/file_utils_spec.rb +203 -0
- data/spec/fixtures/cool-dois-minted/index.html +1 -1
- data/spec/fixtures/cool-dois-minted.html.md +1 -1
- data/spec/fixtures/cool-dois.html.md +1 -1
- data/spec/fixtures/index-minted.html +1 -1
- data/spec/fixtures/index.html +2 -2
- data/spec/fixtures/index.html.erb +1 -1
- data/spec/fixtures/vcr_cassettes/Cirneco_DataCenter/jats/should_generate_jats_for_all_urls.yml +4 -4
- data/spec/fixtures/vcr_cassettes/Cirneco_DataCenter/mint_and_hide_DOIs/should_hide_for_all_urls.yml +3 -44
- data/spec/fixtures/vcr_cassettes/Cirneco_DataCenter/mint_and_hide_DOIs/should_mint_and_hide_for_all_urls.yml +4 -374
- data/spec/fixtures/vcr_cassettes/Cirneco_DataCenter/mint_and_hide_DOIs/should_mint_for_all_urls.yml +4 -4
- data/spec/fixtures/vcr_cassettes/Cirneco_Doi/jats/writes_jats_for_list_of_urls.yml +4 -4
- data/spec/fixtures/vcr_cassettes/Cirneco_Doi/mint_and_hide_DOIs/hides_dois_for_list_of_urls.yml +4 -47
- data/spec/fixtures/vcr_cassettes/Cirneco_Doi/mint_and_hide_DOIs/mints_and_hides_dois_for_list_of_urls.yml +4 -374
- data/spec/fixtures/vcr_cassettes/Cirneco_Doi/mint_and_hide_DOIs/mints_dois_for_list_of_urls.yml +4 -4
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/DOI_API/get/should_get_all_dois.yml +16 -577
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/DOI_API/get/should_get_doi.yml +15 -58
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/DOI_API/get/should_get_doi_not_found.yml +16 -57
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/DOI_API/get/username_missing.yml +16 -16
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/DOI_API/put/should_put_doi.yml +16 -63
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/Media_API/get/should_get_media.yml +16 -61
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/Media_API/post/should_post_media.yml +17 -62
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/Metadata_API/delete/should_delete_metadata.yml +17 -60
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/Metadata_API/get/should_get_metadata.yml +17 -104
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/Metadata_API/post/should_post_metadata.yml +17 -108
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/media/includes_media.yml +17 -17
- data/spec/fixtures/vcr_cassettes/Cirneco_Work/schema/BlogPosting.yml +15 -15
- data/spec/utils_spec.rb +9 -186
- data/spec/work_spec.rb +1 -1
- metadata +4 -12
- data/spec/fixtures/vcr_cassettes/Cirneco_DataCenter/get/should_get_all_dois_by_prefix.yml +0 -562
- data/spec/fixtures/vcr_cassettes/Cirneco_DataCenter/mint_and_hide_DOIs/should_hide_for_url.yml +0 -46
- data/spec/fixtures/vcr_cassettes/Cirneco_DataCenter/mint_and_hide_DOIs/should_hide_metadata_for_work.yml +0 -46
- data/spec/fixtures/vcr_cassettes/Cirneco_DataCenter/mint_and_hide_DOIs/should_mint_and_hide_for_url.yml +0 -182
- data/spec/fixtures/vcr_cassettes/Cirneco_DataCenter/mint_and_hide_DOIs/should_post_metadata_for_work.yml +0 -139
- data/spec/fixtures/vcr_cassettes/Cirneco_Doi/MDS_DOI_API/get/should_get_all_dois.yml +0 -564
- data/spec/fixtures/vcr_cassettes/Cirneco_Doi/MDS_DOI_API/get/should_get_doi.yml +0 -46
- data/spec/fixtures/vcr_cassettes/Cirneco_Doi/MDS_DOI_API/put/should_put_doi.yml +0 -50
- data/spec/fixtures/vcr_cassettes/Cirneco_Doi/mint_and_hide_DOIs/hides_a_doi.yml +0 -46
- data/spec/fixtures/vcr_cassettes/Cirneco_Doi/mint_and_hide_DOIs/mints_and_hides_a_doi.yml +0 -182
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: ca288d421311719334b7a9d38849732549f5f2d3
|
4
|
+
data.tar.gz: 5d9b99265570d9ab485b463e9287f9d5e16fdd01
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: c604f57226dbc7ff543575bf1226605ac887bfb070d0789755fcbe6f0d146d90370798f3618f82e9ec35002a0b520d6a84fb3ca6712698acfec0e4fe6696f305
|
7
|
+
data.tar.gz: 93f67ce51ff86fe67be5c24f61c9bae6a22318e747015e12f6a87978ef3179554bd88337f47c5c245a9570efd91844d32a4e0c1105555cbfeedcd23458f89263
|
data/Gemfile.lock
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
PATH
|
2
2
|
remote: .
|
3
3
|
specs:
|
4
|
-
cirneco (0.9.
|
4
|
+
cirneco (0.9.14)
|
5
5
|
activesupport (>= 4.2.5, < 6)
|
6
6
|
base32-crockford-checksum (~> 0.2.2)
|
7
7
|
bergamasco (~> 0.3)
|
@@ -36,7 +36,7 @@ GEM
|
|
36
36
|
safe_yaml (~> 1.0, >= 1.0.4)
|
37
37
|
bibtex-ruby (4.4.4)
|
38
38
|
latex-decode (~> 0.0)
|
39
|
-
bolognese (0.9.
|
39
|
+
bolognese (0.9.34)
|
40
40
|
activesupport (>= 4.2.5, < 6)
|
41
41
|
benchmark_methods (~> 0.7)
|
42
42
|
bibtex-ruby (~> 4.1)
|
data/lib/cirneco/data_center.rb
CHANGED
@@ -1,8 +1,10 @@
|
|
1
1
|
require 'active_support/all'
|
2
2
|
require 'nokogiri'
|
3
|
+
require 'bolognese'
|
3
4
|
|
4
5
|
require_relative 'api'
|
5
6
|
require_relative 'utils'
|
7
|
+
require_relative 'file_utils'
|
6
8
|
require_relative 'base'
|
7
9
|
|
8
10
|
module Cirneco
|
@@ -10,6 +12,9 @@ module Cirneco
|
|
10
12
|
include Cirneco::Base
|
11
13
|
include Cirneco::Api
|
12
14
|
include Cirneco::Utils
|
15
|
+
include Cirneco::FileUtils
|
16
|
+
include Bolognese::Utils
|
17
|
+
include Bolognese::DoiUtils
|
13
18
|
|
14
19
|
attr_accessor :prefix, :username, :password
|
15
20
|
|
data/lib/cirneco/doi.rb
CHANGED
@@ -1,7 +1,9 @@
|
|
1
|
-
require
|
1
|
+
require 'thor'
|
2
|
+
require 'bolognese'
|
2
3
|
|
3
4
|
require_relative 'api'
|
4
5
|
require_relative 'utils'
|
6
|
+
require_relative 'file_utils'
|
5
7
|
require_relative 'base'
|
6
8
|
|
7
9
|
module Cirneco
|
@@ -9,6 +11,9 @@ module Cirneco
|
|
9
11
|
include Cirneco::Base
|
10
12
|
include Cirneco::Api
|
11
13
|
include Cirneco::Utils
|
14
|
+
include Cirneco::FileUtils
|
15
|
+
include Bolognese::Utils
|
16
|
+
include Bolognese::DoiUtils
|
12
17
|
|
13
18
|
desc "get DOI", "get handle url for DOI"
|
14
19
|
method_option :username, :default => ENV['MDS_USERNAME']
|
@@ -0,0 +1,368 @@
|
|
1
|
+
require 'bergamasco'
|
2
|
+
require 'open-uri'
|
3
|
+
require 'time'
|
4
|
+
|
5
|
+
module Cirneco
|
6
|
+
module FileUtils
|
7
|
+
JATS_SCHEMA = File.expand_path("../../../resources/jats-1.1/JATS-journalpublishing1.xsd", __FILE__)
|
8
|
+
|
9
|
+
LICENSES = {
|
10
|
+
"https://creativecommons.org/licenses/by/4.0/" => "Creative Commons Attribution (CC-BY 4.0)",
|
11
|
+
"https://creativecommons.org/publicdomain/zero/1.0/" => "Creative Commons Public Domain Dedication (CC0 1.0)"
|
12
|
+
}
|
13
|
+
|
14
|
+
def get_accession_number(filepath)
|
15
|
+
metadata = Bergamasco::Markdown.read_yaml_for_doi_metadata(filepath)
|
16
|
+
metadata.fetch("accession_number", nil).to_s.scan(/\d+/).first.to_i
|
17
|
+
end
|
18
|
+
|
19
|
+
def get_all_accession_numbers(folderpath)
|
20
|
+
Dir.glob("#{folderpath}/*.md").sort.map do |filepath|
|
21
|
+
get_accession_number(filepath)
|
22
|
+
end.select { |a| a > 0 }.sort
|
23
|
+
end
|
24
|
+
|
25
|
+
def update_accession_number(filepath, options={})
|
26
|
+
filename = File.basename(filepath)
|
27
|
+
return "File #{filename} ignored: not a markdown or html file" unless %w(.md .html .erb).include?(File.extname(filepath))
|
28
|
+
|
29
|
+
old_metadata = Bergamasco::Markdown.read_yaml_for_doi_metadata(filepath)
|
30
|
+
return "File #{filename} ignored: no yaml front matter" unless old_metadata.present?
|
31
|
+
|
32
|
+
return "File #{filename} ignored: no empty accession_number" if options[:opt_in] && !old_metadata.key?("accession_number")
|
33
|
+
|
34
|
+
return "Accession number #{old_metadata["accession_number"]} not changed for #{filename}" if old_metadata["accession_number"]
|
35
|
+
|
36
|
+
if old_metadata["doi"].present?
|
37
|
+
number = decode_doi(old_metadata["doi"])
|
38
|
+
options[:number] = number if number > 0
|
39
|
+
else
|
40
|
+
folderpath = File.dirname(filepath)
|
41
|
+
options[:registered_numbers] = get_all_accession_numbers(folderpath)
|
42
|
+
end
|
43
|
+
|
44
|
+
accession_number = generate_accession_number(options)
|
45
|
+
|
46
|
+
new_metadata = Bergamasco::Markdown.update_file(filepath, "accession_number" => accession_number)
|
47
|
+
"Accession number #{new_metadata["accession_number"]} generated for #{filename}"
|
48
|
+
end
|
49
|
+
|
50
|
+
def update_all_accession_numbers(folderpath, options={})
|
51
|
+
Dir.glob("#{folderpath}/*.md").sort.map do |filepath|
|
52
|
+
update_accession_number(filepath, options)
|
53
|
+
end
|
54
|
+
end
|
55
|
+
|
56
|
+
# fetch schema.org metadata in JSON-LD format to mint DOI
|
57
|
+
def mint_doi_for_url(url, options={})
|
58
|
+
filename, build_path, source_path = filepath_from_url(url, options)
|
59
|
+
|
60
|
+
json = get_json_ld_from_work(build_path, options)
|
61
|
+
metadata = JSON.parse(json)
|
62
|
+
return "DOI #{metadata["doi"]} not changed for #{filename}" if metadata["doi"] && metadata["dateModified"] == metadata["datePublished"] && options[:force].blank?
|
63
|
+
|
64
|
+
prefix = options[:prefix] || ENV['PREFIX']
|
65
|
+
doi = encode_doi(prefix, number: metadata["alternateName"]) if metadata["doi"].blank?
|
66
|
+
|
67
|
+
new_metadata = Bergamasco::Markdown.update_file(source_path, "doi" => doi, "date" => metadata["dateModified"])
|
68
|
+
"DOI #{new_metadata["doi"]} minted for #{filename}"
|
69
|
+
end
|
70
|
+
|
71
|
+
# fetch schema.org metadata in JSON-LD format to mint DOI
|
72
|
+
def mint_and_hide_doi_for_url(url, options={})
|
73
|
+
filename, build_path, source_path = filepath_from_url(url, options)
|
74
|
+
|
75
|
+
json = get_json_ld_from_work(build_path, options)
|
76
|
+
metadata = JSON.parse(json)
|
77
|
+
metadata["doi"] = doi_from_url(metadata["@id"])
|
78
|
+
return "DOI #{metadata["doi"]} not changed for #{filename}" if metadata["doi"] && metadata["dateModified"] == metadata["datePublished"] && options[:force].blank?
|
79
|
+
|
80
|
+
prefix = options[:prefix] || ENV['PREFIX']
|
81
|
+
metadata["doi"] = encode_doi(prefix, number: metadata["alternateName"]) if metadata["doi"].blank?
|
82
|
+
|
83
|
+
response = post_metadata_for_work(json, options.merge(doi: metadata["doi"]))
|
84
|
+
return "Errors for DOI #{metadata["doi"]}: #{response.body['errors'].first['title']}\n" if response.body['errors'].present?
|
85
|
+
|
86
|
+
response = hide_metadata_for_work(json, options.merge(doi: metadata["doi"]))
|
87
|
+
return "Errors for DOI #{metadata["doi"]}: #{response.body['errors'].first['title']}\n" if response.body['errors'].present?
|
88
|
+
|
89
|
+
new_metadata = Bergamasco::Markdown.update_file(source_path, "published" => false)
|
90
|
+
"DOI #{metadata["doi"]} minted and hidden for #{filename}"
|
91
|
+
end
|
92
|
+
|
93
|
+
# fetch schema.org metadata in JSON-LD format to mint DOI
|
94
|
+
# DOIs are never deleted, but we can remove the metadata from the DataCite index
|
95
|
+
def hide_doi_for_url(url, options={})
|
96
|
+
filename, build_path, source_path = filepath_from_url(url, options)
|
97
|
+
|
98
|
+
json = get_json_ld_from_work(build_path, options)
|
99
|
+
metadata = JSON.parse(json)
|
100
|
+
metadata["doi"] = doi_from_url(metadata["@id"])
|
101
|
+
return "No DOI for #{filename}" unless metadata["doi"]
|
102
|
+
return "DOI #{metadata["doi"]} not active for #{filename}" unless metadata["datePublished"] || options[:force].present?
|
103
|
+
|
104
|
+
response = hide_metadata_for_work(json, options)
|
105
|
+
return "Errors for DOI #{metadata["doi"]}: #{response.body['errors'].first['title']}\n" if response.body['errors'].present?
|
106
|
+
|
107
|
+
new_metadata = Bergamasco::Markdown.update_file(source_path, "published" => false)
|
108
|
+
"DOI #{metadata["doi"]} hidden for #{filename}"
|
109
|
+
end
|
110
|
+
|
111
|
+
def mint_dois_for_all_urls(url, options={})
|
112
|
+
urls = get_urls_for_works(url)
|
113
|
+
urls.map do |u|
|
114
|
+
mint_doi_for_url(u, options)
|
115
|
+
end.join("\n")
|
116
|
+
end
|
117
|
+
|
118
|
+
def mint_and_hide_dois_for_all_urls(url, options={})
|
119
|
+
urls = get_urls_for_works(url)
|
120
|
+
urls.map do |u|
|
121
|
+
mint_and_hide_doi_for_url(u, options)
|
122
|
+
end.join("\n")
|
123
|
+
end
|
124
|
+
|
125
|
+
def hide_dois_for_all_urls(url, options={})
|
126
|
+
urls = get_urls_for_works(url)
|
127
|
+
urls.map do |u|
|
128
|
+
hide_doi_for_url(u, options)
|
129
|
+
end.join("\n")
|
130
|
+
end
|
131
|
+
|
132
|
+
def get_urls_for_works(url)
|
133
|
+
doc = Nokogiri::HTML(open(url))
|
134
|
+
json = doc.at_xpath("//script[@type='application/ld+json']")
|
135
|
+
return [url] unless json.present?
|
136
|
+
|
137
|
+
metadata = ActiveSupport::JSON.decode(json.text)
|
138
|
+
metadata.fetch("hasPart", []).map { |p| p["@id"] } + [url]
|
139
|
+
end
|
140
|
+
|
141
|
+
def get_json_ld_from_work(url, options={})
|
142
|
+
doc = Nokogiri::HTML(open(url))
|
143
|
+
json = doc.at_xpath("//script[@type='application/ld+json']")
|
144
|
+
return { "error" => "Error: no schema.org metadata found" } unless json.present?
|
145
|
+
|
146
|
+
json.text
|
147
|
+
end
|
148
|
+
|
149
|
+
def post_metadata_for_work(input, options={})
|
150
|
+
metadata = JSON.parse(input)
|
151
|
+
|
152
|
+
prefix = options[:prefix] || ENV['PREFIX']
|
153
|
+
doi = encode_doi(prefix, number: metadata["alternateName"]) if metadata["doi"].blank?
|
154
|
+
|
155
|
+
work = Cirneco::Work.new(input: input, doi: doi)
|
156
|
+
return work.errors if work.errors.present?
|
157
|
+
|
158
|
+
response = work.post_metadata(work.datacite, options)
|
159
|
+
return response unless response.status == 201
|
160
|
+
|
161
|
+
response = work.put_doi(work.doi, options.merge(url: metadata["url"]))
|
162
|
+
return response unless response.status == 201
|
163
|
+
|
164
|
+
if work.media.present?
|
165
|
+
work.post_media(metadata.doi, options.merge(media: work.media))
|
166
|
+
else
|
167
|
+
response
|
168
|
+
end
|
169
|
+
end
|
170
|
+
|
171
|
+
def post_and_hide_metadata_for_work(input, options={})
|
172
|
+
metadata = JSON.parse(input)
|
173
|
+
|
174
|
+
prefix = options[:prefix] || ENV['PREFIX']
|
175
|
+
doi = encode_doi(prefix, number: metadata["alternateName"]) if metadata["doi"].blank?
|
176
|
+
|
177
|
+
work = Cirneco::Work.new(input: input, doi: doi)
|
178
|
+
return work.errors if work.errors.present?
|
179
|
+
|
180
|
+
response = work.post_metadata(work.datacite, options)
|
181
|
+
return response unless response.status == 201
|
182
|
+
|
183
|
+
response = work.put_doi(work.doi, options.merge(url: metadata["url"]))
|
184
|
+
return response unless response.status == 201
|
185
|
+
|
186
|
+
response = work.delete_metadata(doi, options)
|
187
|
+
return response unless response.status == 201
|
188
|
+
|
189
|
+
if work.media.present?
|
190
|
+
work.post_media(metadata["doi"], options.merge(media: work.media))
|
191
|
+
else
|
192
|
+
response
|
193
|
+
end
|
194
|
+
end
|
195
|
+
|
196
|
+
def hide_metadata_for_work(input, options={})
|
197
|
+
metadata = JSON.parse(input)
|
198
|
+
metadata["doi"] = doi_from_url(metadata["@id"])
|
199
|
+
|
200
|
+
prefix = options[:prefix] || ENV['PREFIX']
|
201
|
+
metadata["doi"] = encode_doi(prefix, number: metadata["alternateName"]) if metadata["doi"].blank?
|
202
|
+
|
203
|
+
work = Cirneco::Work.new(input: input, doi: metadata["doi"])
|
204
|
+
return work.errors if work.errors.present?
|
205
|
+
|
206
|
+
work.delete_metadata(metadata["doi"], options)
|
207
|
+
end
|
208
|
+
|
209
|
+
def generate_metadata_for_jats(url, options={})
|
210
|
+
doc = Nokogiri::HTML(open(url))
|
211
|
+
json = doc.at_xpath("//script[@type='application/ld+json']")
|
212
|
+
return { "error" => "Error: no schema.org metadata found" } unless json.present?
|
213
|
+
|
214
|
+
metadata = ActiveSupport::JSON.decode(json.text)
|
215
|
+
return { "error" => "Error: blog post not published" } if metadata["published"].to_s == "false"
|
216
|
+
return { "error" => "Error: required metadata missing" } unless ["name", "author", "publisher", "datePublished", "@type"].all? { |k| metadata.key? k }
|
217
|
+
|
218
|
+
# required metadata
|
219
|
+
if /(http|https):\/\/(dx\.)?doi\.org\/(\w+)/.match(metadata["@id"])
|
220
|
+
uri = Addressable::URI.parse(metadata["@id"])
|
221
|
+
metadata["doi"] = uri.path[1..-1].upcase
|
222
|
+
end
|
223
|
+
|
224
|
+
metadata["title"] = metadata["name"]
|
225
|
+
metadata["author"] = format_authors(metadata["author"]).map do |a|
|
226
|
+
{ "given_name" => a[:given_name],
|
227
|
+
"family_name" => a[:family_name],
|
228
|
+
"name" => a[:name],
|
229
|
+
"orcid" => a[:orcid] }.compact
|
230
|
+
end
|
231
|
+
|
232
|
+
metadata["publisher"] = metadata.fetch("publisher", {}).fetch("name", nil)
|
233
|
+
metadata["tags"] = metadata["keywords"].to_s.split(", ").select { |k| k != "featured" }
|
234
|
+
metadata["date"] = metadata.fetch("datePublished", "")
|
235
|
+
metadata["publication_year"] = metadata.fetch("date", "")[0..3].to_i
|
236
|
+
metadata["publication_month"] = metadata.fetch("date", "")[5..6].to_i
|
237
|
+
metadata["publication_day"] = metadata.fetch("date", "")[8..9].to_i
|
238
|
+
|
239
|
+
# metadata["accession_number"] = metadata["alternateName"]
|
240
|
+
metadata["journal_title"] = metadata.fetch("isPartOf", {}).fetch("name", nil)
|
241
|
+
|
242
|
+
if metadata["description"].present?
|
243
|
+
metadata["descriptions"] = [{ value: metadata["description"], description_type: "Abstract" }]
|
244
|
+
end
|
245
|
+
|
246
|
+
# use default version 1.0
|
247
|
+
metadata["version"] ||= "1.0"
|
248
|
+
|
249
|
+
# fetch reference metadata if available
|
250
|
+
metadata["related_identifiers"] = get_related_identifiers(metadata)
|
251
|
+
|
252
|
+
if metadata["license"].present?
|
253
|
+
metadata["license_name"] = LICENSES.fetch(metadata["license"], nil)
|
254
|
+
metadata["license_url"] = metadata["license"]
|
255
|
+
end
|
256
|
+
|
257
|
+
metadata = metadata.extract!(*%w(publisher doi tags title author date
|
258
|
+
publication_year publication_month publication_day license_name
|
259
|
+
license_url accession_number journal_title))
|
260
|
+
end
|
261
|
+
|
262
|
+
def generate_jats_for_url(url, options={})
|
263
|
+
filename, build_path, source_path = filepath_from_url(url, options)
|
264
|
+
metadata = generate_metadata_for_jats(build_path, options)
|
265
|
+
return "No JATS XML written for #{filename}" if metadata["published"].to_s == "false"
|
266
|
+
|
267
|
+
file = IO.read(source_path)
|
268
|
+
content = Bergamasco::Markdown.split_yaml_frontmatter(file).last
|
269
|
+
text = Bergamasco::Markdown.join_yaml_frontmatter(metadata, content)
|
270
|
+
|
271
|
+
xml = Bergamasco::Pandoc.convert_to_jats(text, options)
|
272
|
+
xmlname = filename.gsub(/\.html\.(erb|md)/, ".xml")
|
273
|
+
xmlpath = build_path.gsub("index.html", xmlname)
|
274
|
+
IO.write(xmlpath, xml)
|
275
|
+
|
276
|
+
"JATS XML written for #{filename}"
|
277
|
+
end
|
278
|
+
|
279
|
+
def generate_jats_for_all_urls(url, options={})
|
280
|
+
urls = get_urls_for_works(url)
|
281
|
+
urls.map do |u|
|
282
|
+
generate_jats_for_url(u, options)
|
283
|
+
end.join("\n")
|
284
|
+
end
|
285
|
+
|
286
|
+
def validate_jats(xml)
|
287
|
+
schema = Nokogiri::XML::Schema(open(JATS_SCHEMA))
|
288
|
+
puts schema.validate(Nokogiri::XML(xml))
|
289
|
+
|
290
|
+
OpenStruct.new(body: { "errors" => schema.validate(Nokogiri::XML(xml)).map { |error| { "title" => error.to_s } } })
|
291
|
+
end
|
292
|
+
|
293
|
+
def url_from_path(site_url, filepath)
|
294
|
+
site_url.to_s.chomp("\\") + "/" + File.basename(filepath)[0..-9] + "/"
|
295
|
+
end
|
296
|
+
|
297
|
+
def format_authors(authors)
|
298
|
+
Array(authors).map do |author|
|
299
|
+
orcid = validate_orcid(author["@id"])
|
300
|
+
name = (author["givenName"].present? || author["familyName"].present?) ? nil : author["name"]
|
301
|
+
|
302
|
+
{ :given_name => author["givenName"],
|
303
|
+
:family_name => author["familyName"],
|
304
|
+
:name => name,
|
305
|
+
:orcid => orcid }.compact
|
306
|
+
end
|
307
|
+
end
|
308
|
+
|
309
|
+
def get_related_identifiers(metadata)
|
310
|
+
citations = Array(metadata["citation"])
|
311
|
+
parent = [metadata["isPartOf"]].compact.map { |r| r.merge("relation_type" => "IsPartOf") }
|
312
|
+
children = Array(metadata["hasPart"]).map { |r| r.merge("relation_type" => "HasPart") }
|
313
|
+
|
314
|
+
(citations + parent + children).map do |r|
|
315
|
+
id = r.fetch("@id", "")
|
316
|
+
relation_type = r.fetch("relation_type", "References")
|
317
|
+
|
318
|
+
if /(http|https):\/\/(dx\.)?doi\.org\/(\w+)/.match(id)
|
319
|
+
uri = Addressable::URI.parse(id)
|
320
|
+
value = uri.path[1..-1].upcase
|
321
|
+
type = "DOI"
|
322
|
+
elsif /(http|https):\/\//.match(id)
|
323
|
+
uri = Addressable::URI.parse(id)
|
324
|
+
value = uri.normalize.to_s
|
325
|
+
type = "URL"
|
326
|
+
else
|
327
|
+
type = nil
|
328
|
+
end
|
329
|
+
|
330
|
+
{
|
331
|
+
value: value,
|
332
|
+
related_identifier_type: type,
|
333
|
+
relation_type: relation_type
|
334
|
+
}
|
335
|
+
end.select { |t| t[:related_identifier_type].present? }
|
336
|
+
end
|
337
|
+
|
338
|
+
def format_media(metadata)
|
339
|
+
[metadata["encoding"]].compact.map do |m|
|
340
|
+
{ mime_type: m["fileFormat"],
|
341
|
+
url: m["@id"] }
|
342
|
+
end
|
343
|
+
end
|
344
|
+
|
345
|
+
def filepath_from_url(url, options={})
|
346
|
+
if doi_from_url(url)
|
347
|
+
response = Maremma.head(url, limit: 0)
|
348
|
+
url = response.headers.present? ? response.headers.fetch("Location", "") : ""
|
349
|
+
end
|
350
|
+
|
351
|
+
uri = Addressable::URI.parse(url.gsub(Dir.pwd + options[:build_dir].to_s, "")
|
352
|
+
.gsub(/index\.html$/, "")
|
353
|
+
.gsub(/\/$/, ""))
|
354
|
+
basename = File.basename(uri.path, ".html").presence || "index"
|
355
|
+
|
356
|
+
if basename.starts_with?("index")
|
357
|
+
filename = basename + ".html.erb"
|
358
|
+
source_path = Dir.pwd + options[:source_dir].to_s + filename
|
359
|
+
build_path = Dir.pwd + options[:build_dir].to_s + basename + ".html"
|
360
|
+
else
|
361
|
+
filename = basename + ".html.md"
|
362
|
+
source_path = Dir.pwd + options[:source_dir].to_s + options[:posts_dir].to_s + filename
|
363
|
+
build_path = Dir.pwd + options[:build_dir].to_s + basename + "/index.html"
|
364
|
+
end
|
365
|
+
[filename, build_path, source_path]
|
366
|
+
end
|
367
|
+
end
|
368
|
+
end
|
data/lib/cirneco/media.rb
CHANGED
@@ -3,6 +3,7 @@ require 'active_support/all'
|
|
3
3
|
|
4
4
|
require_relative 'api'
|
5
5
|
require_relative 'utils'
|
6
|
+
require_relative 'file_utils'
|
6
7
|
require_relative 'base'
|
7
8
|
|
8
9
|
module Cirneco
|
@@ -10,7 +11,8 @@ module Cirneco
|
|
10
11
|
include Cirneco::Base
|
11
12
|
include Cirneco::Api
|
12
13
|
include Cirneco::Utils
|
13
|
-
|
14
|
+
include Cirneco::FileUtils
|
15
|
+
|
14
16
|
desc "get DOI", "get media for DOI"
|
15
17
|
method_option :username, :default => ENV['MDS_USERNAME']
|
16
18
|
method_option :password, :default => ENV['MDS_PASSWORD']
|
data/lib/cirneco/metadata.rb
CHANGED
@@ -3,6 +3,7 @@ require 'active_support/all'
|
|
3
3
|
|
4
4
|
require_relative 'api'
|
5
5
|
require_relative 'utils'
|
6
|
+
require_relative 'file_utils'
|
6
7
|
require_relative 'base'
|
7
8
|
|
8
9
|
module Cirneco
|
@@ -10,6 +11,7 @@ module Cirneco
|
|
10
11
|
include Cirneco::Base
|
11
12
|
include Cirneco::Api
|
12
13
|
include Cirneco::Utils
|
14
|
+
include Cirneco::FileUtils
|
13
15
|
|
14
16
|
desc "get DOI", "get metadata for DOI"
|
15
17
|
method_option :username, :default => ENV['MDS_USERNAME']
|