relaton-bipm 1.11.9 → 1.12.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.rubocop.yml +1 -1
- data/README.adoc +28 -0
- data/grammars/biblio.rng +134 -39
- data/grammars/bipm.rng +3 -0
- data/grammars/isodoc.rng +77 -0
- data/lib/relaton_bipm/bipm_bibliography.rb +4 -4
- data/lib/relaton_bipm/data_fetcher.rb +323 -0
- data/lib/relaton_bipm/hash_converter.rb +1 -1
- data/lib/relaton_bipm/index.rb +2 -2
- data/lib/relaton_bipm/processor.rb +14 -0
- data/lib/relaton_bipm/version.rb +1 -1
- data/lib/relaton_bipm.rb +1 -16
- data/relaton_bipm.gemspec +3 -5
- metadata +19 -5
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 34c3cc924311be1a1be33e18480b8521f168e1c72a46976045e9ccba69c38459
|
4
|
+
data.tar.gz: 4d929d15878f08fd5b696ecbc7fc3b0e5e94ad05e245c5fbc74125cc77ae0b8e
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 020f3459446a2a48e0d571933811888bfe4fbcd91263c482609de7a34a98df38a27fe30af2dfedaba79dbd796b9cd161049573fc2209093b73c78ead725f0389
|
7
|
+
data.tar.gz: f638c858d2384822f59dbe04d39fa0ab6f9cb9bbd42fffb2e9e57c917a5f8d64782ff7176abe98a3968b73eca59be0ed95d0822be32349c549c9b2d8b6213db9
|
data/.rubocop.yml
CHANGED
data/README.adoc
CHANGED
@@ -259,6 +259,34 @@ RelatonBipm::BipmBibliographicItem.from_hash hash
|
|
259
259
|
...
|
260
260
|
----
|
261
261
|
|
262
|
+
=== Fetch data
|
263
|
+
|
264
|
+
This gem uses following datasets as data sources:
|
265
|
+
- `bipm-data-outcomes` - looking for local directory with the repository https://github.com/metanorma/bipm-data-outcomes
|
266
|
+
- `si-brochute` - looking for local directory with the repository https://github.com/metanorma/bipm-si-brochure
|
267
|
+
|
268
|
+
The method `RelatonBipm::DataFetcher.fetch(sourece, output: "data", format: "yaml")` fetches all the documents from the datast and save them to the `./data` folder in YAML format.
|
269
|
+
Arguments:
|
270
|
+
|
271
|
+
- `source` - name of the source dataset
|
272
|
+
- `output` - folder to save documents (default './data').
|
273
|
+
- `format` - format in which the documents are saved. Possimle formats are: `yaml`, `xml`, `bibxxml` (default `yaml`).
|
274
|
+
|
275
|
+
[source,ruby]
|
276
|
+
----
|
277
|
+
RelatonBipm::DataFetcher.fetch "bipm-data-outcomes"
|
278
|
+
Started at: 2022-06-23 09:36:55 +0200
|
279
|
+
Stopped at: 2022-06-23 09:36:58 +0200
|
280
|
+
Done in: 2 sec.
|
281
|
+
=> nil
|
282
|
+
|
283
|
+
RelatonBipm::DataFetcher.fetch "si-brochure"
|
284
|
+
Started at: 2022-06-23 09:37:12 +0200
|
285
|
+
Stopped at: 2022-06-23 09:37:12 +0200
|
286
|
+
Done in: 0 sec.
|
287
|
+
=> nil
|
288
|
+
----
|
289
|
+
|
262
290
|
== Development
|
263
291
|
|
264
292
|
After checking out the repo, run `bin/setup` to install dependencies. Then, run `rake spec` to run the tests. You can also run `bin/console` for an interactive prompt that will allow you to experiment.
|
data/grammars/biblio.rng
CHANGED
@@ -614,12 +614,103 @@
|
|
614
614
|
<optional>
|
615
615
|
<ref name="fetched"/>
|
616
616
|
</optional>
|
617
|
-
<
|
618
|
-
<oneOrMore>
|
619
|
-
<ref name="btitle"/>
|
620
|
-
</oneOrMore>
|
617
|
+
<optional>
|
621
618
|
<ref name="formattedref"/>
|
622
|
-
</
|
619
|
+
</optional>
|
620
|
+
<oneOrMore>
|
621
|
+
<ref name="btitle"/>
|
622
|
+
</oneOrMore>
|
623
|
+
<zeroOrMore>
|
624
|
+
<ref name="bsource"/>
|
625
|
+
</zeroOrMore>
|
626
|
+
<oneOrMore>
|
627
|
+
<ref name="docidentifier"/>
|
628
|
+
</oneOrMore>
|
629
|
+
<optional>
|
630
|
+
<ref name="docnumber"/>
|
631
|
+
</optional>
|
632
|
+
<zeroOrMore>
|
633
|
+
<ref name="bdate"/>
|
634
|
+
</zeroOrMore>
|
635
|
+
<zeroOrMore>
|
636
|
+
<ref name="contributor"/>
|
637
|
+
</zeroOrMore>
|
638
|
+
<optional>
|
639
|
+
<ref name="edition"/>
|
640
|
+
</optional>
|
641
|
+
<zeroOrMore>
|
642
|
+
<ref name="version"/>
|
643
|
+
</zeroOrMore>
|
644
|
+
<zeroOrMore>
|
645
|
+
<ref name="biblionote"/>
|
646
|
+
</zeroOrMore>
|
647
|
+
<zeroOrMore>
|
648
|
+
<ref name="language"/>
|
649
|
+
</zeroOrMore>
|
650
|
+
<zeroOrMore>
|
651
|
+
<ref name="script"/>
|
652
|
+
</zeroOrMore>
|
653
|
+
<zeroOrMore>
|
654
|
+
<ref name="bibabstract"/>
|
655
|
+
</zeroOrMore>
|
656
|
+
<optional>
|
657
|
+
<ref name="status"/>
|
658
|
+
</optional>
|
659
|
+
<zeroOrMore>
|
660
|
+
<ref name="copyright"/>
|
661
|
+
</zeroOrMore>
|
662
|
+
<zeroOrMore>
|
663
|
+
<ref name="docrelation"/>
|
664
|
+
</zeroOrMore>
|
665
|
+
<zeroOrMore>
|
666
|
+
<ref name="series"/>
|
667
|
+
</zeroOrMore>
|
668
|
+
<optional>
|
669
|
+
<ref name="medium"/>
|
670
|
+
</optional>
|
671
|
+
<zeroOrMore>
|
672
|
+
<ref name="bplace"/>
|
673
|
+
</zeroOrMore>
|
674
|
+
<zeroOrMore>
|
675
|
+
<ref name="bprice"/>
|
676
|
+
</zeroOrMore>
|
677
|
+
<zeroOrMore>
|
678
|
+
<ref name="extent"/>
|
679
|
+
</zeroOrMore>
|
680
|
+
<optional>
|
681
|
+
<ref name="bibliographic_size"/>
|
682
|
+
</optional>
|
683
|
+
<zeroOrMore>
|
684
|
+
<ref name="accesslocation"/>
|
685
|
+
</zeroOrMore>
|
686
|
+
<zeroOrMore>
|
687
|
+
<ref name="license"/>
|
688
|
+
</zeroOrMore>
|
689
|
+
<zeroOrMore>
|
690
|
+
<ref name="bclassification"/>
|
691
|
+
</zeroOrMore>
|
692
|
+
<zeroOrMore>
|
693
|
+
<ref name="bkeyword"/>
|
694
|
+
</zeroOrMore>
|
695
|
+
<optional>
|
696
|
+
<ref name="validity"/>
|
697
|
+
</optional>
|
698
|
+
</define>
|
699
|
+
<define name="ReducedBibliographicItem">
|
700
|
+
<optional>
|
701
|
+
<attribute name="type">
|
702
|
+
<ref name="BibItemType"/>
|
703
|
+
</attribute>
|
704
|
+
</optional>
|
705
|
+
<optional>
|
706
|
+
<ref name="fetched"/>
|
707
|
+
</optional>
|
708
|
+
<optional>
|
709
|
+
<ref name="formattedref"/>
|
710
|
+
</optional>
|
711
|
+
<zeroOrMore>
|
712
|
+
<ref name="btitle"/>
|
713
|
+
</zeroOrMore>
|
623
714
|
<zeroOrMore>
|
624
715
|
<ref name="bsource"/>
|
625
716
|
</zeroOrMore>
|
@@ -638,9 +729,9 @@
|
|
638
729
|
<optional>
|
639
730
|
<ref name="edition"/>
|
640
731
|
</optional>
|
641
|
-
<
|
732
|
+
<zeroOrMore>
|
642
733
|
<ref name="version"/>
|
643
|
-
</
|
734
|
+
</zeroOrMore>
|
644
735
|
<zeroOrMore>
|
645
736
|
<ref name="biblionote"/>
|
646
737
|
</zeroOrMore>
|
@@ -833,6 +924,12 @@
|
|
833
924
|
<data type="boolean"/>
|
834
925
|
</attribute>
|
835
926
|
</optional>
|
927
|
+
<optional>
|
928
|
+
<attribute name="language"/>
|
929
|
+
</optional>
|
930
|
+
<optional>
|
931
|
+
<attribute name="script"/>
|
932
|
+
</optional>
|
836
933
|
<text/>
|
837
934
|
</element>
|
838
935
|
</define>
|
@@ -986,36 +1083,34 @@
|
|
986
1083
|
<ref name="SeriesType"/>
|
987
1084
|
</attribute>
|
988
1085
|
</optional>
|
989
|
-
<
|
1086
|
+
<optional>
|
990
1087
|
<ref name="formattedref"/>
|
991
|
-
|
992
|
-
|
993
|
-
|
994
|
-
|
995
|
-
|
996
|
-
|
997
|
-
|
998
|
-
|
999
|
-
|
1000
|
-
|
1001
|
-
|
1002
|
-
|
1003
|
-
|
1004
|
-
|
1005
|
-
|
1006
|
-
|
1007
|
-
|
1008
|
-
|
1009
|
-
|
1010
|
-
|
1011
|
-
|
1012
|
-
|
1013
|
-
|
1014
|
-
|
1015
|
-
|
1016
|
-
|
1017
|
-
</group>
|
1018
|
-
</choice>
|
1088
|
+
</optional>
|
1089
|
+
<ref name="btitle"/>
|
1090
|
+
<optional>
|
1091
|
+
<ref name="bplace"/>
|
1092
|
+
</optional>
|
1093
|
+
<optional>
|
1094
|
+
<ref name="seriesorganization"/>
|
1095
|
+
</optional>
|
1096
|
+
<optional>
|
1097
|
+
<ref name="abbreviation"/>
|
1098
|
+
</optional>
|
1099
|
+
<optional>
|
1100
|
+
<ref name="seriesfrom"/>
|
1101
|
+
</optional>
|
1102
|
+
<optional>
|
1103
|
+
<ref name="seriesto"/>
|
1104
|
+
</optional>
|
1105
|
+
<optional>
|
1106
|
+
<ref name="seriesnumber"/>
|
1107
|
+
</optional>
|
1108
|
+
<optional>
|
1109
|
+
<ref name="seriespartnumber"/>
|
1110
|
+
</optional>
|
1111
|
+
<optional>
|
1112
|
+
<ref name="seriesrun"/>
|
1113
|
+
</optional>
|
1019
1114
|
</element>
|
1020
1115
|
</define>
|
1021
1116
|
<define name="SeriesType">
|
@@ -1174,7 +1269,7 @@
|
|
1174
1269
|
</element>
|
1175
1270
|
</optional>
|
1176
1271
|
<element name="bibitem">
|
1177
|
-
<ref name="
|
1272
|
+
<ref name="ReducedBibliographicItem"/>
|
1178
1273
|
</element>
|
1179
1274
|
<choice>
|
1180
1275
|
<zeroOrMore>
|
@@ -1199,9 +1294,9 @@
|
|
1199
1294
|
<optional>
|
1200
1295
|
<ref name="revision-date"/>
|
1201
1296
|
</optional>
|
1202
|
-
<
|
1297
|
+
<optional>
|
1203
1298
|
<ref name="draft"/>
|
1204
|
-
</
|
1299
|
+
</optional>
|
1205
1300
|
</element>
|
1206
1301
|
</define>
|
1207
1302
|
<define name="vedition">
|
data/grammars/bipm.rng
CHANGED
data/grammars/isodoc.rng
CHANGED
@@ -32,6 +32,43 @@
|
|
32
32
|
<ref name="DocumentType"/>
|
33
33
|
</element>
|
34
34
|
</define>
|
35
|
+
<define name="admonition">
|
36
|
+
<element name="admonition">
|
37
|
+
<attribute name="type">
|
38
|
+
<ref name="AdmonitionType"/>
|
39
|
+
</attribute>
|
40
|
+
<optional>
|
41
|
+
<attribute name="class"/>
|
42
|
+
</optional>
|
43
|
+
<attribute name="id">
|
44
|
+
<data type="ID"/>
|
45
|
+
</attribute>
|
46
|
+
<optional>
|
47
|
+
<attribute name="uri">
|
48
|
+
<data type="anyURI"/>
|
49
|
+
</attribute>
|
50
|
+
</optional>
|
51
|
+
<optional>
|
52
|
+
<attribute name="coverpage">
|
53
|
+
<data type="boolean"/>
|
54
|
+
</attribute>
|
55
|
+
</optional>
|
56
|
+
<optional>
|
57
|
+
<attribute name="notag">
|
58
|
+
<data type="boolean"/>
|
59
|
+
</attribute>
|
60
|
+
</optional>
|
61
|
+
<optional>
|
62
|
+
<ref name="tname"/>
|
63
|
+
</optional>
|
64
|
+
<zeroOrMore>
|
65
|
+
<ref name="paragraph-with-footnote"/>
|
66
|
+
</zeroOrMore>
|
67
|
+
<zeroOrMore>
|
68
|
+
<ref name="note"/>
|
69
|
+
</zeroOrMore>
|
70
|
+
</element>
|
71
|
+
</define>
|
35
72
|
<define name="index">
|
36
73
|
<element name="index">
|
37
74
|
<optional>
|
@@ -79,6 +116,11 @@
|
|
79
116
|
<data type="boolean"/>
|
80
117
|
</attribute>
|
81
118
|
</optional>
|
119
|
+
<optional>
|
120
|
+
<attribute name="suppress_identifier">
|
121
|
+
<data type="boolean"/>
|
122
|
+
</attribute>
|
123
|
+
</optional>
|
82
124
|
<ref name="BibliographicItem"/>
|
83
125
|
</element>
|
84
126
|
</define>
|
@@ -245,6 +287,9 @@
|
|
245
287
|
<ref name="MultilingualRenderingType"/>
|
246
288
|
</attribute>
|
247
289
|
</optional>
|
290
|
+
<optional>
|
291
|
+
<ref name="tname"/>
|
292
|
+
</optional>
|
248
293
|
<oneOrMore>
|
249
294
|
<ref name="ul_li"/>
|
250
295
|
</oneOrMore>
|
@@ -287,6 +332,9 @@
|
|
287
332
|
</choice>
|
288
333
|
</attribute>
|
289
334
|
</optional>
|
335
|
+
<optional>
|
336
|
+
<ref name="tname"/>
|
337
|
+
</optional>
|
290
338
|
<oneOrMore>
|
291
339
|
<ref name="li"/>
|
292
340
|
</oneOrMore>
|
@@ -323,6 +371,9 @@
|
|
323
371
|
<ref name="MultilingualRenderingType"/>
|
324
372
|
</attribute>
|
325
373
|
</optional>
|
374
|
+
<optional>
|
375
|
+
<ref name="tname"/>
|
376
|
+
</optional>
|
326
377
|
<oneOrMore>
|
327
378
|
<ref name="dt"/>
|
328
379
|
<ref name="dd"/>
|
@@ -657,6 +708,9 @@
|
|
657
708
|
<optional>
|
658
709
|
<attribute name="tag"/>
|
659
710
|
</optional>
|
711
|
+
<optional>
|
712
|
+
<attribute name="type"/>
|
713
|
+
</optional>
|
660
714
|
<optional>
|
661
715
|
<attribute name="multilingual-rendering">
|
662
716
|
<ref name="MultilingualRenderingType"/>
|
@@ -692,6 +746,9 @@
|
|
692
746
|
<optional>
|
693
747
|
<attribute name="tag"/>
|
694
748
|
</optional>
|
749
|
+
<optional>
|
750
|
+
<attribute name="type"/>
|
751
|
+
</optional>
|
695
752
|
<optional>
|
696
753
|
<attribute name="multilingual-rendering">
|
697
754
|
<ref name="MultilingualRenderingType"/>
|
@@ -851,6 +908,16 @@
|
|
851
908
|
<ref name="MultilingualRenderingType"/>
|
852
909
|
</attribute>
|
853
910
|
</optional>
|
911
|
+
<optional>
|
912
|
+
<attribute name="coverpage">
|
913
|
+
<data type="boolean"/>
|
914
|
+
</attribute>
|
915
|
+
</optional>
|
916
|
+
<optional>
|
917
|
+
<attribute name="notag">
|
918
|
+
<data type="boolean"/>
|
919
|
+
</attribute>
|
920
|
+
</optional>
|
854
921
|
<oneOrMore>
|
855
922
|
<choice>
|
856
923
|
<ref name="paragraph"/>
|
@@ -1178,6 +1245,7 @@
|
|
1178
1245
|
<ref name="concept"/>
|
1179
1246
|
<ref name="add"/>
|
1180
1247
|
<ref name="del"/>
|
1248
|
+
<ref name="span"/>
|
1181
1249
|
</choice>
|
1182
1250
|
</define>
|
1183
1251
|
<define name="add">
|
@@ -1204,6 +1272,14 @@
|
|
1204
1272
|
</choice>
|
1205
1273
|
</element>
|
1206
1274
|
</define>
|
1275
|
+
<define name="span">
|
1276
|
+
<element name="span">
|
1277
|
+
<attribute name="class"/>
|
1278
|
+
<oneOrMore>
|
1279
|
+
<ref name="TextElement"/>
|
1280
|
+
</oneOrMore>
|
1281
|
+
</element>
|
1282
|
+
</define>
|
1207
1283
|
<define name="concept">
|
1208
1284
|
<element name="concept">
|
1209
1285
|
<optional>
|
@@ -1994,6 +2070,7 @@
|
|
1994
2070
|
<value>compare</value>
|
1995
2071
|
<value>contrast</value>
|
1996
2072
|
<value>see</value>
|
2073
|
+
<value>seealso</value>
|
1997
2074
|
</choice>
|
1998
2075
|
</define>
|
1999
2076
|
<define name="deprecates">
|
@@ -4,13 +4,12 @@ module RelatonBipm
|
|
4
4
|
class BipmBibliography
|
5
5
|
GH_ENDPOINT = "https://raw.githubusercontent.com/relaton/relaton-data-bipm/master/".freeze
|
6
6
|
IOP_DOMAIN = "https://iopscience.iop.org".freeze
|
7
|
-
|
7
|
+
TRANSLATIONS = {
|
8
8
|
"Déclaration" => "Declaration",
|
9
9
|
"Réunion" => "Meeting",
|
10
10
|
"Recommandation" => "Recommendation",
|
11
11
|
"Résolution" => "Resolution",
|
12
12
|
"Décision" => "Decision",
|
13
|
-
"CCDS" => "CCTF",
|
14
13
|
}.freeze
|
15
14
|
|
16
15
|
class << self
|
@@ -51,7 +50,8 @@ module RelatonBipm
|
|
51
50
|
rf = ref.sub(/(?:(\d{1,2})\s)?\(?(\d{4})(?!-)\)?/) do
|
52
51
|
"#{$2}-#{$1.to_s.rjust(2, '0')}"
|
53
52
|
end
|
54
|
-
|
53
|
+
rf.sub!("CCDS", "CCTF")
|
54
|
+
TRANSLATIONS.each { |fr, en| rf.sub! fr, en }
|
55
55
|
path = Index.new.search rf
|
56
56
|
return unless path
|
57
57
|
|
@@ -60,7 +60,7 @@ module RelatonBipm
|
|
60
60
|
check_response resp
|
61
61
|
return unless resp.code == "200"
|
62
62
|
|
63
|
-
yaml =
|
63
|
+
yaml = RelatonBib.parse_yaml resp.body, [Date]
|
64
64
|
bib_hash = HashConverter.hash_to_bib yaml
|
65
65
|
BipmBibliographicItem.new(**bib_hash)
|
66
66
|
end
|
@@ -1,9 +1,332 @@
|
|
1
1
|
module RelatonBipm
|
2
2
|
class DataFetcher
|
3
|
+
#
|
4
|
+
# Initialize fetcher
|
5
|
+
#
|
6
|
+
# @param [String] output output directory to save files
|
7
|
+
# @param [String] format format of output files (xml, yaml, bibxml)
|
8
|
+
#
|
3
9
|
def initialize(output, format)
|
4
10
|
@output = output
|
5
11
|
@format = format
|
6
12
|
@ext = format.sub(/^bib/, "")
|
13
|
+
@files = []
|
14
|
+
@index_path = "index.yaml"
|
15
|
+
@index = File.exist?(@index_path) ? YAML.load_file(@index_path) : {}
|
16
|
+
end
|
17
|
+
|
18
|
+
#
|
19
|
+
# Initialize fetcher and run fetching
|
20
|
+
#
|
21
|
+
# @param [String] source Source name
|
22
|
+
# @param [Strin] output directory to save files, default: "data"
|
23
|
+
# @param [Strin] format format of output files (xml, yaml, bibxml), default: yaml
|
24
|
+
#
|
25
|
+
def self.fetch(source, output: "data", format: "yaml")
|
26
|
+
t1 = Time.now
|
27
|
+
puts "Started at: #{t1}"
|
28
|
+
FileUtils.mkdir_p output unless Dir.exist? output
|
29
|
+
new(output, format).fetch(source)
|
30
|
+
t2 = Time.now
|
31
|
+
puts "Stopped at: #{t2}"
|
32
|
+
puts "Done in: #{(t2 - t1).round} sec."
|
33
|
+
end
|
34
|
+
|
35
|
+
#
|
36
|
+
# Fetch bipm-data-outcomes or si-brochure
|
37
|
+
#
|
38
|
+
# @param [String] source Source name
|
39
|
+
#
|
40
|
+
def fetch(source)
|
41
|
+
case source
|
42
|
+
when "bipm-data-outcomes" then parse_bipm_data_outcomes
|
43
|
+
when "si-brochure" then parse_si_brochure
|
44
|
+
end
|
45
|
+
File.write @index_path, @index.to_yaml, encoding: "UTF-8"
|
46
|
+
end
|
47
|
+
|
48
|
+
#
|
49
|
+
# Parse BIPM meeting and write them to YAML files
|
50
|
+
#
|
51
|
+
def parse_bipm_data_outcomes
|
52
|
+
source_path = File.join "bipm-data-outcomes", "{cctf,cgpm,cipm}"
|
53
|
+
Dir[source_path].each { |body_dir| fetch_body(body_dir) }
|
54
|
+
end
|
55
|
+
|
56
|
+
#
|
57
|
+
# Parse SI brochure and write them to YAML files
|
58
|
+
#
|
59
|
+
def parse_si_brochure # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
|
60
|
+
Dir["bipm-si-brochure/site/documents/*.rxl"].each do |f|
|
61
|
+
docstd = Nokogiri::XML File.read f
|
62
|
+
doc = docstd.at "/bibdata"
|
63
|
+
hash1 = RelatonBipm::XMLParser.from_xml(doc.to_xml).to_hash
|
64
|
+
hash1["fetched"] = Date.today.to_s
|
65
|
+
hash1["docid"].detect { |id| id["type"] == "BIPM" }["primary"] = true
|
66
|
+
outfile = File.join @output, File.basename(f).sub(/(?:-(?:en|fr))?\.rxl$/, ".yaml")
|
67
|
+
@index[[hash1["docnumber"] || File.basename(outfile, ".yaml")]] = outfile
|
68
|
+
hash = if File.exist? outfile
|
69
|
+
warn_duplicate = false
|
70
|
+
hash2 = YAML.load_file outfile
|
71
|
+
deep_merge hash1, hash2
|
72
|
+
else
|
73
|
+
warn_duplicate = true
|
74
|
+
hash1
|
75
|
+
end
|
76
|
+
item = RelatonBipm::BipmBibliographicItem.from_hash(**hash)
|
77
|
+
write_file outfile, item, warn_duplicate: warn_duplicate
|
78
|
+
end
|
79
|
+
end
|
80
|
+
|
81
|
+
#
|
82
|
+
# Deep merge two hashes
|
83
|
+
#
|
84
|
+
# @param [Hash] hash1
|
85
|
+
# @param [Hash] hash2
|
86
|
+
#
|
87
|
+
# @return [Hash] Merged hash
|
88
|
+
#
|
89
|
+
def deep_merge(hash1, hash2) # rubocop:disable Metrics/PerceivedComplexity, Metrics/CyclomaticComplexity
|
90
|
+
hash1.merge(hash2) do |_, oldval, newval|
|
91
|
+
if oldval.is_a?(Hash) && newval.is_a?(Hash)
|
92
|
+
deep_merge(oldval, newval)
|
93
|
+
elsif oldval.is_a?(Array) && newval.is_a?(Array)
|
94
|
+
oldval | newval
|
95
|
+
else
|
96
|
+
newval || oldval
|
97
|
+
end
|
98
|
+
end
|
99
|
+
end
|
100
|
+
|
101
|
+
#
|
102
|
+
# Search for English meetings in the body directory
|
103
|
+
#
|
104
|
+
# @param [String] dir body directory
|
105
|
+
#
|
106
|
+
def fetch_body(dir)
|
107
|
+
body = dir.split("/").last.upcase
|
108
|
+
Dir[File.join(dir, "*-en")].each { |type_dir| fetch_type type_dir, body }
|
109
|
+
end
|
110
|
+
|
111
|
+
#
|
112
|
+
# Search for meetings
|
113
|
+
#
|
114
|
+
# @param [String] dir meeting directory
|
115
|
+
# @param [String] body name of body
|
116
|
+
#
|
117
|
+
def fetch_type(dir, body) # rubocop:disable Metrics/AbcSize
|
118
|
+
type = dir.split("/").last.split("-").first.sub(/s$/, "")
|
119
|
+
body_dir = File.join @output, body.downcase
|
120
|
+
Dir.mkdir body_dir unless Dir.exist? body_dir
|
121
|
+
outdir = File.join body_dir, type.downcase
|
122
|
+
Dir.mkdir outdir unless Dir.exist? outdir
|
123
|
+
Dir[File.join(dir, "*.{yml,yaml}")].each { |en_file| fetch_meeting en_file, body, type, outdir }
|
124
|
+
end
|
125
|
+
|
126
|
+
#
|
127
|
+
# Create and write BIPM meeting/resolution
|
128
|
+
#
|
129
|
+
# @param [String] en_file Path to English file
|
130
|
+
# @param [String] body Body name
|
131
|
+
# @param [String] type Type of Recommendation/Decision/Resolution
|
132
|
+
# @param [String] dir output directory
|
133
|
+
#
|
134
|
+
def fetch_meeting(en_file, body, type, dir) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
|
135
|
+
en = RelatonBib.parse_yaml File.read(en_file, encoding: "UTF-8"), [Date]
|
136
|
+
en_md = en["metadata"]
|
137
|
+
fr_file = en_file.sub "en", "fr"
|
138
|
+
fr = RelatonBib.parse_yaml File.read(fr_file, encoding: "UTF-8"), [Date]
|
139
|
+
fr_md = fr["metadata"]
|
140
|
+
gh_src = "https://raw.githubusercontent.com/metanorma/"
|
141
|
+
src_en = gh_src + en_file.split("/")[1..].insert(1, "main").join("/")
|
142
|
+
src_fr = gh_src + fr_file.split("/")[1..].insert(1, "main").join("/")
|
143
|
+
src = [{ type: "src", content: src_en }, { type: "src", content: src_fr }]
|
144
|
+
|
145
|
+
/^(?<num>\d+)(?:-_(?<part>\d+))?-\d{4}$/ =~ en_md["url"].split("/").last
|
146
|
+
id = "#{body} #{type.capitalize} #{num}"
|
147
|
+
file = "#{num}.yaml"
|
148
|
+
path = File.join dir, file
|
149
|
+
link = "https://raw.githubusercontent.com/relaton/relaton-data-bipm/master/#{path}"
|
150
|
+
hash = bibitem type: type, en: en_md, fr: fr_md, id: id, num: num, src: src, pdf: en["pdf"]
|
151
|
+
if @files.include?(path) && part
|
152
|
+
add_part hash, part
|
153
|
+
item = RelatonBipm::BipmBibliographicItem.new(**hash)
|
154
|
+
yaml = RelatonBib.parse_yaml(File.read(path, encoding: "UTF-8"), [Date])
|
155
|
+
has_part_item = RelatonBipm::BipmBibliographicItem.from_hash(yaml)
|
156
|
+
has_part_item.relation << RelatonBib::DocumentRelation.new(type: "partOf", bibitem: item)
|
157
|
+
write_file path, has_part_item, warn_duplicate: false
|
158
|
+
path = File.join dir, "#{num}-#{part}.yaml"
|
159
|
+
elsif part
|
160
|
+
hash[:title].each { |t| t[:content] = t[:content].sub(/\s\(.+\)$/, "") }
|
161
|
+
hash[:link] = [{ type: "src", content: link }]
|
162
|
+
h = bibitem type: type, en: en_md, fr: fr_md, id: id, num: num, src: src, pdf: en["pdf"]
|
163
|
+
add_part h, part
|
164
|
+
part_item = RelatonBipm::BipmBibliographicItem.new(**h)
|
165
|
+
part_item_path = File.join dir, "#{num}-#{part}.yaml"
|
166
|
+
write_file part_item_path, part_item
|
167
|
+
@index[[h[:docnumber]]] = part_item_path
|
168
|
+
hash[:relation] = [RelatonBib::DocumentRelation.new(type: "partOf", bibitem: part_item)]
|
169
|
+
item = RelatonBipm::BipmBibliographicItem.new(**hash)
|
170
|
+
else
|
171
|
+
item = RelatonBipm::BipmBibliographicItem.new(**hash)
|
172
|
+
end
|
173
|
+
write_file path, item
|
174
|
+
@index[[hash[:docnumber]]] = path
|
175
|
+
fetch_resolution body: body, en: en, fr: fr, dir: dir, src: src, num: num
|
176
|
+
end
|
177
|
+
|
178
|
+
#
|
179
|
+
# Parse BIPM resolutions and write them to YAML files
|
180
|
+
#
|
181
|
+
# @param [String] body body name
|
182
|
+
# @param [Hash] eng English metadata
|
183
|
+
# @param [Hash] frn French metadata
|
184
|
+
# @param [String] dir output directory
|
185
|
+
# @param [Array<Hash>] src links to bipm-data-outcomes
|
186
|
+
# @param [String] num number of meeting
|
187
|
+
#
|
188
|
+
def fetch_resolution(**args) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
|
189
|
+
args[:en]["resolutions"].each.with_index do |r, i| # rubocop:disable Metrics/BlockLength
|
190
|
+
hash = { fetched: Date.today.to_s, title: [], doctype: r["type"] }
|
191
|
+
hash[:title] << title(r["title"], "en") if r["title"]
|
192
|
+
fr_title = args[:fr]["resolutions"][i]["title"]
|
193
|
+
hash[:title] << title(fr_title, "fr") if fr_title
|
194
|
+
date = r["dates"].first.to_s
|
195
|
+
hash[:date] = [{ type: "published", on: date }]
|
196
|
+
num = r["identifier"].to_s.split("-").last
|
197
|
+
year = date.split("-").first
|
198
|
+
num = "0" if num == year
|
199
|
+
num_justed = num.rjust 2, "0"
|
200
|
+
type = r["type"].capitalize
|
201
|
+
id = "#{args[:body]} #{type}"
|
202
|
+
hash[:id] = "#{args[:body]}-#{type}-#{year}"
|
203
|
+
if num.to_i.positive?
|
204
|
+
id += " #{num}"
|
205
|
+
hash[:id] += "-#{num_justed}"
|
206
|
+
end
|
207
|
+
id += " (#{year})"
|
208
|
+
hash[:docid] = [
|
209
|
+
make_docid(id: id, type: "BIPM", primary: true),
|
210
|
+
make_docid(id: id, type: "BIPM", primary: true, language: "en", script: "Latn"),
|
211
|
+
id_fr(id),
|
212
|
+
]
|
213
|
+
hash[:docnumber] = id
|
214
|
+
hash[:link] = [{ type: "src", content: r["url"] }] + args[:src]
|
215
|
+
hash[:link] << { type: "pdf", content: r["reference"] } if r["reference"]
|
216
|
+
hash[:language] = %w[en fr]
|
217
|
+
hash[:script] = ["Latn"]
|
218
|
+
hash[:contributor] = [{
|
219
|
+
entity: { url: "www.bipm.org", name: "Bureau International des Poids et Mesures", abbreviation: "BIPM" },
|
220
|
+
role: [{ type: "publisher" }],
|
221
|
+
}]
|
222
|
+
hash[:structuredidentifier] = RelatonBipm::StructuredIdentifier.new docnumber: num
|
223
|
+
item = RelatonBipm::BipmBibliographicItem.new(**hash)
|
224
|
+
file = year
|
225
|
+
file += "-#{num_justed}" if num.size < 4
|
226
|
+
file += ".yaml"
|
227
|
+
out_dir = File.join args[:dir], r["type"].downcase
|
228
|
+
Dir.mkdir out_dir unless Dir.exist? out_dir
|
229
|
+
path = File.join out_dir, file
|
230
|
+
write_file path, item
|
231
|
+
@index[["#{args[:body]} #{type} #{year}-#{num_justed}", "#{args[:body]} #{type} #{args[:num]}-#{num_justed}"]] = path
|
232
|
+
end
|
233
|
+
end
|
234
|
+
|
235
|
+
def title(content, language)
|
236
|
+
{ content: content, language: language, script: "Latn" }
|
237
|
+
end
|
238
|
+
|
239
|
+
#
|
240
|
+
# Add part to ID and structured identifier
|
241
|
+
#
|
242
|
+
# @param [Hash] hash Hash of BIPM meeting
|
243
|
+
# @param [String] session number of meeting
|
244
|
+
#
|
245
|
+
def add_part(hash, part)
|
246
|
+
hash[:id] += "-#{part}"
|
247
|
+
hash[:docnumber] += "-#{part}"
|
248
|
+
id = hash[:docid][0].instance_variable_get(:@id)
|
249
|
+
id += "-#{part}"
|
250
|
+
hash[:docid][0].instance_variable_set(:@id, id)
|
251
|
+
hash[:structuredidentifier].instance_variable_set :@part, part
|
252
|
+
end
|
253
|
+
|
254
|
+
#
|
255
|
+
# Create hash from BIPM meeting/resolution
|
256
|
+
#
|
257
|
+
# @param [Hash] **args Hash of arguments
|
258
|
+
# @option args [String] :type Type of meeting/resolution
|
259
|
+
# @option args [Hash] :en Hash of English metadata
|
260
|
+
# @option args [Hash] :fr Hash of French metadata
|
261
|
+
# @option args [String] :id ID of meeting/resolution
|
262
|
+
# @option args [String] :num Number of meeting/resolution
|
263
|
+
# @option args [Array<Hash>] :src Array of links to bipm-data-outcomes
|
264
|
+
# @option args [String] :pdf link to PDF
|
265
|
+
#
|
266
|
+
# @return [Hash] Hash of BIPM meeting/resolution
|
267
|
+
#
|
268
|
+
def bibitem(**args) # rubocop:disable Metrics/MethodLength, Metrics/AbcSize, Metrics/CyclomaticComplexity
|
269
|
+
hash = { title: [], doctype: args[:type], fetched: Date.today.to_s }
|
270
|
+
hash[:title] << title(args[:en]["title"], "en") if args[:en]["title"]
|
271
|
+
hash[:title] << title(args[:fr]["title"], "fr") if args[:fr]["title"]
|
272
|
+
hash[:date] = [{ type: "published", on: args[:en]["date"] }]
|
273
|
+
hash[:docid] = [
|
274
|
+
make_docid(id: args[:id], type: "BIPM", primary: true),
|
275
|
+
make_docid(id: args[:id], type: "BIPM", primary: true, language: "en", script: "Latn"),
|
276
|
+
id_fr(args[:id]),
|
277
|
+
]
|
278
|
+
hash[:id] = args[:id].gsub " ", "-"
|
279
|
+
hash[:docnumber] = args[:id]
|
280
|
+
hash[:link] = [{ type: "src", content: args[:en]["url"] }]
|
281
|
+
hash[:link] << { type: "pdf", content: args[:pdf] } if args[:pdf]
|
282
|
+
hash[:link] += args[:src] if args[:src]&.any?
|
283
|
+
hash[:language] = %w[en fr]
|
284
|
+
hash[:script] = ["Latn"]
|
285
|
+
hash[:contributor] = [{
|
286
|
+
entity: { url: "www.bipm.org", name: "Bureau International des Poids et Mesures", abbreviation: "BIPM" },
|
287
|
+
role: [{ type: "publisher" }],
|
288
|
+
}]
|
289
|
+
hash[:structuredidentifier] = RelatonBipm::StructuredIdentifier.new docnumber: args[:num]
|
290
|
+
hash
|
291
|
+
end
|
292
|
+
|
293
|
+
def id_fr(en_id)
|
294
|
+
tr = BipmBibliography::TRANSLATIONS.detect { |_, v| en_id.include? v }
|
295
|
+
id = en_id.sub tr[1], tr[0]
|
296
|
+
make_docid(id: id, type: "BIPM", primary: true, language: "fr", script: "Latn")
|
297
|
+
end
|
298
|
+
|
299
|
+
#
|
300
|
+
# Create doucment ID
|
301
|
+
#
|
302
|
+
# @param [String] id ID of document
|
303
|
+
# @param [String] type Type of document
|
304
|
+
# @param [Boolean] primary Primary document
|
305
|
+
# @param [String] language Language of document
|
306
|
+
# @param [String] script Script of document
|
307
|
+
#
|
308
|
+
# @return [RelatonBib::DocumentIdentifier] Document ID
|
309
|
+
#
|
310
|
+
def make_docid(**args)
|
311
|
+
RelatonBib::DocumentIdentifier.new(**args)
|
312
|
+
end
|
313
|
+
|
314
|
+
#
|
315
|
+
# Save document to file
|
316
|
+
#
|
317
|
+
# @param [String] path Path to file
|
318
|
+
# @param [RelatonBipm::BipmBibliographicItem] item document to save
|
319
|
+
# @param [Boolean, nil] warn_duplicate Warn if document already exists
|
320
|
+
#
|
321
|
+
# @return [<Type>] <description>
|
322
|
+
#
|
323
|
+
def write_file(path, item, warn_duplicate: true)
|
324
|
+
if @files.include?(path)
|
325
|
+
warn "File #{path} already exists" if warn_duplicate
|
326
|
+
else
|
327
|
+
@files << path
|
328
|
+
end
|
329
|
+
File.write path, item.to_hash.to_yaml, encoding: "UTF-8"
|
7
330
|
end
|
8
331
|
end
|
9
332
|
end
|
data/lib/relaton_bipm/index.rb
CHANGED
@@ -41,7 +41,7 @@ module RelatonBipm
|
|
41
41
|
def read_index_file
|
42
42
|
return if !File.exist?(path) || File.ctime(path).to_date < Date.today
|
43
43
|
|
44
|
-
@index =
|
44
|
+
@index = RelatonBib.parse_yaml File.read(path, encoding: "UTF-8")
|
45
45
|
end
|
46
46
|
|
47
47
|
#
|
@@ -61,7 +61,7 @@ module RelatonBipm
|
|
61
61
|
def get_index_from_gh # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
|
62
62
|
resp = Zip::InputStream.new URI("#{BipmBibliography::GH_ENDPOINT}index.zip").open
|
63
63
|
zip = resp.get_next_entry
|
64
|
-
@index =
|
64
|
+
@index = RelatonBib.parse_yaml zip.get_input_stream.read
|
65
65
|
save_index_file
|
66
66
|
end
|
67
67
|
end
|
@@ -9,6 +9,7 @@ module RelatonBipm
|
|
9
9
|
@prefix = "BIPM"
|
10
10
|
@defaultprefix = %r{^(?:BIPM|CCTF|CCDS|CGPM|CIPM)(?!\w)}
|
11
11
|
@idtype = "BIPM"
|
12
|
+
@datasets = %w[bipm-data-outcomes si-brochure]
|
12
13
|
end
|
13
14
|
|
14
15
|
# @param code [String]
|
@@ -19,6 +20,19 @@ module RelatonBipm
|
|
19
20
|
::RelatonBipm::BipmBibliography.get(code, date, opts)
|
20
21
|
end
|
21
22
|
|
23
|
+
#
|
24
|
+
# Fetch all the documents from https://github.com/metanorma/bipm-data-outcomes
|
25
|
+
# or https://github.com/metanorma/bipm-si-brochure
|
26
|
+
#
|
27
|
+
# @param [String] source source name
|
28
|
+
# @param [Hash] opts
|
29
|
+
# @option opts [String] :output directory to output documents
|
30
|
+
# @option opts [String] :format
|
31
|
+
#
|
32
|
+
def fetch_data(source, opts)
|
33
|
+
DataFetcher.fetch(source, **opts)
|
34
|
+
end
|
35
|
+
|
22
36
|
# @param xml [String]
|
23
37
|
# @return [RelatonBipm::BipmBibliographicItem]
|
24
38
|
def from_xml(xml)
|
data/lib/relaton_bipm/version.rb
CHANGED
data/lib/relaton_bipm.rb
CHANGED
@@ -14,6 +14,7 @@ require "relaton_bipm/bipm_bibliography"
|
|
14
14
|
require "relaton_bipm/hash_converter"
|
15
15
|
require "relaton_bipm/xml_parser"
|
16
16
|
require "relaton_bipm/index"
|
17
|
+
require "relaton_bipm/data_fetcher"
|
17
18
|
|
18
19
|
module RelatonBipm
|
19
20
|
class Error < StandardError; end
|
@@ -27,21 +28,5 @@ module RelatonBipm
|
|
27
28
|
Digest::MD5.hexdigest grammars
|
28
29
|
end
|
29
30
|
|
30
|
-
#
|
31
|
-
# Parse yaml content
|
32
|
-
#
|
33
|
-
# @param [String] yaml content
|
34
|
-
#
|
35
|
-
# @return [Hash] data
|
36
|
-
#
|
37
|
-
def parse_yaml(yaml, classes = [])
|
38
|
-
# Newer versions of Psych uses the `permitted_classes:` parameter
|
39
|
-
if YAML.method(:safe_load).parameters.map(&:last).include? :permitted_classes
|
40
|
-
YAML.safe_load(yaml, permitted_classes: classes)
|
41
|
-
else
|
42
|
-
YAML.safe_load(yaml, classes)
|
43
|
-
end
|
44
|
-
end
|
45
|
-
|
46
31
|
extend self
|
47
32
|
end
|
data/relaton_bipm.gemspec
CHANGED
@@ -12,7 +12,7 @@ Gem::Specification.new do |spec| # rubocop:disable Metrics/BlockLength
|
|
12
12
|
"bibliographic use using the BibliographicItem model"
|
13
13
|
spec.homepage = "https://github.com/relaton/relaton-bipm"
|
14
14
|
spec.license = "BSD-2-Clause"
|
15
|
-
spec.required_ruby_version = Gem::Requirement.new(">= 2.
|
15
|
+
spec.required_ruby_version = Gem::Requirement.new(">= 2.6.0")
|
16
16
|
|
17
17
|
# spec.metadata["allowed_push_host"] = "TODO: Set to 'http://mygemserver.com'"
|
18
18
|
|
@@ -33,18 +33,16 @@ Gem::Specification.new do |spec| # rubocop:disable Metrics/BlockLength
|
|
33
33
|
spec.require_paths = ["lib"]
|
34
34
|
|
35
35
|
spec.add_development_dependency "byebug"
|
36
|
-
# spec.add_development_dependency "debase"
|
37
36
|
spec.add_development_dependency "equivalent-xml", "~> 0.6"
|
38
37
|
spec.add_development_dependency "pry-byebug"
|
39
|
-
# spec.add_development_dependency "rake", "~> 13.0"
|
40
|
-
# spec.add_development_dependency "ruby-debug-ide"
|
41
38
|
spec.add_development_dependency "ruby-jing"
|
42
39
|
spec.add_development_dependency "simplecov"
|
43
40
|
spec.add_development_dependency "vcr"
|
44
41
|
spec.add_development_dependency "webmock"
|
45
42
|
|
43
|
+
spec.add_dependency "faraday", "~> 1.0"
|
46
44
|
spec.add_dependency "mechanize", "~> 2.8.0"
|
47
|
-
spec.add_dependency "relaton-bib", "~> 1.
|
45
|
+
spec.add_dependency "relaton-bib", "~> 1.12.0"
|
48
46
|
spec.add_dependency "rubyzip", "~> 2.3.0"
|
49
47
|
spec.add_dependency "serrano", "~> 1.0"
|
50
48
|
end
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: relaton-bipm
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 1.
|
4
|
+
version: 1.12.0
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Ribose Inc.
|
8
8
|
autorequire:
|
9
9
|
bindir: exe
|
10
10
|
cert_chain: []
|
11
|
-
date: 2022-06-
|
11
|
+
date: 2022-06-25 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: byebug
|
@@ -108,6 +108,20 @@ dependencies:
|
|
108
108
|
- - ">="
|
109
109
|
- !ruby/object:Gem::Version
|
110
110
|
version: '0'
|
111
|
+
- !ruby/object:Gem::Dependency
|
112
|
+
name: faraday
|
113
|
+
requirement: !ruby/object:Gem::Requirement
|
114
|
+
requirements:
|
115
|
+
- - "~>"
|
116
|
+
- !ruby/object:Gem::Version
|
117
|
+
version: '1.0'
|
118
|
+
type: :runtime
|
119
|
+
prerelease: false
|
120
|
+
version_requirements: !ruby/object:Gem::Requirement
|
121
|
+
requirements:
|
122
|
+
- - "~>"
|
123
|
+
- !ruby/object:Gem::Version
|
124
|
+
version: '1.0'
|
111
125
|
- !ruby/object:Gem::Dependency
|
112
126
|
name: mechanize
|
113
127
|
requirement: !ruby/object:Gem::Requirement
|
@@ -128,14 +142,14 @@ dependencies:
|
|
128
142
|
requirements:
|
129
143
|
- - "~>"
|
130
144
|
- !ruby/object:Gem::Version
|
131
|
-
version: 1.
|
145
|
+
version: 1.12.0
|
132
146
|
type: :runtime
|
133
147
|
prerelease: false
|
134
148
|
version_requirements: !ruby/object:Gem::Requirement
|
135
149
|
requirements:
|
136
150
|
- - "~>"
|
137
151
|
- !ruby/object:Gem::Version
|
138
|
-
version: 1.
|
152
|
+
version: 1.12.0
|
139
153
|
- !ruby/object:Gem::Dependency
|
140
154
|
name: rubyzip
|
141
155
|
requirement: !ruby/object:Gem::Requirement
|
@@ -220,7 +234,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
|
|
220
234
|
requirements:
|
221
235
|
- - ">="
|
222
236
|
- !ruby/object:Gem::Version
|
223
|
-
version: 2.
|
237
|
+
version: 2.6.0
|
224
238
|
required_rubygems_version: !ruby/object:Gem::Requirement
|
225
239
|
requirements:
|
226
240
|
- - ">="
|