ddi-parser 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- data/lib/ddi-parser.rb +161 -0
- data/lib/models/catalog.rb +10 -0
- data/lib/models/category.rb +11 -0
- data/lib/models/category_statistic.rb +11 -0
- data/lib/models/study.rb +11 -0
- data/lib/models/study_date.rb +11 -0
- data/lib/models/summary_stat.rb +11 -0
- data/lib/models/variable.rb +11 -0
- data/lib/models/version.rb +5 -0
- metadata +91 -0
data/lib/ddi-parser.rb
ADDED
@@ -0,0 +1,161 @@
|
|
1
|
+
require 'rubygems'
|
2
|
+
require 'libxml'
|
3
|
+
require 'models/catalog'
|
4
|
+
require 'models/category'
|
5
|
+
require 'models/category_statistic'
|
6
|
+
require 'models/study'
|
7
|
+
require 'models/study_date'
|
8
|
+
require 'models/variable'
|
9
|
+
require 'models/summary_stat'
|
10
|
+
|
11
|
+
module DDI
|
12
|
+
class Parser
|
13
|
+
|
14
|
+
#Given a DDI metadata file, parse it and return study information
|
15
|
+
#
|
16
|
+
#Returns a Nesstar::Study object
|
17
|
+
def parse ddi_file
|
18
|
+
catalog = DDI::Catalog.new
|
19
|
+
study = DDI::Study.new
|
20
|
+
study_info_hash = Hash.new
|
21
|
+
parser = LibXML::XML::Parser.file(ddi_file)
|
22
|
+
doc = parser.parse
|
23
|
+
studynodes = doc.find('//stdyDscr')
|
24
|
+
abstracts = studynodes[0].find('//abstract')
|
25
|
+
abstract = ""
|
26
|
+
abstracts.each do |ab|
|
27
|
+
abstract << ab.content.strip
|
28
|
+
end
|
29
|
+
abstract.strip!
|
30
|
+
study.abstract = abstract
|
31
|
+
study.title = studynodes[0].find('//stdyDscr/citation/titlStmt/titl')[0].first.content.strip
|
32
|
+
study.id = studynodes[0].find('//IDNo')[0].first.content.strip
|
33
|
+
|
34
|
+
#start and finish dates for study
|
35
|
+
dates = []
|
36
|
+
date = studynodes[0].find('//sumDscr/collDate')
|
37
|
+
date.each do |d|
|
38
|
+
a = d.attributes
|
39
|
+
study_date = DDI::StudyDate.new
|
40
|
+
study_date.type = a.get_attribute('event').value.strip
|
41
|
+
study_date.date = a.get_attribute('date').value.strip
|
42
|
+
dates.push(study_date)
|
43
|
+
end
|
44
|
+
study.dates = dates
|
45
|
+
study.sampling_procedure = studynodes[0].find('//sampProc')[0].first.content.strip unless studynodes[0].find('//sampProc')[0] == nil
|
46
|
+
# study.weight = studynodes[0].find('//sampProc')[0].first.content
|
47
|
+
study.variables = get_variable_information doc
|
48
|
+
return study
|
49
|
+
end
|
50
|
+
|
51
|
+
private
|
52
|
+
|
53
|
+
#information about the variables
|
54
|
+
def get_variable_information doc
|
55
|
+
variables = []
|
56
|
+
variable_info_hash = Hash.new
|
57
|
+
docnodes = doc.find('//dataDscr')
|
58
|
+
vargroups = docnodes[0].find('//dataDscr/varGrp')
|
59
|
+
vargroups.each do |vargroup|
|
60
|
+
#hash which holds all the variable groups
|
61
|
+
a = vargroup.attributes
|
62
|
+
groups = a.get_attribute('var')
|
63
|
+
if groups != nil
|
64
|
+
groups = a.get_attribute('var')
|
65
|
+
variable_info_hash[vargroup.find('./labl')[0].first.content] = groups.value.split(' ')
|
66
|
+
# else
|
67
|
+
# variable_info_hash[vargroup.find('./labl')[0].first.content] = groups.value.split(' ')
|
68
|
+
end
|
69
|
+
end
|
70
|
+
vars = docnodes[0].find('//dataDscr/var')
|
71
|
+
vars.each do |var|
|
72
|
+
variable = DDI::Variable.new
|
73
|
+
var_attr = var.attributes
|
74
|
+
variable.id = var_attr.get_attribute('ID').value.strip unless var_attr.get_attribute('ID') == nil
|
75
|
+
variable.name = var_attr.get_attribute('name').value.strip unless var_attr.get_attribute('name') == nil
|
76
|
+
variable.file = var_attr.get_attribute('files').value.strip unless var_attr.get_attribute('files') == nil
|
77
|
+
variable.interval = var_attr.get_attribute('intrvl').value.strip unless var_attr.get_attribute('intrvl') == nil
|
78
|
+
variable.label = var.find('./labl')[0].content.strip unless var.find('./labl')[0] == nil
|
79
|
+
rng = var.find('./valrng')
|
80
|
+
if rng != nil
|
81
|
+
if rng[0] != nil
|
82
|
+
range_attr = rng[0].first.attributes
|
83
|
+
max_val = range_attr.get_attribute('max')
|
84
|
+
variable.max = max_val.value.strip unless max_val == nil
|
85
|
+
min_val = range_attr.get_attribute('min')
|
86
|
+
variable.min = min_val.value.strip unless min_val == nil
|
87
|
+
end
|
88
|
+
end
|
89
|
+
q = var.find('./qstn')
|
90
|
+
if q[0] != nil
|
91
|
+
ql = q[0].find('./qstnLit')
|
92
|
+
if ql != nil
|
93
|
+
if ql[0] != nil
|
94
|
+
variable.question = ql[0].first.content.strip
|
95
|
+
end
|
96
|
+
end
|
97
|
+
iv = q[0].find('./ivuInstr')
|
98
|
+
if iv != nil
|
99
|
+
if iv[0] != nil
|
100
|
+
variable.interview_instruction = iv[0].first.content.strip
|
101
|
+
end
|
102
|
+
end
|
103
|
+
end
|
104
|
+
stats = var.find('./sumStat')
|
105
|
+
summary_stats = []
|
106
|
+
stats.each do |stat|
|
107
|
+
a = stat.attributes
|
108
|
+
# summary_stats[a.get_attribute('type').value] = stat.first.content
|
109
|
+
statistic = DDI::SummaryStat.new
|
110
|
+
statistic.type = a.get_attribute('type').value.strip
|
111
|
+
statistic.value = stat.first.content.strip
|
112
|
+
summary_stats.push(statistic)
|
113
|
+
end
|
114
|
+
variable.summary_stats = summary_stats
|
115
|
+
catgry = var.find('./catgry')
|
116
|
+
categories = []
|
117
|
+
#categories in ddi are value domains in mb
|
118
|
+
catgry.each do |cat|
|
119
|
+
category = DDI::Category.new
|
120
|
+
valxml = cat.find('./catValu')
|
121
|
+
if valxml != nil && valxml[0] != nil
|
122
|
+
category.value = valxml[0].first.content.strip unless valxml[0].first == nil
|
123
|
+
else
|
124
|
+
category.value = 'N/A'
|
125
|
+
end
|
126
|
+
labxml = cat.find('./labl')
|
127
|
+
if labxml != nil && labxml[0] != nil
|
128
|
+
category.label = labxml[0].first.content.strip unless labxml[0].first == nil
|
129
|
+
else
|
130
|
+
category.label = 'N/A'
|
131
|
+
end
|
132
|
+
catstats = cat.find('./catStat')
|
133
|
+
category_statistics = []
|
134
|
+
catstats.each do |catstat|
|
135
|
+
category_statistic = DDI::CategoryStatistic.new
|
136
|
+
a = catstat.attributes
|
137
|
+
if a != nil && a.get_attribute('type') != nil
|
138
|
+
category_statistic.type = a.get_attribute('type').value.strip
|
139
|
+
category_statistic.value = catstat.first.content.strip unless catstat.first == nil
|
140
|
+
category_statistics.push(category_statistic)
|
141
|
+
end
|
142
|
+
end
|
143
|
+
category.category_statistics = category_statistics
|
144
|
+
categories.push(category)
|
145
|
+
end
|
146
|
+
#what group is the variable in
|
147
|
+
variable_info_hash.each_key do |key|
|
148
|
+
if variable_info_hash[key].include?(variable.id)
|
149
|
+
variable.group = key.strip
|
150
|
+
break
|
151
|
+
end
|
152
|
+
end
|
153
|
+
|
154
|
+
variable.categories = categories
|
155
|
+
variables.push(variable)
|
156
|
+
end
|
157
|
+
return variables
|
158
|
+
end
|
159
|
+
|
160
|
+
end
|
161
|
+
end
|
@@ -0,0 +1,10 @@
|
|
1
|
+
module DDI
|
2
|
+
#A Nesstar catalog object, can contain 1 or more studies (ie datasets)
|
3
|
+
class Catalog
|
4
|
+
|
5
|
+
attr_reader :studies, :label, :description, :nesstar_id, :nesstar_uri
|
6
|
+
attr_writer :studies, :label, :description, :nesstar_id, :nesstar_uri
|
7
|
+
|
8
|
+
end
|
9
|
+
|
10
|
+
end
|
data/lib/models/study.rb
ADDED
@@ -0,0 +1,11 @@
|
|
1
|
+
module DDI
|
2
|
+
|
3
|
+
#Contains a set of variables and belongs to a catalog
|
4
|
+
class Study
|
5
|
+
|
6
|
+
attr_reader :variables, :abstract, :title, :id, :dates, :sampling_procedure, :weight, :nesstar_id, :nesstar_uri
|
7
|
+
attr_writer :variables, :abstract, :title, :id, :dates, :sampling_procedure, :weight, :nesstar_id, :nesstar_uri
|
8
|
+
|
9
|
+
end
|
10
|
+
|
11
|
+
end
|
@@ -0,0 +1,11 @@
|
|
1
|
+
module DDI
|
2
|
+
|
3
|
+
#Information about a variable/column in a dataset
|
4
|
+
class Variable
|
5
|
+
|
6
|
+
attr_reader :name, :label, :group, :id, :file, :interval, :max, :min, :question, :interview_instruction, :summary_stats, :categories
|
7
|
+
attr_writer :name, :label, :group, :id, :file, :interval, :max, :min, :question, :interview_instruction, :summary_stats, :categories
|
8
|
+
|
9
|
+
end
|
10
|
+
|
11
|
+
end
|
metadata
ADDED
@@ -0,0 +1,91 @@
|
|
1
|
+
--- !ruby/object:Gem::Specification
|
2
|
+
name: ddi-parser
|
3
|
+
version: !ruby/object:Gem::Version
|
4
|
+
hash: 29
|
5
|
+
prerelease: false
|
6
|
+
segments:
|
7
|
+
- 0
|
8
|
+
- 0
|
9
|
+
- 1
|
10
|
+
version: 0.0.1
|
11
|
+
platform: ruby
|
12
|
+
authors:
|
13
|
+
- Ian Dunlop
|
14
|
+
autorequire:
|
15
|
+
bindir: bin
|
16
|
+
cert_chain: []
|
17
|
+
|
18
|
+
date: 2011-05-06 00:00:00 +01:00
|
19
|
+
default_executable:
|
20
|
+
dependencies:
|
21
|
+
- !ruby/object:Gem::Dependency
|
22
|
+
name: libxml-ruby
|
23
|
+
prerelease: false
|
24
|
+
requirement: &id001 !ruby/object:Gem::Requirement
|
25
|
+
none: false
|
26
|
+
requirements:
|
27
|
+
- - "="
|
28
|
+
- !ruby/object:Gem::Version
|
29
|
+
hash: 27
|
30
|
+
segments:
|
31
|
+
- 1
|
32
|
+
- 1
|
33
|
+
- 4
|
34
|
+
version: 1.1.4
|
35
|
+
type: :runtime
|
36
|
+
version_requirements: *id001
|
37
|
+
description: This gem parses ddi metadata files
|
38
|
+
email:
|
39
|
+
- ian.dunlop@manchester.ac.uk
|
40
|
+
executables: []
|
41
|
+
|
42
|
+
extensions: []
|
43
|
+
|
44
|
+
extra_rdoc_files: []
|
45
|
+
|
46
|
+
files:
|
47
|
+
- lib/ddi-parser.rb
|
48
|
+
- lib/models/catalog.rb
|
49
|
+
- lib/models/category.rb
|
50
|
+
- lib/models/category_statistic.rb
|
51
|
+
- lib/models/study.rb
|
52
|
+
- lib/models/study_date.rb
|
53
|
+
- lib/models/summary_stat.rb
|
54
|
+
- lib/models/variable.rb
|
55
|
+
- lib/models/version.rb
|
56
|
+
has_rdoc: true
|
57
|
+
homepage: http://github.com/mygrid/ddi-parser
|
58
|
+
licenses: []
|
59
|
+
|
60
|
+
post_install_message:
|
61
|
+
rdoc_options: []
|
62
|
+
|
63
|
+
require_paths:
|
64
|
+
- lib
|
65
|
+
required_ruby_version: !ruby/object:Gem::Requirement
|
66
|
+
none: false
|
67
|
+
requirements:
|
68
|
+
- - ">="
|
69
|
+
- !ruby/object:Gem::Version
|
70
|
+
hash: 3
|
71
|
+
segments:
|
72
|
+
- 0
|
73
|
+
version: "0"
|
74
|
+
required_rubygems_version: !ruby/object:Gem::Requirement
|
75
|
+
none: false
|
76
|
+
requirements:
|
77
|
+
- - ">="
|
78
|
+
- !ruby/object:Gem::Version
|
79
|
+
hash: 3
|
80
|
+
segments:
|
81
|
+
- 0
|
82
|
+
version: "0"
|
83
|
+
requirements: []
|
84
|
+
|
85
|
+
rubyforge_project: ddi-parser
|
86
|
+
rubygems_version: 1.3.7
|
87
|
+
signing_key:
|
88
|
+
specification_version: 3
|
89
|
+
summary: API for parsing ddi metadata files and returning results
|
90
|
+
test_files: []
|
91
|
+
|