litvinovg/vivo
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

VIVO-125 Create more "Google Spanish" labels

Browse source 
Add more labels by changing the filter in the label_stripper/label_inserter scripts. Improve the structure of the scripts to use an external filter.
This commit is contained in:
j2blake 2013-06-21 14:58:28 -04:00
parent b758f374ad
commit c5acf46316
5 changed files with 589 additions and 423 deletions
Download patch file Download diff file Expand all files Collapse all files

10
utilities/languageSupport/translateLabelsInOntology/label_common.rb
View file

@ -11,6 +11,9 @@ require 'rdf/ntriples'
require 'rdf/n3'
include RDF
class UsageError < StandardError; end
class FilterError < StandardError; end
class LabelCommon
# ------------------------------------------------------------------------------------
@ -21,6 +24,13 @@ class LabelCommon
public
# ------------------------------------------------------------------------------------
def self.load_filter(filter_file)
return lambda{|s| true} unless filter_file
return eval(IO.read(filter_file))
rescue
raise FilterError.new($!.message)
end
def initialize(rdf_file, &filter)
@filter = filter.nil? ? lambda{true} : filter
@graph = Graph.load(rdf_file)

52
utilities/languageSupport/translateLabelsInOntology/label_inserter.rb
View file

@ -24,7 +24,7 @@ require 'label_common'
include RDF
class LabelStripper
class LabelInserter
# ------------------------------------------------------------------------------------
private
# ------------------------------------------------------------------------------------
@ -33,29 +33,32 @@ class LabelStripper
# Parse the arguments and complain if they don't make sense.
#
def sanity_check_arguments(args)
raise "usage is: label_inserter.rb <rdf_file> <labels_input_file> <locale> <n3_output_file> [ok]" unless (4..5).include?(args.length)
if args[4].nil?
ok = false
elsif args[4].downcase == 'ok'
if args[-1].downcase == 'ok'
ok = true
args.pop
else
raise "fifth argument, if present, must be 'ok'"
ok = false
end
raise UsageError, "usage is: label_inserter.rb <rdf_file> <labels_input_file> <locale> [filter_file] <n3_output_file> [ok]" unless (4..5).include?(args.length)
n3_output_file = args.pop
raise UsageError, "File '#{n3_output_file}' already exists. specify 'ok' to overwrite it." if File.exist?(n3_output_file) && !ok
rdf_file = args[0]
raise "File '#{rdf_file}' does not exist." unless File.exist?(rdf_file)
raise UsageError, "File '#{rdf_file}' does not exist." unless File.exist?(rdf_file)
labels_input_file = args[1]
raise "File '#{labels_input_file}' does not exist." unless File.exist?(labels_input_file)
raise UsageError, "File '#{labels_input_file}' does not exist." unless File.exist?(labels_input_file)
locale = args[2]
raise "Locale should be like 'ab' or 'ab-CD'." unless /^[a-z]{2}(-[A-Z]{2})?$/ =~ locale
raise UsageError, "Locale should be like 'ab' or 'ab-CD'." unless /^[a-z]{2}(-[A-Z]{2})?$/ =~ locale
n3_output_file = args[3]
raise "File '#{n3_output_file}' already exists. specify 'ok' to overwrite it." if File.exist?(n3_output_file) && !ok
filter_file = args[3]
raise UsageError, "File '#{filter_file}' does not exist." if filter_file && !File.exist?(filter_file)
filter = LabelCommon.load_filter(filter_file)
return rdf_file, labels_input_file, locale, n3_output_file
return rdf_file, labels_input_file, locale, filter, n3_output_file
end
# ------------------------------------------------------------------------------------
@ -63,19 +66,23 @@ class LabelStripper
# ------------------------------------------------------------------------------------
def initialize(args)
@rdf_file, @labels_input_file, @locale, @n3_output_file = sanity_check_arguments(args)
@rdf_file, @labels_input_file, @locale, @filter, @n3_output_file = sanity_check_arguments(args)
rescue UsageError => e
puts "\n----------------\nUsage error\n----------------\n\n#{e}\n\n----------------\n\n"
exit
rescue FilterError => e
puts "\n----------------\nFilter file is invalid\n----------------\n\n#{e}\n\n----------------\n\n"
exit
end
def process(&filter)
filter = filter || lambda{true}
def process()
query = Query.new({
:prop => {
RDFS.label => :label,
}
})
solutions = LabelCommon.new(@rdf_file).process(query, &filter)
solutions = LabelCommon.new(@rdf_file).process(query, &@filter)
labels = IO.readlines(@labels_input_file)
@ -105,10 +112,7 @@ end
# ------------------------------------------------------------------------------------
#
#vivo_filter = lambda {|s| s.prop.start_with?("http://vivoweb.org/ontology/core#") && !s.label.to_s.strip.empty?}
vivo_filter = lambda {|s| !s.label.to_s.strip.empty?}
if File.expand_path($0) == File.expand_path(__FILE__)
stripper = LabelStripper.new(ARGV)
stripper.process(&vivo_filter)
inserter = LabelInserter.new(ARGV)
inserter.process()
end

44
utilities/languageSupport/translateLabelsInOntology/label_stripper.rb
View file

@ -32,23 +32,26 @@ class LabelStripper
# Parse the arguments and complain if they don't make sense.
#
def sanity_check_arguments(args)
raise "usage is: label_stripper.rb <rdf_file> <labels_output_file> [ok]" unless (2..3).include?(args.length)
if args[2].nil?
ok = false
elsif args[2].downcase == 'ok'
if args[-1].downcase == 'ok'
ok = true
args.pop
else
raise "third argument, if present, must be 'ok'"
ok = false
end
raise UsageError, "usage is: label_stripper.rb <rdf_file> [filter_file] <labels_output_file> [ok]" unless (2..3).include?(args.length)
output_file = args.pop
raise UsageError, "File '#{output_file}' already exists. specify 'ok' to overwrite it." if File.exist?(output_file) && !ok
rdf_file = args[0]
raise "File '#{rdf_file}' does not exist." unless File.exist?(rdf_file)
raise UsageError, "File '#{rdf_file}' does not exist." unless File.exist?(rdf_file)
labels_output_file = args[1]
raise "File '#{labels_output_file}' already exists. specify 'ok' to overwrite it." if File.exist?(labels_output_file) && !ok
filter_file = args[1]
raise UsageError, "File '#{filter_file}' does not exist." if filter_file && !File.exist?(filter_file)
filter = LabelCommon.load_filter(filter_file)
return rdf_file, labels_output_file
return rdf_file, filter, output_file
end
# ------------------------------------------------------------------------------------
@ -56,19 +59,23 @@ class LabelStripper
# ------------------------------------------------------------------------------------
def initialize(args)
@rdf_file, @labels_output_file = sanity_check_arguments(args)
@rdf_file, @filter, @labels_output_file = sanity_check_arguments(args)
rescue UsageError => e
puts "\n----------------\nUsage error\n----------------\n\n#{e}\n\n----------------\n\n"
exit
rescue FilterError => e
puts "\n----------------\nFilter file is invalid\n----------------\n\n#{e}\n\n----------------\n\n"
exit
end
def process(&filter)
filter = filter || lambda{true}
def process()
query = Query.new({
:prop => {
RDFS.label => :label,
}
})
solutions = LabelCommon.new(@rdf_file).process(query, &filter)
solutions = LabelCommon.new(@rdf_file).process(query, &@filter)
File.open(@labels_output_file, 'w') do |f|
solutions.each do |s|
@ -88,10 +95,7 @@ end
# ------------------------------------------------------------------------------------
#
#vivo_filter = lambda {|s| s.prop.start_with?("http://vivoweb.org/ontology/core#") && !s.label.to_s.strip.empty?}
vivo_filter = lambda {|s| !s.label.to_s.strip.empty?}
if File.expand_path($0) == File.expand_path(__FILE__)
stripper = LabelStripper.new(ARGV)
stripper.process(&vivo_filter)
stripper.process()
end

12
utilities/languageSupport/translateLabelsInOntology/vivo_core_filter Normal file
View file

@ -0,0 +1,12 @@
lambda {|s|
(
s.prop.start_with?("http://vivoweb.org/ontology/core#") ||
s.prop.start_with?("http://vitro.mannlib.cornell.edu/ns/") ||
s.prop.start_with?("http://purl.org/ontology/bibo/") ||
s.prop.start_with?("http://xmlns.com/foaf/0.1/") ||
s.prop.start_with?("http://purl.org/dc/terms/") ||
s.prop.start_with?("http://purl.org/dc/elements/1.1/") ||
s.prop.start_with?("http://purl.org/NET/c4dm/event.owl#") ) &&
!s.label.to_s.strip.empty?
}