require 'spec_helper'
module GutenbergRdf
describe Rdf do
let(:xml) do
'
2006-09-28
en
Project Gutenberg
Public domain in the USA.
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects an id" do
expect(rdf.id).to eql "98765"
end
it "expects a published date" do
expect(rdf.published).to eql "2006-09-28"
end
it "expects a publisher" do
expect(rdf.publisher).to eql "Project Gutenberg"
end
it "expects a language" do
expect(rdf.language).to eql "en"
end
it "expects the rights" do
expect(rdf.rights).to eql "Public domain in the USA."
end
describe "#type" do
let(:xml) do
'
Text
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expect the type of entity" do
expect(rdf.type).to eql 'Text'
end
end
describe "Titles" do
let(:xml) do
'
A Great Title
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects a title" do
expect(rdf.title).to eql 'A Great Title'
end
it "expects subtitle to be empty" do
expect(rdf.subtitle).to eql ''
end
context "with a title and subtitle, on separate lines" do
let(:xml) do
'
A Great Multi-Title
Or, a Subtitle
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects the title to be the first line" do
expect(rdf.title).to eql 'A Great Multi-Title'
end
it "expects the subtitle to be the second line" do
expect(rdf.subtitle).to eql 'Or, a Subtitle'
end
end
context "when title:subtitle are separated by a colon" do
let(:xml) do
'
A Great Multi-Title: And a Subtitle
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects a title" do
expect(rdf.title).to eql 'A Great Multi-Title'
end
it "expects a subtitle" do
expect(rdf.subtitle).to eql 'And a Subtitle'
end
end
context "when title; and subtitle are separated by a semi-colon" do
let(:xml) do
'
A Great Multi-Title; Or, a Subtitle
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects a title" do
expect(rdf.title).to eql 'A Great Multi-Title'
end
it "expects a subtitle" do
expect(rdf.subtitle).to eql 'Or, a Subtitle'
end
context "...except when subtitles already exists" do
let(:xml) do
'
A Great Multi-Title; and some other text
Then a Subtitle on a newline
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects a title" do
expect(rdf.title).to eql 'A Great Multi-Title; and some other text'
end
it "expects a subtitle" do
expect(rdf.subtitle).to eql 'Then a Subtitle on a newline'
end
end
end
end
describe "#authors" do
let(:xml) do
'
1830
1905
Dodge, Mary Mapes
Dodge, Mary
Verschillende
Various
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects a Array" do
expect(rdf.authors.class).to be Array
end
it "expects correct number to be returned" do
expect(rdf.authors.count).to be 2
end
it "expects an author object" do
expect(rdf.authors.first.class).to be Rdf::Agent
end
end
describe "#subjects" do
let(:xml) do
%q{
Children's literature -- Periodicals
Children's periodicals, American
PZ
}
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects correct number to be returned" do
expect(rdf.subjects.count).to be 2
end
it "expects the correct data" do
expect(rdf.subjects.first).to eql "Children's literature -- Periodicals"
expect(rdf.subjects.last).to eql "Children's periodicals, American"
end
end
describe "#covers" do
describe "official PG covers" do
let(:xml) do
'
http://www.gutenberg.org/files/12345/12345-h/images/cover.jpg
92652
application/epub+zip
2013-09-21T19:22:32.115259
10856
image/jpeg
2013-09-21T19:22:34.484114
1904
image/jpeg
2013-09-21T19:22:34.379124
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects the correct number of entries returned" do
expect(rdf.covers.count).to be 3
end
it "expects those to be used" do
expect(rdf.covers[0]).to eql 'http://www.gutenberg.org/ebooks/12345.cover.medium'
expect(rdf.covers[1]).to eql 'http://www.gutenberg.org/ebooks/12345.cover.small'
end
it "expects any other images to be listed after the official ones" do
expect(rdf.covers[2]).to eql 'http://www.gutenberg.org/files/12345/12345-h/images/cover.jpg'
end
end
describe "HTML ebook cover image" do
let(:xml) do
'
file:///public/vhost/g/gutenberg/html/files/12345/12345-rst/images/cover.jpg
file:///public/vhost/g/gutenberg/html/files/12345/12345-h/images/cover.jpg
http://www.gutenberg.org/files/12345/12345-h/images/cover.jpg
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects only unique entries" do
expect(rdf.covers.count).to be 2
end
it "should convert File URIs to the Gutenberg URL" do
expect(rdf.covers.first).to match 'http://www.gutenberg.org'
end
it "expects the covers to be listed in the correct order" do
expect(rdf.covers[0]).to eql 'http://www.gutenberg.org/files/12345/12345-h/images/cover.jpg'
expect(rdf.covers[1]).to eql 'http://www.gutenberg.org/files/12345/12345-rst/images/cover.jpg'
end
end
end
describe "#ebook" do
let(:xml) do
'
293684
text/plain; charset=utf-8
2010-02-16T08:29:52.373092
116685
application/zip
text/plain; charset=us-ascii
2006-09-28T12:37:26
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "expects the correct number of entries" do
expect(rdf.ebooks.count).to be 2
end
it "expects an entry Hash to have the correct keys" do
expect(rdf.ebooks.first).to have_key :uri
expect(rdf.ebooks.first).to have_key :mime_type
expect(rdf.ebooks.first).to have_key :encoding
expect(rdf.ebooks.first).to have_key :modified
end
it "expcts the modified value to be a DateTime" do
expect(rdf.ebooks.first[:modified].class).to be DateTime
end
it "should return the URL" do
expect(rdf.ebooks.first[:uri]).to eql 'http://www.gutenberg.org/ebooks/98765.txt.utf-8'
end
it "should return the mime_type" do
expect(rdf.ebooks.first[:mime_type]).to eql 'text/plain'
end
it "should return the encoding" do
expect(rdf.ebooks.first[:encoding]).to eql 'utf-8'
end
it "should return the modified datetime" do
expect(rdf.ebooks.first[:modified].to_s).to eql '2010-02-16T08:29:52-07:00'
end
context "when there are two mime-types" do
let(:xml) do
'
116685
application/zip
text/plain; charset=us-ascii
2006-09-28T12:37:26
'
end
let(:rdf) { Rdf.new(Nokogiri::XML(xml)) }
it "should use just the first one" do
expect(rdf.ebooks.first[:mime_type]).to eql 'application/zip'
end
it "expects the encoding to be an empty string" do
expect(rdf.ebooks.first[:encoding]).to eql ''
end
end
end
end
end