wax_iiif 0.0.2 → 0.1.0

data/lib/{iiif_s3 → wax_iiif}/utilities/pdf_splitter.rb

@@ -1,9 +1,7 @@
-require "mini_magick"
+require 'mini_magick'
 
-module IiifS3
+module WaxIiif
   module Utilities
-
-    #
     # Class PdfSplitter is a utility function designed to convert a PDF into a stack of images.
     #
     # @author David Newbury <david.newbury@gmail.com>
@@ -15,36 +13,35 @@ module IiifS3
       # @param [Hash] options The configuration hash
       #
       # @return [Array<String>] The paths to the generated files
-      # 
-      def self.split(path, options={})
-
-        output_dir = options.fetch(:output_dir, ".")
-        verbose = options.fetch(:verbose, false)
-        puts "processing #{path}" if verbose
-        name = File.basename(path, File.extname(path))
+      def self.split(path, options = {})
+        puts "processing #{path}" if options.fetch(:verbose, false)
 
-        pages = []
-
-        pdf = MiniMagick::Image.open(path)
+        name        = File.basename(path, File.extname(path))
+        output_dir  = "#{options.fetch(:output_dir, '.')}/#{name}"
+        pages       = []
+        pdf         = MiniMagick::Image.open(path)
+        max_digits  = pdf.pages.length.digits.length
 
         pdf.pages.each_with_index do |page, index|
-          page_file_name = "#{output_dir}/#{name}_#{index+1}.jpg"
-          
+          FileUtils.mkdir_p output_dir
+          page_file_name = "#{output_dir}/#{index.to_s.rjust(max_digits, '0')}.jpg"
+
           MiniMagick::Tool::Convert.new do |convert|
-            convert.density("300")
-            convert.units("PixelsPerInch")
-            convert << page.path  
-            convert.quality("80")
-            convert.colorspace("sRGB")
-            convert.interlace("none")
+            convert.density('300')
+            convert.units('PixelsPerInch')
+            convert << page.path
+            convert.quality('80')
+            convert.colorspace('sRGB')
+            convert.interlace('none')
             convert.flatten
             convert << page_file_name
           end
           pages.push(page_file_name)
         end
         GC.start
+
         pages
       end
     end
   end
-end
+end

data/spec/base_properties_spec.rb

@@ -1,22 +1,18 @@
 require 'spec_helper'
 
-shared_examples "base properties" do
-  it "has a description" do
-    test_string = "test description"
+shared_examples 'base properties' do
+  it 'has a description' do
+    test_string = 'test description'
     @object.description = test_string
     expect(@object.description).to eq test_string
   end
-  it "has an @id" do
+  it 'has an @id' do
     expect(@object.id).not_to be_nil
   end
-  it "has a valid @id" do
-    expect(@object.id).to include IiifS3::Config::DEFAULT_URL
+  it 'has a valid @id' do
+    expect(@object.id).to include WaxIiif::Config::DEFAULT_URL
   end
-  it "encodes @ids when set" do
-    @object.id = "http://www.example.com/a space"
-    expect(@object.id).to include "%20"
-  end
-  it "reveals the type" do
+  it 'reveals the type' do
     expect(@object.type).to eq described_class::TYPE
   end
-end
+end

data/spec/shared_contexts.rb

@@ -4,112 +4,48 @@ RSpec.shared_context "fake variants" do
       "full" => OpenStruct.new(:id => "http://www.example.com/images/1", :width => 1000, :height => 1200),
       "thumbnail" => OpenStruct.new(:id => "http://www.example.com/images/1", :width => 100, :height => 120)
     }
-    @fake_image_info = '{
-  "@context": "http://iiif.io/api/image/2/context.json",
-  "@id": "http://www.example.com/test/1",
-  "protocol": "http://iiif.io/api/image",
-  "width": 1000,
-  "height": 1200,
-  "sizes": [
-    {
-      "width": 1000,
-      "height": 1200
-    },
-    {
-      "width": 100,
-      "height": 120
-    }
-  ],
-  "profile": [
-    "http://iiif.io/api/image/2/level0.json",
-    {
-      "supports": [
-        "cors",
-        "sizeByWhListed",
-        "baseUriRedirect"
-      ]
-    }
-  ]
-}'
+    @fake_image_info = '{"@context":"http://iiif.io/api/image/2/context.json","@id":"http://www.example.com/test/1","protocol":"http://iiif.io/api/image","width":1000,"height":1200,"sizes":[{"width":1000,"height":1200},{"width":100,"height":120}],"profile":["http://iiif.io/api/image/2/level0.json",{"supports":["cors","sizeByWhListed","baseUriRedirect"]}]}'
   end
 end
 
 RSpec.shared_context "fake data" do
   include_context("fake variants")
   before(:example) do
-    
-    @fake_collection = '{
-  "@context": "http://iiif.io/api/presentation/2/context.json",
-  "@id": "http://0.0.0.0/collection/name.json",
-  "@type": "sc:Collection",
-  "label": "Test Data",
-  "manifests": [
-    {
-      "@id": "http://0.0.0.0/1/manifest.json",
-      "@type": "sc:Manifest",
-      "label": "test label"
-    }
-  ]
-}'
 
-    @fake_data = ImageRecord.new({
+    @fake_collection = '{"@context":"http://iiif.io/api/presentation/2/context.json","@id":"http://0.0.0.0/collection/test.json","@type":"sc:Collection","label":"test","manifests":[{"@id":"http://0.0.0.0/1/manifest.json","@type":"sc:Manifest","label":"testlabel"}]}'
+
+    @fake_data = [ WaxIiif::ImageRecord.new({
+      "manifest_id" => 1,
       "id" => 1,
-      "page_number" => "1",
       "image_path" => "./spec/data/test.jpg",
       "is_primary" => true,
       "variants" => @fake_variants,
       "label" => "test label",
       "attribution" => "All rights reserved",
       "logo" => "http://www.example.com/logo.jpg"
-    })
-    @fake_manifest = '{
-  "@context": "http://iiif.io/api/presentation/2/context.json",
-  "@id": "http://0.0.0.0/1/manifest.json",
-  "@type": "sc:Manifest",
-  "label": "test label",
-  "attribution": "All rights reserved",
-  "logo": "http://www.example.com/logo.jpg",
-  "thumbnail": "http://www.example.com/images/1/full/100,/0/default.jpg",
-  "viewingDirection": "left-to-right",
-  "viewingHint": "individuals",
-  "sequences": [
-    {
-      "@id": "http://0.0.0.0/1/sequence/default.json",
-      "@type": "sc:Sequence",
-      "canvases": [
-        {
-          "@type": "sc:Canvas",
-          "@id": "http://0.0.0.0/1/canvas/front.json",
-          "label": "front",
-          "width": 2000,
-          "height": 2400,
-          "thumbnail": "http://www.example.com/images/1/full/100,/0/default.jpg",
-          "images": [
-            {
-              "@type": "oa:Annotation",
-              "@id": "http://0.0.0.0/1/annotation/front.json",
-              "motivation": "sc:painting",
-              "resource": {
-                "@id": "http://www.example.com/images/1/full/full/0/default.jpg",
-                "@type": "dcterms:Image",
-                "format": "image/jpeg",
-                "service": {
-                  "@context": "http://iiif.io/api/image/2/context.json",
-                  "@id": "http://www.example.com/images/1",
-                  "profile": "http://iiif.io/api/image/2/level0.json"
-                },
-                "width": 1000,
-                "height": 1200
-              },
-              "on": "http://0.0.0.0/1/canvas/front.json"
-            }
-          ]
-        }
-      ]
-    }
-  ]
-}'
+    }),
+    WaxIiif::ImageRecord.new({
+      "manifest_id" => 1,
+      "id" => 2,
+      "image_path" => "./spec/data/test.jpg",
+      "is_primary" => false,
+      "variants" => @fake_variants,
+      "label" => "test label",
+      "attribution" => "All rights reserved",
+      "logo" => "http://www.example.com/logo.jpg"
+    }),
+    WaxIiif::ImageRecord.new({
+      "id" => 3,
+      "image_path" => "./spec/data/test.jpg",
+      "is_primary" => true,
+      "variants" => @fake_variants,
+      "label" => "test label",
+      "attribution" => "All rights reserved",
+      "logo" => "http://www.example.com/logo.jpg"
+    })]
 
+    @fake_manifest_1 = '{"@context":"http://iiif.io/api/presentation/2/context.json","@id":"http://0.0.0.0/1/manifest.json","@type":"sc:Manifest","label":"testlabel","attribution":"Allrightsreserved","logo":"http://www.example.com/logo.jpg","thumbnail":"http://www.example.com/images/1/full/100,/0/default.jpg","viewingDirection":"left-to-right","viewingHint":"individuals","sequences":[{"@id":"http://0.0.0.0/sequence/1.json","@type":"sc:Sequence","canvases":[{"@type":"sc:Canvas","@id":"http://0.0.0.0/canvas/1.json","label":"front","width":2000,"height":2400,"thumbnail":"http://www.example.com/images/1/full/100,/0/default.jpg","images":[{"@type":"oa:Annotation","@id":"http://0.0.0.0/annotation/1.json","motivation":"sc:painting","resource":{"@id":"http://www.example.com/images/1/full/full/0/default.jpg","@type":"dcterms:Image","format":"image/jpeg","service":{"@context":"http://iiif.io/api/image/2/context.json","@id":"http://www.example.com/images/1","profile":"http://iiif.io/api/image/2/level0.json"},"width":1000,"height":1200},"on":"http://0.0.0.0/canvas/1.json"}]},{"@type":"sc:Canvas","@id":"http://0.0.0.0/canvas/2.json","label":"front","width":2000,"height":2400,"thumbnail":"http://www.example.com/images/1/full/100,/0/default.jpg","images":[{"@type":"oa:Annotation","@id":"http://0.0.0.0/annotation/2.json","motivation":"sc:painting","resource":{"@id":"http://www.example.com/images/1/full/full/0/default.jpg","@type":"dcterms:Image","format":"image/jpeg","service":{"@context":"http://iiif.io/api/image/2/context.json","@id":"http://www.example.com/images/1","profile":"http://iiif.io/api/image/2/level0.json"},"width":1000,"height":1200},"on":"http://0.0.0.0/canvas/2.json"}]}]}]}'
 
+    @fake_manifest_3 = '{"@context":"http://iiif.io/api/presentation/2/context.json","@id":"http://0.0.0.0/3/manifest.json","@type":"sc:Manifest","label":"testlabel","attribution":"Allrightsreserved","logo":"http://www.example.com/logo.jpg","thumbnail":"http://www.example.com/images/1/full/100,/0/default.jpg","viewingDirection":"left-to-right","viewingHint":"individuals","sequences":[{"@id":"http://0.0.0.0/sequence/3.json","@type":"sc:Sequence","canvases":[{"@type":"sc:Canvas","@id":"http://0.0.0.0/canvas/3.json","label":"front","width":2000,"height":2400,"thumbnail":"http://www.example.com/images/1/full/100,/0/default.jpg","images":[{"@type":"oa:Annotation","@id":"http://0.0.0.0/annotation/3.json","motivation":"sc:painting","resource":{"@id":"http://www.example.com/images/1/full/full/0/default.jpg","@type":"dcterms:Image","format":"image/jpeg","service":{"@context":"http://iiif.io/api/image/2/context.json","@id":"http://www.example.com/images/1","profile":"http://iiif.io/api/image/2/level0.json"},"width":1000,"height":1200},"on":"http://0.0.0.0/canvas/3.json"}]}]}]}'
   end
-end
+end

data/spec/spec_helper.rb

@@ -5,8 +5,8 @@ Dotenv.load
 require 'simplecov'
 SimpleCov.start
 
-require "shared_contexts"
+require 'shared_contexts'
 require 'wax_iiif'
 
-ENV["TEST_INTERNET_CONNECTIVITY"] ||= nil
-ENV["SKIP_EXPENSIVE_TESTS"] ||= nil
+ENV['TEST_INTERNET_CONNECTIVITY'] ||= nil
+ENV['SKIP_EXPENSIVE_TESTS'] ||= nil

data/spec/wax_iiif/builder_spec.rb

@@ -0,0 +1,143 @@
+require 'spec_helper'
+require 'ostruct'
+require 'fileutils'
+
+describe WaxIiif::Builder do
+  let (:iiif) { WaxIiif::Builder.new() }
+  let (:test_object_0) { WaxIiif::ImageRecord.new({'id' => 'img1' }) }
+  let (:test_object_1) { WaxIiif::ImageRecord.new({'id' => 'img2', 'manifest_id' => 'm1'}) }
+  let (:test_object_2) { WaxIiif::ImageRecord.new({'id' => 'img3', 'manifest_id' => 'm1' }) }
+  let (:test_object_2) { WaxIiif::ImageRecord.new({'id' => 'img4', 'manifest_id' => 'm2' }) }
+  let (:data) {[test_object_0, test_object_1,test_object_2]}
+
+  context 'When initializing' do
+    it 'generates manifests' do
+      expect(iiif.manifests).to eq(Array.new)
+    end
+    it 'uses the default config' do
+      expect(iiif.config).to eq(WaxIiif::Config.new)
+    end
+    it 'will accept a configuration hash' do
+      opts = {tile_width: 99}
+      iiif2 = WaxIiif::Builder.new(opts)
+      expect(iiif2.config.tile_width).to eq(99)
+    end
+  end
+
+  context 'loading data' do
+    it 'will accept an array of objects' do
+      iiif.load(data)
+      expect(iiif.data).to eq(data)
+    end
+    it 'will accept a single object' do
+      iiif.load(test_object_1)
+      expect(iiif.data).to eq([test_object_1])
+    end
+
+    it 'will error if the data is bad' do
+      expect{iiif.load({random: 'hash'})}.to raise_error(WaxIiif::Error::InvalidImageData)
+      expect{iiif.load([{random: 'hash'}])}.to raise_error(WaxIiif::Error::InvalidImageData)
+    end
+  end
+
+  context 'when processing data' do
+    include_context('fake variants')
+    include_context('fake data')
+
+    before(:example) do
+      @iiif = WaxIiif::Builder.new({base_url: 'http://0.0.0.0', verbose: true, thumbnail_size: 120})
+      @iiif.load(@fake_data)
+      allow(@iiif).to receive(:generate_tiles) {nil}
+      allow(@iiif).to receive(:generate_variants) {@fake_variants}
+    end
+    it 'does not fail with no data' do
+      expect {iiif.process_data}.not_to raise_error
+    end
+
+    it 'does not fail with real data' do
+      expect {@iiif.process_data}.not_to raise_error
+    end
+
+    it ' passes the Temporary Manifest Check' do
+      @iiif.process_data
+      expect(@iiif.manifests.count).to eq 2
+      expect(@iiif.manifests.first.to_json.delete(" \t\r\n")).to eq @fake_manifest_1
+      expect(@iiif.manifests.last.to_json.delete(" \t\r\n")).to eq @fake_manifest_3
+    end
+  end
+
+
+  context 'when dealing with already loaded data' do
+    include_context('fake data')
+    include_context('fake variants')
+
+    before(:example) do
+      @dir = Dir.mktmpdir
+      @iiif = WaxIiif::Builder.new({output_dir: @dir, base_url: 'http://0.0.0.0', verbose: true, thumbnail_size: 120})
+      @iiif.load(@fake_data)
+      @info_json = "#{@dir}/images/1/info.json"
+      allow(@iiif).to receive(:generate_tiles) {nil}
+      allow(@iiif).to receive(:generate_variants) {@fake_variants}
+      @iiif.process_data
+    end
+
+    after(:example) do
+      FileUtils.remove_entry @dir
+    end
+
+    it 'has the temporary file' do
+      expect(File.exist?(@info_json)).to eq true
+    end
+
+   it 'does try to generate images if that file is missing' do
+      File.delete(@info_json)
+      @iiif.process_data
+      expect(@iiif).to have_received(:generate_tiles).exactly(4).times
+      expect(@iiif).to have_received(:generate_variants).exactly(4).times
+    end
+
+    it 'does not try to generate images if that file is present' do
+      @iiif.process_data
+      expect(@iiif).to have_received(:generate_tiles).exactly(3).times
+      expect(@iiif).to have_received(:generate_variants).exactly(3).times
+    end
+
+    it 'generates the correct manifest anyway' do
+      @iiif.process_data
+      expect(@iiif.manifests.count).to eq 2
+      expect(@iiif.manifests.first.to_json.delete(" \t\r\n")).to eq @fake_manifest_1
+    end
+
+  end
+
+  context 'When load_csving CSV files' do
+    it 'accepts a path' do
+      expect{iiif.load_csv('./spec/data/test.csv')}.not_to raise_error()
+    end
+    it 'fails on blank CSV files' do
+      expect{iiif.load_csv('./spec/data/blank.csv')}.to raise_error(WaxIiif::Error::BlankCSV)
+    end
+    it 'fails on invalid CSV files' do
+      expect{iiif.load_csv('./spec/data/invalid.csv')}.to raise_error(WaxIiif::Error::InvalidCSV)
+    end
+    it 'fails on missing CSV files' do
+      expect{iiif.load_csv('./spec/data/i_dont_exist.csv')}.to raise_error(WaxIiif::Error::InvalidCSV)
+    end
+  end
+
+  context 'When loading a CSV file' do
+    it 'saves the data into the @data param' do
+      expect(iiif.data).to be_nil
+      iiif.load_csv('./spec/data/test.csv')
+      expect(iiif.data).not_to be_nil
+    end
+    it 'removes headers' do
+      iiif.load_csv('./spec/data/test.csv')
+      expect(iiif.data[0]['image_path']).to eq('spec/data/test.jpg')
+    end
+    it "doesn't remove headers if not present" do
+      iiif.load_csv('./spec/data/no_header.csv')
+      expect(iiif.data[0]['image_path']).to eq('spec/data/test.jpg')
+    end
+  end
+end

data/spec/wax_iiif/collection_spec.rb

@@ -0,0 +1,53 @@
+require 'spec_helper'
+require 'base_properties_spec'
+
+describe WaxIiif::Collection do
+  let (:labeled_config) { WaxIiif::Config.new(collection_label: 'test') }
+  let (:config) { WaxIiif::Config.new }
+  context 'base' do
+    before(:each) do
+      @object         = WaxIiif::Collection.new(labeled_config)
+      @labeled_object = WaxIiif::Collection.new(config)
+    end
+    it_behaves_like 'base properties'
+  end
+
+  context 'initialization' do
+    it 'initializes without issues when given a label' do
+      collection = nil
+      expect{ collection = WaxIiif::Collection.new(labeled_config) }.not_to raise_error
+      expect(collection.id).to eq('http://0.0.0.0/collection/test.json')
+    end
+    it 'initializes with default label when none is specified' do
+      collection = WaxIiif::Collection.new(config)
+      expect(collection.id).to include 'top.json'
+    end
+  end
+  context 'data init' do
+    include_context('fake data')
+    let(:collection) { WaxIiif::Collection.new(labeled_config) }
+    let(:manifest) { WaxIiif::Manifest.new(@fake_data[0,2], config) }
+    it 'has a label' do
+      expect(collection.label).to eq 'test'
+    end
+    it 'has an id' do
+      expect(collection.id).to eq 'http://0.0.0.0/collection/test.json'
+    end
+
+    it 'allows you to add a collection' do
+      newCollection = collection.clone
+      expect{ collection.add_collection(newCollection) }.not_to raise_error
+    end
+
+    it 'fails if you add something else to a collection' do
+      newCollection = {}
+      expect{ collection.add_collection(newCollection) }.to raise_error(WaxIiif::Error::NotACollection)
+    end
+
+
+    it 'generates correct JSON' do
+      collection.add_manifest(manifest)
+      expect(collection.to_json.delete(" \t\r\n")).to eq @fake_collection
+    end
+  end
+end