Merge pull request #8 from anusharanganathan/feature_iiif_manifest_data

Feature iiif manifest data

app/models/concerns/iiif_manifest_data.rb

@@ -0,0 +1,65 @@
+module IiifManifestData
+  include JsonReader
+  extend ActiveSupport::Concern
+
+  attr_accessor :manifest_url, :manifest, :modsxml
+
+  def read_manifest
+    return nil unless self.manifest_url
+    self.manifest = JsonReader::Reader.new.from_url(self.manifest_url)
+  end
+
+  def title
+    return nil unless self.manifest_url
+    self.read_manifest unless self.manifest 
+    return nil unless self.manifest
+    self.manifest["label"]
+  end
+
+  def druid
+    #druid is a 11 digit alphanumeric sring
+    return nil unless self.manifest_url
+    self.read_manifest unless self.manifest 
+    return nil unless self.manifest
+    self.manifest["@id"].match( /\/([a-zA-Z0-9]{11})\// ).to_s.gsub('/', '')
+  end
+
+  def mods_url
+    return nil unless self.manifest_url
+    self.read_manifest unless self.manifest 
+    return nil unless self.manifest
+    return nil unless self.manifest.has_key?("seeAlso")
+    if self.manifest["seeAlso"]["dcterms:format"] == "application/mods+xml"
+      return self.manifest["seeAlso"]["@id"]
+    elsif self.manifest["seeAlso"]["format"] == "application/mods+xml"
+      return self.manifest["seeAlso"]["@id"]
+    end
+    return nil
+  end
+
+  def annotation_lists
+    return [] unless self.manifest_url
+    self.read_manifest unless self.manifest 
+    return [] unless self.manifest
+    annotations = []
+    self.manifest["sequences"].each do |s|
+      s["canvases"].each do |c|
+        if c.has_key?("otherContent")
+          annotations = annotations + c["otherContent"].select {|item| item["@type"] == "sc:AnnotationList"}.map{|item| item['label'] = c["label"]; item}
+        end
+      end
+    end
+    annotations
+  end
+
+  def get_modsxml
+    self.modsxml = nil
+    url = self.mods_url
+    return nil unless url
+    uri = URI.parse(url)
+    uri.scheme = "https"
+    require 'open-uri'
+    self.modsxml = open(uri.to_s).read
+  end
+
+end

app/models/iiif_manifest.rb

@@ -0,0 +1,7 @@
+class IiifManifest
+  include IiifManifestData
+
+  def initialize(manifest_url=nil)
+    self.manifest_url = manifest_url if manifest_url
+  end
+end

spec/fixtures/iiif_manifest_records/iiif_manifest_fixtures.rb

@@ -0,0 +1,18 @@
+# encoding: UTF-8
+module IiifManifestFixtures
+  def manifest_url_001
+    'http://dms-data.stanford.edu/data/manifests/Stanford/kq131cs7229/manifest.json'
+  end
+
+  def manifest_url_002
+    'http://dms-data.stanford.edu/data/manifests/BnF/jr903ng8662/manifest.json'
+  end
+
+  def manifest_url_003
+    'http://dms-data.stanford.edu/data/manifests/Parker/fh878gz0315/manifest.json'
+  end
+
+  def manifest_url_004
+    'https://purl.stanford.edu/bb389yg4719/iiif/manifest.json'
+  end
+end

spec/fixtures/iiif_manifest_records/manifest_urls.csv

@@ -0,0 +1,6 @@
+https://purl.stanford.edu/bb389yg4719/iiif/manifest.json
+https://purl.stanford.edu/bb475kg5932/iiif/manifest.json
+https://purl.stanford.edu/bc991ym6224/iiif/manifest.json
+https://purl.stanford.edu/bk254dq8652/iiif/manifest.json
+https://purl.stanford.edu/bn235bx5422/iiif/manifest.json
+https://purl.stanford.edu/bn247br1244/iiif/manifest.json

spec/models/concerns/iiif_manifest_data_spec.rb

@@ -0,0 +1,127 @@
+require "spec_helper"
+
+class IiifManifestDataTestClass
+  include IiifManifestData
+end
+
+describe IiifManifestData do
+  include IiifManifestFixtures
+  before(:all) do
+    @document_empty = IiifManifest.new()
+    @document_empty.read_manifest
+    @document = IiifManifest.new(manifest_url=manifest_url_001)
+    @document.read_manifest
+    @document2 = IiifManifest.new(manifest_url=manifest_url_004)
+    @document2.read_manifest
+  end
+
+  describe "#read_manifest" do
+    it "should be nil if no manifest url" do
+      expect(@document_empty.manifest).to be_nil
+    end
+
+    it "should be a Hash" do
+      expect(@document.manifest).to be_a Hash
+      expect(@document2.manifest).to be_a Hash
+    end
+
+    it "should not be an empty Hash" do
+      expect(@document.manifest).not_to be_empty
+      expect(@document2.manifest).not_to be_empty
+    end
+  end
+
+  describe "#title" do
+    it "should be nil if no manifest url" do
+      expect(@document_empty.title).to be_nil
+    end
+
+    it "should be a string" do
+      expect(@document.title).to be_a String
+      expect(@document2.title).to be_a String
+    end
+
+    it "should be a title string" do
+      expect(@document.title).to eq("Manuscript fragment of the Gospels and Canonical Epistles, glossed")
+      expect(@document2.title).to eq("Stanford University Libraries, M1814. Flat Box 2, Folder 02")
+    end
+  end
+
+  describe "#druid" do
+    it "should be nil if no manifest url" do
+      expect(@document_empty.druid).to be_nil
+    end
+
+    it "should be a string" do
+      expect(@document.druid).to be_a String
+      expect(@document2.druid).to be_a String
+    end
+
+    it "should be a druid string" do
+      expect(@document.druid).to eq("kq131cs7229")
+      expect(@document2.druid).to eq("bb389yg4719")
+    end
+  end
+
+  describe "#mods_url" do
+    it "should be nil if no manifest url" do
+      expect(@document_empty.mods_url).to be_nil
+    end
+
+    it "should be a string" do
+      expect(@document.mods_url).to be_a String
+      expect(@document2.mods_url).to be_a String
+    end
+
+    it "should be a mods url string" do
+      expect(@document.mods_url).to eq("http://purl.stanford.edu/kq131cs7229.mods")
+      expect(@document2.mods_url).to eq("https://purl.stanford.edu/bb389yg4719.mods")
+    end
+  end
+
+  describe "#annotation_lists" do
+    it "should be an array" do
+      expect(@document_empty.annotation_lists).to be_a Array
+      expect(@document.annotation_lists).to be_a Array
+      expect(@document2.annotation_lists).to be_a Array
+    end
+
+    it "should be an empty array if no manifest url" do
+      expect(@document_empty.annotation_lists).to be_empty
+      expect(@document2.annotation_lists).to be_empty
+    end
+
+    it "should not be an empty array" do
+      expect(@document.annotation_lists).not_to be_empty
+    end
+
+    it "should be of length 36" do
+      expect(@document.annotation_lists.length).to eq(36)
+    end
+
+    it "should be an array of hashes" do
+      expect(@document.annotation_lists.any? {|hash| hash.keys == ['@id', '@type', 'label']}).to be_true
+    end
+  end
+
+  describe "#get_modsxml" do
+    before(:all) do
+      @document.get_modsxml
+      @document2.get_modsxml
+    end
+
+    it "should be nil if no manifest url" do
+      expect(@document_empty.title).to be_nil
+    end
+
+    it "should be a String" do
+      expect(@document.modsxml).to be_a String
+      expect(@document2.modsxml).to be_a String
+    end
+
+    it "should be a xml doc" do
+      expect(@document.modsxml).to start_with ("<?xml")
+      expect(@document2.modsxml).to start_with ("<?xml")
+    end
+  end
+end

spec/spec_helper.rb

@@ -8,6 +8,7 @@
 #require 'fixtures/marc_records/marc_metadata_fixtures'
 require 'fixtures/mods_records/mods_fixtures'
 require 'fixtures/annotation_records/annotation_fixtures'
+require 'fixtures/iiif_manifest_records/iiif_manifest_fixtures'
 
 Capybara.register_driver :poltergeist do |app|
   Capybara::Poltergeist::Driver.new(app, {timeout: 60})