Merge pull request #15 from anusharanganathan/feature_index_data

Feature index data

.gitignore

@@ -33,3 +33,7 @@ config/environments/*.local.yml
 
 # Ignore coverage directory
 coverage
+
+#Ignore all data files
+data/*
+!data/.keep

Gemfile

@@ -48,6 +48,10 @@ group :development do
   gem 'spring'
 end
 
+group :test do
+  gem 'webmock'
+end
+
 gem 'blacklight', '>= 5.3.0'
 gem 'jettywrapper', '>= 2.0'
 gem 'tzinfo-data', platforms: [:mingw, :mswin, :x64_mingw]

Gemfile.lock

@@ -36,6 +36,7 @@ GEM
       minitest (~> 5.1)
       thread_safe (~> 0.3, >= 0.3.4)
       tzinfo (~> 1.1)
+    addressable (2.4.0)
     arel (6.0.3)
     autoprefixer-rails (6.1.0.1)
       execjs
@@ -110,6 +111,8 @@ GEM
       term-ansicolor (~> 1.3)
       thor (~> 0.19.1)
       tins (~> 1.6.0)
+    crack (0.4.3)
+      safe_yaml (~> 1.0.0)
     debug_inspector (0.0.2)
     deep_merge (1.0.1)
     deprecation (0.2.2)
@@ -134,6 +137,7 @@ GEM
     ffi (1.9.10)
     globalid (0.3.6)
       activesupport (>= 4.1.0)
+    hashdiff (0.2.3)
     http-cookie (1.0.2)
       domain_name (~> 0.5)
     i18n (0.7.0)
@@ -248,6 +252,7 @@ GEM
       rspec-expectations (~> 2.14.0)
       rspec-mocks (~> 2.14.0)
     rubyzip (1.1.7)
+    safe_yaml (1.0.4)
     sass (3.4.19)
     sass-rails (5.0.4)
       railties (>= 4.0.0, < 5.0)
@@ -304,6 +309,10 @@ GEM
       binding_of_caller (>= 0.7.2)
       railties (>= 4.0)
       sprockets-rails (>= 2.0, < 4.0)
+    webmock (1.22.3)
+      addressable (>= 2.3.6)
+      crack (>= 0.3.2)
+      hashdiff
     websocket-driver (0.6.3)
       websocket-extensions (>= 0.1.0)
     websocket-extensions (0.1.2)
@@ -346,6 +355,7 @@ DEPENDENCIES
   tzinfo-data
   uglifier (>= 1.3.0)
   web-console (~> 2.0)
+  webmock
 
 BUNDLED WITH
    1.10.6

app/models/concerns/annotation_data.rb

@@ -2,12 +2,9 @@ module AnnotationData
   include JsonReader
   extend ActiveSupport::Concern
 
-  # attr_accessor :annotation_url, :annotation_list
-
   def read_annotation(url = nil)
     return nil unless url # self[:annotation_url]
     @annotation_list = JsonReader::Reader.new.from_url(url)
-    # self[:annotation_list] = JsonReader::Reader.new.from_str(self[:annotation_url])
   end
 
   def motivation_for_annotations
@@ -18,14 +15,21 @@ def motivation_for_transcriptions
     'sc:painting'
   end
 
+  def resources(annotation_list = nil)
+    return [] unless annotation_list
+    return [] unless annotation_list.key? 'resources'
+    annotation_list['resources']
+  end
+
   def annotations(annotation_list = nil)
     # the motivation for annotations will be: "oa:commenting"
     # return [] unless self[:annotation_url]
     # self.read_annotation unless self[:annotation_list]
     # return [] unless self[:annotation_list]
     # return self[:annotation_list][:resources].select {|anno| anno["motivation"] == "oa:commenting" }
-    return [] unless annotation_list
-    annotation_list['resources'].select { |anno| anno['motivation'] == motivation_for_annotations }
+    al = resources(annotation_list)
+    return [] unless al
+    al.select { |anno| anno['motivation'] == motivation_for_annotations }
   end
 
   def transcriptions(annotation_list = nil)
@@ -34,8 +38,9 @@ def transcriptions(annotation_list = nil)
     # self.read_annotation unless self[:annotation_list]
     # return [] unless self[:annotation_list]
     # return self[:annotation_list][:resources].select {|anno| anno["motivation"] == "sc:painting" }
-    return [] unless annotation_list
-    annotation_list['resources'].select { |anno| anno['motivation'] == motivation_for_transcriptions }
+    al = resources(annotation_list)
+    return [] unless al
+    al.select { |anno| anno['motivation'] == motivation_for_transcriptions }
   end
 
   def map_annotation(annotation = nil)
@@ -61,17 +66,17 @@ def map_annotation(annotation = nil)
 
   def annotation_to_solr(data = {})
     # data.keys = [:annotation, :manuscript, :folio, :url]
-    return {} unless data['annotation']
+    return {} unless data.key?('annotation') || data['annotation']
     anno = map_annotation(data['annotation'])
     return {} unless anno['id']
     solr_doc = {}
     solr_doc['id'] = anno['id']
     solr_doc['druid'] = self['druid']
-    solr_doc['url_sfx'] = data['url']
     solr_doc['manifest_urls'] = self['iiif_manifest']
     solr_doc['collection'] = self['collection']
-    solr_doc['folio'] = data['folio']
-    solr_doc['manuscript_search'] = data['manuscript']
+    solr_doc['url_sfx'] = data['url'] if data.key?('url')
+    solr_doc['folio'] = data['folio'] if data.key?('folio')
+    solr_doc['manuscript_search'] = data['manuscript'] if data.key?('manuscript')
     solr_doc['model'] = anno['model']
     solr_doc['motivation'] = anno['motivation']
     solr_doc['target_url'] = anno['target_url']

app/models/concerns/iiif_manifest_data.rb

@@ -59,6 +59,12 @@ def fetch_modsxml
     uri = URI.parse(url)
     uri.scheme = 'https'
     require 'open-uri'
-    self.modsxml = open(uri.to_s).read
+    begin
+      self.modsxml = open(uri.to_s).read
+    rescue Timeout::Error, Errno::EINVAL, Errno::ECONNRESET, Errno::ETIMEDOUT, EOFError,
+           Net::HTTPBadResponse, Net::HTTPHeaderSyntaxError, Net::ProtocolError, OpenURI::HTTPError => the_error
+      puts "\nOpen URI error for #{uri}\n\t#{the_error.message}" # TODO: Add to log
+      return nil
+    end
   end
 end

app/models/concerns/json_reader.rb

@@ -8,7 +8,8 @@ def from_url(url, _encoding = nil)
       require 'open-uri'
       begin
         JSON.parse(open(url).read)
-      rescue OpenURI::HTTPError => the_error
+      rescue Timeout::Error, Errno::EINVAL, Errno::ECONNRESET, Errno::ETIMEDOUT, EOFError,
+             Net::HTTPBadResponse, Net::HTTPHeaderSyntaxError, Net::ProtocolError, OpenURI::HTTPError => the_error
         puts "\nOpen URI error for #{url}\n\t#{the_error.message}" # TODO: Add to log
         return nil
       end

config/application.rb

@@ -22,5 +22,7 @@ class Application < Rails::Application
 
     # Do not swallow errors in after_commit/after_rollback callbacks.
     config.active_record.raise_in_transactional_callbacks = true
+
+   WebMock.disable! if Rails.env.test?
   end
 end

data/test_manifest_urls.csv

@@ -0,0 +1,4 @@
+http://dms-data.stanford.edu/data/manifests/Stanford/kq131cs7229/manifest.json
+http://dms-data.stanford.edu/data/manifests/BnF/jr903ng8662/manifest.json
+http://dms-data.stanford.edu/data/manifests/Parker/fh878gz0315/manifest.json
+http://dms-data.stanford.edu/data/manifests/Parker/ft757ht3699/manifest.json

lib/data_indexer.rb

@@ -0,0 +1,113 @@
+require 'csv'
+class DataIndexer
+
+  # Index mods and annotations data from a list of manifest urls or each url
+  # Params:  
+  # +collection+:: Name of collection the manifest(s) belong to
+  # +csv_file+:: string containing the path to the csv file. 
+  #    File to have one url per line and no header
+  # +manifest_url+:: url to the mnaifest file
+  # Usage: 
+  #   DataIndexer.new('collection_name', 'file_path').run 
+  #      to index csv file
+  #   DataIndexer.new('collection_name', nil, 'url').run
+  #      to index one manifest at url
+  def initialize(collection = nil, csv_file = nil, manifest_url = nil)
+    @collection = collection
+    @csv_file = csv_file
+    @url = manifest_url
+    @manifest = nil
+    @title = nil
+    @doc = SolrDocument.new
+    @solr = Blacklight.default_index.connection
+  end
+
+  # Index and commit mods and annotations data either 
+  # from a list of manifest urls or each url
+  # depending on the options 
+  def run
+    if @csv_file.present? && File.exist?(@csv_file)
+      index_csv
+      commit
+    elsif @url
+      index
+      commit
+    end
+  end
+
+  # Index mods and annotations data from a list of manifest urls
+  #   csv file to contain one url per line and no header
+  def index_csv
+    return unless @csv_file.present? && File.exist?(@csv_file)
+    CSV.foreach(@csv_file) do |row|
+      @url = row[0]
+      index
+    end
+  end
+
+  # Index MODS and annotation lists fetched from the IIIF manifest url
+  def index
+    return unless @url.present?
+    fetch_manifest
+    if define_doc
+      index_mods
+      index_annotations
+    end
+  end
+
+  # Commit the data indexed in solr
+  def commit
+    @solr.commit
+  end
+
+  protected
+
+  # Get the manifest data
+  def fetch_manifest
+    @manifest = IiifManifest.new(@url)
+    @manifest.read_manifest
+  end
+
+  def define_doc
+    unless @manifest.title.blank? || @manifest.druid.blank?
+      @doc[:collection] = @collection
+      @doc[:druid] = @manifest.druid
+      @doc[:iiif_manifest] = @url
+      @doc[:mods_url] = @manifest.mods_url
+      @doc[:modsxml] = @manifest.fetch_modsxml
+      return true
+    end
+    false
+  end
+
+  # index mods data in solr
+  def index_mods
+    solr_doc = @doc.mods_to_solr
+    unless solr_doc.blank?
+      @title = solr_doc['title_search'] if solr_doc.key?('title_search')
+      @solr.add solr_doc
+    end
+  end
+
+  # index all of the annotations data in solr
+  def index_annotations
+    list_count = 0
+    doc_count = 0
+    add_count = 0
+    @manifest.annotation_lists.each do |al|
+      annotation_list = @doc.read_annotation(al['@id'])
+      @doc.resources(annotation_list).each do |a|
+        data = { "annotation" => a, "manuscript" => @title, "folio" => al['label'], "url" => al['@id'] }
+        solr_doc = @doc.annotation_to_solr(data)
+        unless solr_doc.blank?
+          @solr.add solr_doc
+          add_count += 1
+        end
+        doc_count += 1
+      end 
+      list_count += 1
+    end
+    [list_count, doc_count, add_count]
+  end
+
+end

lib/tasks/colligo.rake

@@ -1,10 +1,12 @@
 require 'jettywrapper'
+require 'data_indexer'
 namespace :colligo do
   desc 'Run Colligo local installation steps'
   task install: [:environment] do
     Rake::Task['db:migrate'].invoke
     Rake::Task['colligo:download_and_unzip_jetty'].invoke
     Rake::Task['colligo:copy_solr_configs'].invoke
+    Rake::Task['colligo:fixtures'].invoke
   end
   desc 'Download and unzip jetty'
   task :download_and_unzip_jetty do
@@ -21,4 +23,8 @@ namespace :colligo do
       cp "#{Rails.root}/config/solr_configs/#{file}.xml", "#{Rails.root}/jetty/solr/blacklight-core/conf/"
     end
   end
+  desc 'Index test fixtures'
+  task :fixtures do
+    DataIndexer.new('Test collection', 'data/test_manifest_urls.csv').run
+  end
 end