Warn about 'ingest.iiif.library.ucla.edu' URLs (#880)

Co-authored-by: Parinita Mulak <[email protected]>

app/uploaders/csv_manifest_validator.rb

@@ -190,6 +190,11 @@ def validate_records
         end
       end
 
+      # Row has an incorrect IIIF url
+      [*CalifornicaMapper.californica_terms_map[:iiif_manifest_url], *CalifornicaMapper.californica_terms_map[:access_copy]].each do |header_term|
+        this_row_warnings << "'#{header_term}' points to ingest.iiif.library.ucla.edu." if row[header_term]&.match?('ingest.iiif.library.ucla.edu')
+      end
+
       # Row has a File Name that doesn't exist
       if @mapper.preservation_copy
         if @mapper.preservation_copy.start_with?('masters.in.library.ucla.edu/')

spec/fixtures/csv_import/csv_files_with_problems/example-ingest_dot_iiif_url.csv

@@ -0,0 +1,2 @@
+Object Type,Project Name,Parent ARK,Item ARK,Subject,Type.typeOfResource,Rights.copyrightStatus,Type.genre,Name.subject,Coverage.geographic,Relation.isPartOf,Publisher.publisherName,Rights.countryCreation,Rights.rightsHolderContact,Name.architect,Name.photographer,Name.repository,Date.normalized,AltIdentifier.local,Title,Date.creation,Format.extent,Format.medium,Format.dimensions,Description.note,Description.fundingNote,Description.longitude,Description.latitude,Description.caption,File Name,AltTitle.other,AltTitle.translated,Place of origin,AltTitle.uniform,Support,Author,Summary,Page layout,Text direction,Binding note,viewingHint,IIIF Range,Illustrations note,Provenance,Table of Contents,Subject.conceptTopic,Subject.descriptiveTopic,Collation,Foliation note,Foliation,Illuminator,Name.illuminator,Name.lyricist,Name.composer,Scribe,Name.scribe,Condition note,Rights.statementLocal,Masthead,Representative image,Featured image,Tagline,Commentator,Name.commentator,Translator,Name.translator,Subject temporal,Opac url,Subject geographic,Colophon,Description.colophon,Finding Aid URL,Alt ID.url,Rubricator,Name.rubricator,Name.creator,Illustrator,Name.illustrator,Calligrapher,Name.calligrapher,Editor,Name.editor,Engraver,Name.engraver,Note,Printmaker,Name.printmaker,Contents note,Description.contents,Thumbnail,IIIF Manifest URL
+Work,Los Angeles Daily News Negatives,21198/zz00294nz8,21198/zz0002nq4w,Eminent domain--California--Los Angeles|~|Demonstrations--California--Los Angeles County|~|Express highways--California--Los Angeles County--Design and construction|~|Government|~|Transportation|~|Activism|~|Interstate 10,still image,copyrighted,news photographs,Los Angeles County (Calif.). $b Board of Supervisors,Los Angeles (Calif.),"Los Angeles Daily News Negatives. Department of Special Collections, Charles E. Young Research Library, University of California at Los Angeles.",Los Angeles Daily News,US,"UCLA Charles E. Young Research Library Department of Special Collections, A1713 Young Research Library, Box 951575, Los Angeles, CA 90095-1575. E-mail: [email protected]. Phone: (310)825-4988",,,,1947-09-17,uclalat_1387_b107_40098,"Protesters with signs in gallery of Los Angeles County Supervisors hearing over eminent domain for construction of Harbor Freeway, Calif., 1947","September 17, 1947",1 photograph,b&w negative,10 x 12.5 cm.,"At the Hall of Records, 220 N. Broadway.|~|Description 2",Info about funding,-118.243865,34.054133,This example does not have a caption.,clusc_1_1_00010432a.tif,"Protesters with signs"," manifestante n común",Los Angeles,Los Angeles Daily News Negatives,UCLA,Famous Author,Protesters with signs,images,left-to-right,Green,paged,iiif-range,illustration-note,history-description,table of contents,concept-topic,descriptive-topic,collated,follated,also follated,illuminated,also illuminated,la la la,Schubert,Amat-Mamu,Sin-liqe-unninni,condition_note,local_statement,local_masthead,local_representative_image,local_featured_image,local_tagline,commentator_1,commentator_2,translator_1,translator_2,subject_temporal_1,example_opac_url,subject_geographic_1,colophon_text,colophon_text_2,finding_aid_url_1,finding_aid_url_2,rubricator_1,rubricator_2,name_creator,Illustrator-1,Illustrator-2,Calligrapher-1,Name-calligrapher,Editor-1,Name-editor,Engraver-1,Name-engraver,Note-1,Printmaker-1,Name-printmaker,contents_note-1,Description-contents,https://fake.url/iiif/ark%3A%2F21198%2Fzz0002nq4w,https://ingest.iiif.library.ucla.edu/id/manifest

spec/uploaders/csv_manifest_validator_spec.rb

@@ -142,6 +142,15 @@
     end
   end
 
+  context 'when the csv has an \'ingest.iiif\' URL' do
+    let(:csv_file) { File.join(fixture_path, 'csv_import', 'csv_files_with_problems', 'example-ingest_dot_iiif_url.csv') }
+
+    it 'gives a warning' do
+      validator.validate
+      expect(validator.warnings).to contain_exactly("Row 2: 'IIIF Manifest URL' points to ingest.iiif.library.ucla.edu.")
+    end
+  end
+
   describe '#valid_headers' do
     it 'Contains all the values from CalifornicaMapper.CALIFORNICA_TERMS_MAP' do
       expect(validator.valid_headers).to contain_exactly(