Merge pull request #387 from ivplus/386-extract-marc-deletes

Don't try to parse delete files as marc records

app/jobs/generate_delta_dump_job.rb

@@ -27,7 +27,7 @@ def perform(organization)
       hash = current_marc_records(uploads)
 
       uploads.each do |upload|
-        upload.each_marc_record_metadata.each do |record|
+        upload.each_marc_record_metadata(checksum: false).each do |record|
           next unless hash.dig(record.marc001, 'file_id') == record.file_id
 
           if hash.dig(record.marc001, 'status') == 'delete'
@@ -57,7 +57,7 @@ def current_marc_records(uploads)
     hash = {}
 
     uploads.each do |upload|
-      upload.each_marc_record_metadata.each do |record|
+      upload.each_marc_record_metadata(checksum: false).each do |record|
         hash[record.marc001] = record.attributes.slice('file_id', 'status')
       end
     end

app/jobs/generate_full_dump_job.rb

@@ -24,7 +24,7 @@ def perform(organization)
       hash = current_marc_records(organization.default_stream.uploads)
 
       organization.default_stream.uploads.each do |upload|
-        upload.each_marc_record_metadata.each do |record|
+        upload.each_marc_record_metadata(checksum: false).each do |record|
           next unless hash.dig(record.marc001, 'file_id') == record.file_id || hash.dig(record.marc001, 'status') == 'delete'
 
           writer.write_marc_record(record)
@@ -68,7 +68,7 @@ def current_marc_records(uploads)
     hash = {}
 
     uploads.each do |upload|
-      upload.each_marc_record_metadata.each do |record|
+      upload.each_marc_record_metadata(checksum: false).each do |record|
         hash[record.marc001] = record.attributes.slice('file_id', 'status')
       end
     end

app/models/upload.rb

@@ -35,19 +35,22 @@ def archive
     files.find_each(&:purge_later)
   end
 
-  def each_marc_record_metadata(&block)
-    return to_enum(:each_marc_record_metadata) unless block
+  def each_marc_record_metadata(**options, &block)
+    return to_enum(:each_marc_record_metadata, **options) unless block
 
     files.each do |file|
       service = MarcRecordService.new(file.blob)
 
       format = service.identify
 
-      extract_marc_record_delete_metadata(file, &block) if format == :delete
-
-      next if format == :unknown
-
-      extract_marc_record_metadata(file, service, &block)
+      case format
+      when :delete
+        extract_marc_record_delete_metadata(file, &block)
+      when :unknown
+        next
+      else
+        extract_marc_record_metadata(file, service, **options, &block)
+      end
     rescue StandardError => e
       Honeybadger.notify(e)
     end
@@ -88,8 +91,8 @@ def extract_marc_record_delete_metadata(file)
     end
   end
 
-  def extract_marc_record_metadata(file, service)
-    return to_enum(:extract_marc_record_metadata, file, service) unless block_given?
+  def extract_marc_record_metadata(file, service, checksum: true)
+    return to_enum(:extract_marc_record_metadata, file, service, checksum: checksum) unless block_given?
 
     service.each_with_metadata do |record, metadata|
       out = MarcRecord.new(
@@ -101,7 +104,7 @@ def extract_marc_record_metadata(file, service)
         json: Zlib::Deflate.new.deflate(record.to_marchash.to_json, Zlib::FINISH)
       )
 
-      out.checksum ||= Digest::MD5.hexdigest(record.to_xml.to_s)
+      out.checksum ||= Digest::MD5.hexdigest(record.to_xml.to_s) if checksum
 
       yield out
     end