Blob entry import/export helper

Rakefile

@@ -34,6 +34,16 @@ task :good_job_cron => :environment do
   end
 end
 
+namespace :blob_entries do
+  task :export_samples => :environment do
+    limit = ENV["limit"]&.to_i || raise("need 'limit' (num)")
+    out = ENV["out"] || raise("need 'out' (file path, .json encoded)")
+    BlobEntrySampleExporter.new.export_samples(limit, out)
+  end
+  task :import_samples => :environment do
+  end
+end
+
 task :good_job do
   proxies = ["direct", "proxy-1", "dedipath-1", "serverhost-1"]
   proxy = ENV["proxy"]

app/lib/blob_entry_sample_exporter.rb

@@ -0,0 +1,40 @@
+class BlobEntrySampleExporter
+  include HasColorLogger
+  include HasMeasureDuration
+
+  def export_samples(limit, file)
+    @num_written = 0
+    @bytes_written = 0
+    measure(proc {
+      "wrote #{@num_written} blob entries, #{HexUtil.humansize(@bytes_written)} to #{file}"
+    }) do
+      File.open(file, "w") do |file|
+        ::BlobEntry.limit(limit).find_each(batch_size: 32) do |blob_entry|
+          write_blob_entry(file, blob_entry)
+          write_blob_entry(file, blob_entry.base) if blob_entry.base
+        end
+      end
+    end
+  end
+
+  def write_blob_entry(file, blob_entry)
+    hash = blob_entry.to_bulk_insert_hash
+    json_hash = {
+      sha256: HexUtil.bin2hex(hash[:sha256]),
+      content_type: hash[:content_type],
+      size: hash[:size],
+      created_at: blob_entry.created_at,
+    }
+    json_hash[:base_sha256] = HexUtil.bin2hex(hash[:base_sha256]) if hash[:base_sha256]
+    file.puts(json_hash.to_json)
+    file.puts(Base64.strict_encode64(blob_entry.read_attribute(:contents)))
+    logger.info(
+      "#{HexUtil.bin2hex(blob_entry.sha256)} - " +
+        "#{blob_entry.base_sha256.present?} - " +
+        "#{blob_entry.content_type} - #{HexUtil.humansize(blob_entry.size)} " +
+        "(#{HexUtil.humansize(blob_entry.bytes_stored)} actual)"
+    )
+    @bytes_written += blob_entry.bytes_stored
+    @num_written += 1
+  end
+end