Merge pull request #3141 from betagouv/frederic/fix_2180/improve_migration_task

Improve PJ migration task

lib/cellar/cellar_adapter.rb

@@ -81,12 +81,24 @@ module Cellar
       end
 
       def list_prefixed(prefix)
-        request = Net::HTTP::Get.new("/?prefix=#{prefix}")
-        @signer.sign(request, "")
-        response = @http.request(request)
-        if response.is_a?(Net::HTTPSuccess)
-          parse_bucket_listing(response.body)
-        end
+        result = []
+        marker = ''
+
+        begin
+          request = Net::HTTP::Get.new("/?prefix=#{prefix}&marker=#{marker}")
+          @signer.sign(request, "")
+          response = @http.request(request)
+          if response.is_a?(Net::HTTPSuccess)
+            (listing, truncated) = parse_bucket_listing(response.body)
+            result += listing
+            marker = listing.last
+          else
+            # TODO: error handling
+            return nil
+          end
+        end while truncated
+
+        result
       end
 
       def delete_keys(keys)
@@ -126,9 +138,11 @@ module Cellar
 
       def parse_bucket_listing(bucket_listing_xml)
         doc = Nokogiri::XML(bucket_listing_xml)
-        doc
+        listing = doc
           .xpath('//xmlns:Contents/xmlns:Key')
           .map(&:text)
+        truncated = doc.xpath('//xmlns:IsTruncated').text == 'true'
+        [listing, truncated]
       end
 
       def bulk_deletion_request_body(keys)

lib/tasks/2018_12_03_finish_piece_jointe_transfer.rake

@@ -4,8 +4,8 @@ namespace :'2018_12_03_finish_piece_jointe_transfer' do
       def run
         notify_dry_run
         refresh_outdated_files
-        fix_openstack_mime_types
         remove_unused_openstack_objects
+        fix_openstack_mime_types
         notify_dry_run
       end
 
@@ -69,35 +69,36 @@ namespace :'2018_12_03_finish_piece_jointe_transfer' do
       def refresh_outdated_files
         rake_puts "Refresh outdated attachments"
 
-        bar = RakeProgressbar.new(ActiveStorage::Blob.count)
         refreshed_keys = []
         missing_keys = []
         old_pj_adapter.session do |old_pjs|
-          ActiveStorage::Blob.find_each do |blob|
-            new_pj_metadata = new_pjs.files.head(blob.key)
+          keys = old_pjs.list_prefixed('')
+          progress = ProgressReport.new(keys.count)
+          keys.each do |key|
+            new_pj_metadata = new_pjs.files.head(key)
 
             refresh_needed = new_pj_metadata.nil?
             if !refresh_needed
               new_pj_last_modified = new_pj_metadata.last_modified.in_time_zone
-              old_pj_last_modified = old_pjs.last_modified(blob.key)
+              old_pj_last_modified = old_pjs.last_modified(key)
               if old_pj_last_modified.nil?
-                missing_keys.push(blob.key)
+                missing_keys.push(key)
               else
                 refresh_needed = new_pj_last_modified < old_pj_last_modified
               end
             end
 
             if refresh_needed
-              refreshed_keys.push(blob.key)
+              refreshed_keys.push(key)
               if force?
-                file = Tempfile.new(blob.key)
+                file = Tempfile.new(key)
                 file.binmode
-                old_pjs.download(blob.key) do |chunk|
+                old_pjs.download(key) do |chunk|
                   file.write(chunk)
                 end
                 file.rewind
                 new_pjs.files.create(
-                  :key    => blob.key,
+                  :key    => key,
                   :body   => file,
                   :public => false
                 )
@@ -105,10 +106,10 @@ namespace :'2018_12_03_finish_piece_jointe_transfer' do
                 file.unlink
               end
             end
-            bar.inc
+            progress.inc
           end
+          progress.finish
         end
-        bar.finished
 
         if verbose?
           rake_puts "Refreshed #{refreshed_keys.count} attachments\n#{refreshed_keys.join(', ')}"
@@ -132,27 +133,30 @@ namespace :'2018_12_03_finish_piece_jointe_transfer' do
         end
         rake_puts "Fix MIME types"
 
-        bar = RakeProgressbar.new(ActiveStorage::Blob.count)
+        progress = ProgressReport.new(new_pjs.count.to_i)
         failed_keys = []
         updated_keys = []
-        ActiveStorage::Blob.find_each do |blob|
-          if blob.identified? && blob.content_type.present?
-            updated_keys.push(blob.key)
+        new_pjs.files.each do |file|
+          blob = ActiveStorage::Blob.find_by(key: file.key)
+          if blob.nil?
+            failed_keys.push(file.key)
+          elsif blob.identified? && blob.content_type.present?
+            updated_keys.push(file.key)
             if force?
-              if !blob.service.change_content_type(blob.key, blob.content_type)
-                failed_keys.push(blob.key)
+              if !blob.service.change_content_type(file.key, blob.content_type)
+                failed_keys.push(file.key)
               end
             end
           end
-          bar.inc
+          progress.inc
         end
-        bar.finished
+        progress.finish
 
         if verbose?
           rake_puts "Updated MIME Type for #{updated_keys.count} keys\n#{updated_keys.join(', ')}"
         end
         if failed_keys.present?
-          rake_puts "failed to update #{failed_keys.count} keys (dangling blob?)\n#{failed_keys.join(', ')}"
+          rake_puts "failed to update #{failed_keys.count} keys\n#{failed_keys.join(', ')}"
         end
       end
 
@@ -160,7 +164,7 @@ namespace :'2018_12_03_finish_piece_jointe_transfer' do
       def remove_unused_openstack_objects
         rake_puts "Remove unused files"
 
-        bar = RakeProgressbar.new(new_pjs.count.to_i)
+        progress = ProgressReport.new(new_pjs.count.to_i)
         removed_keys = []
         new_pjs.files.each do |file|
           if !ActiveStorage::Blob.exists?(key: file.key)
@@ -170,9 +174,9 @@ namespace :'2018_12_03_finish_piece_jointe_transfer' do
             end
           end
 
-          bar.inc
+          progress.inc
         end
-        bar.finished
+        progress.finish
 
         if verbose?
           rake_puts "Removed #{removed_keys.count} unused objects\n#{removed_keys.join(', ')}"

lib/tasks/task_helper.rb

@@ -8,3 +8,55 @@ def rake_puts(*args)
     puts(*args)
   end
 end
+
+def rake_print(*args)
+  if Rake.verbose
+    print(*args)
+  end
+end
+
+class ProgressReport
+  def initialize(total)
+    @start = Time.zone.now
+    rake_puts
+    set_progress(total: total, count: 0)
+  end
+
+  def inc
+    set_progress(count: @count + 1)
+    if @per_10_000 % 10 == 0
+      print_progress
+    end
+  end
+
+  def finish
+    if @count > 0 && @per_10_000 != 10_000
+      set_progress(total: @count)
+      print_progress
+    end
+    rake_puts
+  end
+
+  def set_progress(total: nil, count: nil)
+    if total.present?
+      @total = total
+    end
+    if count.present?
+      @count = count
+      @total = [@count, @total].max
+    end
+    @per_10_000 = 10_000 * @count / @total
+  end
+
+  def print_progress
+    elapsed = Time.zone.now - @start
+    percent = sprintf('%5.1f%%', @per_10_000 / 100.0)
+    total = @total.to_s
+    count = @count.to_s.rjust(total.length)
+    rake_print("\r#{percent} (#{count}/#{total}) [#{format_duration(elapsed)}/#{format_duration(elapsed * 10_000 / @per_10_000)}]")
+  end
+
+  def format_duration(seconds)
+    Time.at(seconds).utc.strftime('%H:%M:%S')
+  end
+end

spec/lib/cellar/cellar_adapter_spec.rb

@@ -54,7 +54,7 @@ describe 'CellarAdapter' do
 
     subject { session.send(:parse_bucket_listing, response) }
 
-    it { is_expected.to eq(["sample1.jpg", "sample2.jpg"]) }
+    it { is_expected.to eq([["sample1.jpg", "sample2.jpg"], false]) }
   end
 
   describe 'bulk_deletion_request_body' do