Add a rake task to re-extract any missing attachmentsreextract-attachments

You can see the number of emails that would be reparsed
by doing:

  bundle exec rake temp:reextract_missing_attachments

To actually reparse the incoming emails and rextract
attachments for any that were missing, you would do:

  bundle exec rake temp:reextract_missing_attachments[commit]

(In fact, the 'commit' can be any non-empty string.)

1 files changed, 22 insertions, 0 deletions

diff --git a/lib/tasks/temp.rake b/lib/tasks/temp.rake
index fd0dd069f..9eda530fd 100644
--- a/lib/tasks/temp.rake
+++ b/lib/tasks/temp.rake
@@ -21,6 +21,28 @@ namespace :temp do
         end
     end
 
+    desc "Re-extract any missing cached attachments"
+    task :reextract_missing_attachments, [:commit] => :environment do |t, args|
+        dry_run = args.commit.nil? || args.commit.empty?
+        total_messages = 0
+        messages_to_reparse = 0
+        IncomingMessage.find_each :include => :foi_attachments do |im|
+            reparse = im.foi_attachments.any? { |fa| ! File.exists? fa.filepath }
+            total_messages += 1
+            messages_to_reparse += 1 if reparse
+            if total_messages % 1000 == 0
+                puts "Considered #{total_messages} received emails."
+            end
+            unless dry_run
+                im.parse_raw_email! true if reparse
+                sleep 2
+            end
+        end
+        message = dry_run ? "Would reparse" : "Reparsed"
+        message += " #{messages_to_reparse} out of #{total_messages} received emails."
+        puts message
+    end
+
     desc 'Remove file caches for requests that are not publicly visible or have been destroyed'
     task :remove_obsolete_info_request_caches => :environment do
         dryrun = ENV['DRYRUN'] == '0' ? false : true