services/api/db/migrate/20150303210106_fix_collection_portable_data_hash_with_hinted_manifest.rb

   1 require 'has_uuid'
   2 require 'kind_and_etag'
   3
   4 class FixCollectionPortableDataHashWithHintedManifest < ActiveRecord::Migration
   5   include CurrentApiClient
   6
   7   class ArvadosModel < ActiveRecord::Base
   8     self.abstract_class = true
   9     extend HasUuid::ClassMethods
  10     include CurrentApiClient
  11     include KindAndEtag
  12     before_create do |record|
  13       record.uuid ||= record.class.generate_uuid
  14       record.owner_uuid ||= system_user_uuid
  15     end
  16     serialize :properties, Hash
  17
  18     def self.to_s
  19       # Clean up the name of the stub model class so we generate correct UUIDs.
  20       super.sub("FixCollectionPortableDataHashWithHintedManifest::", "")
  21     end
  22   end
  23
  24   class Collection < ArvadosModel
  25   end
  26
  27   class Log < ArvadosModel
  28     def self.log_for(thing, age="old")
  29       { "#{age}_etag" => thing.etag,
  30         "#{age}_attributes" => thing.attributes,
  31       }
  32     end
  33
  34     def self.log_create(thing)
  35       new_log("create", thing, log_for(thing, "new"))
  36     end
  37
  38     def self.log_update(thing, start_state)
  39       new_log("update", thing, start_state.merge(log_for(thing, "new")))
  40     end
  41
  42     def self.log_destroy(thing)
  43       new_log("destroy", thing, log_for(thing, "old"))
  44     end
  45
  46     private
  47
  48     def self.new_log(event_type, thing, properties)
  49       create!(event_type: event_type,
  50               event_at: Time.now,
  51               object_uuid: thing.uuid,
  52               object_owner_uuid: thing.owner_uuid,
  53               properties: properties)
  54     end
  55   end
  56
  57   def each_bad_collection
  58     Collection.find_each do |coll|
  59       next unless (coll.manifest_text =~ /\+[A-Z]/)
  60       stripped_manifest = coll.manifest_text.
  61         gsub(/( [0-9a-f]{32}(\+\d+)?)(\+\S+)/, '\1')
  62       stripped_pdh = sprintf("%s+%i",
  63                              Digest::MD5.hexdigest(stripped_manifest),
  64                              stripped_manifest.bytesize)
  65       yield [coll, stripped_pdh] if (coll.portable_data_hash != stripped_pdh)
  66     end
  67   end
  68
  69   def up
  70     Collection.reset_column_information
  71     Log.reset_column_information
  72     copied_attr_names =
  73       [:owner_uuid, :created_at, :modified_by_client_uuid, :manifest_text,
  74        :modified_by_user_uuid, :modified_at, :updated_at, :name,
  75        :description, :portable_data_hash, :replication_desired,
  76        :replication_confirmed, :replication_confirmed_at, :expires_at]
  77     new_expiry = Date.new(2038, 1, 31)
  78
  79     each_bad_collection do |coll, stripped_pdh|
  80       # Create a copy of the collection including bad portable data hash,
  81       # with an expiration.  This makes it possible to resolve the bad
  82       # portable data hash, but the expiration can hide the Collection
  83       # from more user-friendly interfaces like Workbench.
  84       start_log = Log.log_for(coll)
  85       attributes = Hash[copied_attr_names.map { |key| [key, coll.send(key)] }]
  86       attributes[:expires_at] ||= new_expiry
  87       attributes[:properties] = (coll.properties.dup rescue {})
  88       attributes[:properties]["migrated_from"] ||= coll.uuid
  89       coll_copy = Collection.create!(attributes)
  90       Log.log_create(coll_copy)
  91       coll.update_attributes(portable_data_hash: stripped_pdh)
  92       Log.log_update(coll, start_log)
  93     end
  94   end
  95
  96   def down
  97     Collection.reset_column_information
  98     Log.reset_column_information
  99     each_bad_collection do |coll, stripped_pdh|
 100       if ((src_uuid = coll.properties["migrated_from"]) and
 101           (src_coll = Collection.where(uuid: src_uuid).first) and
 102           (src_coll.portable_data_hash == stripped_pdh))
 103         start_log = Log.log_for(src_coll)
 104         src_coll.portable_data_hash = coll.portable_data_hash
 105         src_coll.save!
 106         Log.log_update(src_coll, start_log)
 107         coll.destroy or raise Exception.new("failed to destroy old collection")
 108         Log.log_destroy(coll)
 109       end
 110     end
 111   end
 112 end