X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/be8fbfd594717232c4dc7c4d16cdf3bde1137ee0..797885e26f85f04123bcba751addcd415ed3db2e:/services/api/app/models/collection.rb

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index 61946f82b0..775ebdb494 100644
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -14,9 +14,11 @@ class Collection < ArvadosModel
   include CommonApiTemplate
   include Trashable
 
-  serialize :properties, Hash
-  serialize :storage_classes_desired, Array
-  serialize :storage_classes_confirmed, Array
+  # Posgresql JSONB columns should NOT be declared as serialized, Rails 5
+  # already know how to properly treat them.
+  attribute :properties, :jsonbHash, default: {}
+  attribute :storage_classes_desired, :jsonbArray, default: ["default"]
+  attribute :storage_classes_confirmed, :jsonbArray, default: []
 
   before_validation :default_empty_manifest
   before_validation :default_storage_classes, on: :create
@@ -27,8 +29,11 @@ class Collection < ArvadosModel
   validate :ensure_pdh_matches_manifest_text
   validate :ensure_storage_classes_desired_is_not_empty
   validate :ensure_storage_classes_contain_non_empty_strings
+  validate :versioning_metadata_updates, on: :update
+  validate :past_versions_cannot_be_updated, on: :update
+  after_validation :set_file_count_and_total_size
   before_save :set_file_names
-  before_create :set_current_version_uuid
+  around_update :manage_versioning, unless: :is_past_version?
 
   api_accessible :user, extend: :common do |t|
     t.add :name
@@ -46,6 +51,11 @@ class Collection < ArvadosModel
     t.add :delete_at
     t.add :trash_at
     t.add :is_trashed
+    t.add :version
+    t.add :current_version_uuid
+    t.add :preserve_version
+    t.add :file_count
+    t.add :file_size_total
   end
 
   after_initialize do
@@ -78,7 +88,7 @@ class Collection < ArvadosModel
 
   FILE_TOKEN = /^[[:digit:]]+:[[:digit:]]+:/
   def check_signatures
-    return false if self.manifest_text.nil?
+    throw(:abort) if self.manifest_text.nil?
 
     return true if current_user.andand.is_admin
 
@@ -95,7 +105,7 @@ class Collection < ArvadosModel
       # Check permissions on the collection manifest.
       # If any signature cannot be verified, raise PermissionDeniedError
       # which will return 403 Permission denied to the client.
-      api_token = current_api_client_authorization.andand.api_token
+      api_token = Thread.current[:token]
       signing_opts = {
         api_token: api_token,
         now: @validation_timestamp.to_i,
@@ -115,7 +125,7 @@ class Collection < ArvadosModel
             # Signature provided, but verify_signature did not like it.
             logger.warn "Invalid signature on locator #{tok}"
             raise ArvadosModel::PermissionDeniedError
-          elsif Rails.configuration.permit_create_collection_with_unsigned_manifest
+          elsif !Rails.configuration.Collections.BlobSigning
             # No signature provided, but we are running in insecure mode.
             logger.debug "Missing signature on locator #{tok} ignored"
           elsif Blob.new(tok).empty?
@@ -190,14 +200,33 @@ class Collection < ArvadosModel
     true
   end
 
+  def set_file_count_and_total_size
+    # Only update the file stats if the manifest changed
+    if self.manifest_text_changed?
+      m = Keep::Manifest.new(self.manifest_text)
+      self.file_size_total = m.files_size
+      self.file_count = m.files_count
+    # If the manifest didn't change but the attributes did, ignore the changes
+    elsif self.file_count_changed? || self.file_size_total_changed?
+      self.file_count = self.file_count_was
+      self.file_size_total = self.file_size_total_was
+    end
+    true
+  end
+
   def manifest_files
     return '' if !self.manifest_text
 
+    done = {}
     names = ''
     self.manifest_text.scan(/ \d+:\d+:(\S+)/) do |name|
+      next if done[name]
+      done[name] = true
       names << name.first.gsub('\040',' ') + "\n"
     end
     self.manifest_text.scan(/^\.\/(\S+)/m) do |stream_name|
+      next if done[stream_name]
+      done[stream_name] = true
       names << stream_name.first.gsub('\040',' ') + "\n"
     end
     names
@@ -215,75 +244,109 @@ class Collection < ArvadosModel
     ['current_version_uuid']
   end
 
-  def set_current_version_uuid
-    self.current_version_uuid ||= self.uuid
-  end
+  def manage_versioning
+    should_preserve_version = should_preserve_version? # Time sensitive, cache value
+    return(yield) unless (should_preserve_version || syncable_updates.any?)
 
-  def save!
-    if !Rails.configuration.collection_versioning || new_record? || (!self.changes.include?('uuid') && current_version_uuid != uuid)
-      return super
-    end
+    # Put aside the changes because with_lock forces a record reload
     changes = self.changes
-    # Updates that will be synced with older versions
-    synced_updates = ['uuid', 'owner_uuid', 'delete_at', 'trash_at', 'is_trashed',
-                      'replication_desired', 'storage_classes_desired'] & changes.keys
-    # Updates that will produce a new version
-    versionable_updates = ['manifest_text', 'description', 'properties', 'name'] & changes.keys
-
-    if versionable_updates.empty? && synced_updates.empty?
-      # Updates don't include interesting attributes, so don't save a new snapshot nor
-      # sync older versions.
-      return super
-    end
-
-    # Does row locking (transaction is implicit) because 'version'
-    # may be incremented and the older versions synced.
-    # Note that 'with_lock' reloads the object after locking.
+    snapshot = nil
     with_lock do
-      # Sync older versions.
-      if !synced_updates.empty?
-        updates = {}
-        synced_updates.each do |attr|
-          if attr == 'uuid'
-            # Point old versions to current version's new UUID
-            updates['current_version_uuid'] = changes[attr].last
-          else
-            updates[attr] = changes[attr].last
-          end
-        end
-        Collection.where('current_version_uuid = ? AND uuid != ?', uuid, uuid).each do |c|
-          c.update_attributes!(updates)
-        end
-        # Also update current object just in case a new version will be created,
-        # as it has to receive the same values for the synced attributes.
-        self.attributes = updates
-      end
-      snapshot = nil
-      # Make a new version if applicable.
-      if !versionable_updates.empty?
-        # Create a snapshot of the original collection
+      # Copy the original state to save it as old version
+      if should_preserve_version
         snapshot = self.dup
         snapshot.uuid = nil # Reset UUID so it's created as a new record
-        # Update current version number
-        self.version += 1
+        snapshot.created_at = self.created_at
       end
+
       # Restore requested changes on the current version
       changes.keys.each do |attr|
-        next if attr == 'version'
+        if attr == 'preserve_version' && changes[attr].last == false
+          next # Ignore false assignment, once true it'll be true until next version
+        end
         self.attributes = {attr => changes[attr].last}
+        if attr == 'uuid'
+          # Also update the current version reference
+          self.attributes = {'current_version_uuid' => changes[attr].last}
+        end
+      end
+
+      if should_preserve_version
+        self.version += 1
+        self.preserve_version = false
+      end
+
+      yield
+
+      sync_past_versions if syncable_updates.any?
+      if snapshot
+        snapshot.attributes = self.syncable_updates
+        leave_modified_by_user_alone do
+          act_as_system_user do
+            snapshot.save
+          end
+        end
+      end
+    end
+  end
+
+  def syncable_updates
+    updates = {}
+    (syncable_attrs & self.changes.keys).each do |attr|
+      if attr == 'uuid'
+        # Point old versions to current version's new UUID
+        updates['current_version_uuid'] = self.changes[attr].last
+      else
+        updates[attr] = self.changes[attr].last
       end
-      # Save current version first to avoid index collision
-      super
-      # Save the snapshot with previous state (if applicable)
-      snapshot.andand.save!
-      return true
     end
+    return updates
+  end
+
+  def sync_past_versions
+    updates = self.syncable_updates
+    Collection.where('current_version_uuid = ? AND uuid != ?', self.uuid_was, self.uuid_was).each do |c|
+      c.attributes = updates
+      # Use a different validation context to skip the 'past_versions_cannot_be_updated'
+      # validator, as on this case it is legal to update some fields.
+      leave_modified_by_user_alone do
+        leave_modified_at_alone do
+          c.save(context: :update_old_versions)
+        end
+      end
+    end
+  end
+
+  def versionable_updates?(attrs)
+    (['manifest_text', 'description', 'properties', 'name'] & attrs).any?
+  end
+
+  def syncable_attrs
+    ['uuid', 'owner_uuid', 'delete_at', 'trash_at', 'is_trashed', 'replication_desired', 'storage_classes_desired']
+  end
+
+  def is_past_version?
+    # Check for the '_was' values just in case the update operation
+    # includes a change on current_version_uuid or uuid.
+    !(new_record? || self.current_version_uuid_was == self.uuid_was)
+  end
+
+  def should_preserve_version?
+    return false unless (Rails.configuration.Collections.CollectionVersioning && versionable_updates?(self.changes.keys))
+
+    return false if self.is_trashed
+
+    idle_threshold = Rails.configuration.Collections.PreserveVersionIfIdle
+    if !self.preserve_version_was &&
+      (idle_threshold < 0 ||
+        (idle_threshold > 0 && self.modified_at_was > db_current_time-idle_threshold.seconds))
+      return false
+    end
+    return true
   end
 
   def check_encoding
-    if manifest_text.encoding.name == 'UTF-8' and manifest_text.valid_encoding?
-      true
-    else
+    if !(manifest_text.encoding.name == 'UTF-8' and manifest_text.valid_encoding?)
       begin
         # If Ruby thinks the encoding is something else, like 7-bit
         # ASCII, but its stored bytes are equal to the (valid) UTF-8
@@ -298,7 +361,7 @@ class Collection < ArvadosModel
       rescue
       end
       errors.add :manifest_text, "must use UTF-8 encoding"
-      false
+      throw(:abort)
     end
   end
 
@@ -308,7 +371,7 @@ class Collection < ArvadosModel
       true
     rescue ArgumentError => e
       errors.add :manifest_text, e.message
-      false
+      throw(:abort)
     end
   end
 
@@ -318,8 +381,8 @@ class Collection < ArvadosModel
     elsif is_trashed
       return manifest_text
     else
-      token = current_api_client_authorization.andand.api_token
-      exp = [db_current_time.to_i + Rails.configuration.blob_signature_ttl,
+      token = Thread.current[:token]
+      exp = [db_current_time.to_i + Rails.configuration.Collections.BlobSigningTTL.to_i,
              trash_at].compact.map(&:to_i).min
       self.class.sign_manifest manifest_text, token, exp
     end
@@ -327,7 +390,7 @@ class Collection < ArvadosModel
 
   def self.sign_manifest manifest, token, exp=nil
     if exp.nil?
-      exp = db_current_time.to_i + Rails.configuration.blob_signature_ttl
+      exp = db_current_time.to_i + Rails.configuration.Collections.BlobSigningTTL.to_i
     end
     signing_opts = {
       api_token: token,
@@ -437,7 +500,7 @@ class Collection < ArvadosModel
   #
   # If filter_compatible_format is true (the default), only return image
   # collections which are support by the installation as indicated by
-  # Rails.configuration.docker_image_formats.  Will follow
+  # Rails.configuration.Containers.SupportedDockerImageFormats.  Will follow
   # 'docker_image_migration' links if search_term resolves to an incompatible
   # image, but an equivalent compatible image is available.
   def self.find_all_for_docker_image(search_term, search_tag=nil, readers=nil, filter_compatible_format: true)
@@ -448,15 +511,17 @@ class Collection < ArvadosModel
       joins("JOIN collections ON links.head_uuid = collections.uuid").
       order("links.created_at DESC")
 
-    if (Rails.configuration.docker_image_formats.include? 'v1' and
-        Rails.configuration.docker_image_formats.include? 'v2') or filter_compatible_format == false
+    docker_image_formats = Rails.configuration.Containers.SupportedDockerImageFormats
+
+    if (docker_image_formats.include? 'v1' and
+        docker_image_formats.include? 'v2') or filter_compatible_format == false
       pattern = /^(sha256:)?[0-9A-Fa-f]{64}\.tar$/
-    elsif Rails.configuration.docker_image_formats.include? 'v2'
+    elsif docker_image_formats.include? 'v2'
       pattern = /^(sha256:)[0-9A-Fa-f]{64}\.tar$/
-    elsif Rails.configuration.docker_image_formats.include? 'v1'
+    elsif docker_image_formats.include? 'v1'
       pattern = /^[0-9A-Fa-f]{64}\.tar$/
     else
-      raise "Unrecognized configuration for docker_image_formats #{Rails.configuration.docker_image_formats}"
+      raise "Unrecognized configuration for docker_image_formats #{docker_image_formats}"
     end
 
     # If the search term is a Collection locator that contains one file
@@ -464,7 +529,16 @@ class Collection < ArvadosModel
     if loc = Keep::Locator.parse(search_term)
       loc.strip_hints!
       coll_match = readable_by(*readers).where(portable_data_hash: loc.to_s).limit(1)
-      return get_compatible_images(readers, pattern, coll_match)
+      rc = Rails.configuration.RemoteClusters.select{ |k|
+        k != :"*" && k != Rails.configuration.ClusterID}
+      if coll_match.any? or rc.length == 0
+        return get_compatible_images(readers, pattern, coll_match)
+      else
+        # Allow bare pdh that doesn't exist in the local database so
+        # that federated container requests which refer to remotely
+        # stored containers will validate.
+        return [Collection.new(portable_data_hash: loc.to_s)]
+      end
     end
 
     if search_tag.nil? and (n = search_term.index(":"))
@@ -508,7 +582,7 @@ class Collection < ArvadosModel
   end
 
   def self.searchable_columns operator
-    super - ["manifest_text", "current_version_uuid"]
+    super - ["manifest_text"]
   end
 
   def self.full_text_searchable_columns
@@ -585,4 +659,30 @@ class Collection < ArvadosModel
       end
     end
   end
+
+  def past_versions_cannot_be_updated
+    if is_past_version?
+      errors.add(:base, "past versions cannot be updated")
+      false
+    end
+  end
+
+  def versioning_metadata_updates
+    valid = true
+    if !is_past_version? && current_version_uuid_changed?
+      errors.add(:current_version_uuid, "cannot be updated")
+      valid = false
+    end
+    if version_changed?
+      errors.add(:version, "cannot be updated")
+      valid = false
+    end
+    valid
+  end
+
+  def assign_uuid
+    super
+    self.current_version_uuid ||= self.uuid
+    true
+  end
 end