X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/be4852ec32e5eeed1af9a62017cfc39ed66ac186..2a17d214467d5302e97008618ef5f560ff1fd45b:/services/api/app/models/collection.rb

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index 33f6bc2524..2cebd5438e 100644
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -1,5 +1,10 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: AGPL-3.0
+
 require 'arvados/keep'
-require 'sweep_trashed_collections'
+require 'sweep_trashed_objects'
+require 'trashable'
 
 class Collection < ArvadosModel
   extend CurrentApiClient
@@ -7,24 +12,26 @@ class Collection < ArvadosModel
   include HasUuid
   include KindAndEtag
   include CommonApiTemplate
+  include Trashable
 
   serialize :properties, Hash
+  serialize :storage_classes_desired, Array
+  serialize :storage_classes_confirmed, Array
 
-  before_validation :set_validation_timestamp
   before_validation :default_empty_manifest
+  before_validation :default_storage_classes, on: :create
   before_validation :check_encoding
   before_validation :check_manifest_validity
   before_validation :check_signatures
   before_validation :strip_signatures_and_update_replication_confirmed
-  before_validation :ensure_trash_at_not_in_past
-  before_validation :sync_trash_state
-  before_validation :default_trash_interval
   validate :ensure_pdh_matches_manifest_text
-  validate :validate_trash_and_delete_timing
+  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
   before_save :set_file_names
-
-  # Query only untrashed collections by default.
-  default_scope { where("is_trashed = false") }
+  before_save :set_file_count_and_total_size
+  around_update :manage_versioning
 
   api_accessible :user, extend: :common do |t|
     t.add :name
@@ -36,9 +43,17 @@ class Collection < ArvadosModel
     t.add :replication_desired
     t.add :replication_confirmed
     t.add :replication_confirmed_at
+    t.add :storage_classes_desired
+    t.add :storage_classes_confirmed
+    t.add :storage_classes_confirmed_at
     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
@@ -65,6 +80,10 @@ class Collection < ArvadosModel
     super + ["updated_at", "file_names"]
   end
 
+  def self.limit_index_columns_read
+    ["manifest_text"]
+  end
+
   FILE_TOKEN = /^[[:digit:]]+:[[:digit:]]+:/
   def check_signatures
     return false if self.manifest_text.nil?
@@ -84,7 +103,7 @@ 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,
@@ -179,27 +198,133 @@ class Collection < ArvadosModel
     true
   end
 
+  def set_file_count_and_total_size
+    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
+    end
+    true
+  end
+
   def manifest_files
+    return '' if !self.manifest_text
+
+    done = {}
     names = ''
-    if self.manifest_text
-      self.manifest_text.scan(/ \d+:\d+:(\S+)/) do |name|
-        names << name.first.gsub('\040',' ') + "\n"
-        break if names.length > 2**12
+    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
+  end
+
+  def default_empty_manifest
+    self.manifest_text ||= ''
+  end
+
+  def skip_uuid_existence_check
+    # Avoid checking the existence of current_version_uuid, as it's
+    # assigned on creation of a new 'current version' collection, so
+    # the collection's UUID only lives on memory when the validation check
+    # is performed.
+    ['current_version_uuid']
+  end
+
+  def manage_versioning
+    should_preserve_version = should_preserve_version? # Time sensitive, cache value
+    return(yield) unless (should_preserve_version || syncable_updates.any?)
+
+    # Put aside the changes because with_lock forces a record reload
+    changes = self.changes
+    snapshot = nil
+    with_lock do
+      # 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
+        snapshot.created_at = self.created_at
+      end
+
+      # Restore requested changes on the current version
+      changes.keys.each do |attr|
+        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
+        snapshot.manifest_text = snapshot.signed_manifest_text
+        snapshot.save
       end
     end
+  end
 
-    if self.manifest_text and names.length < 2**12
-      self.manifest_text.scan(/^\.\/(\S+)/m) do |stream_name|
-        names << stream_name.first.gsub('\040',' ') + "\n"
-        break if names.length > 2**12
+  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
     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 'old_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
 
-    names[0,2**12]
+  def versionable_updates?(attrs)
+    (['manifest_text', 'description', 'properties', 'name'] & attrs).any?
   end
 
-  def default_empty_manifest
-    self.manifest_text ||= ''
+  def syncable_attrs
+    ['uuid', 'owner_uuid', 'delete_at', 'trash_at', 'is_trashed', 'replication_desired', 'storage_classes_desired']
+  end
+
+  def should_preserve_version?
+    return false unless (Rails.configuration.collection_versioning && versionable_updates?(self.changes.keys))
+
+    idle_threshold = Rails.configuration.preserve_version_if_idle
+    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
@@ -240,7 +365,7 @@ class Collection < ArvadosModel
     elsif is_trashed
       return manifest_text
     else
-      token = current_api_client_authorization.andand.api_token
+      token = Thread.current[:token]
       exp = [db_current_time.to_i + Rails.configuration.blob_signature_ttl,
              trash_at].compact.map(&:to_i).min
       self.class.sign_manifest manifest_text, token, exp
@@ -386,7 +511,14 @@ 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)
+      if coll_match.any? or Rails.configuration.remote_hosts.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(":"))
@@ -434,15 +566,26 @@ class Collection < ArvadosModel
   end
 
   def self.full_text_searchable_columns
-    super - ["manifest_text"]
+    super - ["manifest_text", "storage_classes_desired", "storage_classes_confirmed", "current_version_uuid"]
   end
 
   def self.where *args
-    SweepTrashedCollections.sweep_if_stale
+    SweepTrashedObjects.sweep_if_stale
     super
   end
 
   protected
+
+  # Although the defaults for these columns is already set up on the schema,
+  # collection creation from an API client seems to ignore them, making the
+  # validation on empty desired storage classes return an error.
+  def default_storage_classes
+    if self.storage_classes_desired.nil? || self.storage_classes_desired.empty?
+      self.storage_classes_desired = ["default"]
+    end
+    self.storage_classes_confirmed ||= []
+  end
+
   def portable_manifest_text
     self.class.munge_manifest_locators(manifest_text) do |match|
       if match[2] # size
@@ -470,89 +613,58 @@ class Collection < ArvadosModel
   end
 
   def ensure_permission_to_save
-    if (not current_user.andand.is_admin and
-        (replication_confirmed_at_changed? or replication_confirmed_changed?) and
-        not (replication_confirmed_at.nil? and replication_confirmed.nil?))
-      raise ArvadosModel::PermissionDeniedError.new("replication_confirmed and replication_confirmed_at attributes cannot be changed, except by setting both to nil")
+    if (not current_user.andand.is_admin)
+      if (replication_confirmed_at_changed? or replication_confirmed_changed?) and
+        not (replication_confirmed_at.nil? and replication_confirmed.nil?)
+        raise ArvadosModel::PermissionDeniedError.new("replication_confirmed and replication_confirmed_at attributes cannot be changed, except by setting both to nil")
+      end
+      if (storage_classes_confirmed_changed? or storage_classes_confirmed_at_changed?) and
+        not (storage_classes_confirmed == [] and storage_classes_confirmed_at.nil?)
+        raise ArvadosModel::PermissionDeniedError.new("storage_classes_confirmed and storage_classes_confirmed_at attributes cannot be changed, except by setting them to [] and nil respectively")
+      end
     end
     super
   end
 
-  # Use a single timestamp for all validations, even though each
-  # validation runs at a different time.
-  def set_validation_timestamp
-    @validation_timestamp = db_current_time
-  end
-
-  # If trash_at is being changed to a time in the past, change it to
-  # now. This allows clients to say "expires {client-current-time}"
-  # without failing due to clock skew, while avoiding odd log entries
-  # like "expiry date changed to {1 year ago}".
-  def ensure_trash_at_not_in_past
-    if trash_at_changed? && trash_at
-      self.trash_at = [@validation_timestamp, trash_at].max
+  def ensure_storage_classes_desired_is_not_empty
+    if self.storage_classes_desired.empty?
+      raise ArvadosModel::InvalidStateTransitionError.new("storage_classes_desired shouldn't be empty")
     end
   end
 
-  # Caller can move into/out of trash by setting/clearing is_trashed
-  # -- however, if the caller also changes trash_at, then any changes
-  # to is_trashed are ignored.
-  def sync_trash_state
-    if is_trashed_changed? && !trash_at_changed?
-      if is_trashed
-        self.trash_at = @validation_timestamp
-      else
-        self.trash_at = nil
-        self.delete_at = nil
+  def ensure_storage_classes_contain_non_empty_strings
+    (self.storage_classes_desired + self.storage_classes_confirmed).each do |c|
+      if !c.is_a?(String) || c == ''
+        raise ArvadosModel::InvalidStateTransitionError.new("storage classes should only be non-empty strings")
       end
     end
-    self.is_trashed = trash_at && trash_at <= @validation_timestamp || false
-    true
   end
 
-  def default_trash_interval
-    if trash_at_changed? && !delete_at_changed?
-      # If trash_at is updated without touching delete_at,
-      # automatically update delete_at to a sensible value.
-      if trash_at.nil?
-        self.delete_at = nil
-      else
-        self.delete_at = trash_at + Rails.configuration.default_trash_lifetime.seconds
-      end
-    elsif !trash_at || !delete_at || trash_at > delete_at
-      # Not trash, or bogus arguments? Just validate in
-      # validate_trash_and_delete_timing.
-    elsif delete_at_changed? && delete_at >= trash_at
-      # Fix delete_at if needed, so it's not earlier than the expiry
-      # time on any permission tokens that might have been given out.
-
-      # In any case there are no signatures expiring after now+TTL.
-      # Also, if the existing trash_at time has already passed, we
-      # know we haven't given out any signatures since then.
-      earliest_delete = [
-        @validation_timestamp,
-        trash_at_was,
-      ].compact.min + Rails.configuration.blob_signature_ttl.seconds
-
-      # The previous value of delete_at is also an upper bound on the
-      # longest-lived permission token. For example, if TTL=14,
-      # trash_at_was=now-7, delete_at_was=now+7, then it is safe to
-      # set trash_at=now+6, delete_at=now+8.
-      earliest_delete = [earliest_delete, delete_at_was].compact.min
-
-      # If delete_at is too soon, use the earliest possible time.
-      if delete_at < earliest_delete
-        self.delete_at = earliest_delete
-      end
+  def past_versions_cannot_be_updated
+    # We check for the '_was' values just in case the update operation
+    # includes a change on current_version_uuid or uuid.
+    if current_version_uuid_was != uuid_was
+      errors.add(:base, "past versions cannot be updated")
+      false
     end
   end
 
-  def validate_trash_and_delete_timing
-    if trash_at.nil? != delete_at.nil?
-      errors.add :delete_at, "must be set if trash_at is set, and must be nil otherwise"
-    elsif delete_at && delete_at < trash_at
-      errors.add :delete_at, "must not be earlier than trash_at"
+  def versioning_metadata_updates
+    valid = true
+    if (current_version_uuid_was == uuid_was) && 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