X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/b9e5c8b32858338850da3e12ce27570b828898b3..6c78b28f9f54664babc57a4b4372c502065ed5d1:/services/api/app/models/arvados_model.rb

diff --git a/services/api/app/models/arvados_model.rb b/services/api/app/models/arvados_model.rb
index aed0309591..339bc9e23f 100644
--- a/services/api/app/models/arvados_model.rb
+++ b/services/api/app/models/arvados_model.rb
@@ -1,17 +1,21 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: AGPL-3.0
+
+require 'arvados_model_updates'
 require 'has_uuid'
 require 'record_filters'
+require 'serializers'
+require 'request_error'
 
-class ArvadosModel < ActiveRecord::Base
+class ArvadosModel < ApplicationRecord
   self.abstract_class = true
 
+  include ArvadosModelUpdates
   include CurrentApiClient      # current_user, current_api_client, etc.
   include DbCurrentTime
   extend RecordFilters
 
-  attr_protected :created_at
-  attr_protected :modified_by_user_uuid
-  attr_protected :modified_by_client_uuid
-  attr_protected :modified_at
   after_initialize :log_start_state
   before_save :ensure_permission_to_save
   before_save :ensure_owner_uuid_is_permitted
@@ -26,39 +30,58 @@ class ArvadosModel < ActiveRecord::Base
   after_find :convert_serialized_symbols_to_strings
   before_validation :normalize_collection_uuids
   before_validation :set_default_owner
-  validate :ensure_serialized_attribute_type
   validate :ensure_valid_uuids
 
   # Note: This only returns permission links. It does not account for
   # permissions obtained via user.is_admin or
   # user.uuid==object.owner_uuid.
-  has_many :permissions, :foreign_key => :head_uuid, :class_name => 'Link', :primary_key => :uuid, :conditions => "link_class = 'permission'"
+  has_many(:permissions,
+           ->{where(link_class: 'permission')},
+           foreign_key: :head_uuid,
+           class_name: 'Link',
+           primary_key: :uuid)
+
+  # If async is true at create or update, permission graph
+  # update is deferred allowing making multiple calls without the performance
+  # penalty.
+  attr_accessor :async_permissions_update
+
+  # Ignore listed attributes on mass assignments
+  def self.protected_attributes
+    []
+  end
 
-  class PermissionDeniedError < StandardError
+  class PermissionDeniedError < RequestError
     def http_status
       403
     end
   end
 
-  class AlreadyLockedError < StandardError
+  class AlreadyLockedError < RequestError
+    def http_status
+      422
+    end
+  end
+
+  class LockFailedError < RequestError
     def http_status
       422
     end
   end
 
-  class InvalidStateTransitionError < StandardError
+  class InvalidStateTransitionError < RequestError
     def http_status
       422
     end
   end
 
-  class UnauthorizedError < StandardError
+  class UnauthorizedError < RequestError
     def http_status
       401
     end
   end
 
-  class UnresolvableContainerError < StandardError
+  class UnresolvableContainerError < RequestError
     def http_status
       422
     end
@@ -72,6 +95,59 @@ class ArvadosModel < ActiveRecord::Base
     "#{current_api_base}/#{self.class.to_s.pluralize.underscore}/#{self.uuid}"
   end
 
+  def self.permit_attribute_params raw_params
+    # strong_parameters does not provide security: permissions are
+    # implemented with before_save hooks.
+    #
+    # The following permit! is necessary even with
+    # "ActionController::Parameters.permit_all_parameters = true",
+    # because permit_all does not permit nested attributes.
+    raw_params ||= {}
+
+    if raw_params
+      raw_params = raw_params.to_hash
+      raw_params.delete_if { |k, _| self.protected_attributes.include? k }
+      serialized_attributes.each do |colname, coder|
+        param = raw_params[colname.to_sym]
+        if param.nil?
+          # ok
+        elsif !param.is_a?(coder.object_class)
+          raise ArgumentError.new("#{colname} parameter must be #{coder.object_class}, not #{param.class}")
+        elsif has_nonstring_keys?(param)
+          raise ArgumentError.new("#{colname} parameter cannot have non-string hash keys")
+        end
+      end
+      # Check JSONB columns that aren't listed on serialized_attributes
+      columns.select{|c| c.type == :jsonb}.collect{|j| j.name}.each do |colname|
+        if serialized_attributes.include? colname || raw_params[colname.to_sym].nil?
+          next
+        end
+        if has_nonstring_keys?(raw_params[colname.to_sym])
+          raise ArgumentError.new("#{colname} parameter cannot have non-string hash keys")
+        end
+      end
+    end
+    ActionController::Parameters.new(raw_params).permit!
+  end
+
+  def initialize raw_params={}, *args
+    super(self.class.permit_attribute_params(raw_params), *args)
+  end
+
+  # Reload "old attributes" for logging, too.
+  def reload(*args)
+    super
+    log_start_state
+  end
+
+  def self.create raw_params={}, *args
+    super(permit_attribute_params(raw_params), *args)
+  end
+
+  def update_attributes raw_params={}, *args
+    super(self.class.permit_attribute_params(raw_params), *args)
+  end
+
   def self.selectable_attributes(template=:user)
     # Return an array of attribute name strings that can be selected
     # in the given template.
@@ -150,6 +226,14 @@ class ArvadosModel < ActiveRecord::Base
     ["id", "uuid"]
   end
 
+  def self.limit_index_columns_read
+    # This method returns a list of column names.
+    # If an index request reads that column from the database,
+    # APIs that return lists will only fetch objects until reaching
+    # max_index_database_read bytes of data from those columns.
+    []
+  end
+
   # If current user can manage the object, return an array of uuids of
   # users and groups that have permission to write the object. The
   # first two elements are always [self.owner_uuid, current user's
@@ -179,7 +263,7 @@ class ArvadosModel < ActiveRecord::Base
     end.compact.uniq
   end
 
-  # Return a query with read permissions restricted to the union of of the
+  # Return a query with read permissions restricted to the union of the
   # permissions of the members of users_list, i.e. if something is readable by
   # any user in users_list, it will be readable in the query returned by this
   # function.
@@ -194,70 +278,155 @@ class ArvadosModel < ActiveRecord::Base
       kwargs = {}
     end
 
-    # Check if any of the users are admin.  If so, we're done.
-    if users_list.select { |u| u.is_admin }.any?
-      return self
-    end
-
     # Collect the UUIDs of the authorized users.
+    sql_table = kwargs.fetch(:table_name, table_name)
+    include_trash = kwargs.fetch(:include_trash, false)
+    include_old_versions = kwargs.fetch(:include_old_versions, false)
+
+    sql_conds = nil
     user_uuids = users_list.map { |u| u.uuid }
 
-    # Collect the UUIDs of all groups readable by any of the
-    # authorized users. If one of these (or the UUID of one of the
-    # authorized users themselves) is an object's owner_uuid, that
-    # object is readable.
-    owner_uuids = user_uuids + users_list.flat_map { |u| u.groups_i_can(:read) }
-    owner_uuids.uniq!
+    exclude_trashed_records = ""
+    if !include_trash and (sql_table == "groups" or sql_table == "collections") then
+      # Only include records that are not explicitly trashed
+      exclude_trashed_records = "AND #{sql_table}.is_trashed = false"
+    end
 
-    sql_conds = []
-    sql_table = kwargs.fetch(:table_name, table_name)
+    if users_list.select { |u| u.is_admin }.any?
+      # Admin skips most permission checks, but still want to filter on trashed items.
+      if !include_trash
+        if sql_table != "api_client_authorizations"
+          # Only include records where the owner is not trashed
+          sql_conds = "#{sql_table}.owner_uuid NOT IN (SELECT target_uuid FROM #{PERMISSION_VIEW} "+
+                      "WHERE trashed = 1) #{exclude_trashed_records}"
+        end
+      end
+    else
+      trashed_check = ""
+      if !include_trash then
+        trashed_check = "AND trashed = 0"
+      end
 
-    # Match any object (evidently a group or user) whose UUID is
-    # listed explicitly in owner_uuids.
-    sql_conds += ["#{sql_table}.uuid in (:owner_uuids)"]
+      # Note: it is possible to combine the direct_check and
+      # owner_check into a single EXISTS() clause, however it turns
+      # out query optimizer doesn't like it and forces a sequential
+      # table scan.  Constructing the query with separate EXISTS()
+      # clauses enables it to use the index.
+      #
+      # see issue 13208 for details.
+
+      # Match a direct read permission link from the user to the record uuid
+      direct_check = "#{sql_table}.uuid IN (SELECT target_uuid FROM #{PERMISSION_VIEW} "+
+                     "WHERE user_uuid IN (:user_uuids) AND perm_level >= 1 #{trashed_check})"
+
+      # Match a read permission link from the user to the record's owner_uuid
+      owner_check = ""
+      if sql_table != "api_client_authorizations" and sql_table != "groups" then
+        owner_check = "OR #{sql_table}.owner_uuid IN (SELECT target_uuid FROM #{PERMISSION_VIEW} "+
+          "WHERE user_uuid IN (:user_uuids) AND perm_level >= 1 #{trashed_check} AND target_owner_uuid IS NOT NULL) "
+      end
 
-    # Match any object whose owner is listed explicitly in
-    # owner_uuids.
-    sql_conds += ["#{sql_table}.owner_uuid IN (:owner_uuids)"]
+      links_cond = ""
+      if sql_table == "links"
+        # Match any permission link that gives one of the authorized
+        # users some permission _or_ gives anyone else permission to
+        # view one of the authorized users.
+        links_cond = "OR (#{sql_table}.link_class IN (:permission_link_classes) AND "+
+                       "(#{sql_table}.head_uuid IN (:user_uuids) OR #{sql_table}.tail_uuid IN (:user_uuids)))"
+      end
+
+      sql_conds = "(#{direct_check} #{owner_check} #{links_cond}) #{exclude_trashed_records}"
 
-    # Match the head of any permission link whose tail is listed
-    # explicitly in owner_uuids.
-    sql_conds += ["#{sql_table}.uuid IN (SELECT head_uuid FROM links WHERE link_class='permission' AND tail_uuid IN (:owner_uuids))"]
+    end
 
-    if sql_table == "links"
-      # Match any permission link that gives one of the authorized
-      # users some permission _or_ gives anyone else permission to
-      # view one of the authorized users.
-      sql_conds += ["(#{sql_table}.link_class in (:permission_link_classes) AND "+
-                    "(#{sql_table}.head_uuid IN (:user_uuids) OR #{sql_table}.tail_uuid IN (:user_uuids)))"]
+    if !include_old_versions && sql_table == "collections"
+      exclude_old_versions = "#{sql_table}.uuid = #{sql_table}.current_version_uuid"
+      if sql_conds.nil?
+        sql_conds = exclude_old_versions
+      else
+        sql_conds += " AND #{exclude_old_versions}"
+      end
     end
 
-    where(sql_conds.join(' OR '),
-          owner_uuids: owner_uuids,
-          user_uuids: user_uuids,
-          permission_link_classes: ['permission', 'resources'])
+    self.where(sql_conds,
+               user_uuids: user_uuids,
+               permission_link_classes: ['permission', 'resources'])
+  end
+
+  def save_with_unique_name!
+    uuid_was = uuid
+    name_was = name
+    max_retries = 2
+    transaction do
+      conn = ActiveRecord::Base.connection
+      conn.exec_query 'SAVEPOINT save_with_unique_name'
+      begin
+        save!
+      rescue ActiveRecord::RecordNotUnique => rn
+        raise if max_retries == 0
+        max_retries -= 1
+
+        conn.exec_query 'ROLLBACK TO SAVEPOINT save_with_unique_name'
+
+        # Dig into the error to determine if it is specifically calling out a
+        # (owner_uuid, name) uniqueness violation.  In this specific case, and
+        # the client requested a unique name with ensure_unique_name==true,
+        # update the name field and try to save again.  Loop as necessary to
+        # discover a unique name.  It is necessary to handle name choosing at
+        # this level (as opposed to the client) to ensure that record creation
+        # never fails due to a race condition.
+        err = rn.cause
+        raise unless err.is_a?(PG::UniqueViolation)
+
+        # Unfortunately ActiveRecord doesn't abstract out any of the
+        # necessary information to figure out if this the error is actually
+        # the specific case where we want to apply the ensure_unique_name
+        # behavior, so the following code is specialized to Postgres.
+        detail = err.result.error_field(PG::Result::PG_DIAG_MESSAGE_DETAIL)
+        raise unless /^Key \(owner_uuid, name\)=\([a-z0-9]{5}-[a-z0-9]{5}-[a-z0-9]{15}, .*?\) already exists\./.match detail
+
+        new_name = "#{name_was} (#{db_current_time.utc.iso8601(3)})"
+        if new_name == name
+          # If the database is fast enough to do two attempts in the
+          # same millisecond, we need to wait to ensure we try a
+          # different timestamp on each attempt.
+          sleep 0.002
+          new_name = "#{name_was} (#{db_current_time.utc.iso8601(3)})"
+        end
+
+        self[:name] = new_name
+        if uuid_was.nil? && !uuid.nil?
+          self[:uuid] = nil
+          if self.is_a? Collection
+            # Reset so that is assigned to the new UUID
+            self[:current_version_uuid] = nil
+          end
+        end
+        conn.exec_query 'SAVEPOINT save_with_unique_name'
+        retry
+      ensure
+        conn.exec_query 'RELEASE SAVEPOINT save_with_unique_name'
+      end
+    end
   end
 
   def logged_attributes
-    attributes.except(*Rails.configuration.unlogged_attributes)
+    attributes.except(*Rails.configuration.AuditLogs.UnloggedAttributes)
   end
 
   def self.full_text_searchable_columns
     self.columns.select do |col|
-      col.type == :string or col.type == :text
+      [:string, :text, :jsonb].include?(col.type)
     end.map(&:name)
   end
 
   def self.full_text_tsvector
     parts = full_text_searchable_columns.collect do |column|
-      "coalesce(#{column},'')"
+      is_jsonb = self.columns.select{|x|x.name == column}[0].type == :jsonb
+      cast = (is_jsonb || serialized_attributes[column]) ? '::text' : ''
+      "coalesce(#{column}#{cast},'')"
     end
-    # We prepend a space to the tsvector() argument here. Otherwise,
-    # it might start with a column that has its own (non-full-text)
-    # index, which causes Postgres to use the column index instead of
-    # the tsvector index, which causes full text queries to be just as
-    # slow as if we had no index at all.
-    "to_tsvector('english', ' ' || #{parts.join(" || ' ' || ")})"
+    "to_tsvector('english', substr(#{parts.join(" || ' ' || ")}, 0, 8000))"
   end
 
   def self.apply_filters query, filters
@@ -299,7 +468,7 @@ class ArvadosModel < ActiveRecord::Base
           end
         rescue ActiveRecord::RecordNotFound => e
           errors.add :owner_uuid, "is not owned by any user: #{e}"
-          return false
+          throw(:abort)
         end
         if uuid_in_path[x]
           if x == owner_uuid
@@ -307,7 +476,7 @@ class ArvadosModel < ActiveRecord::Base
           else
             errors.add :owner_uuid, "has an ownership cycle"
           end
-          return false
+          throw(:abort)
         end
         uuid_in_path[x] = true
       end
@@ -335,36 +504,31 @@ class ArvadosModel < ActiveRecord::Base
       raise PermissionDeniedError
     end
 
-    # Verify "write" permission on old owner
-    # default fail unless one of:
-    # owner_uuid did not change
-    # previous owner_uuid is nil
-    # current user is the old owner
-    # current user is this object
-    # current user can_write old owner
-    unless !owner_uuid_changed? or
-        owner_uuid_was.nil? or
-        current_user.uuid == self.owner_uuid_was or
-        current_user.uuid == self.uuid or
-        current_user.can? write: self.owner_uuid_was
-      logger.warn "User #{current_user.uuid} tried to modify #{self.class.to_s} #{uuid} but does not have permission to write old owner_uuid #{owner_uuid_was}"
-      errors.add :owner_uuid, "cannot be changed without write permission on old owner"
-      raise PermissionDeniedError
-    end
-
-    # Verify "write" permission on new owner
-    # default fail unless one of:
-    # current_user is this object
-    # current user can_write new owner, or this object if owner unchanged
-    if new_record? or owner_uuid_changed? or is_a?(ApiClientAuthorization)
-      write_target = owner_uuid
+    if new_record? || owner_uuid_changed?
+      # Permission on owner_uuid_was is needed to move an existing
+      # object away from its previous owner (which implies permission
+      # to modify this object itself, so we don't need to check that
+      # separately). Permission on the new owner_uuid is also needed.
+      [['old', owner_uuid_was],
+       ['new', owner_uuid]
+      ].each do |which, check_uuid|
+        if check_uuid.nil?
+          # old_owner_uuid is nil? New record, no need to check.
+        elsif !current_user.can?(write: check_uuid)
+          logger.warn "User #{current_user.uuid} tried to set ownership of #{self.class.to_s} #{self.uuid} but does not have permission to write #{which} owner_uuid #{check_uuid}"
+          errors.add :owner_uuid, "cannot be set or changed without write permission on #{which} owner"
+          raise PermissionDeniedError
+        end
+      end
     else
-      write_target = uuid
-    end
-    unless current_user == self or current_user.can? write: write_target
-      logger.warn "User #{current_user.uuid} tried to modify #{self.class.to_s} #{uuid} but does not have permission to write new owner_uuid #{owner_uuid}"
-      errors.add :owner_uuid, "cannot be changed without write permission on new owner"
-      raise PermissionDeniedError
+      # If the object already existed and we're not changing
+      # owner_uuid, we only need write permission on the object
+      # itself.
+      if !current_user.can?(write: self.uuid)
+        logger.warn "User #{current_user.uuid} tried to modify #{self.class.to_s} #{self.uuid} without write permission"
+        errors.add :uuid, "is not writable"
+        raise PermissionDeniedError
+      end
     end
 
     true
@@ -412,14 +576,32 @@ class ArvadosModel < ActiveRecord::Base
 
   def update_modified_by_fields
     current_time = db_current_time
+    self.created_at ||= created_at_was || current_time
     self.updated_at = current_time
     self.owner_uuid ||= current_default_owner if self.respond_to? :owner_uuid=
-    self.modified_at = current_time
-    self.modified_by_user_uuid = current_user ? current_user.uuid : nil
+    if !anonymous_updater
+      self.modified_by_user_uuid = current_user ? current_user.uuid : nil
+    end
+    if !timeless_updater
+      self.modified_at = current_time
+    end
     self.modified_by_client_uuid = current_api_client ? current_api_client.uuid : nil
     true
   end
 
+  def self.has_nonstring_keys? x
+    if x.is_a? Hash
+      x.each do |k,v|
+        return true if !(k.is_a?(String) || k.is_a?(Symbol)) || has_nonstring_keys?(v)
+      end
+    elsif x.is_a? Array
+      x.each do |v|
+        return true if has_nonstring_keys?(v)
+      end
+    end
+    false
+  end
+
   def self.has_symbols? x
     if x.is_a? Hash
       x.each do |k,v|
@@ -455,22 +637,44 @@ class ArvadosModel < ActiveRecord::Base
     end
   end
 
-  def ensure_serialized_attribute_type
-    # Specifying a type in the "serialize" declaration causes rails to
-    # raise an exception if a different data type is retrieved from
-    # the database during load().  The validation preventing such
-    # crash-inducing records from being inserted in the database in
-    # the first place seems to have been left as an exercise to the
-    # developer.
-    self.class.serialized_attributes.each do |colname, attr|
-      if attr.object_class
-        if self.attributes[colname].class != attr.object_class
-          self.errors.add colname.to_sym, "must be a #{attr.object_class.to_s}, not a #{self.attributes[colname].class.to_s}"
-        elsif self.class.has_symbols? attributes[colname]
-          self.errors.add colname.to_sym, "must not contain symbols: #{attributes[colname].inspect}"
-        end
-      end
+  def self.where_serialized(colname, value, md5: false)
+    colsql = colname.to_s
+    if md5
+      colsql = "md5(#{colsql})"
+    end
+    if value.empty?
+      # rails4 stores as null, rails3 stored as serialized [] or {}
+      sql = "#{colsql} is null or #{colsql} IN (?)"
+      sorted = value
+    else
+      sql = "#{colsql} IN (?)"
+      sorted = deep_sort_hash(value)
     end
+    params = [sorted.to_yaml, SafeJSON.dump(sorted)]
+    if md5
+      params = params.map { |x| Digest::MD5.hexdigest(x) }
+    end
+    where(sql, params)
+  end
+
+  Serializer = {
+    Hash => HashSerializer,
+    Array => ArraySerializer,
+  }
+
+  def self.serialize(colname, type)
+    coder = Serializer[type]
+    @serialized_attributes ||= {}
+    @serialized_attributes[colname.to_s] = coder
+    super(colname, coder)
+  end
+
+  def self.serialized_attributes
+    @serialized_attributes ||= {}
+  end
+
+  def serialized_attributes
+    self.class.serialized_attributes
   end
 
   def convert_serialized_symbols_to_strings
@@ -480,11 +684,12 @@ class ArvadosModel < ActiveRecord::Base
     # we'll convert symbols to strings when loading from the
     # database. (Otherwise, loading and saving an object with existing
     # symbols in a serialized field will crash.)
-    self.class.serialized_attributes.each do |colname, attr|
+    jsonb_cols = self.class.columns.select{|c| c.type == :jsonb}.collect{|j| j.name}
+    (jsonb_cols + self.class.serialized_attributes.keys).uniq.each do |colname|
       if self.class.has_symbols? attributes[colname]
         attributes[colname] = self.class.recursive_stringify attributes[colname]
-        self.send(colname + '=',
-                  self.class.recursive_stringify(attributes[colname]))
+        send(colname + '=',
+             self.class.recursive_stringify(attributes[colname]))
       end
     end
   end
@@ -530,7 +735,7 @@ class ArvadosModel < ActiveRecord::Base
   end
 
   def self.uuid_like_pattern
-    "#{Rails.configuration.uuid_prefix}-#{uuid_prefix}-_______________"
+    "#{Rails.configuration.ClusterID}-#{uuid_prefix}-_______________"
   end
 
   def self.uuid_regex
@@ -608,36 +813,51 @@ class ArvadosModel < ActiveRecord::Base
     end
   end
 
+  def is_audit_logging_enabled?
+    return !(Rails.configuration.AuditLogs.MaxAge.to_i == 0 &&
+             Rails.configuration.AuditLogs.MaxDeleteBatch.to_i > 0)
+  end
+
   def log_start_state
-    @old_attributes = Marshal.load(Marshal.dump(attributes))
-    @old_logged_attributes = Marshal.load(Marshal.dump(logged_attributes))
+    if is_audit_logging_enabled?
+      @old_attributes = Marshal.load(Marshal.dump(attributes))
+      @old_logged_attributes = Marshal.load(Marshal.dump(logged_attributes))
+    end
   end
 
   def log_change(event_type)
-    log = Log.new(event_type: event_type).fill_object(self)
-    yield log
-    log.save!
-    log_start_state
+    if is_audit_logging_enabled?
+      log = Log.new(event_type: event_type).fill_object(self)
+      yield log
+      log.save!
+      log_start_state
+    end
   end
 
   def log_create
-    log_change('create') do |log|
-      log.fill_properties('old', nil, nil)
-      log.update_to self
+    if is_audit_logging_enabled?
+      log_change('create') do |log|
+        log.fill_properties('old', nil, nil)
+        log.update_to self
+      end
     end
   end
 
   def log_update
-    log_change('update') do |log|
-      log.fill_properties('old', etag(@old_attributes), @old_logged_attributes)
-      log.update_to self
+    if is_audit_logging_enabled?
+      log_change('update') do |log|
+        log.fill_properties('old', etag(@old_attributes), @old_logged_attributes)
+        log.update_to self
+      end
     end
   end
 
   def log_destroy
-    log_change('delete') do |log|
-      log.fill_properties('old', etag(@old_attributes), @old_logged_attributes)
-      log.update_to nil
+    if is_audit_logging_enabled?
+      log_change('delete') do |log|
+        log.fill_properties('old', etag(@old_attributes), @old_logged_attributes)
+        log.update_to nil
+      end
     end
   end
 end