gitlab-org--gitlab-foss/app/models/integration.rb

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

609 lines
18 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
# To add new integration you should build a class inherited from Integration
2013-05-22 13:58:44 +00:00
# and implement a set of methods
class Integration < ApplicationRecord
include Sortable
2018-01-25 15:21:23 +00:00
include Importable
include Integrations::Loggable
include Integrations::HasDataFields
include Integrations::ResetSecretFields
include FromUnion
include EachBatch
include IgnorableColumns
extend ::Gitlab::Utils::Override
UnknownType = Class.new(StandardError)
self.inheritance_column = :type_new
INTEGRATION_NAMES = %w[
asana assembla bamboo bugzilla buildkite campfire confluence custom_issue_tracker datadog discord
drone_ci emails_on_push ewm external_wiki flowdock hangouts_chat harbor irker jira
mattermost mattermost_slash_commands microsoft_teams packagist pipelines_email
pivotaltracker prometheus pushover redmine slack slack_slash_commands teamcity unify_circuit webex_teams youtrack zentao
].freeze
# TODO Shimo is temporary disabled on group and instance-levels.
# See: https://gitlab.com/gitlab-org/gitlab/-/issues/345677
PROJECT_SPECIFIC_INTEGRATION_NAMES = %w[
jenkins shimo
].freeze
# Fake integrations to help with local development.
DEV_INTEGRATION_NAMES = %w[
mock_ci mock_monitoring
].freeze
# Base classes which aren't actual integrations.
BASE_CLASSES = %w[
Integrations::BaseChatNotification
Integrations::BaseCi
Integrations::BaseIssueTracker
Integrations::BaseMonitoring
Integrations::BaseSlashCommands
].freeze
SECTION_TYPE_CONFIGURATION = 'configuration'
SECTION_TYPE_CONNECTION = 'connection'
SECTION_TYPE_TRIGGER = 'trigger'
attr_encrypted :properties,
mode: :per_attribute_iv,
key: Settings.attr_encrypted_db_key_base_32,
algorithm: 'aes-256-gcm',
marshal: true,
marshaler: ::Gitlab::Json,
encode: false,
encode_iv: false
# Handle assignment of props with symbol keys.
# To do this correctly, we need to call the method generated by attr_encrypted.
alias_method :attr_encrypted_props=, :properties=
private :attr_encrypted_props=
def properties=(props)
self.attr_encrypted_props = props&.with_indifferent_access&.freeze
end
alias_attribute :type, :type_new
default_value_for :active, false
default_value_for :alert_events, true
default_value_for :category, 'common'
default_value_for :commit_events, true
default_value_for :confidential_issues_events, true
default_value_for :confidential_note_events, true
default_value_for :issues_events, true
default_value_for :job_events, true
default_value_for :merge_requests_events, true
default_value_for :note_events, true
default_value_for :pipeline_events, true
default_value_for :push_events, true
default_value_for :tag_push_events, true
default_value_for :wiki_page_events, true
2014-09-11 15:48:29 +00:00
after_initialize :initialize_properties
after_commit :reset_updated_properties
belongs_to :project, inverse_of: :integrations
belongs_to :group, inverse_of: :integrations
has_one :service_hook, inverse_of: :integration, foreign_key: :service_id
2012-11-19 18:14:05 +00:00
validates :project_id, presence: true, unless: -> { instance_level? || group_level? }
validates :group_id, presence: true, unless: -> { instance_level? || project_level? }
validates :project_id, :group_id, absence: true, if: -> { instance_level? }
validates :type, presence: true, exclusion: BASE_CLASSES
validates :type, uniqueness: { scope: :instance }, if: :instance_level?
validates :type, uniqueness: { scope: :project_id }, if: :project_level?
validates :type, uniqueness: { scope: :group_id }, if: :group_level?
validate :validate_belongs_to_project_or_group
2013-01-03 07:52:14 +00:00
scope :external_issue_trackers, -> { where(category: 'issue_tracker').active }
# TODO: Will be modified in 15.0
# Details: https://gitlab.com/gitlab-org/gitlab/-/merge_requests/74501#note_744393645
scope :third_party_wikis, -> { where(type: %w[Integrations::Confluence Integrations::Shimo]).active }
scope :by_name, ->(name) { by_type(integration_name_to_type(name)) }
scope :external_wikis, -> { by_name(:external_wiki).active }
Greatly improve external_issue_tracker performance This greatly improves the performance of Project#external_issue_tracker by moving most of the fields queried in Ruby to the database and letting the database handle all logic. Prior to this change the process of finding an external issue tracker was along the lines of the following: 1. Load all project services into memory. 2. Reduce the list to only services where "issue_tracker?" returns true 3. Reduce the list from step 2 to service where "default?" returns false 4. Find the first service where "activated?" returns true This has to two big problems: 1. Loading all services into memory only to reduce the list down to a single item later on is a waste of memory (and slow timing wise). 2. Calling Array#select followed by Array#reject followed by Array#find allocates extra objects when this really isn't needed. To work around this the following service fields have been moved to the database (instead of being hardcoded): * category * default This in turn means we can get the external issue tracker using the following query: SELECT * FROM services WHERE active IS TRUE AND default IS FALSE AND category = 'issue_tracker' AND project_id = XXX LIMIT 1 This coupled with memoizing the result (just as before this commit) greatly reduces the time it takes for Project#external_issue_tracker to complete. The exact reduction depends on one's environment, but locally the execution time is reduced from roughly 230 ms to only 2 ms (= a reduction of almost 180x). Fixes gitlab-org/gitlab-ce#10771
2016-01-19 12:48:07 +00:00
scope :active, -> { where(active: true) }
scope :by_type, ->(type) { where(type: type) } # INTERNAL USE ONLY: use by_name instead
scope :by_active_flag, -> (flag) { where(active: flag) }
scope :inherit_from_id, -> (id) { where(inherit_from_id: id) }
scope :with_default_settings, -> { where.not(inherit_from_id: nil) }
scope :with_custom_settings, -> { where(inherit_from_id: nil) }
scope :for_group, -> (group) { where(group_id: group, type: available_integration_types(include_project_specific: false)) }
scope :for_instance, -> { where(instance: true, type: available_integration_types(include_project_specific: false)) }
scope :push_hooks, -> { where(push_events: true, active: true) }
scope :tag_push_hooks, -> { where(tag_push_events: true, active: true) }
scope :issue_hooks, -> { where(issues_events: true, active: true) }
scope :confidential_issue_hooks, -> { where(confidential_issues_events: true, active: true) }
scope :merge_request_hooks, -> { where(merge_requests_events: true, active: true) }
scope :note_hooks, -> { where(note_events: true, active: true) }
scope :confidential_note_hooks, -> { where(confidential_note_events: true, active: true) }
scope :job_hooks, -> { where(job_events: true, active: true) }
scope :archive_trace_hooks, -> { where(archive_trace_events: true, active: true) }
scope :pipeline_hooks, -> { where(pipeline_events: true, active: true) }
scope :wiki_page_hooks, -> { where(wiki_page_events: true, active: true) }
scope :deployment_hooks, -> { where(deployment_events: true, active: true) }
scope :alert_hooks, -> { where(alert_events: true, active: true) }
scope :deployment, -> { where(category: 'deployment') }
class << self
private
attr_writer :field_storage
def field_storage
@field_storage || :properties
end
end
# :nocov: Tested on subclasses.
def self.field(name, storage: field_storage, **attrs)
fields << ::Integrations::Field.new(name: name, integration_class: self, **attrs)
case storage
when :properties
prop_accessor(name)
when :data_fields
data_field(name)
else
raise ArgumentError, "Unknown field storage: #{storage}"
end
boolean_accessor(name) if attrs[:type] == 'checkbox'
end
# :nocov:
def self.fields
@fields ||= []
end
def fields
self.class.fields.dup
end
# Provide convenient accessor methods for each serialized property.
# Also keep track of updated properties in a similar way as ActiveModel::Dirty
def self.prop_accessor(*args)
args.each do |arg|
class_eval <<~RUBY, __FILE__, __LINE__ + 1
unless method_defined?(arg)
def #{arg}
properties['#{arg}'] if properties.present?
end
end
def #{arg}=(value)
self.properties ||= {}
updated_properties['#{arg}'] = #{arg} unless #{arg}_changed?
self.properties = self.properties.merge('#{arg}' => value)
end
def #{arg}_changed?
#{arg}_touched? && #{arg} != #{arg}_was
end
def #{arg}_touched?
updated_properties.include?('#{arg}')
end
def #{arg}_was
updated_properties['#{arg}']
end
RUBY
end
end
# Provide convenient boolean accessor methods for each serialized property.
# Also keep track of updated properties in a similar way as ActiveModel::Dirty
def self.boolean_accessor(*args)
args.each do |arg|
# TODO: Allow legacy usage of `.boolean_accessor`, once all integrations
# are converted to the field DSL we can remove this and only call
# `.boolean_accessor` through `.field`.
#
# See https://gitlab.com/groups/gitlab-org/-/epics/7652
prop_accessor(arg) unless method_defined?(arg)
class_eval <<~RUBY, __FILE__, __LINE__ + 1
# Make the original getter available as a private method.
alias_method :#{arg}_before_type_cast, :#{arg}
private(:#{arg}_before_type_cast)
def #{arg}
Gitlab::Utils.to_boolean(#{arg}_before_type_cast)
end
def #{arg}?
# '!!' is used because nil or empty string is converted to nil
!!#{arg}
end
RUBY
end
end
def self.to_param
raise NotImplementedError
end
def self.event_names
self.supported_events.map { |event| IntegrationsHelper.integration_event_field_name(event) }
end
def self.supported_events
%w[commit push tag_push issue confidential_issue merge_request wiki_page]
end
def self.default_test_event
'push'
end
def self.event_description(event)
IntegrationsHelper.integration_event_description(event)
end
def self.find_or_initialize_non_project_specific_integration(name, instance: false, group_id: nil)
return unless name.in?(available_integration_names(include_project_specific: false))
integration_name_to_model(name).find_or_initialize_by(instance: instance, group_id: group_id)
end
def self.find_or_initialize_all_non_project_specific(scope)
scope + build_nonexistent_integrations_for(scope)
end
def self.build_nonexistent_integrations_for(scope)
nonexistent_integration_types_for(scope).map do |type|
integration_type_to_model(type).new
end
end
private_class_method :build_nonexistent_integrations_for
# Returns a list of integration types that do not exist in the given scope.
# Example: ["AsanaService", ...]
def self.nonexistent_integration_types_for(scope)
# Using #map instead of #pluck to save one query count. This is because
# ActiveRecord loaded the object here, so we don't need to query again later.
available_integration_types(include_project_specific: false) - scope.map(&:type)
end
private_class_method :nonexistent_integration_types_for
# Returns a list of available integration names.
# Example: ["asana", ...]
# @deprecated
def self.available_integration_names(include_project_specific: true, include_dev: true)
names = integration_names
names += project_specific_integration_names if include_project_specific
names += dev_integration_names if include_dev
names.sort_by(&:downcase)
end
def self.integration_names
INTEGRATION_NAMES
end
def self.dev_integration_names
return [] unless Gitlab.dev_or_test_env?
DEV_INTEGRATION_NAMES
end
def self.project_specific_integration_names
PROJECT_SPECIFIC_INTEGRATION_NAMES
end
# Returns a list of available integration types.
# Example: ["Integrations::Asana", ...]
def self.available_integration_types(include_project_specific: true, include_dev: true)
available_integration_names(include_project_specific: include_project_specific, include_dev: include_dev).map do
integration_name_to_type(_1)
end
end
# Returns the model for the given integration name.
# Example: :asana => Integrations::Asana
def self.integration_name_to_model(name)
type = integration_name_to_type(name)
integration_type_to_model(type)
end
# Returns the STI type for the given integration name.
# Example: "asana" => "Integrations::Asana"
def self.integration_name_to_type(name)
name = name.to_s
if available_integration_names.exclude?(name)
Gitlab::ErrorTracking.track_and_raise_for_dev_exception(UnknownType.new(name.inspect))
else
"Integrations::#{name.camelize}"
end
end
# Returns the model for the given STI type.
# Example: "Integrations::Asana" => Integrations::Asana
def self.integration_type_to_model(type)
type.constantize
end
private_class_method :integration_type_to_model
def self.build_from_integration(integration, project_id: nil, group_id: nil)
new_integration = integration.dup
new_integration.instance = false
new_integration.project_id = project_id
new_integration.group_id = group_id
new_integration.inherit_from_id = integration.id if integration.inheritable?
new_integration
end
# Duplicating an integration also duplicates the data fields. Duped records have different ciphertexts.
override :dup
def dup
new_integration = super
new_integration.assign_attributes(reencrypt_properties)
if supports_data_fields?
fields = data_fields.dup
fields.integration = new_integration
end
new_integration
end
def inheritable?
instance_level? || group_level?
end
def self.instance_exists_for?(type)
exists?(instance: true, type: type)
end
def self.default_integration(type, scope)
closest_group_integration(type, scope) || instance_level_integration(type)
end
Greatly improve external_issue_tracker performance This greatly improves the performance of Project#external_issue_tracker by moving most of the fields queried in Ruby to the database and letting the database handle all logic. Prior to this change the process of finding an external issue tracker was along the lines of the following: 1. Load all project services into memory. 2. Reduce the list to only services where "issue_tracker?" returns true 3. Reduce the list from step 2 to service where "default?" returns false 4. Find the first service where "activated?" returns true This has to two big problems: 1. Loading all services into memory only to reduce the list down to a single item later on is a waste of memory (and slow timing wise). 2. Calling Array#select followed by Array#reject followed by Array#find allocates extra objects when this really isn't needed. To work around this the following service fields have been moved to the database (instead of being hardcoded): * category * default This in turn means we can get the external issue tracker using the following query: SELECT * FROM services WHERE active IS TRUE AND default IS FALSE AND category = 'issue_tracker' AND project_id = XXX LIMIT 1 This coupled with memoizing the result (just as before this commit) greatly reduces the time it takes for Project#external_issue_tracker to complete. The exact reduction depends on one's environment, but locally the execution time is reduced from roughly 230 ms to only 2 ms (= a reduction of almost 180x). Fixes gitlab-org/gitlab-ce#10771
2016-01-19 12:48:07 +00:00
def self.closest_group_integration(type, scope)
group_ids = scope.ancestors(hierarchy_order: :asc).reselect(:id)
array = group_ids.to_sql.present? ? "array(#{group_ids.to_sql})" : 'ARRAY[]'
where(type: type, group_id: group_ids, inherit_from_id: nil)
.order(Arel.sql("array_position(#{array}::bigint[], #{table_name}.group_id)"))
.first
end
private_class_method :closest_group_integration
def self.instance_level_integration(type)
find_by(type: type, instance: true)
end
private_class_method :instance_level_integration
# Returns the number of successfully saved integrations
# Duplicate integrations are excluded from this count by their validations.
def self.create_from_active_default_integrations(owner, association)
group_ids = sorted_ancestors(owner).select(:id)
array = group_ids.to_sql.present? ? "array(#{group_ids.to_sql})" : 'ARRAY[]'
order = Arel.sql("type_new ASC, array_position(#{array}::bigint[], #{table_name}.group_id), instance DESC")
from_union([active.where(instance: true), active.where(group_id: group_ids, inherit_from_id: nil)])
.order(order)
.group_by(&:type)
.count { |type, parents| build_from_integration(parents.first, association => owner.id).save }
end
def self.inherited_descendants_from_self_or_ancestors_from(integration)
inherit_from_ids =
where(type: integration.type, group: integration.group.self_and_ancestors)
.or(where(type: integration.type, instance: true)).select(:id)
from_union([
where(type: integration.type, inherit_from_id: inherit_from_ids, group: integration.group.descendants),
where(type: integration.type, inherit_from_id: inherit_from_ids, project: Project.in_namespace(integration.group.self_and_descendants))
])
end
2013-01-03 07:52:14 +00:00
def activated?
active
end
2013-05-22 13:58:44 +00:00
def operating?
active && persisted?
end
def show_active_box?
true
end
def editable?
true
end
def activate_disabled_reason
nil
end
def category
Greatly improve external_issue_tracker performance This greatly improves the performance of Project#external_issue_tracker by moving most of the fields queried in Ruby to the database and letting the database handle all logic. Prior to this change the process of finding an external issue tracker was along the lines of the following: 1. Load all project services into memory. 2. Reduce the list to only services where "issue_tracker?" returns true 3. Reduce the list from step 2 to service where "default?" returns false 4. Find the first service where "activated?" returns true This has to two big problems: 1. Loading all services into memory only to reduce the list down to a single item later on is a waste of memory (and slow timing wise). 2. Calling Array#select followed by Array#reject followed by Array#find allocates extra objects when this really isn't needed. To work around this the following service fields have been moved to the database (instead of being hardcoded): * category * default This in turn means we can get the external issue tracker using the following query: SELECT * FROM services WHERE active IS TRUE AND default IS FALSE AND category = 'issue_tracker' AND project_id = XXX LIMIT 1 This coupled with memoizing the result (just as before this commit) greatly reduces the time it takes for Project#external_issue_tracker to complete. The exact reduction depends on one's environment, but locally the execution time is reduced from roughly 230 ms to only 2 ms (= a reduction of almost 180x). Fixes gitlab-org/gitlab-ce#10771
2016-01-19 12:48:07 +00:00
read_attribute(:category).to_sym
end
2014-09-11 15:48:29 +00:00
def initialize_properties
self.properties = {} if has_attribute?(:encrypted_properties) && encrypted_properties.nil?
end
2013-05-22 13:58:44 +00:00
def title
# implement inside child
2013-05-22 13:58:44 +00:00
end
def description
# implement inside child
2013-05-22 13:58:44 +00:00
end
def help
# implement inside child
end
2013-05-22 13:58:44 +00:00
def to_param
# implement inside child
self.class.to_param
2013-05-22 13:58:44 +00:00
end
def sections
[]
end
# TODO: Once all integrations use `Integrations::Field` we can
# use `#secret?` here.
# See: https://gitlab.com/groups/gitlab-org/-/epics/7652
def secret_fields
fields.select { |f| f[:type] == 'password' }.pluck(:name)
end
# Expose a list of fields in the JSON endpoint.
#
# This list is used in `Integration#as_json(only: json_fields)`.
def json_fields
%w[active]
end
# properties is always nil - ignore it.
override :attributes
def attributes
super.except('properties')
end
# Returns a hash of attributes (columns => values) used for inserting into the database.
def to_database_hash
column = self.class.attribute_aliases.fetch('type', 'type')
as_json(
except: %w[id instance project_id group_id created_at updated_at]
).merge(column => type)
.merge(reencrypt_properties)
end
def reencrypt_properties
unless properties.nil? || properties.empty?
alg = self.class.encrypted_attributes[:properties][:algorithm]
iv = generate_iv(alg)
ep = self.class.encrypt(:properties, properties, { iv: iv })
end
{ 'encrypted_properties' => ep, 'encrypted_properties_iv' => iv }
end
def event_channel_names
[]
end
def event_names
self.class.event_names
end
def api_field_names
fields.reject { _1[:type] == 'password' }.pluck(:name)
end
def form_fields
fields.reject { _1[:api_only] == true }
2016-07-13 19:49:47 +00:00
end
def configurable_events
events = supported_events
# No need to disable individual triggers when there is only one
if events.count == 1
[]
else
events
end
end
def supported_events
self.class.supported_events
end
def default_test_event
self.class.default_test_event
end
def execute(data)
# implement inside child
end
def test(data)
# default implementation
result = execute(data)
{ success: result.present?, result: result }
end
# Disable test for instance-level and group-level integrations.
# https://gitlab.com/gitlab-org/gitlab/-/issues/213138
def testable?
project_level?
end
2014-09-08 00:54:18 +00:00
def project_level?
project_id.present?
end
def group_level?
group_id.present?
end
def instance_level?
instance?
end
def parent
project || group
end
# Returns a hash of the properties that have been assigned a new value since last save,
# indicating their original values (attr => original value).
Greatly improve external_issue_tracker performance This greatly improves the performance of Project#external_issue_tracker by moving most of the fields queried in Ruby to the database and letting the database handle all logic. Prior to this change the process of finding an external issue tracker was along the lines of the following: 1. Load all project services into memory. 2. Reduce the list to only services where "issue_tracker?" returns true 3. Reduce the list from step 2 to service where "default?" returns false 4. Find the first service where "activated?" returns true This has to two big problems: 1. Loading all services into memory only to reduce the list down to a single item later on is a waste of memory (and slow timing wise). 2. Calling Array#select followed by Array#reject followed by Array#find allocates extra objects when this really isn't needed. To work around this the following service fields have been moved to the database (instead of being hardcoded): * category * default This in turn means we can get the external issue tracker using the following query: SELECT * FROM services WHERE active IS TRUE AND default IS FALSE AND category = 'issue_tracker' AND project_id = XXX LIMIT 1 This coupled with memoizing the result (just as before this commit) greatly reduces the time it takes for Project#external_issue_tracker to complete. The exact reduction depends on one's environment, but locally the execution time is reduced from roughly 230 ms to only 2 ms (= a reduction of almost 180x). Fixes gitlab-org/gitlab-ce#10771
2016-01-19 12:48:07 +00:00
# ActiveRecord does not provide a mechanism to track changes in serialized keys,
# so we need a specific implementation for integration properties.
# This allows to track changes to properties set with the accessor methods,
# but not direct manipulation of properties hash.
def updated_properties
@updated_properties ||= ActiveSupport::HashWithIndifferentAccess.new
end
def reset_updated_properties
@updated_properties = nil
end
Greatly improve external_issue_tracker performance This greatly improves the performance of Project#external_issue_tracker by moving most of the fields queried in Ruby to the database and letting the database handle all logic. Prior to this change the process of finding an external issue tracker was along the lines of the following: 1. Load all project services into memory. 2. Reduce the list to only services where "issue_tracker?" returns true 3. Reduce the list from step 2 to service where "default?" returns false 4. Find the first service where "activated?" returns true This has to two big problems: 1. Loading all services into memory only to reduce the list down to a single item later on is a waste of memory (and slow timing wise). 2. Calling Array#select followed by Array#reject followed by Array#find allocates extra objects when this really isn't needed. To work around this the following service fields have been moved to the database (instead of being hardcoded): * category * default This in turn means we can get the external issue tracker using the following query: SELECT * FROM services WHERE active IS TRUE AND default IS FALSE AND category = 'issue_tracker' AND project_id = XXX LIMIT 1 This coupled with memoizing the result (just as before this commit) greatly reduces the time it takes for Project#external_issue_tracker to complete. The exact reduction depends on one's environment, but locally the execution time is reduced from roughly 230 ms to only 2 ms (= a reduction of almost 180x). Fixes gitlab-org/gitlab-ce#10771
2016-01-19 12:48:07 +00:00
def async_execute(data)
return unless supported_events.include?(data[:object_kind])
Integrations::ExecuteWorker.perform_async(id, data)
end
# override if needed
def supports_data_fields?
false
end
private
# Ancestors sorted by hierarchy depth in bottom-top order.
def self.sorted_ancestors(scope)
if scope.root_ancestor.use_traversal_ids?
Namespace.from(scope.ancestors(hierarchy_order: :asc))
else
scope.ancestors
end
end
def validate_belongs_to_project_or_group
errors.add(:project_id, 'The service cannot belong to both a project and a group') if project_level? && group_level?
end
def validate_recipients?
2018-01-26 13:48:47 +00:00
activated? && !importing?
end
2012-11-19 18:14:05 +00:00
end
Integration.prepend_mod_with('Integration')