gitlab-org--gitlab-foss/lib/banzai/reference_parser/base_parser.rb
Yorick Peterse daad7144ec
Support Markdown rendering using multiple projects
This refactors the Markdown pipeline so it supports the rendering of
multiple documents that may belong to different projects. An example of
where this happens is when displaying the event feed of a group. In this
case we retrieve events for all projects in the group. Previously we
would group events per project and render these chunks separately, but
this would result in many SQL queries being executed. By extending the
Markdown pipeline to support this out of the box we can drastically
reduce the number of SQL queries.

To achieve this we introduce a new object to the pipeline:
Banzai::RenderContext. This object simply wraps two other objects: an
optional Project instance, and an optional User instance. On its own
this wouldn't be very helpful, but a RenderContext can also be used to
associate HTML documents with specific Project instances. This work is
done in Banzai::ObjectRenderer and allows us to reuse as many queries
(and results) as possible.
2018-04-11 14:10:19 +02:00

257 lines
8 KiB
Ruby

module Banzai
module ReferenceParser
# Base class for reference parsing classes.
#
# Each parser should also specify its reference type by calling
# `self.reference_type = ...` in the body of the class. The value of this
# method should be a symbol such as `:issue` or `:merge_request`. For
# example:
#
# class IssueParser < BaseParser
# self.reference_type = :issue
# end
#
# The reference type is used to determine what nodes to pass to the
# `referenced_by` method.
#
# Parser classes should either implement the instance method
# `references_relation` or overwrite `referenced_by`. The
# `references_relation` method is supposed to return an
# ActiveRecord::Relation used as a base relation for retrieving the objects
# referenced in a set of HTML nodes.
#
# Each class can implement two additional methods:
#
# * `nodes_user_can_reference`: returns an Array of nodes the given user can
# refer to.
# * `nodes_visible_to_user`: returns an Array of nodes that are visible to
# the given user.
#
# You only need to overwrite these methods if you want to tweak who can see
# which references. For example, the IssueParser class defines its own
# `nodes_visible_to_user` method so it can ensure users can only see issues
# they have access to.
class BaseParser
class << self
attr_accessor :reference_type, :reference_options
end
# Returns the attribute name containing the value for every object to be
# parsed by the current parser.
#
# For example, for a parser class that returns "Animal" objects this
# attribute would be "data-animal".
def self.data_attribute
@data_attribute ||= "data-#{reference_type.to_s.dasherize}"
end
# context - An instance of `Banzai::RenderContext`.
def initialize(context)
@context = context
end
def project_for_node(node)
context.project_for_node(node)
end
# Returns all the nodes containing references that the user can refer to.
def nodes_user_can_reference(user, nodes)
nodes
end
# Returns all the nodes that are visible to the given user.
def nodes_visible_to_user(user, nodes)
projects = lazy { projects_for_nodes(nodes) }
project_attr = 'data-project'
nodes.select do |node|
if node.has_attribute?(project_attr)
can_read_reference?(user, projects[node], node)
else
true
end
end
end
# Returns an Array of objects referenced by any of the given HTML nodes.
def referenced_by(nodes)
ids = unique_attribute_values(nodes, self.class.data_attribute)
if ids.empty?
references_relation.none
else
references_relation.where(id: ids)
end
end
# Returns the ActiveRecord::Relation to use for querying references in the
# DB.
def references_relation
raise NotImplementedError,
"#{self.class} does not implement #{__method__}"
end
# Returns a Hash containing attribute values per project ID.
#
# The returned Hash uses the following format:
#
# { project id => [value1, value2, ...] }
#
# nodes - An Array of HTML nodes to process.
# attribute - The name of the attribute (as a String) for which to gather
# values.
#
# Returns a Hash.
def gather_attributes_per_project(nodes, attribute)
per_project = Hash.new { |hash, key| hash[key] = Set.new }
nodes.each do |node|
project_id = node.attr('data-project').to_i
id = node.attr(attribute)
per_project[project_id] << id if id
end
per_project
end
# Returns a Hash containing objects for an attribute grouped per the
# nodes that reference them.
#
# The returned Hash uses the following format:
#
# { node => row }
#
# nodes - An Array of HTML nodes to process.
#
# collection - The model or ActiveRecord relation to use for retrieving
# rows from the database.
#
# attribute - The name of the attribute containing the primary key values
# for every row.
#
# Returns a Hash.
def grouped_objects_for_nodes(nodes, collection, attribute)
return {} if nodes.empty?
ids = unique_attribute_values(nodes, attribute)
collection_objects = collection_objects_for_ids(collection, ids)
objects_by_id = collection_objects.index_by(&:id)
nodes.each_with_object({}) do |node, hash|
if node.has_attribute?(attribute)
obj = objects_by_id[node.attr(attribute).to_i]
hash[node] = obj if obj
end
end
end
# Returns an Array containing all unique values of an attribute of the
# given nodes.
def unique_attribute_values(nodes, attribute)
values = Set.new
nodes.each do |node|
if node.has_attribute?(attribute)
values << node.attr(attribute)
end
end
values.to_a
end
# Queries the collection for the objects with the given IDs.
#
# If the RequestStore module is enabled this method will only query any
# objects that have not yet been queried. For objects that have already
# been queried the object is returned from the cache.
def collection_objects_for_ids(collection, ids)
if RequestStore.active?
ids = ids.map(&:to_i)
cache = collection_cache[collection_cache_key(collection)]
to_query = ids - cache.keys
unless to_query.empty?
collection.where(id: to_query).each { |row| cache[row.id] = row }
end
cache.values_at(*ids).compact
else
collection.where(id: ids)
end
end
# Returns the cache key to use for a collection.
def collection_cache_key(collection)
collection.respond_to?(:model) ? collection.model : collection
end
# Processes the list of HTML documents and returns an Array containing all
# the references.
def process(documents)
type = self.class.reference_type
reference_options = self.class.reference_options
nodes = documents.flat_map do |document|
Querying.css(document, "a[data-reference-type='#{type}'].gfm", reference_options).to_a
end
gather_references(nodes)
end
# Gathers the references for the given HTML nodes.
def gather_references(nodes)
nodes = nodes_user_can_reference(current_user, nodes)
nodes = nodes_visible_to_user(current_user, nodes)
referenced_by(nodes)
end
# Returns a Hash containing the projects for a given list of HTML nodes.
#
# The returned Hash uses the following format:
#
# { node => project }
#
def projects_for_nodes(nodes)
@projects_for_nodes ||=
grouped_objects_for_nodes(nodes, Project, 'data-project')
end
def can?(user, permission, subject = :global)
Ability.allowed?(user, permission, subject)
end
def find_projects_for_hash_keys(hash)
collection_objects_for_ids(Project, hash.keys)
end
private
attr_reader :context
def current_user
context.current_user
end
# When a feature is disabled or visible only for
# team members we should not allow team members
# see reference comments.
# Override this method on subclasses
# to check if user can read resource
def can_read_reference?(user, ref_project, node)
raise NotImplementedError
end
def lazy(&block)
Gitlab::Lazy.new(&block)
end
def collection_cache
RequestStore[:banzai_collection_cache] ||= Hash.new do |hash, key|
hash[key] = {}
end
end
end
end
end