debian-mirror-gitlab/lib/banzai/filter/abstract_reference_filter.rb

447 lines
13 KiB
Ruby
Raw Normal View History

2018-11-18 11:00:15 +05:30
# frozen_string_literal: true
2015-12-23 02:04:40 +05:30
module Banzai
module Filter
# Issues, Merge Requests, Snippets, Commits and Commit Ranges share
# similar functionality in reference filtering.
class AbstractReferenceFilter < ReferenceFilter
include CrossProjectReference
2019-09-04 21:01:54 +05:30
# REFERENCE_PLACEHOLDER is used for re-escaping HTML text except found
# reference (which we replace with placeholder during re-scaping). The
# random number helps ensure it's pretty close to unique. Since it's a
# transitory value (it never gets saved) we can initialize once, and it
# doesn't matter if it changes on a restart.
REFERENCE_PLACEHOLDER = "_reference_#{SecureRandom.hex(16)}_"
REFERENCE_PLACEHOLDER_PATTERN = %r{#{REFERENCE_PLACEHOLDER}(\d+)}.freeze
2015-12-23 02:04:40 +05:30
def self.object_class
# Implement in child class
# Example: MergeRequest
end
def self.object_name
2016-06-02 11:05:42 +05:30
@object_name ||= object_class.name.underscore
2015-12-23 02:04:40 +05:30
end
def self.object_sym
2016-06-02 11:05:42 +05:30
@object_sym ||= object_name.to_sym
2015-12-23 02:04:40 +05:30
end
# Public: Find references in text (like `!123` for merge requests)
#
# AnyReferenceFilter.references_in(text) do |match, id, project_ref, matches|
# object = find_object(project_ref, id)
# "<a href=...>#{object.to_reference}</a>"
# end
#
# text - String text to search.
#
# Yields the String match, the Integer referenced object ID, an optional String
# of the external project reference, and all of the matchdata.
#
# Returns a String replaced with the return of the block.
def self.references_in(text, pattern = object_class.reference_pattern)
text.gsub(pattern) do |match|
2020-03-13 15:44:24 +05:30
if ident = identifier($~)
yield match, ident, $~[:project], $~[:namespace], $~
2017-08-17 22:00:37 +05:30
else
match
end
2015-12-23 02:04:40 +05:30
end
end
2020-03-13 15:44:24 +05:30
def self.identifier(match_data)
symbol = symbol_from_match(match_data)
parse_symbol(symbol, match_data) if object_class.reference_valid?(symbol)
end
def identifier(match_data)
self.class.identifier(match_data)
end
def self.symbol_from_match(match)
key = object_sym
match[key] if match.names.include?(key.to_s)
end
# Transform a symbol extracted from the text to a meaningful value
# In most cases these will be integers, so we call #to_i by default
#
# This method has the contract that if a string `ref` refers to a
# record `record`, then `parse_symbol(ref) == record_identifier(record)`.
def self.parse_symbol(symbol, match_data)
symbol.to_i
end
# We assume that most classes are identifying records by ID.
#
# This method has the contract that if a string `ref` refers to a
# record `record`, then `class.parse_symbol(ref) == record_identifier(record)`.
def record_identifier(record)
record.id
end
def object_class
self.class.object_class
end
def object_sym
self.class.object_sym
end
def references_in(*args, &block)
self.class.references_in(*args, &block)
end
2015-12-23 02:04:40 +05:30
2017-09-10 17:25:29 +05:30
# Implement in child class
# Example: project.merge_requests.find
2018-05-09 12:01:36 +05:30
def find_object(parent_object, id)
2015-12-23 02:04:40 +05:30
end
2017-09-10 17:25:29 +05:30
# Override if the link reference pattern produces a different ID (global
# ID vs internal ID, for instance) to the regular reference pattern.
2018-05-09 12:01:36 +05:30
def find_object_from_link(parent_object, id)
find_object(parent_object, id)
2017-09-10 17:25:29 +05:30
end
2016-06-02 11:05:42 +05:30
2017-09-10 17:25:29 +05:30
# Implement in child class
# Example: project_merge_request_url
2018-05-09 12:01:36 +05:30
def url_for_object(object, parent_object)
2017-09-10 17:25:29 +05:30
end
2018-05-09 12:01:36 +05:30
def find_object_cached(parent_object, id)
cached_call(:banzai_find_object, id, path: [object_class, parent_object.id]) do
find_object(parent_object, id)
2016-06-02 11:05:42 +05:30
end
end
2018-05-09 12:01:36 +05:30
def find_object_from_link_cached(parent_object, id)
cached_call(:banzai_find_object_from_link, id, path: [object_class, parent_object.id]) do
find_object_from_link(parent_object, id)
2016-06-02 11:05:42 +05:30
end
end
2018-03-17 18:26:18 +05:30
def from_ref_cached(ref)
cached_call("banzai_#{parent_type}_refs".to_sym, ref) do
parent_from_ref(ref)
2017-09-10 17:25:29 +05:30
end
2015-12-23 02:04:40 +05:30
end
2018-05-09 12:01:36 +05:30
def url_for_object_cached(object, parent_object)
cached_call(:banzai_url_for_object, object, path: [object_class, parent_object.id]) do
url_for_object(object, parent_object)
2015-12-23 02:04:40 +05:30
end
2016-06-02 11:05:42 +05:30
end
2015-12-23 02:04:40 +05:30
2016-06-02 11:05:42 +05:30
def call
2018-03-17 18:26:18 +05:30
return doc unless project || group
2016-06-02 11:05:42 +05:30
ref_pattern = object_class.reference_pattern
link_pattern = object_class.link_reference_pattern
2018-11-08 19:23:39 +05:30
# Compile often used regexps only once outside of the loop
ref_pattern_anchor = /\A#{ref_pattern}\z/
link_pattern_start = /\A#{link_pattern}/
link_pattern_anchor = /\A#{link_pattern}\z/
2016-06-22 15:30:34 +05:30
nodes.each do |node|
2016-06-02 11:05:42 +05:30
if text_node?(node) && ref_pattern
replace_text_when_pattern_matches(node, ref_pattern) do |content|
object_link_filter(content, ref_pattern)
end
2015-12-23 02:04:40 +05:30
2016-06-02 11:05:42 +05:30
elsif element_node?(node)
2017-08-17 22:00:37 +05:30
yield_valid_link(node) do |link, inner_html|
2018-11-08 19:23:39 +05:30
if ref_pattern && link =~ ref_pattern_anchor
2016-06-02 11:05:42 +05:30
replace_link_node_with_href(node, link) do
2017-08-17 22:00:37 +05:30
object_link_filter(link, ref_pattern, link_content: inner_html)
2016-06-02 11:05:42 +05:30
end
next
end
next unless link_pattern
2018-11-08 19:23:39 +05:30
if link == inner_html && inner_html =~ link_pattern_start
2016-06-02 11:05:42 +05:30
replace_link_node_with_text(node, link) do
2017-09-10 17:25:29 +05:30
object_link_filter(inner_html, link_pattern, link_reference: true)
2016-06-02 11:05:42 +05:30
end
next
end
2018-11-08 19:23:39 +05:30
if link =~ link_pattern_anchor
2016-06-02 11:05:42 +05:30
replace_link_node_with_href(node, link) do
2017-09-10 17:25:29 +05:30
object_link_filter(link, link_pattern, link_content: inner_html, link_reference: true)
2016-06-02 11:05:42 +05:30
end
next
end
end
end
2015-12-23 02:04:40 +05:30
end
2016-06-02 11:05:42 +05:30
doc
2015-12-23 02:04:40 +05:30
end
# Replace references (like `!123` for merge requests) in text with links
# to the referenced object's details page.
#
# text - String text to replace references in.
# pattern - Reference pattern to match against.
2017-08-17 22:00:37 +05:30
# link_content - Original content of the link being replaced.
2017-09-10 17:25:29 +05:30
# link_reference - True if this was using the link reference pattern,
# false otherwise.
2015-12-23 02:04:40 +05:30
#
# Returns a String with references replaced with links. All links
# have `gfm` and `gfm-OBJECT_NAME` class names attached for styling.
2017-09-10 17:25:29 +05:30
def object_link_filter(text, pattern, link_content: nil, link_reference: false)
2017-08-17 22:00:37 +05:30
references_in(text, pattern) do |match, id, project_ref, namespace_ref, matches|
2018-03-17 18:26:18 +05:30
parent_path = if parent_type == :group
full_group_path(namespace_ref)
else
full_project_path(namespace_ref, project_ref)
end
2015-12-23 02:04:40 +05:30
2018-03-17 18:26:18 +05:30
parent = from_ref_cached(parent_path)
if parent
2017-09-10 17:25:29 +05:30
object =
if link_reference
2018-03-17 18:26:18 +05:30
find_object_from_link_cached(parent, id)
2017-09-10 17:25:29 +05:30
else
2018-03-17 18:26:18 +05:30
find_object_cached(parent, id)
2017-09-10 17:25:29 +05:30
end
end
if object
2018-05-09 12:01:36 +05:30
title = object_link_title(object, matches)
2015-12-23 02:04:40 +05:30
klass = reference_class(object_sym)
2019-07-07 11:18:12 +05:30
data_attributes = data_attributes_for(link_content || match, parent, object,
link_content: !!link_content,
link_reference: link_reference)
data = data_attribute(data_attributes)
2015-12-23 02:04:40 +05:30
2017-08-17 22:00:37 +05:30
url =
if matches.names.include?("url") && matches[:url]
matches[:url]
else
2018-03-17 18:26:18 +05:30
url_for_object_cached(object, parent)
2017-08-17 22:00:37 +05:30
end
2015-12-23 02:04:40 +05:30
2017-08-17 22:00:37 +05:30
content = link_content || object_link_text(object, matches)
2015-12-23 02:04:40 +05:30
2019-07-07 11:18:12 +05:30
link = %(<a href="#{url}" #{data}
title="#{escape_once(title)}"
class="#{klass}">#{content}</a>)
wrap_link(link, object)
2015-12-23 02:04:40 +05:30
else
match
end
end
end
2019-07-07 11:18:12 +05:30
def wrap_link(link, object)
link
end
2018-05-09 12:01:36 +05:30
def data_attributes_for(text, parent, object, link_content: false, link_reference: false)
object_parent_type = parent.is_a?(Group) ? :group : :project
2019-07-07 11:18:12 +05:30
{
2018-05-09 12:01:36 +05:30
original: text,
link: link_content,
link_reference: link_reference,
object_parent_type => parent.id,
object_sym => object.id
2019-07-07 11:18:12 +05:30
}
2016-08-24 12:49:21 +05:30
end
2015-12-23 02:04:40 +05:30
def object_link_text_extras(object, matches)
extras = []
if matches.names.include?("anchor") && matches[:anchor] && matches[:anchor] =~ /\A\#note_(\d+)\z/
extras << "comment #{$1}"
end
2018-05-09 12:01:36 +05:30
extension = matches[:extension] if matches.names.include?("extension")
extras << extension if extension
2015-12-23 02:04:40 +05:30
extras
end
2018-05-09 12:01:36 +05:30
def object_link_title(object, matches)
2016-09-29 09:46:39 +05:30
object.title
2015-12-23 02:04:40 +05:30
end
def object_link_text(object, matches)
2018-03-17 18:26:18 +05:30
parent = context[:project] || context[:group]
text = object.reference_link_text(parent)
extras = object_link_text_extras(object, matches)
text += " (#{extras.join(", ")})" if extras.any?
text
end
2016-06-02 11:05:42 +05:30
2016-06-22 15:30:34 +05:30
# Returns a Hash containing all object references (e.g. issue IDs) per the
# project they belong to.
2018-03-17 18:26:18 +05:30
def references_per_parent
@references_per ||= {}
@references_per[parent_type] ||= begin
2016-06-22 15:30:34 +05:30
refs = Hash.new { |hash, key| hash[key] = Set.new }
2020-03-13 15:44:24 +05:30
regex = [
object_class.link_reference_pattern,
object_class.reference_pattern
].compact.reduce { |a, b| Regexp.union(a, b) }
2016-06-22 15:30:34 +05:30
nodes.each do |node|
node.to_html.scan(regex) do
2018-03-17 18:26:18 +05:30
path = if parent_type == :project
full_project_path($~[:namespace], $~[:project])
else
full_group_path($~[:group])
end
2020-03-13 15:44:24 +05:30
if ident = identifier($~)
refs[path] << ident
end
2016-06-22 15:30:34 +05:30
end
end
refs
end
end
# Returns a Hash containing referenced projects grouped per their full
# path.
2018-03-17 18:26:18 +05:30
def parent_per_reference
@per_reference ||= {}
@per_reference[parent_type] ||= begin
2016-06-22 15:30:34 +05:30
refs = Set.new
2018-03-17 18:26:18 +05:30
references_per_parent.each do |ref, _|
refs << ref
2016-06-22 15:30:34 +05:30
end
2018-03-17 18:26:18 +05:30
find_for_paths(refs.to_a).index_by(&:full_path)
2016-06-22 15:30:34 +05:30
end
end
2018-03-17 18:26:18 +05:30
def relation_for_paths(paths)
klass = parent_type.to_s.camelize.constantize
result = klass.where_full_path_in(paths)
return result if parent_type == :group
result.includes(:namespace) if parent_type == :project
2016-06-22 15:30:34 +05:30
end
2016-11-03 12:29:30 +05:30
# Returns projects for the given paths.
2018-03-17 18:26:18 +05:30
def find_for_paths(paths)
2018-12-05 23:21:45 +05:30
if Gitlab::SafeRequestStore.active?
2018-03-17 18:26:18 +05:30
cache = refs_cache
2017-08-17 22:00:37 +05:30
to_query = paths - cache.keys
2016-11-03 12:29:30 +05:30
unless to_query.empty?
2018-03-17 18:26:18 +05:30
records = relation_for_paths(to_query)
2017-08-17 22:00:37 +05:30
found = []
2018-03-17 18:26:18 +05:30
records.each do |record|
ref = record.full_path
get_or_set_cache(cache, ref) { record }
2017-08-17 22:00:37 +05:30
found << ref
end
not_found = to_query - found
not_found.each do |ref|
get_or_set_cache(cache, ref) { nil }
2016-11-03 12:29:30 +05:30
end
end
2017-08-17 22:00:37 +05:30
cache.slice(*paths).values.compact
2016-11-03 12:29:30 +05:30
else
2018-03-17 18:26:18 +05:30
relation_for_paths(paths)
2016-11-03 12:29:30 +05:30
end
end
2018-03-17 18:26:18 +05:30
def current_parent_path
@current_parent_path ||= parent&.full_path
2016-06-22 15:30:34 +05:30
end
2017-08-17 22:00:37 +05:30
def current_project_namespace_path
2018-03-17 18:26:18 +05:30
@current_project_namespace_path ||= project&.namespace&.full_path
2017-08-17 22:00:37 +05:30
end
2019-10-12 21:52:04 +05:30
def records_per_parent
@_records_per_project ||= {}
@_records_per_project[object_class.to_s.underscore] ||= begin
hash = Hash.new { |h, k| h[k] = {} }
parent_per_reference.each do |path, parent|
record_ids = references_per_parent[path]
parent_records(parent, record_ids).each do |record|
hash[parent][record_identifier(record)] = record
end
end
hash
end
end
2016-06-02 11:05:42 +05:30
private
2017-08-17 22:00:37 +05:30
def full_project_path(namespace, project_ref)
2018-03-17 18:26:18 +05:30
return current_parent_path unless project_ref
2017-08-17 22:00:37 +05:30
namespace_ref = namespace || current_project_namespace_path
"#{namespace_ref}/#{project_ref}"
end
2018-03-17 18:26:18 +05:30
def refs_cache
2018-12-05 23:21:45 +05:30
Gitlab::SafeRequestStore["banzai_#{parent_type}_refs".to_sym] ||= {}
2016-06-02 11:05:42 +05:30
end
2018-03-17 18:26:18 +05:30
def parent_type
:project
2016-06-02 11:05:42 +05:30
end
2018-03-17 18:26:18 +05:30
def parent
parent_type == :project ? project : group
2016-06-02 11:05:42 +05:30
end
2018-05-09 12:01:36 +05:30
def full_group_path(group_ref)
return current_parent_path unless group_ref
group_ref
end
2019-09-04 21:01:54 +05:30
def unescape_html_entities(text)
CGI.unescapeHTML(text.to_s)
end
def escape_html_entities(text)
CGI.escapeHTML(text.to_s)
end
def escape_with_placeholders(text, placeholder_data)
escaped = escape_html_entities(text)
escaped.gsub(REFERENCE_PLACEHOLDER_PATTERN) do |match|
placeholder_data[$1.to_i]
end
end
2015-12-23 02:04:40 +05:30
end
end
end
2019-12-04 20:38:33 +05:30
Banzai::Filter::AbstractReferenceFilter.prepend_if_ee('EE::Banzai::Filter::AbstractReferenceFilter')