gitlab-org--gitlab-foss/app/helpers/search_helper.rb
Jan Provaznik 090ca9c33e Use limit for search count queries
Search query is especially slow if a user searches a generic string
which matches many records, in such case search can take tens of
seconds or time out. To speed up the search query, we search only for
first 1000 records, if there is >1000 matching records we just display
"1000+" instead of precise total count supposing that with such amount
the exact count is not so important for the user.

Because for issues even limited search was not fast enough, 2-phase
approach is used for issues: first we use simpler/faster query to get
all public issues, if this exceeds the limit, we just return the limit.
If the amount of matching results is lower than limit, we re-run more
complex search query (which includes also confidential issues).
Re-running the complex query should be fast enough in such case because the
amount of matching issues is lower than limit.

Because exact total_count is now limited, this patch also switches to
to "prev/next" pagination.

Related #40540
2018-01-23 22:33:42 +01:00

177 lines
6.1 KiB
Ruby

module SearchHelper
def search_autocomplete_opts(term)
return unless current_user
resources_results = [
groups_autocomplete(term),
projects_autocomplete(term)
].flatten
search_pattern = Regexp.new(Regexp.escape(term), "i")
generic_results = project_autocomplete + default_autocomplete + help_autocomplete
generic_results.concat(default_autocomplete_admin) if current_user.admin?
generic_results.select! { |result| result[:label] =~ search_pattern }
[
resources_results,
generic_results
].flatten.uniq do |item|
item[:label]
end
end
def search_entries_info(collection, scope, term)
return unless collection.count > 0
from = collection.offset_value + 1
to = collection.offset_value + collection.length
count = collection.total_count
"Showing #{from} - #{to} of #{count} #{scope.humanize(capitalize: false)} for \"#{term}\""
end
def parse_search_result(result)
Gitlab::ProjectSearchResults.parse_search_result(result)
end
private
# Autocomplete results for various settings pages
def default_autocomplete
[
{ category: "Settings", label: "User settings", url: profile_path },
{ category: "Settings", label: "SSH Keys", url: profile_keys_path },
{ category: "Settings", label: "Dashboard", url: root_path }
]
end
# Autocomplete results for settings pages, for admins
def default_autocomplete_admin
[
{ category: "Settings", label: "Admin Section", url: admin_root_path }
]
end
# Autocomplete results for internal help pages
def help_autocomplete
[
{ category: "Help", label: "API Help", url: help_page_path("api/README") },
{ category: "Help", label: "Markdown Help", url: help_page_path("user/markdown") },
{ category: "Help", label: "Permissions Help", url: help_page_path("user/permissions") },
{ category: "Help", label: "Public Access Help", url: help_page_path("public_access/public_access") },
{ category: "Help", label: "Rake Tasks Help", url: help_page_path("raketasks/README") },
{ category: "Help", label: "SSH Keys Help", url: help_page_path("ssh/README") },
{ category: "Help", label: "System Hooks Help", url: help_page_path("system_hooks/system_hooks") },
{ category: "Help", label: "Webhooks Help", url: help_page_path("user/project/integrations/webhooks") },
{ category: "Help", label: "Workflow Help", url: help_page_path("workflow/README") }
]
end
# Autocomplete results for the current project, if it's defined
def project_autocomplete
if @project && @project.repository.exists? && @project.repository.root_ref
ref = @ref || @project.repository.root_ref
[
{ category: "Current Project", label: "Files", url: project_tree_path(@project, ref) },
{ category: "Current Project", label: "Commits", url: project_commits_path(@project, ref) },
{ category: "Current Project", label: "Network", url: project_network_path(@project, ref) },
{ category: "Current Project", label: "Graph", url: project_graph_path(@project, ref) },
{ category: "Current Project", label: "Issues", url: project_issues_path(@project) },
{ category: "Current Project", label: "Merge Requests", url: project_merge_requests_path(@project) },
{ category: "Current Project", label: "Milestones", url: project_milestones_path(@project) },
{ category: "Current Project", label: "Snippets", url: project_snippets_path(@project) },
{ category: "Current Project", label: "Members", url: project_project_members_path(@project) },
{ category: "Current Project", label: "Wiki", url: project_wikis_path(@project) }
]
else
[]
end
end
# Autocomplete results for the current user's groups
def groups_autocomplete(term, limit = 5)
current_user.authorized_groups.order_id_desc.search(term).limit(limit).map do |group|
{
category: "Groups",
id: group.id,
label: "#{search_result_sanitize(group.full_name)}",
url: group_path(group)
}
end
end
# Autocomplete results for the current user's projects
def projects_autocomplete(term, limit = 5)
current_user.authorized_projects.order_id_desc.search_by_title(term)
.sorted_by_stars.non_archived.limit(limit).map do |p|
{
category: "Projects",
id: p.id,
value: "#{search_result_sanitize(p.name)}",
label: "#{search_result_sanitize(p.name_with_namespace)}",
url: project_path(p)
}
end
end
def search_result_sanitize(str)
Sanitize.clean(str)
end
def search_filter_path(options = {})
exist_opts = {
search: params[:search],
project_id: params[:project_id],
group_id: params[:group_id],
scope: params[:scope],
repository_ref: params[:repository_ref]
}
options = exist_opts.merge(options)
search_path(options)
end
def search_filter_input_options(type)
opts =
{
id: "filtered-search-#{type}",
placeholder: 'Search or filter results...',
data: {
'username-params' => UserSerializer.new.represent(@users)
},
autocomplete: 'off'
}
if @project.present?
opts[:data]['project-id'] = @project.id
opts[:data]['base-endpoint'] = project_path(@project)
else
# Group context
opts[:data]['group-id'] = @group.id
opts[:data]['base-endpoint'] = group_canonical_path(@group)
end
opts
end
# Sanitize a HTML field for search display. Most tags are stripped out and the
# maximum length is set to 200 characters.
def search_md_sanitize(object, field)
html = markdown_field(object, field)
html = Truncato.truncate(
html,
count_tags: false,
count_tail: false,
max_length: 200
)
# Truncato's filtered_tags and filtered_attributes are not quite the same
sanitize(html, tags: %w(a p ol ul li pre code))
end
def limited_count(count, limit = 1000)
count > limit ? "#{limit}+" : count
end
end