services/api/app/controllers/arvados/v1/collections_controller.rb

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: AGPL-3.0
   4
   5 require "arvados/keep"
   6 require "trashable"
   7
   8 class Arvados::V1::CollectionsController < ApplicationController
   9   include DbCurrentTime
  10   include TrashableController
  11
  12   def self._index_requires_parameters
  13     (super rescue {}).
  14       merge({
  15         include_trash: {
  16           type: 'boolean', required: false, description: "Include collections whose is_trashed attribute is true."
  17         },
  18         include_old_versions: {
  19           type: 'boolean', required: false, description: "Include past collection versions."
  20         },
  21       })
  22   end
  23
  24   def self._show_requires_parameters
  25     (super rescue {}).
  26       merge({
  27         include_trash: {
  28           type: 'boolean', required: false, description: "Show collection even if its is_trashed attribute is true."
  29         },
  30         include_old_versions: {
  31           type: 'boolean', required: false, description: "Include past collection versions."
  32         },
  33       })
  34   end
  35
  36   def create
  37     if resource_attrs[:uuid] and (loc = Keep::Locator.parse(resource_attrs[:uuid]))
  38       resource_attrs[:portable_data_hash] = loc.to_s
  39       resource_attrs.delete :uuid
  40     end
  41     resource_attrs.delete :version
  42     resource_attrs.delete :current_version_uuid
  43     super
  44   end
  45
  46   def find_objects_for_index
  47     opts = {}
  48     if params[:include_trash] || ['destroy', 'trash', 'untrash'].include?(action_name)
  49       opts.update({include_trash: true})
  50     end
  51     if params[:include_old_versions] || @include_old_versions
  52       opts.update({include_old_versions: true})
  53     end
  54     @objects = Collection.readable_by(*@read_users, opts) if !opts.empty?
  55     super
  56   end
  57
  58   def find_object_by_uuid
  59     @include_old_versions = true
  60
  61     if loc = Keep::Locator.parse(params[:id])
  62       loc.strip_hints!
  63
  64       # It matters which Collection object we pick because we use it to get signed_manifest_text,
  65       # the value of which is affected by the value of trash_at.
  66       #
  67       # From postgres doc: "By default, null values sort as if larger than any non-null
  68       # value; that is, NULLS FIRST is the default for DESC order, and
  69       # NULLS LAST otherwise."
  70       #
  71       # "trash_at desc" sorts null first, then latest to earliest, so
  72       # it will select the Collection object with the longest
  73       # available lifetime.
  74
  75       if c = Collection.readable_by(*@read_users).where({ portable_data_hash: loc.to_s }).order("trash_at desc").limit(1).first
  76         @object = {
  77           uuid: c.portable_data_hash,
  78           portable_data_hash: c.portable_data_hash,
  79           manifest_text: c.signed_manifest_text,
  80         }
  81       end
  82       true
  83     else
  84       super
  85     end
  86   end
  87
  88   def show
  89     if @object.is_a? Collection
  90       # Omit unsigned_manifest_text
  91       @select ||= model_class.selectable_attributes - ["unsigned_manifest_text"]
  92       super
  93     else
  94       send_json @object
  95     end
  96   end
  97
  98
  99   def find_collections(visited, sp, &b)
 100     case sp
 101     when ArvadosModel
 102       sp.class.columns.each do |c|
 103         find_collections(visited, sp[c.name.to_sym], &b) if c.name != "log"
 104       end
 105     when Hash
 106       sp.each do |k, v|
 107         find_collections(visited, v, &b)
 108       end
 109     when Array
 110       sp.each do |v|
 111         find_collections(visited, v, &b)
 112       end
 113     when String
 114       if m = /[a-f0-9]{32}\+\d+/.match(sp)
 115         yield m[0], nil
 116       elsif m = Collection.uuid_regex.match(sp)
 117         yield nil, m[0]
 118       end
 119     end
 120   end
 121
 122   def search_edges(visited, uuid, direction)
 123     if uuid.nil? or uuid.empty? or visited[uuid]
 124       return
 125     end
 126
 127     if loc = Keep::Locator.parse(uuid)
 128       loc.strip_hints!
 129       return if visited[loc.to_s]
 130     end
 131
 132     logger.debug "visiting #{uuid}"
 133
 134     if loc
 135       # uuid is a portable_data_hash
 136       collections = Collection.readable_by(*@read_users).where(portable_data_hash: loc.to_s)
 137       c = collections.limit(2).all
 138       if c.size == 1
 139         visited[loc.to_s] = c[0]
 140       elsif c.size > 1
 141         name = collections.limit(1).where("name <> ''").first
 142         if name
 143           visited[loc.to_s] = {
 144             portable_data_hash: c[0].portable_data_hash,
 145             name: "#{name.name} + #{collections.count-1} more"
 146           }
 147         else
 148           visited[loc.to_s] = {
 149             portable_data_hash: c[0].portable_data_hash,
 150             name: loc.to_s
 151           }
 152         end
 153       end
 154
 155       if direction == :search_up
 156         # Search upstream for jobs where this locator is the output of some job
 157         if !Rails.configuration.API.DisabledAPIs.include?("jobs.list")
 158           Job.readable_by(*@read_users).where(output: loc.to_s).each do |job|
 159             search_edges(visited, job.uuid, :search_up)
 160           end
 161
 162           Job.readable_by(*@read_users).where(log: loc.to_s).each do |job|
 163             search_edges(visited, job.uuid, :search_up)
 164           end
 165         end
 166
 167         Container.readable_by(*@read_users).where(output: loc.to_s).each do |c|
 168           search_edges(visited, c.uuid, :search_up)
 169         end
 170
 171         Container.readable_by(*@read_users).where(log: loc.to_s).each do |c|
 172           search_edges(visited, c.uuid, :search_up)
 173         end
 174       elsif direction == :search_down
 175         if loc.to_s == "d41d8cd98f00b204e9800998ecf8427e+0"
 176           # Special case, don't follow the empty collection.
 177           return
 178         end
 179
 180         # Search downstream for jobs where this locator is in script_parameters
 181         if !Rails.configuration.API.DisabledAPIs.include?("jobs.list")
 182           Job.readable_by(*@read_users).where(["jobs.script_parameters like ?", "%#{loc.to_s}%"]).each do |job|
 183             search_edges(visited, job.uuid, :search_down)
 184           end
 185
 186           Job.readable_by(*@read_users).where(["jobs.docker_image_locator = ?", "#{loc.to_s}"]).each do |job|
 187             search_edges(visited, job.uuid, :search_down)
 188           end
 189         end
 190
 191         Container.readable_by(*@read_users).where(["mounts like ?", "%#{loc.to_s}%"]).each do |c|
 192           search_edges(visited, c.uuid, :search_down)
 193         end
 194
 195         Container.readable_by(*@read_users).where(["container_image = '#{loc.to_s}'"]).each do |c|
 196           search_edges(visited, c.uuid, :search_down)
 197         end
 198
 199       end
 200     else
 201       # uuid is a regular Arvados UUID
 202       rsc = ArvadosModel::resource_class_for_uuid uuid
 203       if rsc == Job
 204         Job.readable_by(*@read_users).where(uuid: uuid).each do |job|
 205           visited[uuid] = job.as_api_response
 206           if direction == :search_up
 207             # Follow upstream collections referenced in the script parameters
 208             find_collections(visited, job) do |hash, col_uuid|
 209               search_edges(visited, hash, :search_up) if hash
 210               search_edges(visited, col_uuid, :search_up) if col_uuid
 211             end
 212           elsif direction == :search_down
 213             # Follow downstream job output
 214             search_edges(visited, job.output, direction)
 215           end
 216         end
 217       elsif rsc == Container
 218         Container.readable_by(*@read_users).where(uuid: uuid).each do |c|
 219           visited[uuid] = c.as_api_response
 220           if direction == :search_up
 221             # Follow upstream collections referenced in the script parameters
 222             find_collections(visited, c) do |hash, col_uuid|
 223               search_edges(visited, hash, :search_up) if hash
 224               search_edges(visited, col_uuid, :search_up) if col_uuid
 225             end
 226           elsif direction == :search_down
 227             # Follow downstream job output
 228             search_edges(visited, c.output, direction)
 229           end
 230         end
 231       elsif rsc == Collection
 232         if c = Collection.readable_by(*@read_users).where(uuid: uuid).limit(1).first
 233           search_edges(visited, c.portable_data_hash, direction)
 234           visited[c.portable_data_hash] = c.as_api_response
 235         end
 236       elsif rsc != nil
 237         rsc.where(uuid: uuid).each do |r|
 238           visited[uuid] = r.as_api_response
 239         end
 240       end
 241     end
 242
 243     if direction == :search_up
 244       # Search for provenance links pointing to the current uuid
 245       Link.readable_by(*@read_users).
 246         where(head_uuid: uuid, link_class: "provenance").
 247         each do |link|
 248         visited[link.uuid] = link.as_api_response
 249         search_edges(visited, link.tail_uuid, direction)
 250       end
 251     elsif direction == :search_down
 252       # Search for provenance links emanating from the current uuid
 253       Link.readable_by(current_user).
 254         where(tail_uuid: uuid, link_class: "provenance").
 255         each do |link|
 256         visited[link.uuid] = link.as_api_response
 257         search_edges(visited, link.head_uuid, direction)
 258       end
 259     end
 260   end
 261
 262   def provenance
 263     visited = {}
 264     search_edges(visited, @object[:portable_data_hash], :search_up)
 265     search_edges(visited, @object[:uuid], :search_up)
 266     send_json visited
 267   end
 268
 269   def used_by
 270     visited = {}
 271     search_edges(visited, @object[:uuid], :search_down)
 272     search_edges(visited, @object[:portable_data_hash], :search_down)
 273     send_json visited
 274   end
 275
 276   protected
 277
 278   def load_limit_offset_order_params *args
 279     super
 280     if action_name == 'index'
 281       # Omit manifest_text and unsigned_manifest_text from index results unless expressly selected.
 282       @select ||= model_class.selectable_attributes - ["manifest_text", "unsigned_manifest_text"]
 283     end
 284   end
 285 end