X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/ed1580a38c1aae0910cef83605622a41d927fba3..763af5413e6195fe004846abf86a7ae134168990:/services/api/app/models/collection.rb

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index ac845f50ad..5dd760feee 100644
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -1,141 +1,147 @@
+require 'arvados/keep'
+
 class Collection < ArvadosModel
   include HasUuid
   include KindAndEtag
   include CommonApiTemplate
 
+  before_validation :check_encoding
+  before_validation :check_signatures
+  before_validation :strip_manifest_text
+  before_validation :set_portable_data_hash
+  validate :ensure_hash_matches_manifest_text
+
+  # Query only undeleted collections by default.
+  default_scope where("expires_at IS NULL or expires_at > CURRENT_TIMESTAMP")
+
   api_accessible :user, extend: :common do |t|
-    t.add :data_size
-    t.add :files
+    t.add :name
+    t.add :description
+    t.add :properties
+    t.add :portable_data_hash
     t.add :manifest_text
   end
 
-  def self.attributes_required_columns
-    super.merge({ "data_size" => ["manifest_text"],
-                  "files" => ["manifest_text"],
-                })
+  def check_signatures
+    return false if self.manifest_text.nil?
+
+    return true if current_user.andand.is_admin
+
+    if self.manifest_text_changed?
+      # Check permissions on the collection manifest.
+      # If any signature cannot be verified, raise PermissionDeniedError
+      # which will return 403 Permission denied to the client.
+      api_token = current_api_client_authorization.andand.api_token
+      signing_opts = {
+        key: Rails.configuration.blob_signing_key,
+        api_token: api_token,
+        ttl: Rails.configuration.blob_signing_ttl,
+      }
+      self.manifest_text.lines.each do |entry|
+        entry.split[1..-1].each do |tok|
+          if /^[[:digit:]]+:[[:digit:]]+:/.match tok
+            # This is a filename token, not a blob locator. Note that we
+            # keep checking tokens after this, even though manifest
+            # format dictates that all subsequent tokens will also be
+            # filenames. Safety first!
+          elsif Blob.verify_signature tok, signing_opts
+            # OK.
+          elsif Keep::Locator.parse(tok).andand.signature
+            # Signature provided, but verify_signature did not like it.
+            logger.warn "Invalid signature on locator #{tok}"
+            raise ArvadosModel::PermissionDeniedError
+          elsif Rails.configuration.permit_create_collection_with_unsigned_manifest
+            # No signature provided, but we are running in insecure mode.
+            logger.debug "Missing signature on locator #{tok} ignored"
+          elsif Blob.new(tok).empty?
+            # No signature provided -- but no data to protect, either.
+          else
+            logger.warn "Missing signature on locator #{tok}"
+            raise ArvadosModel::PermissionDeniedError
+          end
+        end
+      end
+    end
+    true
   end
 
-  def redundancy_status
-    if redundancy_confirmed_as.nil?
-      'unconfirmed'
-    elsif redundancy_confirmed_as < redundancy
-      'degraded'
-    else
-      if redundancy_confirmed_at.nil?
-        'unconfirmed'
-      elsif Time.now - redundancy_confirmed_at < 7.days
-        'OK'
-      else
-        'stale'
+  def strip_manifest_text
+    if self.manifest_text_changed?
+      # Remove any permission signatures from the manifest.
+      Collection.munge_manifest_locators(self[:manifest_text]) do |loc|
+        loc.without_signature.to_s
       end
     end
+    true
   end
 
-  def assign_uuid
-    if not self.manifest_text
-      errors.add :manifest_text, 'not supplied'
-      return false
-    end
-    expect_uuid = Digest::MD5.hexdigest(self.manifest_text)
-    if self.uuid
-      self.uuid.gsub! /\+.*/, ''
-      if self.uuid != expect_uuid
-        errors.add :uuid, 'must match checksum of manifest_text'
+  def set_portable_data_hash
+    if (self.portable_data_hash.nil? or (self.portable_data_hash == "") or (manifest_text_changed? and !portable_data_hash_changed?))
+      self.portable_data_hash = "#{Digest::MD5.hexdigest(manifest_text)}+#{manifest_text.length}"
+    elsif portable_data_hash_changed?
+      begin
+        loc = Keep::Locator.parse!(self.portable_data_hash)
+        loc.strip_hints!
+        if loc.size
+          self.portable_data_hash = loc.to_s
+        else
+          self.portable_data_hash = "#{loc.hash}+#{self.manifest_text.length}"
+        end
+      rescue ArgumentError => e
+        errors.add(:portable_data_hash, "#{e}")
         return false
       end
-    else
-      self.uuid = expect_uuid
     end
-    self.uuid.gsub! /$/, '+' + self.manifest_text.length.to_s
     true
   end
 
-  # TODO (#3036/tom) replace above assign_uuid method with below assign_uuid and self.generate_uuid
-  # def assign_uuid
-  #   # Even admins cannot assign collection uuids.
-  #   self.uuid = self.class.generate_uuid
-  # end
-  # def self.generate_uuid
-  #   # The last 10 characters of a collection uuid are the last 10
-  #   # characters of the base-36 SHA256 digest of manifest_text.
-  #   [Server::Application.config.uuid_prefix,
-  #    self.uuid_prefix,
-  #    rand(2**256).to_s(36)[-5..-1] + Digest::SHA256.hexdigest(self.manifest_text).to_i(16).to_s(36)[-10..-1],
-  #   ].join '-'
-  # end
-
-  def data_size
-    inspect_manifest_text if @data_size.nil? or manifest_text_changed?
-    @data_size
-  end
-
-  def files
-    inspect_manifest_text if @files.nil? or manifest_text_changed?
-    @files
+  def ensure_hash_matches_manifest_text
+    if manifest_text_changed? or portable_data_hash_changed?
+      computed_hash = "#{Digest::MD5.hexdigest(manifest_text)}+#{manifest_text.length}"
+      unless computed_hash == portable_data_hash
+        logger.debug "(computed) '#{computed_hash}' != '#{portable_data_hash}' (provided)"
+        errors.add(:portable_data_hash, "does not match hash of manifest_text")
+        return false
+      end
+    end
+    true
   end
 
-  def inspect_manifest_text
-    if !manifest_text
-      @data_size = false
-      @files = []
-      return
+  def check_encoding
+    if manifest_text.encoding.name == 'UTF-8' and manifest_text.valid_encoding?
+      true
+    else
+      errors.add :manifest_text, "must use UTF-8 encoding"
+      false
     end
+  end
 
-    @data_size = 0
-    tmp = {}
-
-    manifest_text.split("\n").each do |stream|
-      toks = stream.split(" ")
-
-      stream = toks[0].gsub /\\(\\|[0-7]{3})/ do |escape_sequence|
-        case $1
-        when '\\' '\\'
-        else $1.to_i(8).chr
-        end
-      end
-
-      toks[1..-1].each do |tok|
-        if (re = tok.match /^[0-9a-f]{32}/)
-          blocksize = nil
-          tok.split('+')[1..-1].each do |hint|
-            if !blocksize and hint.match /^\d+$/
-              blocksize = hint.to_i
-            end
-            if (re = hint.match /^GS(\d+)$/)
-              blocksize = re[1].to_i
-            end
-          end
-          @data_size = false if !blocksize
-          @data_size += blocksize if @data_size
-        else
-          if (re = tok.match /^(\d+):(\d+):(\S+)$/)
-            filename = re[3].gsub /\\(\\|[0-7]{3})/ do |escape_sequence|
-              case $1
-              when '\\' '\\'
-              else $1.to_i(8).chr
-              end
-            end
-            fn = stream + '/' + filename
-            i = re[2].to_i
-            if tmp[fn]
-              tmp[fn] += i
-            else
-              tmp[fn] = i
-            end
-          end
-        end
+  def redundancy_status
+    if redundancy_confirmed_as.nil?
+      'unconfirmed'
+    elsif redundancy_confirmed_as < redundancy
+      'degraded'
+    else
+      if redundancy_confirmed_at.nil?
+        'unconfirmed'
+      elsif Time.now - redundancy_confirmed_at < 7.days
+        'OK'
+      else
+        'stale'
       end
     end
-
-    @files = []
-    tmp.each do |k, v|
-      re = k.match(/^(.+)\/(.+)/)
-      @files << [re[1], re[2], v]
-    end
   end
 
-  def self.uuid_like_pattern
-    "________________________________+%"
+  def self.munge_manifest_locators(manifest)
+    # Given a manifest text and a block, yield each locator,
+    # and replace it with whatever the block returns.
+    manifest.andand.gsub!(/ [[:xdigit:]]{32}(\+[[:digit:]]+)?(\+\S+)/) do |word|
+      if loc = Keep::Locator.parse(word.strip)
+        " " + yield(loc)
+      else
+        " " + word
+      end
+    end
   end
 
   def self.normalize_uuid uuid
@@ -154,7 +160,8 @@ class Collection < ArvadosModel
     [hash_part, size_part].compact.join '+'
   end
 
-  def self.uuids_for_docker_image(search_term, search_tag=nil, readers=nil)
+  # Return array of Collection objects
+  def self.find_all_for_docker_image(search_term, search_tag=nil, readers=nil)
     readers ||= [Thread.current[:user]]
     base_search = Link.
       readable_by(*readers).
@@ -164,15 +171,25 @@ class Collection < ArvadosModel
 
     # If the search term is a Collection locator that contains one file
     # that looks like a Docker image, return it.
-    if loc = Locator.parse(search_term)
+    if loc = Keep::Locator.parse(search_term)
       loc.strip_hints!
-      coll_match = readable_by(*readers).where(uuid: loc.to_s).first
-      if coll_match and (coll_match.files.size == 1) and
-          (coll_match.files[0][1] =~ /^[0-9A-Fa-f]{64}\.tar$/)
-        return [loc.to_s]
+      coll_match = readable_by(*readers).where(portable_data_hash: loc.to_s).limit(1).first
+      if coll_match
+        # Check if the Collection contains exactly one file whose name
+        # looks like a saved Docker image.
+        manifest = Keep::Manifest.new(coll_match.manifest_text)
+        if manifest.exact_file_count?(1) and
+            (manifest.files[0][1] =~ /^[0-9A-Fa-f]{64}\.tar$/)
+          return [coll_match]
+        end
       end
     end
 
+    if search_tag.nil? and (n = search_term.index(":"))
+      search_tag = search_term[n+1..-1]
+      search_term = search_term[0..n-1]
+    end
+
     # Find Collections with matching Docker image repository+tag pairs.
     matches = base_search.
       where(link_class: "docker_image_repo+tag",
@@ -181,7 +198,7 @@ class Collection < ArvadosModel
     # If that didn't work, find Collections with matching Docker image hashes.
     if matches.empty?
       matches = base_search.
-        where("link_class = ? and name LIKE ?",
+        where("link_class = ? and links.name LIKE ?",
               "docker_image_hash", "#{search_term}%")
     end
 
@@ -189,20 +206,14 @@ class Collection < ArvadosModel
     # so that anything with an image timestamp is considered more recent than
     # anything without; then we use the link's created_at as a tiebreaker.
     uuid_timestamps = {}
-    matches.find_each do |link|
-      uuid_timestamps[link.head_uuid] =
-        [(-link.properties["image_timestamp"].to_datetime.to_i rescue 0),
-         -link.created_at.to_i]
+    matches.all.map do |link|
+      uuid_timestamps[link.head_uuid] = [(-link.properties["image_timestamp"].to_datetime.to_i rescue 0),
+       -link.created_at.to_i]
     end
-    uuid_timestamps.keys.sort_by { |uuid| uuid_timestamps[uuid] }
+    Collection.where('uuid in (?)', uuid_timestamps.keys).sort_by { |c| uuid_timestamps[c.uuid] }
   end
 
   def self.for_latest_docker_image(search_term, search_tag=nil, readers=nil)
-    image_uuid = uuids_for_docker_image(search_term, search_tag, readers).first
-    if image_uuid.nil?
-      nil
-    else
-      find_by_uuid(image_uuid)
-    end
+    find_all_for_docker_image(search_term, search_tag, readers).first
   end
 end