X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/a17b176ea55dc3820ef1bde4b99cf33c628ffbbe..20fe67d073424f5c277fbd13557ffe5ae2b15fd9:/services/api/app/models/collection.rb

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index d25536ec43..f1e7b4f1e1 100644
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -1,54 +1,80 @@
+require 'arvados/keep'
+
 class Collection < ArvadosModel
+  extend DbCurrentTime
   include HasUuid
   include KindAndEtag
   include CommonApiTemplate
 
+  serialize :properties, Hash
+
+  before_validation :default_empty_manifest
+  before_validation :check_encoding
+  before_validation :check_manifest_validity
   before_validation :check_signatures
-  before_validation :strip_manifest_text
-  before_validation :set_portable_data_hash
-  validate :ensure_hash_matches_manifest_text
+  before_validation :strip_signatures_and_update_replication_confirmed
+  validate :ensure_pdh_matches_manifest_text
+  before_save :set_file_names
+
+  # 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
+    t.add :signed_manifest_text, as: :manifest_text
+    t.add :replication_desired
+    t.add :replication_confirmed
+    t.add :replication_confirmed_at
   end
 
   def self.attributes_required_columns
-    super.merge({ "data_size" => ["manifest_text"],
-                  "files" => ["manifest_text"],
-                })
+    super.merge(
+                # If we don't list manifest_text explicitly, the
+                # params[:select] code gets confused by the way we
+                # expose signed_manifest_text as manifest_text in the
+                # API response, and never let clients select the
+                # manifest_text column.
+                'manifest_text' => ['manifest_text'],
+                )
   end
 
+  FILE_TOKEN = /^[[:digit:]]+:[[:digit:]]+:/
   def check_signatures
     return false if self.manifest_text.nil?
 
     return true if current_user.andand.is_admin
 
+    # Provided the manifest_text hasn't changed materially since an
+    # earlier validation, it's safe to pass this validation on
+    # subsequent passes without checking any signatures. This is
+    # important because the signatures have probably been stripped off
+    # by the time we get to a second validation pass!
+    return true if @signatures_checked and @signatures_checked == computed_pdh
+
     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,
+        now: db_current_time.to_i,
       }
-      self.manifest_text.lines.each do |entry|
-        entry.split[1..-1].each do |tok|
-          if /^[[:digit:]]+:[[:digit:]]+:/.match tok
+      self.manifest_text.each_line do |entry|
+        entry.split.each do |tok|
+          if tok == '.' or tok.starts_with? './'
+            # Stream name token.
+          elsif tok =~ FILE_TOKEN
             # 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 Locator.parse(tok).andand.signature
+          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
@@ -64,145 +90,177 @@ class Collection < ArvadosModel
         end
       end
     end
-    true
+    @signatures_checked = computed_pdh
   end
 
-  def strip_manifest_text
+  def strip_signatures_and_update_replication_confirmed
     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
+      in_old_manifest = {}
+      if not self.replication_confirmed.nil?
+        self.class.each_manifest_locator(manifest_text_was) do |match|
+          in_old_manifest[match[1]] = true
+        end
       end
-    end
-    true
-  end
 
-  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 = 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}"
+      stripped_manifest = self.class.munge_manifest_locators(manifest_text) do |match|
+        if not self.replication_confirmed.nil? and not in_old_manifest[match[1]]
+          # If the new manifest_text contains locators whose hashes
+          # weren't in the old manifest_text, storage replication is no
+          # longer confirmed.
+          self.replication_confirmed_at = nil
+          self.replication_confirmed = nil
         end
-      rescue ArgumentError => e
-        errors.add(:portable_data_hash, "#{e}")
-        return false
+
+        # Return the locator with all permission signatures removed,
+        # but otherwise intact.
+        match[0].gsub(/\+A[^+]*/, '')
       end
+
+      if @computed_pdh_for_manifest_text == manifest_text
+        # If the cached PDH was valid before stripping, it is still
+        # valid after stripping.
+        @computed_pdh_for_manifest_text = stripped_manifest.dup
+      end
+
+      self[:manifest_text] = stripped_manifest
     end
     true
   end
 
-  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
+  def ensure_pdh_matches_manifest_text
+    if not manifest_text_changed? and not portable_data_hash_changed?
+      true
+    elsif portable_data_hash.nil? or not portable_data_hash_changed?
+      self.portable_data_hash = computed_pdh
+    elsif portable_data_hash !~ Keep::Locator::LOCATOR_REGEXP
+      errors.add(:portable_data_hash, "is not a valid locator")
+      false
+    elsif portable_data_hash[0..31] != computed_pdh[0..31]
+      errors.add(:portable_data_hash,
+                 "does not match computed hash #{computed_pdh}")
+      false
+    else
+      # Ignore the client-provided size part: always store
+      # computed_pdh in the database.
+      self.portable_data_hash = computed_pdh
+    end
+  end
+
+  def set_file_names
+    if self.manifest_text_changed?
+      self.file_names = manifest_files
     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 manifest_files
+    names = ''
+    if self.manifest_text
+      self.manifest_text.scan(/ \d+:\d+:(\S+)/) do |name|
+        names << name.first.gsub('\040',' ') + "\n"
+        break if names.length > 2**12
+      end
+    end
+
+    if self.manifest_text and names.length < 2**12
+      self.manifest_text.scan(/^\.\/(\S+)/m) do |stream_name|
+        names << stream_name.first.gsub('\040',' ') + "\n"
+        break if names.length > 2**12
       end
     end
+
+    names[0,2**12]
   end
 
-  def data_size
-    inspect_manifest_text if @data_size.nil? or manifest_text_changed?
-    @data_size
+  def default_empty_manifest
+    self.manifest_text ||= ''
   end
 
-  def files
-    inspect_manifest_text if @files.nil? or manifest_text_changed?
-    @files
+  def check_encoding
+    if manifest_text.encoding.name == 'UTF-8' and manifest_text.valid_encoding?
+      true
+    else
+      begin
+        # If Ruby thinks the encoding is something else, like 7-bit
+        # ASCII, but its stored bytes are equal to the (valid) UTF-8
+        # encoding of the same string, we declare it to be a UTF-8
+        # string.
+        utf8 = manifest_text
+        utf8.force_encoding Encoding::UTF_8
+        if utf8.valid_encoding? and utf8 == manifest_text.encode(Encoding::UTF_8)
+          manifest_text = utf8
+          return true
+        end
+      rescue
+      end
+      errors.add :manifest_text, "must use UTF-8 encoding"
+      false
+    end
   end
 
-  def inspect_manifest_text
-    if !manifest_text
-      @data_size = false
-      @files = []
-      return
+  def check_manifest_validity
+    begin
+      Keep::Manifest.validate! manifest_text
+      true
+    rescue ArgumentError => e
+      errors.add :manifest_text, e.message
+      false
     end
+  end
 
-    @data_size = 0
-    tmp = {}
+  def signed_manifest_text
+    if has_attribute? :manifest_text
+      token = current_api_client_authorization.andand.api_token
+      @signed_manifest_text = self.class.sign_manifest manifest_text, token
+    end
+  end
 
-    manifest_text.split("\n").each do |stream|
-      toks = stream.split(" ")
+  def self.sign_manifest manifest, token
+    signing_opts = {
+      api_token: token,
+      expire: db_current_time.to_i + Rails.configuration.blob_signature_ttl,
+    }
+    m = munge_manifest_locators(manifest) do |match|
+      Blob.sign_locator(match[0], signing_opts)
+    end
+    return m
+  end
 
-      stream = toks[0].gsub /\\(\\|[0-7]{3})/ do |escape_sequence|
-        case $1
-        when '\\' '\\'
-        else $1.to_i(8).chr
-        end
-      end
+  def self.munge_manifest_locators manifest
+    # Given a manifest text and a block, yield the regexp MatchData
+    # for each locator. Return a new manifest in which each locator
+    # has been replaced by the block's return value.
+    return nil if !manifest
+    return '' if manifest == ''
 
-      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
+    new_lines = []
+    manifest.each_line do |line|
+      line.rstrip!
+      new_words = []
+      line.split(' ').each do |word|
+        if new_words.empty?
+          new_words << word
+        elsif match = Keep::Locator::LOCATOR_REGEXP.match(word)
+          new_words << yield(match)
         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
+          new_words << word
         end
       end
+      new_lines << new_words.join(' ')
     end
-
-    @files = []
-    tmp.each do |k, v|
-      re = k.match(/^(.+)\/(.+)/)
-      @files << [re[1], re[2], v]
-    end
+    new_lines.join("\n") + "\n"
   end
 
-  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 = Locator.parse(word.strip)
-        " " + yield(loc)
-      else
-        " " + word
+  def self.each_manifest_locator manifest
+    # Given a manifest text and a block, yield the regexp match object
+    # for each locator.
+    manifest.each_line do |line|
+      # line will have a trailing newline, but the last token is never
+      # a locator, so it's harmless here.
+      line.split(' ').each do |word|
+        if match = Keep::Locator::LOCATOR_REGEXP.match(word)
+          yield(match)
+        end
       end
     end
   end
@@ -234,12 +292,17 @@ 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(portable_data_hash: loc.to_s).limit(1).first
-      if coll_match and (coll_match.files.size == 1) and
-          (coll_match.files[0][1] =~ /^[0-9A-Fa-f]{64}\.tar$/)
-        return [coll_match]
+      if coll_match
+        # Check if the Collection contains exactly one file whose name
+        # looks like a saved Docker image.
+        manifest = Keep::Manifest.new(coll_match.manifest_text)
+        if manifest.exact_file_count?(1) and
+            (manifest.files[0][1] =~ /^[0-9A-Fa-f]{64}\.tar$/)
+          return [coll_match]
+        end
       end
     end
 
@@ -274,4 +337,48 @@ class Collection < ArvadosModel
   def self.for_latest_docker_image(search_term, search_tag=nil, readers=nil)
     find_all_for_docker_image(search_term, search_tag, readers).first
   end
+
+  def self.searchable_columns operator
+    super - ["manifest_text"]
+  end
+
+  def self.full_text_searchable_columns
+    super - ["manifest_text"]
+  end
+
+  protected
+  def portable_manifest_text
+    self.class.munge_manifest_locators(manifest_text) do |match|
+      if match[2] # size
+        match[1] + match[2]
+      else
+        match[1]
+      end
+    end
+  end
+
+  def compute_pdh
+    portable_manifest = portable_manifest_text
+    (Digest::MD5.hexdigest(portable_manifest) +
+     '+' +
+     portable_manifest.bytesize.to_s)
+  end
+
+  def computed_pdh
+    if @computed_pdh_for_manifest_text == manifest_text
+      return @computed_pdh
+    end
+    @computed_pdh = compute_pdh
+    @computed_pdh_for_manifest_text = manifest_text.dup
+    @computed_pdh
+  end
+
+  def ensure_permission_to_save
+    if (not current_user.andand.is_admin and
+        (replication_confirmed_at_changed? or replication_confirmed_changed?) and
+        not (replication_confirmed_at.nil? and replication_confirmed.nil?))
+      raise ArvadosModel::PermissionDeniedError.new("replication_confirmed and replication_confirmed_at attributes cannot be changed, except by setting both to nil")
+    end
+    super
+  end
 end