X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/5f3df47f1369ca700f2405fca4231055989765c5..39292137fab0ea9720d2abc18c001d0e2ff8ce11:/services/api/app/models/collection.rb

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index 91d1d1e108..248c8b9824 100644
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -8,15 +8,17 @@ class Collection < ArvadosModel
 
   serialize :properties, Hash
 
+  before_validation :default_empty_manifest
   before_validation :check_encoding
+  before_validation :check_manifest_validity
   before_validation :check_signatures
   before_validation :strip_signatures_and_update_replication_confirmed
-  before_validation :set_portable_data_hash
   validate :ensure_pdh_matches_manifest_text
   before_save :set_file_names
+  before_save :expires_at_not_in_past
 
   # Query only undeleted collections by default.
-  default_scope where("expires_at IS NULL or expires_at > CURRENT_TIMESTAMP")
+  default_scope where("expires_at IS NULL or expires_at > statement_timestamp()")
 
   api_accessible :user, extend: :common do |t|
     t.add :name
@@ -27,10 +29,9 @@ class Collection < ArvadosModel
     t.add :replication_desired
     t.add :replication_confirmed
     t.add :replication_confirmed_at
+    t.add :expires_at
   end
 
-  LOCATOR_REGEXP = /^([[:xdigit:]]{32})(\+([[:digit:]]+))?(\+([[:upper:]][[:alnum:]+@_-]*))?$/
-
   def self.attributes_required_columns
     super.merge(
                 # If we don't list manifest_text explicitly, the
@@ -38,7 +39,10 @@ class Collection < ArvadosModel
                 # expose signed_manifest_text as manifest_text in the
                 # API response, and never let clients select the
                 # manifest_text column.
-                'manifest_text' => ['manifest_text'],
+                #
+                # We need expires_at to determine the correct
+                # timestamp in signed_manifest_text.
+                'manifest_text' => ['manifest_text', 'expires_at'],
                 )
   end
 
@@ -64,9 +68,11 @@ class Collection < ArvadosModel
         api_token: api_token,
         now: db_current_time.to_i,
       }
-      self.manifest_text.lines.each do |entry|
-        entry.split[1..-1].each do |tok|
-          if tok =~ FILE_TOKEN
+      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
@@ -95,66 +101,53 @@ class Collection < ArvadosModel
   def strip_signatures_and_update_replication_confirmed
     if self.manifest_text_changed?
       in_old_manifest = {}
-      self.class.munge_manifest_locators!(manifest_text_was) do |match|
-        in_old_manifest[match[1]] = true
+      if not self.replication_confirmed.nil?
+        self.class.each_manifest_locator(manifest_text_was) do |match|
+          in_old_manifest[match[1]] = true
+        end
       end
 
-      cleared_replication_confirmed = false
-
-      # Remove any permission signatures from the manifest.
-      self[:manifest_text] = self.class.munge_manifest_locators!(self[:manifest_text]) do |match|
-        if not cleared_replication_confirmed  and not in_old_manifest[match[1]]
+      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
-          cleared_replication_confirmed = true
         end
-        self.class.locator_without_signature(match)
-      end
-    end
-    @computed_pdh_for_manifest_text = self[:manifest_text] if @computed_pdh_for_manifest_text
-    true
-  end
 
-  def self.locator_without_signature match
-    without_signature = match[1]
-    without_signature += match[2] if match[2]
-    if match[4]
-      hints = match[4].split('+').reject { |hint| hint.start_with?("A") }
-      without_signature += hints.join('+')
-    end
-    without_signature
-  end
+        # Return the locator with all permission signatures removed,
+        # but otherwise intact.
+        match[0].gsub(/\+A[^+]*/, '')
+      end
 
-  def set_portable_data_hash
-    if (portable_data_hash.nil? or
-        portable_data_hash == "" or
-        (manifest_text_changed? and !portable_data_hash_changed?))
-      self.portable_data_hash = computed_pdh
-    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}+#{portable_manifest_text.bytesize}"
-        end
-      rescue ArgumentError => e
-        errors.add(:portable_data_hash, "#{e}")
-        return false
+      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_pdh_matches_manifest_text
-    return true unless manifest_text_changed? or portable_data_hash_changed?
-    # No need verify it if :set_portable_data_hash just computed it!
-    expect_pdh = computed_pdh
-    if expect_pdh != portable_data_hash
+    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 #{expect_pdh}")
-      return false
+                 "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
 
@@ -184,6 +177,10 @@ class Collection < ArvadosModel
     names[0,2**12]
   end
 
+  def default_empty_manifest
+    self.manifest_text ||= ''
+  end
+
   def check_encoding
     if manifest_text.encoding.name == 'UTF-8' and manifest_text.valid_encoding?
       true
@@ -206,50 +203,76 @@ class Collection < ArvadosModel
     end
   end
 
+  def check_manifest_validity
+    begin
+      Keep::Manifest.validate! manifest_text
+      true
+    rescue ArgumentError => e
+      errors.add :manifest_text, e.message
+      false
+    end
+  end
+
   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
+      exp = [db_current_time.to_i + Rails.configuration.blob_signature_ttl,
+             expires_at].compact.map(&:to_i).min
+      @signed_manifest_text = self.class.sign_manifest manifest_text, token, exp
     end
   end
 
-  def self.sign_manifest manifest, token
+  def self.sign_manifest manifest, token, exp=nil
+    if exp.nil?
+      exp = db_current_time.to_i + Rails.configuration.blob_signature_ttl
+    end
     signing_opts = {
       api_token: token,
-      expire: db_current_time.to_i + Rails.configuration.blob_signature_ttl,
+      expire: exp,
     }
-    m = manifest.dup
-    m = munge_manifest_locators!(m) do |match|
-      Blob.sign_locator(locator_without_signature(match), signing_opts)
+    m = munge_manifest_locators(manifest) do |match|
+      Blob.sign_locator(match[0], signing_opts)
     end
     return m
   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.
+  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 == ''
+
     new_lines = []
-    lines = manifest.andand.split("\n")
-    lines.andand.each do |line|
-      words = line.split(' ')
+    manifest.each_line do |line|
+      line.rstrip!
       new_words = []
-      words.each do |word|
-        if match = LOCATOR_REGEXP.match(word.strip)
+      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
-          new_words << word.strip
+          new_words << word
         end
       end
       new_lines << new_words.join(' ')
     end
+    new_lines.join("\n") + "\n"
+  end
 
-    if !new_lines.empty?
-      ends_with_newline = manifest.end_with?("\n")
-      manifest = new_lines.join("\n")
-      manifest += "\n" if ends_with_newline
+  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
-
-    manifest
   end
 
   def self.normalize_uuid uuid
@@ -333,17 +356,21 @@ class Collection < ArvadosModel
     super - ["manifest_text"]
   end
 
+  def logged_attributes
+    attrs = attributes.dup
+    attrs.delete('manifest_text')
+    attrs
+  end
+
   protected
   def portable_manifest_text
-    portable_manifest = self[:manifest_text].dup
-    portable_manifest = self.class.munge_manifest_locators!(portable_manifest) do |match|
+    self.class.munge_manifest_locators(manifest_text) do |match|
       if match[2] # size
         match[1] + match[2]
       else
         match[1]
       end
     end
-    portable_manifest
   end
 
   def compute_pdh
@@ -370,4 +397,14 @@ class Collection < ArvadosModel
     end
     super
   end
+
+  # If expires_at is being changed to a time in the past, change it to
+  # now. This allows clients to say "expires {client-current-time}"
+  # without failing due to clock skew, while avoiding odd log entries
+  # like "expiry date changed to {1 year ago}".
+  def expires_at_not_in_past
+    if expires_at_changed? and expires_at
+      self.expires_at = [db_current_time, expires_at].max
+    end
+  end
 end