longleaf 0.1.0.pre.3 → 1.1.0

data/lib/longleaf/events/deregister_event.rb

@@ -0,0 +1,53 @@
+require 'longleaf/errors'
+require 'longleaf/events/event_names'
+require 'longleaf/events/event_status_tracking'
+require 'longleaf/services/metadata_serializer'
+
+module Longleaf
+  # Event to deregister a file from longleaf
+  class DeregisterEvent
+    include Longleaf::EventStatusTracking
+
+    # @param file_rec [FileRecord] file record
+    # @param app_manager [ApplicationConfigManager] the application configuration
+    # @param force [boolean] if true, then already deregistered files will be deregistered again
+    def initialize(file_rec:, app_manager:, force: false)
+      raise ArgumentError.new('Must provide a file_rec parameter') if file_rec.nil?
+      raise ArgumentError.new('Parameter file_rec must be a FileRecord') \
+          unless file_rec.is_a?(FileRecord)
+      raise ArgumentError.new('Must provide an ApplicationConfigManager') if app_manager.nil?
+      raise ArgumentError.new('Parameter app_manager must be an ApplicationConfigManager') \
+          unless app_manager.is_a?(ApplicationConfigManager)
+
+      @app_manager = app_manager
+      @file_rec = file_rec
+      @force = force
+    end
+
+    # Perform a deregistration event on the given file record
+    # @raise DeregistrationError if a file cannot be deregistered
+    def perform
+      begin
+        md_rec = @file_rec.metadata_record
+
+        # Only need to deregister a deregistered file if the force flag is provided
+        if md_rec.deregistered? && !@force
+          raise DeregistrationError.new("Unable to deregister '#{@file_rec.path}', it is already deregistered.")
+        end
+
+        md_rec.deregistered = Time.now.utc.iso8601(3)
+
+        # persist the metadata
+        @app_manager.md_manager.persist(@file_rec)
+
+        record_success(EventNames::DEREGISTER, @file_rec.path)
+      rescue DeregistrationError => err
+        record_failure(EventNames::DEREGISTER, @file_rec.path, err.message)
+      rescue InvalidStoragePathError => err
+        record_failure(EventNames::DEREGISTER, @file_rec.path, err.message)
+      end
+
+      return_status
+    end
+  end
+end

data/lib/longleaf/events/event_names.rb

@@ -0,0 +1,9 @@
+module Longleaf
+  # Constants for preservation event names
+  class EventNames
+    REGISTER = 'register'
+    PRESERVE = 'preserve'
+    CLEANUP = 'cleanup'
+    DEREGISTER = 'deregister'
+  end
+end

data/lib/longleaf/events/event_status_tracking.rb

@@ -0,0 +1,59 @@
+require 'longleaf/logging'
+
+module Longleaf
+  # Helper methods for tracking and recording the overall outcome of a preservation event.
+  module EventStatusTracking
+    include Longleaf::Logging
+
+    # Record a successful operation to the output and the overall status of this event.
+    # @param args [Array] arguments to pass to logger
+    def record_success(*args)
+      logger.success(*args)
+      track_success
+    end
+
+    # Update the status of this action with a success outcome.
+    def track_success
+      if @return_status.nil? || @return_status == 0
+        @return_status = 0
+      else
+        @return_status = 2
+      end
+    end
+
+    # Record a failed operation to the output and the overall status of this event.
+    # @param args [Array] arguments to pass to logger
+    def record_failure(*args)
+      logger.failure(*args)
+      track_failure
+    end
+
+    # Update the status of this action with a failure outcome.
+    def track_failure
+      if @return_status.nil? || @return_status == 1
+        @return_status = 1
+      else
+        @return_status = 2
+      end
+    end
+
+    # Update the status of this action with the provided outcome status number.
+    # @param status [Integer] outcome status
+    def track_status(status)
+      if status == 2
+        @return_status = 2
+      elsif status == 0
+        track_success
+      elsif status == 1
+        track_failure
+      end
+    end
+
+    # @return [Integer] the return status for this event, where 0 indicates success,
+    # 1 indicates failure, and 2 indicates partial failure
+    def return_status
+      @return_status = 0 if @return_status.nil?
+      @return_status
+    end
+  end
+end

data/lib/longleaf/events/preserve_event.rb

@@ -0,0 +1,82 @@
+require 'longleaf/services/service_manager'
+require 'longleaf/events/event_names'
+require 'longleaf/events/event_status_tracking'
+require 'longleaf/logging'
+
+module Longleaf
+  # Verify event for a single file
+  class PreserveEvent
+    include Longleaf::Logging
+    include Longleaf::EventStatusTracking
+
+    # @param file_rec [FileRecord] file record
+    # @param app_manager [ApplicationConfigManager] the application configuration
+    # @param force [boolean] if true, then services run regardless of whether they are flagged as needed
+    def initialize(file_rec:, app_manager:, force: false)
+      raise ArgumentError.new('Must provide a file_rec parameter') if file_rec.nil?
+      raise ArgumentError.new('Must provide an ApplicationConfigManager') if app_manager.nil?
+
+      @app_manager = app_manager
+      @file_rec = file_rec
+      @force = force
+    end
+
+    # Perform a preserve event on the given file, updating its metadata record if any services were executed.
+    def perform
+      storage_loc = @file_rec.storage_location
+      service_manager = @app_manager.service_manager
+      md_rec = @file_rec.metadata_record
+      f_path = @file_rec.path
+      phys_path = @file_rec.physical_path
+
+      logger.info("Performing preserve event on #{f_path}")
+
+      needs_persist = false
+      begin
+        if !File.exist?(phys_path)
+          # Need to persist metadata to avoid repeating processing of this file too soon.
+          needs_persist = true
+          record_failure(EventNames::PRESERVE, f_path, "File is registered but missing.")
+          return return_status
+        end
+
+        # get the list of services applicable to this location and event
+        service_manager.list_services(location: storage_loc.name, event: EventNames::PRESERVE).each do |service_name|
+          # Skip over this service if it does not need to be run, unless force flag active
+          unless @force || service_manager.service_needed?(service_name, md_rec)
+            logger.debug("Service #{service_name} not needed for file '#{@file_rec.path}', skipping")
+            next
+          end
+
+          begin
+            logger.info("Performing preserve service #{service_name} for #{@file_rec.path}")
+            needs_persist = true
+            # execute the service
+            service_manager.perform_service(service_name, @file_rec, EventNames::PRESERVE)
+
+            # record the outcome
+            @file_rec.metadata_record.update_service_as_performed(service_name)
+            record_success(EventNames::PRESERVE, f_path, nil, service_name)
+          rescue PreservationServiceError => e
+            @file_rec.metadata_record.update_service_as_failed(service_name)
+            record_failure(EventNames::PRESERVE, f_path, e.message, service_name)
+          rescue StorageLocationUnavailableError => e
+            raise e
+          rescue StandardError => e
+            @file_rec.metadata_record.update_service_as_failed(service_name)
+            record_failure(EventNames::PRESERVE, f_path, nil, service_name, error: e)
+            return return_status
+          end
+        end
+      ensure
+        # persist the metadata out to file if any services were executed
+        if needs_persist
+          # persist the metadata
+          @app_manager.md_manager.persist(@file_rec)
+        end
+      end
+
+      return_status
+    end
+  end
+end

data/lib/longleaf/events/register_event.rb

@@ -1,92 +1,100 @@
 require 'longleaf/errors'
+require 'longleaf/events/event_names'
+require 'longleaf/events/event_status_tracking'
 require 'longleaf/models/metadata_record'
 require 'longleaf/services/metadata_deserializer'
 require 'longleaf/services/metadata_serializer'
 require 'time'
 
-# Event to register a file with longleaf
 module Longleaf
+  # Event to register a file with longleaf
   class RegisterEvent
-    EVENT_NAME = 'register'
-    
+    include Longleaf::EventStatusTracking
+
     # @param file_rec [FileRecord] file record
     # @param app_manager [ApplicationConfigManager] the application configuration
     # @param force [boolean] if true, then already registered files will be re-registered
-    def initialize(file_rec:, app_manager:, force: false, checksums: nil)
+    # @param digest_provider [#get_digests] object which provides digests for files being registered
+    def initialize(file_rec:, app_manager:, force: false, digest_provider: nil)
       raise ArgumentError.new('Must provide a file_rec parameter') if file_rec.nil?
       raise ArgumentError.new('Parameter file_rec must be a FileRecord') \
           unless file_rec.is_a?(FileRecord)
       raise ArgumentError.new('Must provide an ApplicationConfigManager') if app_manager.nil?
       raise ArgumentError.new('Parameter app_manager must be an ApplicationConfigManager') \
           unless app_manager.is_a?(ApplicationConfigManager)
-      
+
       @app_manager = app_manager
       @file_rec = file_rec
       @force = force
-      @checksums = checksums
+      @digest_provider = digest_provider
     end
-    
+
     # Perform a registration event on the given file
-    # @raises RegistrationError if a file cannot be registered 
+    # @raise RegistrationError if a file cannot be registered
     def perform
-      metadata_exists = File.file?(@file_rec.metadata_path)
-      # If the file's metadata exists, only need to register it if the force flag is provided
-      if metadata_exists && !@force
-        raise RegistrationError.new("Unable to register '#{@file_rec.path}', it is already registered.")
-      end
-      
-      # create metadata record
-      md_rec = MetadataRecord.new(registered: Time.now.utc.iso8601)
-      @file_rec.metadata_record = md_rec
-      
-      # retain significant details from former record
-      if metadata_exists
-        retain_existing_properties
+      begin
+        # Only need to re-register file if the force flag is provided
+        if @file_rec.metadata_present? && !@force
+          raise RegistrationError.new("Unable to register '#{@file_rec.path}', it is already registered.")
+        end
+
+        # create metadata record
+        md_rec = MetadataRecord.new(registered: Time.now.utc.iso8601(3))
+        @file_rec.metadata_record = md_rec
+
+        # retain significant details from former record
+        if @file_rec.metadata_present?
+          retain_existing_properties
+        end
+
+        populate_file_properties
+
+        if !@digest_provider.nil?
+          checksums = @digest_provider.get_digests(@file_rec.path)
+          md_rec.checksums.merge!(checksums) unless checksums.nil?
+        end
+
+        # persist the metadata
+        @app_manager.md_manager.persist(@file_rec)
+
+        record_success(EventNames::REGISTER, @file_rec.path)
+      rescue RegistrationError => err
+        record_failure(EventNames::REGISTER, @file_rec.path, err.message)
+      rescue InvalidStoragePathError => err
+        record_failure(EventNames::REGISTER, @file_rec.path, err.message)
       end
-      
-      populate_file_properties
-      
-      md_rec.checksums.merge!(@checksums) unless @checksums.nil?
-      
-      populate_services
-      
-      # persist the metadata out to file
-      MetadataSerializer::write(metadata: md_rec, file_path: @file_rec.metadata_path)
+
+      return_status
     end
-    
+
     private
     def populate_file_properties
       md_rec = @file_rec.metadata_record
-      
+      physical_path = @file_rec.physical_path
+
       # Set file properties
-      md_rec.last_modified = File.mtime(@file_rec.path).utc.iso8601
-      md_rec.file_size = File.size(@file_rec.path)
-    end
-    
-    def populate_services
-      md_rec = @file_rec.metadata_record
-      
-      service_manager = @app_manager.service_manager
-      definitions = service_manager.list_service_definitions(location: @file_rec.storage_location.name)
-      
-      # Add service section
-      definitions.each do |serv_def|
-        serv_name = serv_def.name
-        md_rec.add_service(serv_name)
+      md_rec.last_modified = File.mtime(physical_path).utc.iso8601(3)
+      md_rec.file_size = File.size(physical_path)
+
+      if physical_path != @file_rec.path
+        md_rec.physical_path = physical_path
+      else
+        md_rec.physical_path = nil
       end
     end
-    
+
     # Copy a subset of properties from an existing metadata record to the new record
     def retain_existing_properties
       md_rec = @file_rec.metadata_record
-      
-      old_md = MetadataDeserializer.deserialize(file_path: @file_rec.metadata_path)
+
+      old_md = MetadataDeserializer.deserialize(file_path: @file_rec.metadata_path,
+              digest_algs: @file_rec.storage_location.metadata_location.digests)
       # Copy custom properties
       old_md.properties.each { |name, value| md_rec.properties[name] = value }
       # Copy stale-replicas flag per service
       old_md.list_services.each do |serv_name|
         serv_rec = old_md.service(serv_name)
-        
+
         stale_replicas = serv_rec.stale_replicas
         if stale_replicas
           new_service = md_rec.service(serv_name)
@@ -95,4 +103,4 @@ module Longleaf
       end
     end
   end
-end
+end

data/lib/longleaf/helpers/case_insensitive_hash.rb

@@ -0,0 +1,38 @@
+module Longleaf
+  # Hash subclass which provides case insensitive keys, where keys are always downcased.
+  class CaseInsensitiveHash < Hash
+    def [](key)
+      super _insensitive(key)
+    end
+
+    def []=(key, value)
+      super _insensitive(key), value
+    end
+
+    def delete(key)
+      super _insensitive(key)
+    end
+
+    def has_key?(key)
+      super _insensitive(key)
+    end
+
+    def merge(other_hash)
+      super other_hash.map {|k, v| [_insensitive(k), v] }.to_h
+    end
+
+    def merge!(other_hash)
+      super other_hash.map {|k, v| [_insensitive(k), v] }.to_h
+    end
+
+    # Cause this hash to serialize as a regular hash to avoid deserialization failures
+    def encode_with coder
+      coder.represent_map nil, self
+    end
+
+    protected
+    def _insensitive(key)
+      key.respond_to?(:downcase) ? key.downcase : key
+    end
+  end
+end

data/lib/longleaf/helpers/digest_helper.rb

@@ -0,0 +1,56 @@
+require 'longleaf/errors'
+require 'digest'
+
+module Longleaf
+  # Helper methods for generating digests
+  class DigestHelper
+    KNOWN_DIGESTS ||= ['md5', 'sha1', 'sha2', 'sha256', 'sha384', 'sha512', 'rmd160']
+
+    # @param algs Either a string containing one or an array containing zero or more digest
+    #    algorithm names.
+    # @raise [InvalidDigestAlgorithmError] thrown if any of the digest algorithms listed are not
+    #    known to the system.
+    def self.validate_algorithms(algs)
+      return if algs.nil?
+      if algs.is_a?(String)
+        unless self.is_known_algorithm?(algs)
+          raise InvalidDigestAlgorithmError.new("Unknown digest algorithm #{algs}")
+        end
+      else
+        unknown = algs.select { |alg| !KNOWN_DIGESTS.include?(alg) }
+        unless unknown.empty?
+          raise InvalidDigestAlgorithmError.new("Unknown digest algorithm(s): #{unknown}")
+        end
+      end
+    end
+
+    # @param alg [String] identifier of digest algorithm
+    # @return [Boolean] true if the digest is a valid known algorithm
+    def self.is_known_algorithm?(alg)
+      KNOWN_DIGESTS.include?(alg)
+    end
+
+    # Get a Digest class for the specified algorithm
+    # @param alg [String] name of the digest algorithm
+    # @return [Digest] A digest class for the requested algorithm
+    # @raise [InvalidDigestAlgorithmError] if an unknown digest algorithm is requested
+    def self.start_digest(alg)
+      case alg
+      when 'md5'
+        return Digest::MD5.new
+      when 'sha1'
+        return Digest::SHA1.new
+      when 'sha2', 'sha256'
+        return Digest::SHA2.new
+      when 'sha384'
+        return Digest::SHA2.new(384)
+      when 'sha512'
+        return Digest::SHA2.new(512)
+      when 'rmd160'
+        return Digest::RMD160.new
+      else
+        raise InvalidDigestAlgorithmError.new("Cannot produce digest for unknown algorithm '#{alg}'.")
+      end
+    end
+  end
+end