merritt-manifest 0.1.0

data/lib/merritt/manifest/file.rb

@@ -0,0 +1,33 @@
+require 'merritt/manifest/fields'
+
+module Merritt
+  class Manifest
+    # A marker interface for file-like objects. Each field
+    # may or may not be relevant to a given manifest format.
+    module File
+      # @return [nil, URI] the URL at which to retrieve the file
+      attr_reader :file_url
+
+      # @return [nil, String] the hash algorithm used to hash the file
+      attr_reader :hash_algorithm
+
+      # @return [nil, String] the hash value
+      attr_reader :hash_value
+
+      # @return [nil, Integer] the file size in bytes.
+      attr_reader :file_size
+
+      # @return [nil, DateTime] date and time the file was last modified.
+      #   Note that according to the [Merritt Ingest Service docs](https://confluence.ucop.edu/download/attachments/16744573/Merritt-ingest-service-latest.pdf),
+      #   “modification time field SHOULD NOT be specified, and will be ignored if provided.”
+      attr_reader :file_last_modified
+
+      # @return [nil, String] the file name. If not present, will be inferred from
+      #   {#file_url}.
+      attr_reader :file_name
+
+      # @return [nil, MIME::Type] the mime type
+      attr_reader :mime_type
+    end
+  end
+end

data/lib/merritt/manifest/module_info.rb

@@ -0,0 +1,12 @@
+module Merritt
+  class Manifest
+    # The name of this gem
+    NAME = 'merritt-manifest'.freeze
+
+    # The version of this gem
+    VERSION = '0.1.0'.freeze
+
+    # The copyright notice for this gem
+    COPYRIGHT = 'Copyright (c) 2017 The Regents of the University of California'.freeze
+  end
+end

data/lib/merritt/manifest/object.rb

@@ -0,0 +1,32 @@
+module Merritt
+  class Manifest
+    # A specialization of {Manifest} for Merritt object submissions.
+    class Object < Manifest
+
+      # Creates a new {Manifest::Object}
+      # @param files [Array<Manifest::File>] an array of data files to be converted to entries.
+      #   (Note that these not be actual {Manifest::File} objects so long as they respond to,
+      #   at minimum, `#file_url`. The other {Manifest::File} fields are optional.)
+      def initialize(files:)
+        super(
+          profile: 'http://uc3.cdlib.org/registry/ingest/manifest/mrt-ingest-manifest',
+          prefixes: Merritt::Manifest::Fields::Object.prefixes,
+          fields: Merritt::Manifest::Fields::Object.fields,
+          entries: to_entries(files)
+        )
+      end
+
+      private
+
+      def to_entries(files)
+        files.map do |file|
+          Merritt::Manifest::Fields::Object.map do |field|
+            field_name = field.field_name
+            field_value = field.value_from(file)
+            [field_name, field_value]
+          end.to_h
+        end
+      end
+    end
+  end
+end

data/lib/merritt/manifest.rb

@@ -0,0 +1,128 @@
+Dir.glob(File.expand_path('../manifest/*.rb', __FILE__)).sort.each(&method(:require))
+
+module Merritt
+  # A Merritt manifest file
+  class Manifest
+
+    # Base for all recognized profile URIs
+    PROFILE_BASE_URI = 'http://uc3.cdlib.org/registry/ingest/manifest/'.freeze
+
+    # Checkm 0.7 conformance level
+    CHECKM_0_7 = 'checkm_0.7'.freeze
+
+    # @return [String] the conformance level
+    attr_reader :conformance
+
+    # @return [URI] the profile URI
+    attr_reader :profile
+
+    # @return [Hash{Symbol => URI}] a map from namespace prefixes to their URIs
+    attr_reader :prefixes
+
+    # @return [Array<String>] the field names, in the form prefix:fieldname
+    attr_reader :fields
+
+    # @return [Array<Hash<String, Object>>] the entries
+    attr_reader :entries
+
+    # Creates a new manifest. Note that the prefix, field, and entry arrays are
+    # copied on initialization, as are the individual entry hashes.
+    #
+    # @param conformance [String] the conformance level. Defaults to {CHECKM_0_7}.
+    # @param profile [URI, String] the profile URI. Must begin with
+    # @param prefixes [Hash{String,Symbol => URI, String}] a map from namespace prefixes to their URIs
+    # @param fields Array<String> a list of field names, in the form prefix:fieldname
+    # @param entries [Array<Hash<String, Object><] A list of entries, each of which is a hash keyed by
+    #   a prefixed fieldname defined in `fields`. Nil values are allowed.
+    # @raise [ArgumentError] if `profile` does not begin with {PROFILE_BASE_URI}
+    # @raise [ArgumentError] if `fields` cannot be parsed as prefix:fieldname, or if one or more prefixes
+    #   is not mapped to a URI in `prefixes`
+    # @raise [URI::InvalidURIError] if `profile` cannot be parsed as a URI
+    def initialize(conformance: CHECKM_0_7, profile:, prefixes: {}, fields: [], entries: [])
+      @conformance = conformance
+      @profile = normalize_profile_uri(profile).freeze
+      @prefixes = normalize_prefixes(prefixes).freeze
+      @fields = validate_fields(fields).freeze
+      @entries = normalize_entries(entries).freeze
+    end
+
+    # Writes this manifest to the specified IO
+    # @param io [IO] the IO to write to
+    def write_to(io)
+      write_sc(io, conformance)
+      write_sc(io, 'profile', profile)
+      prefixes.each { |prefix, url| write_sc(io, 'prefix', "#{prefix}:", url) }
+      write_sc(io, 'fields', *fields)
+      entries.each { |entry| io.puts(entry_line(entry)) }
+      write_sc(io, 'eof')
+    end
+
+    # Writes this manifest as a string
+    # @return [String] the manifest file contents as a string
+    def write_to_string
+      io = StringIO.new
+      write_to(io)
+      io.string
+    end
+
+    private
+
+    # checkm column separator
+    COLSEP = ' | '.freeze
+
+    def entry_line(entry)
+      fields.map { |f| entry[f] }.join(COLSEP).sub(/[| ]+\z/, '')
+    end
+
+    # writes a checkm "structured comment"
+    # @param io [IO] the IO to write to
+    # @param comment [String] the comment
+    # @param columns [nil, Array<String>] columns to follow the initial comment
+    def write_sc(io, comment, *columns)
+      io << '#%' << comment
+      io << COLSEP << columns.join(COLSEP) unless columns.empty?
+      io << "\n"
+    end
+
+    def normalize_entries(entries)
+      entries.each_with_index.map do |entry, i|
+        raise ArgumentError, "Nil entry at index #{i}" unless entry
+        normalize_entry(entry)
+      end
+    end
+
+    def normalize_entry(entry)
+      normalized = {}
+      fields.each do |f|
+        next unless (value = entry[f])
+        normalized[f] = value
+      end
+      raise ArgumentError, "No fields found in entry #{entry}" if normalized.empty?
+      normalized
+    end
+
+    def validate_fields(fields)
+      fields.map { |f| validate_field(f) }
+    end
+
+    def validate_field(field)
+      prefix, fieldname = field.split(':')
+      raise ArgumentError "Unknown prefix in field '#{field}': #{prefix}" unless prefixes.key?(prefix.to_sym)
+      raise ArgumentError "Field '#{field}' cannot be parsed as prefix:fieldname" unless fieldname
+      field
+    end
+
+    def normalize_prefixes(prefixes)
+      return {} unless prefixes
+      prefixes.map { |k, v| [k.to_sym, Util.to_uri(v)] }.to_h
+    end
+
+    def normalize_profile_uri(profile)
+      profile_uri = Util.to_uri(profile)
+      raise ArgumentError, "Invalid profile: #{profile || 'nil'}" unless profile_uri &&
+        profile_uri.to_s.start_with?(PROFILE_BASE_URI)
+      profile_uri.clone # defensive copy
+    end
+
+  end
+end

data/lib/merritt/util.rb

@@ -0,0 +1,19 @@
+module Merritt
+  # Miscellaneous utility methods
+  module Util
+    class << self
+      # Ensures that the specified argument is a URI.
+      # @param url [String, URI] The argument. If the argument is already
+      #   a URI, it is returned unchanged; otherwise, the argument's string
+      #   form (as returned by +`to_s`+) is parsed as a URI.
+      # @return [nil, URI] +`nil`+ if +`url`+ is nil, otherwise the URI.
+      # @raise [URI::InvalidURIError] if `url` is a string that is not a valid URI
+      def to_uri(url)
+        return nil unless url
+        return url if url.is_a? URI
+        stripped = url.respond_to?(:strip) ? url.strip : url.to_s.strip
+        URI.parse(stripped)
+      end
+    end
+  end
+end

data/lib/merritt.rb

@@ -0,0 +1,4 @@
+# Supermodule for Merritt-related code
+module Merritt
+  Dir.glob(File.expand_path('../merritt/*.rb', __FILE__)).sort.each(&method(:require))
+end

data/merritt-manifest.gemspec

@@ -0,0 +1,37 @@
+# coding: utf-8
+
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+
+require 'merritt/manifest/module_info'
+require 'uri'
+
+Gem::Specification.new do |spec|
+  spec.name          = Merritt::Manifest::NAME
+  spec.version       = Merritt::Manifest::VERSION
+  spec.authors       = ['David Moles']
+  spec.email         = ['david.moles@ucop.edu']
+  spec.summary       = 'Merritt manifest generation library'
+  spec.description   = 'Merritt manifest generation library'
+  spec.license       = 'MIT'
+
+  origin = `git config --get remote.origin.url`.chomp
+  origin_uri = origin.start_with?('http') ? URI(origin) : URI(origin.gsub(%r{git@([^:]+)(.com|.org)[^\/]+}, 'http://\1\2'))
+  spec.homepage = URI::HTTP.build(host: origin_uri.host, path: origin_uri.path.chomp('.git')).to_s
+
+  spec.files         = `git ls-files -z`.split("\x0")
+  spec.executables   = spec.files.grep(%r{^bin/}) { |f| File.basename(f) }
+
+  spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
+  spec.require_paths = ['lib']
+
+  spec.add_dependency 'typesafe_enum', '~> 0.1.7'
+
+  spec.add_development_dependency 'bundler', '~> 1.14'
+  spec.add_development_dependency 'rake', '~> 12.0'
+  spec.add_development_dependency 'rspec', '~> 3.5'
+  spec.add_development_dependency 'simplecov', '~> 0.14'
+  spec.add_development_dependency 'simplecov-console', '~> 0.4'
+  spec.add_development_dependency 'rubocop', '~> 0.47'
+  spec.add_development_dependency 'yard', '~> 0.9'
+end

data/merritt-manifest.iml

@@ -0,0 +1,37 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<module type="RUBY_MODULE" version="4">
+  <component name="NewModuleRootManager" inherit-compiler-output="true">
+    <exclude-output />
+    <content url="file://$MODULE_DIR$" />
+    <orderEntry type="jdk" jdkName="ruby-2.2.5-p319" jdkType="RUBY_SDK" />
+    <orderEntry type="sourceFolder" forTests="false" />
+    <orderEntry type="library" scope="PROVIDED" name="ansi (v1.5.0, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="ast (v2.3.0, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="bundler (v1.14.6, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="diff-lcs (v1.3, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="docile (v1.1.5, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="hirb (v0.7.3, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="json (v2.0.4, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="parser (v2.4.0.0, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="powerpack (v0.1.1, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="rainbow (v2.2.1, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="rake (v12.0.0, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="rspec (v3.5.0, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="rspec-core (v3.5.4, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="rspec-expectations (v3.5.0, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="rspec-mocks (v3.5.0, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="rspec-support (v3.5.0, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="rubocop (v0.48.1, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="ruby-progressbar (v1.8.1, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="simplecov (v0.14.1, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="simplecov-console (v0.4.1, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="simplecov-html (v0.10.0, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="typesafe_enum (v0.1.7, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="unicode-display_width (v1.2.1, ruby-2.2.5-p319) [gem]" level="application" />
+    <orderEntry type="library" scope="PROVIDED" name="yard (v0.9.8, ruby-2.2.5-p319) [gem]" level="application" />
+  </component>
+  <component name="RModuleSettingsStorage">
+    <LOAD_PATH number="2" string0="$MODULE_DIR$/lib" string1="$MODULE_DIR$/spec" />
+    <I18N_FOLDERS number="0" />
+  </component>
+</module>

data/spec/.rubocop.yml

@@ -0,0 +1,10 @@
+inherit_from: ../.rubocop.yml
+
+Metrics/MethodLength:
+  Enabled: false
+
+Metrics/ModuleLength:
+  Enabled: false
+
+Metrics/BlockLength:
+  Enabled: false

data/spec/data/4blocks.checkm

@@ -0,0 +1,8 @@
+#%checkm_0.7
+#%profile | http://uc3.cdlib.org/registry/ingest/manifest/mrt-ingest-manifest
+#%prefix | mrt: | http://merritt.cdlib.org/terms#
+#%prefix | nfo: | http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#
+#%fields | nfo:fileUrl | nfo:hashAlgorithm | nfo:hashValue | nfo:fileSize | nfo:fileLastModified | nfo:fileName | mrt:mimeType
+http://merritt.cdlib.org/samples/4blocks.jpg | md5 | 0b21c6d48e815dd537d42dc1cfac0111 |  |  | 4blocks.jpg
+http://merritt.cdlib.org/samples/4blocks.txt | md5 | ed04a855f89f31f8dc8e9bb946f5f159 |  |  | 4blocks.txt
+#%eof

data/spec/data/mrt-dataone-manifest.txt

@@ -0,0 +1,32 @@
+#%dataonem_0.1
+#%profile | http://uc3.cdlib.org/registry/ingest/manifest/mrt-dataone-manifest
+#%prefix | dom: | http://uc3.cdlib.org/ontology/dataonem
+#%prefix | mrt: | http://uc3.cdlib.org/ontology/mom
+#%fields | dom:scienceMetadataFile | dom:scienceMetadataFormat | dom:scienceDataFile | mrt:mimeType
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | Laney_300394_Exempt_Determination_Letter.pdf | application/pdf
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | Laney_300394_Exempt_Determination_Letter.pdf | application/pdf
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | Laney_IRBProposal.docx | application/vnd.openxmlformats-officedocument.wordprocessingml.document
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | Laney_IRBProposal.docx | application/vnd.openxmlformats-officedocument.wordprocessingml.document
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | SensorSurvey_Printout.pdf | application/pdf
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | SensorSurvey_Printout.pdf | application/pdf
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | clean_survey_data_no_ids.csv | text/csv
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | clean_survey_data_no_ids.csv | text/csv
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | research_coords.csv | text/csv
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | research_coords.csv | text/csv
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | survey/Q10/research_sites.R | text/plain
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | survey/Q10/research_sites.R | text/plain
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | survey/Q11-23/sensors_platforms.R | text/plain
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | survey/Q11-23/sensors_platforms.R | text/plain
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | survey/Q24/limitsToExpansion.R | text/plain
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | survey/Q24/limitsToExpansion.R | text/plain
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | survey/Q25-32/data_metadata_management.R | text/plain
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | survey/Q25-32/data_metadata_management.R | text/plain
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | survey/Q3-9/respondent_info.R | text/plain
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | survey/Q3-9/respondent_info.R | text/plain
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | survey/Q33-37/networking.R | text/plain
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | survey/Q33-37/networking.R | text/plain
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | survey/Q38-42/publications.R | text/plain
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | survey/Q38-42/publications.R | text/plain
+mrt-datacite.xml | http://datacite.org/schema/kernel-3.1 | survey_data_prep.R | text/plain
+mrt-oaidc.xml | http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd | survey_data_prep.R | text/plain
+#%eof

data/spec/data/sampleBatchOfContainers.checkm

@@ -0,0 +1,9 @@
+#%checkm_0.7
+#%profile | http://uc3.cdlib.org/registry/ingest/manifest/mrt-container-batch-manifest
+#%prefix | mrt: | http://merritt.cdlib.org/terms#
+#%prefix | nfo: | http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#
+#%fields | nfo:fileUrl | nfo:hashAlgorithm | nfo:hashValue | nfo:fileSize | nfo:fileLastModified | nfo:fileName | mrt:primaryIdentifier | mrt:localIdentifier | mrt:creator | mrt:title | mrt:date
+http://merritt.cdlib.org/samples/huskyChicken.zip |  |  |  |  | huskyChicken.zip |  |  | Tracy Seneca | Husky Chicken | 2009-09-01T00:00:00
+http://merritt.cdlib.org/samples/outdoorStore.zip |  |  |  |  | outdoorStore.zip |  |  |  | The Outdoor Store, Portland OR
+http://merritt.cdlib.org/samples/souvenirs.zip |  |  |  |  | souvenirs.zip |  |  | Tracy Seneca | SouvenirsLast Summer
+#%eof

data/spec/data/sampleBatchOfFiles.checkm

@@ -0,0 +1,9 @@
+#%checkm_0.7
+#%profile | http://uc3.cdlib.org/registry/ingest/manifest/mrt-single-file-batch-manifest
+#%prefix | mrt: | http://merritt.cdlib.org/terms#
+#%prefix | nfo: | http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#
+#%fields | nfo:fileUrl | nfo:hashAlgorithm | nfo:hashValue | nfo:fileSize | nfo:fileLastModified | nfo:fileName | mrt:primaryIdentifier | mrt:localIdentifier | mrt:creator | mrt:title | mrt:date
+http://merritt.cdlib.org/samples/goldenDragon.jpg | md5 | aa59e145dfb2237ecd8cb5ce4f4953ea |  |  | goldenDragon.jpg |  |  | Tracy Seneca | Golden Dragon Restaurant Sign | 2009-09-01T00:00:00
+http://merritt.cdlib.org/samples/tumbleBug.jpg | md5 | 61224ef4f2a8dad552ff57980ed0bf49 |  |  | tumbleBug.jpg |  |  | Tracy Seneca | Tumble Bug, Luna Park.  Sydney
+http://merritt.cdlib.org/samples/generalDrapery.jpg |  |  |  |  | generalDrapery.jpg |  |  |  | General Drapery
+#%eof

data/spec/data/sampleBatchOfManifests.checkm

@@ -0,0 +1,9 @@
+#%checkm_0.7
+#%profile | http://uc3.cdlib.org/registry/ingest/manifest/mrt-batch-manifest
+#%prefix | mrt: | http://merritt.cdlib.org/terms#
+#%prefix | nfo: | http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#
+#%fields | nfo:fileUrl | nfo:hashAlgorithm | nfo:hashValue | nfo:fileSize | nfo:fileLastModified | nfo:fileName | mrt:primaryIdentifier | mrt:localIdentifier | mrt:creator | mrt:title | mrt:date
+http://merritt.cdlib.org/samples/manifests/bigHunt.checkm |  |  |  |  | bigHunt.checkm |  |  | Tracy Seneca |  | 2009-09-01T00:00:00
+http://merritt.cdlib.org/samples/manifests/4blocks.checkm |  |  |  |  | 4blocks.checkm |  |  | Tracy Seneca | Motel Capri - manifest style | 2009-09-01T00:00:00
+http://merritt.cdlib.org/samples/manifests/call911.checkm |  |  |  |  | call911.checkm |  |  | Tracy Seneca | If This Bell Rings Call 911 - manifest style | 2009-09-01T00:00:00
+#%eof

data/spec/spec_helper.rb

@@ -0,0 +1,29 @@
+# ------------------------------------------------------------
+# SimpleCov setup
+
+if ENV['COVERAGE']
+  require 'simplecov'
+  require 'simplecov-console'
+
+  SimpleCov.minimum_coverage 100
+  SimpleCov.start do
+    add_filter '/spec/'
+    SimpleCov.formatter = SimpleCov::Formatter::MultiFormatter[
+      SimpleCov::Formatter::HTMLFormatter,
+      SimpleCov::Formatter::Console,
+    ]
+  end
+end
+
+# ------------------------------------------------------------
+# Rspec configuration
+
+RSpec.configure do |config|
+  config.raise_errors_for_deprecations!
+  config.mock_with :rspec
+end
+
+# ------------------------------------------------------------
+# Merritt::Manifest
+
+require 'merritt'

data/spec/unit/merritt/manifest/data_one_spec.rb

@@ -0,0 +1,122 @@
+require 'spec_helper'
+require 'ostruct'
+
+module Merritt
+  describe Manifest::DataONE do
+    attr_reader :files
+    attr_reader :manifest
+
+    before(:each) do
+      @files = {
+        'Laney_300394_Exempt_Determination_Letter.pdf' => 'application/pdf',
+        'Laney_IRBProposal.docx' => 'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
+        'SensorSurvey_Printout.pdf' => 'application/pdf',
+        'clean_survey_data_no_ids.csv' => 'text/csv',
+        'research_coords.csv' => 'text/csv',
+        'survey/Q10/research_sites.R' => 'text/plain',
+        'survey/Q11-23/sensors_platforms.R' => 'text/plain',
+        'survey/Q24/limitsToExpansion.R' => 'text/plain',
+        'survey/Q25-32/data_metadata_management.R' => 'text/plain',
+        'survey/Q3-9/respondent_info.R' => 'text/plain',
+        'survey/Q33-37/networking.R' => 'text/plain',
+        'survey/Q38-42/publications.R' => 'text/plain',
+        'survey_data_prep.R' => 'text/plain'
+      }.map { |name, type| OpenStruct.new(name: name, type: type) }
+      @manifest = Manifest::DataONE.new(files: files)
+    end
+
+    describe :conformance do
+      it 'returns DataONE 0.1' do
+        expect(manifest.conformance).to eq('dataonem_0.1')
+      end
+    end
+
+    describe :profile do
+      it 'returns the DataONE manifest profile' do
+        expect(manifest.profile).to eq(URI('http://uc3.cdlib.org/registry/ingest/manifest/mrt-dataone-manifest'))
+      end
+    end
+
+    describe :prefixes do
+      attr_reader :prefixes
+      before(:each) do
+        @prefixes = manifest.prefixes
+      end
+      it 'includes :dom' do
+        expect(prefixes[:dom]).to eq(URI('http://uc3.cdlib.org/ontology/dataonem'))
+      end
+      it 'includes :mrt' do
+        expect(prefixes[:mrt]).to eq(URI('http://uc3.cdlib.org/ontology/mom'))
+      end
+    end
+
+    describe :fields do
+      it 'returns the expected fields' do
+        expected = %w[dom:scienceMetadataFile dom:scienceMetadataFormat dom:scienceDataFile mrt:mimeType]
+        expect(manifest.fields).to eq(expected)
+      end
+    end
+
+    describe :entries do
+      attr_reader :entries
+      before(:each) do
+        @entries = manifest.entries
+      end
+
+      it 'converts files to entries' do
+        expect(entries.size).to eq(2 * files.size)
+        files.each_with_index do |file, index|
+          dcs_index = 2 * index
+          dcs_entry = entries[dcs_index]
+          oai_entry = entries[1 + dcs_index]
+          expect(dcs_entry['dom:scienceMetadataFile']).to eq('mrt-datacite.xml')
+          expect(oai_entry['dom:scienceMetadataFile']).to eq('mrt-oaidc.xml')
+          expect(dcs_entry['dom:scienceMetadataFormat']).to eq('http://datacite.org/schema/kernel-3.1')
+          expect(oai_entry['dom:scienceMetadataFormat']).to eq('http://dublincore.org/schemas/xmls/qdc/2008/02/11/qualifieddc.xsd')
+
+          [dcs_entry, oai_entry].each do |entry|
+            expect(entry['dom:scienceDataFile']).to eq(file.name)
+            expect(entry['mrt:mimeType']).to eq(file.type)
+          end
+        end
+      end
+    end
+
+    describe :write_to_string do
+      it 'writes a DataONE manifest' do
+        path = 'mrt-dataone-manifest.txt'
+        expected = File.read("spec/data/#{path}")
+        actual = manifest.write_to_string
+        if actual != expected
+          now = Time.now.to_i
+          FileUtils.mkdir('tmp') unless File.directory?('tmp')
+          File.open("tmp/#{now}-expected-#{path}", 'w') { |f| f.write(expected) }
+          File.open("tmp/#{now}-actual-#{path}", 'w') { |f| f.write(actual) }
+        end
+        expect(actual).to eq(expected)
+      end
+    end
+
+    describe :write_to_file do
+      it 'writes to a file' do
+        file = Tempfile.new('manifest.txt')
+        begin
+          manifest.write_to(file)
+          file.close
+          actual = IO.read(file.path)
+          path = 'mrt-dataone-manifest.txt'
+          expected = File.read("spec/data/#{path}")
+          actual = manifest.write_to_string
+          if actual != expected
+            now = Time.now.to_i
+            FileUtils.mkdir('tmp') unless File.directory?('tmp')
+            File.open("tmp/#{now}-expected-#{path}", 'w') { |f| f.write(expected) }
+            File.open("tmp/#{now}-actual-#{path}", 'w') { |f| f.write(actual) }
+          end
+        ensure
+          file.delete
+        end
+      end
+    end
+  end
+end

data/spec/unit/merritt/manifest/fields_spec.rb

@@ -0,0 +1,55 @@
+require 'spec_helper'
+require 'ostruct'
+
+module Merritt
+  describe Manifest::Fields::Object do
+    describe :value_from do
+      describe :FILE_URL do
+        it 'returns a URI even if given a String' do
+          url_str = 'http://example.org/example.txt'
+          obj = OpenStruct.new(file_url: url_str)
+          expected = URI(url_str)
+          actual = Manifest::Fields::Object::FILE_URL.value_from(obj)
+          expect(actual).to eq(expected)
+        end
+
+        it 'fails if not present' do
+          obj = OpenStruct.new
+          expect { Manifest::Fields::Object::FILE_URL.value_from(obj) }.to raise_error(ArgumentError)
+        end
+      end
+
+      describe :FILE_SIZE do
+        it 'returns nil for nil' do
+          obj = OpenStruct.new
+          actual = Manifest::Fields::Object::FILE_SIZE.value_from(obj)
+          expect(actual).to be_nil
+        end
+        it 'returns an integer even if given a String' do
+          obj = OpenStruct.new(file_size: '1234')
+          expected = 1234
+          actual = Manifest::Fields::Object::FILE_SIZE.value_from(obj)
+          expect(actual).to eq(expected)
+        end
+      end
+
+      describe :FILE_NAME do
+        it 'prefers the filename to the URI' do
+          url = URI('http://example.org/example.txt')
+          name = 'EXAMPLE.TXT'
+          obj = OpenStruct.new(file_url: url, file_name: name)
+          actual = Manifest::Fields::Object::FILE_NAME.value_from(obj)
+          expect(actual).to eq(name)
+        end
+
+        it 'extracts the filename from the URI if not present' do
+          url = URI('http://example.org/example.txt')
+          obj = OpenStruct.new(file_url: url)
+          expected = 'example.txt'
+          actual = Manifest::Fields::Object::FILE_NAME.value_from(obj)
+          expect(actual).to eq(expected)
+        end
+      end
+    end
+  end
+end