duracloud-client 0.3.0 → 0.4.0

data/lib/duracloud/content_manifest.rb

@@ -0,0 +1,88 @@
+require 'nokogiri'
+require 'active_model'
+
+module Duracloud
+  class ContentManifest
+    include ActiveModel::Model
+
+    validates_presence_of :space_id, :manifest_id
+
+    attr_accessor :space_id, :manifest_id, :store_id
+
+    def self.find(**kwargs)
+      new(**kwargs).tap do |manifest|
+        manifest.content
+      end
+    end
+
+    def content
+      @content ||= Content.new(space_id: space_id, content_id: manifest_id, store_id: store_id).tap do |c|
+        c.load_properties
+      end
+    end
+
+    def source
+      @source ||= Source.new(self)
+    end
+
+    def xml
+      @xml ||= content.download.body
+    end
+
+    protected
+
+    def method_missing(name, *args, &block)
+      if content.respond_to?(name)
+        content.send(name, *args, &block)
+      else
+        super
+      end
+    end
+
+    class Source
+      attr_reader :manifest
+
+      def initialize(manifest)
+        @manifest = manifest
+      end
+
+      def doc
+        @doc ||= Nokogiri::XML(manifest.xml)
+      end
+
+      def md5
+        doc.css("sourceContent md5").text
+      end
+
+      def content_id
+        doc.css("sourceContent").first["contentId"]
+      end
+
+      def size
+        doc.css("sourceContent byteSize").text.to_i
+      end
+
+      def content_type
+        doc.css("sourceContent mimetype").text
+      end
+
+      def download(&block)
+        chunks.each do |chunk|
+          chunk.download(&block)
+        end
+      end
+
+      def chunks
+        Enumerator.new do |e|
+          doc.css("chunk").each do |chunk_xml|
+            e << Content.find(space_id: manifest.space_id,
+                              content_id: chunk_xml["chunkId"],
+                              store_id: manifest.store_id,
+                              md5: chunk_xml.css("md5").text)
+          end
+        end
+      end
+    end
+
+  end
+end

data/lib/duracloud/manifest.rb

@@ -2,35 +2,54 @@ module Duracloud
   class Manifest
     include TSV
 
+    TSV_FORMAT = "TSV"
+    BAGIT_FORMAT = "BAGIT"
+
     attr_reader :space_id, :store_id
 
     def initialize(space_id, store_id = nil)
       @space_id = space_id
       @store_id = store_id
-      @tsv_response = nil
-      @bagit_response = nil
     end
 
-    def tsv
-      super || tsv_response.body
+    # Returns the manifest in TSV format,
+    #   downloading from DuraCloud is not pre-loaded.
+    # @yield [String] chunk of the manifest, if block given.
+    # @return [Duracloud::Response, String, IO] the response,
+    #   if downloaded, or the pre-loaded TSV.
+    # @raise [Duracloud::NotFoundError]
+    def tsv(&block)
+      tsv_source? ? super : download(TSV_FORMAT, &block)
     end
 
-    def bagit
-      bagit_response.body
+    # Downloads the manifest in BAGIT format.
+    # @yield [String] chunk of the manifest, if block given.
+    # @return [Duracloud::Response] the response.
+    # @raise [Duracloud::NotFoundError]
+    def bagit(&block)
+      download(BAGIT_FORMAT, &block)
     end
 
-    private
-
-    def tsv_response
-      @tsv_response ||= get_response("TSV")
+    # Downloads the manifest
+    # @yield [String] chunk of the manifest, if block given.
+    # @param format [Symbol, String] the format of the manifest.
+    #    Defaults to "TSV".
+    # @return [Duracloud::Response, String] the response, if block
+    #   given, or the manifest content, if no block.
+    # @raise [Duracloud::NotFoundError]
+    def download(format = TSV_FORMAT, &block)
+      fmt = format.to_s.upcase
+      if block_given?
+        get_response(fmt, &block)
+      else
+        get_response(fmt).body
+      end
     end
 
-    def bagit_response
-      @bagit_response ||= get_response("BAGIT")
-    end
+    private
 
-    def get_response(format)
-      Client.get_manifest(space_id, query(format))
+    def get_response(format, &block)
+      Client.get_manifest(space_id, query(format), &block)
     end
 
     def query(format)

data/lib/duracloud/properties.rb

@@ -33,7 +33,7 @@ module Duracloud
     # @param prop [String] the property name
     # @return [Boolean]
     def self.property?(prop)
-      duracloud_property?(prop) && !internal_property?(prop)
+      duracloud_property?(prop)
     end
 
     # Filter the hash of properties, selecting only the properties valid

data/lib/duracloud/request.rb

@@ -16,24 +16,25 @@ module Duracloud
       set_options(options.dup)
     end
 
-    def execute
-      response_class.new(original_response)
+    def execute(&block)
+      response_class.new original_response(&block)
     end
 
     private
 
-    def original_response
+    def original_response(&block)
       connection.send(http_method,
                       url,
                       body: body,
                       query: query,
-                      header: headers)
+                      header: headers,
+                      &block)
     end
 
     def set_options(options)
       @body    = options.delete(:body)
       @headers = options.delete(:headers)
-      query   = options.delete(:query) || {}
+      query    = options.delete(:query) || {}
       # Treat other keywords args as query params and ignore empty params
       @query = query.merge(options).reject { |k, v| v.to_s.empty? }
     end

data/lib/duracloud/rest_methods.rb

@@ -33,8 +33,8 @@ module Duracloud
       durastore(:delete, space_id, **query)
     end
 
-    def get_content(space_id, content_id, **options)
-      durastore_content(:get, space_id, content_id, **options)
+    def get_content(space_id, content_id, **options, &block)
+      durastore_content(:get, space_id, content_id, **options, &block)
     end
 
     def get_content_properties(space_id, content_id, **options)
@@ -61,8 +61,8 @@ module Duracloud
       durastore(:get, "audit/#{space_id}", **query)
     end
 
-    def get_manifest(space_id, **query)
-      durastore(:get, "manifest/#{space_id}", **query)
+    def get_manifest(space_id, **query, &block)
+      durastore(:get, "manifest/#{space_id}", **query, &block)
     end
 
     def get_bit_integrity_report(space_id, **query)
@@ -85,14 +85,14 @@ module Duracloud
 
     private
 
-    def durastore(*args)
-      execute(DurastoreRequest, *args)
+    def durastore(*args, &block)
+      execute(DurastoreRequest, *args, &block)
     end
 
-    def durastore_content(http_method, space_id, content_id, **options)
+    def durastore_content(http_method, space_id, content_id, **options, &block)
       escaped_content_id = content_id.gsub(/%/, "%25").gsub(/ /, "%20")
       url = [ space_id, escaped_content_id ].join("/")
-      durastore(http_method, url, **options)
+      durastore(http_method, url, **options, &block)
     end
 
   end

data/lib/duracloud/space.rb

@@ -5,9 +5,7 @@ module Duracloud
   #
   # A "space" within a DuraCloud account.
   #
-  class Space
-    include Persistence
-    include HasProperties
+  class Space < AbstractEntity
 
     after_save :reset_acls
 
@@ -119,7 +117,7 @@ module Duracloud
       end
     end
 
-    attr_reader :space_id, :store_id
+    attr_accessor :space_id, :store_id
     alias_method :id, :space_id
 
     after_save :reset_acls
@@ -128,8 +126,7 @@ module Duracloud
     # @param space_id [String] the space ID
     # @param store_id [String] the store ID (optional)
     def initialize(space_id, store_id = nil)
-      @space_id = space_id
-      @store_id = store_id
+      super(space_id: space_id, store_id: store_id)
       yield self if block_given?
     end
 
@@ -226,6 +223,11 @@ module Duracloud
 
     private
 
+    def do_load_properties
+      response = Client.get_space_properties(id, **query)
+      self.properties = response.headers
+    end
+
     def reset_acls
       @acls = nil
     end
@@ -244,10 +246,6 @@ module Duracloud
       SpaceProperties
     end
 
-    def get_properties_response
-      Client.get_space_properties(id, **query)
-    end
-
     def do_delete
       Client.delete_space(id, **query)
     end

data/lib/duracloud/tsv.rb

@@ -3,6 +3,8 @@ require "csv"
 module Duracloud
   module TSV
 
+    CHUNK_SIZE = 1024 * 16
+
     def csv
       @csv ||= CSV.new(tsv, csv_options)
     end
@@ -20,16 +22,36 @@ module Duracloud
       csv.rewind
     end
 
-    def tsv
-      @tsv
+    def tsv(&block)
+      return unless tsv_source?
+      begin
+        tsv_source.rewind
+        if block_given?
+          while chunk = tsv_source.read(CHUNK_SIZE)
+            yield chunk
+          end
+        else
+          tsv_source.read
+        end
+      ensure
+        tsv_source.rewind
+      end
     end
 
     def load_tsv(io_or_str)
-      @tsv = io_or_str
+      @tsv_source = io_or_str.is_a?(String) ? StringIO.new(io_or_str, "rb") : io_or_str
+    end
+
+    def tsv_source
+      @tsv_source
+    end
+
+    def tsv_source?
+      !!@tsv_source
     end
 
     def load_tsv_file(path)
-      load_tsv(File.new(path, "rb"))
+      load_tsv File.new(path, "rb")
     end
 
     def to_s
@@ -42,6 +64,7 @@ module Duracloud
       { col_sep: "\t",
         quote_char: "`",
         headers: true,
+        return_headers: false,
         header_converters: header_converters,
       }
     end

data/lib/duracloud/version.rb

@@ -1,3 +1,3 @@
 module Duracloud
-  VERSION = "0.3.0"
+  VERSION = "0.4.0"
 end

data/lib/duracloud.rb

@@ -2,12 +2,15 @@ require "duracloud/version"
 require "duracloud/error"
 
 module Duracloud
+  autoload :AbstractEntity, "duracloud/abstract_entity"
   autoload :AuditLog, "duracloud/audit_log"
   autoload :BitIntegrityReport, "duracloud/bit_integrity_report"
+  autoload :ChunkedContent, "duracloud/chunked_content"
   autoload :Client, "duracloud/client"
   autoload :Configuration, "duracloud/configuration"
   autoload :Connection, "duracloud/connection"
   autoload :Content, "duracloud/content"
+  autoload :ContentManifest, "duracloud/content_manifest"
   autoload :ContentProperties, "duracloud/content_properties"
   autoload :DurastoreRequest, "duracloud/durastore_request"
   autoload :ErrorHandler, "duracloud/error_handler"

data/spec/fixtures/content_manifest.xml

@@ -0,0 +1,32 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<dur:chunksManifest xmlns:dur="duracloud.org">
+  <header schemaVersion="0.2">
+    <sourceContent contentId="datastreamStore/8/b/d4/info%3Afedora%2Fduke%3A447146%2Fcontent%2Fcontent.0">
+      <mimetype>application/octet-stream</mimetype>
+      <byteSize>4227858432</byteSize>
+      <md5>164e9aee34c0c42915716e11d5d539b5</md5>
+    </sourceContent>
+  </header>
+  <chunks>
+    <chunk chunkId="datastreamStore/8/b/d4/info%3Afedora%2Fduke%3A447146%2Fcontent%2Fcontent.0.dura-chunk-0000" index="0">
+      <byteSize>1000000000</byteSize>
+      <md5>8a7d5beb2523fb5e4d7c921096be50a9</md5>
+    </chunk>
+    <chunk chunkId="datastreamStore/8/b/d4/info%3Afedora%2Fduke%3A447146%2Fcontent%2Fcontent.0.dura-chunk-0001" index="1">
+      <byteSize>1000000000</byteSize>
+      <md5>e37115d4da0e187130ab645dee4f14ed</md5>
+    </chunk>
+    <chunk chunkId="datastreamStore/8/b/d4/info%3Afedora%2Fduke%3A447146%2Fcontent%2Fcontent.0.dura-chunk-0002" index="2">
+      <byteSize>1000000000</byteSize>
+      <md5>e37115d4da0e187130ab645dee4f14ed</md5>
+    </chunk>
+    <chunk chunkId="datastreamStore/8/b/d4/info%3Afedora%2Fduke%3A447146%2Fcontent%2Fcontent.0.dura-chunk-0003" index="3">
+      <byteSize>1000000000</byteSize>
+      <md5>93e9a4d242a9fb89796b98060094910d</md5>
+    </chunk>
+    <chunk chunkId="datastreamStore/8/b/d4/info%3Afedora%2Fduke%3A447146%2Fcontent%2Fcontent.0.dura-chunk-0004" index="4">
+      <byteSize>227858432</byteSize>
+      <md5>db0124ee56298ff7c7ac17be4ef14871</md5>
+    </chunk>
+  </chunks>
+</dur:chunksManifest>

data/spec/support/shared_examples_for_tsv.rb

@@ -0,0 +1,23 @@
+RSpec.shared_examples "a TSV" do
+
+  describe "#load_tsv" do
+    it "loads a string" do
+      tsv = File.read(path)
+      subject.load_tsv(tsv)
+      expect(subject.tsv).to eq(tsv)
+    end
+    it "loads an IO" do
+      tsv = File.read(path)
+      subject.load_tsv(tsv)
+      expect(subject.tsv).to eq(tsv)
+    end
+  end
+
+  describe "#load_tsv_file" do
+    specify {
+      subject.load_tsv_file(path)
+      expect(subject.tsv).to eq(File.read(path))
+    }
+  end
+
+end

data/spec/unit/audit_log_spec.rb

@@ -1,9 +1,13 @@
+require 'support/shared_examples_for_tsv'
+
 module Duracloud
   RSpec.describe AuditLog do
 
+    subject { described_class.new("myspace") }
+
     let(:path) { File.expand_path('../../fixtures/audit_log.tsv', __FILE__) }
 
-    subject { described_class.new("myspace") }
+    it_behaves_like "a TSV"
 
     describe "#csv" do
       before {
@@ -17,19 +21,5 @@ module Duracloud
       }
     end
 
-    describe "#load_tsv" do
-      it "loads a string" do
-        tsv = File.read(path)
-        subject.load_tsv(tsv)
-        expect(subject.tsv).to eq(tsv)
-      end
-      it "loads an IO" do
-        tsv = File.read(path)
-        tsv_io = File.new(path, "rb")
-        subject.load_tsv(tsv)
-        expect(subject.tsv.to_s).to eq(tsv)
-      end
-    end
-
   end
 end