Class: Stanford::ContentInventory

Inherits:
Object
  • Object
show all
Defined in:
lib/stanford/content_inventory.rb

Overview

Note:

Copyright © 2012 by The Board of Trustees of the Leland Stanford Junior University. All rights reserved. See LICENSE for details.

Stanford-specific utility methods for transforming contentMetadata to versionInventory and doing

Data Model

  • DorMetadata = utility methods for interfacing with Stanford metadata files (esp contentMetadata)

    • ContentInventory [1..1] = utilities for transforming contentMetadata to versionInventory and doing comparsions

    • ActiveFedoraObject [1..*] = utility for extracting content or other information from a Fedora Instance

Instance Method Summary collapse

Instance Method Details

#generate_content_metadata(file_group, object_id, version_id) ⇒ String

Returns The contentMetadata instance generated from the FileGroup.

Examples:

Parameters:

Returns:

  • (String) —

    The contentMetadata instance generated from the FileGroup



96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
 
# File 'lib/stanford/content_inventory.rb', line 96

def generate_content_metadata(file_group, object_id, version_id)
  cm = Nokogiri::XML::Builder.new do |xml|
    xml.contentMetadata(:type=>"sample", :objectId=>object_id) {
      xml.resource(:type=>"version", :sequence=>"1", :id=>"version-#{version_id.to_s}") {
        file_group.files.each do |file_manifestation|
          signature = file_manifestation.signature
          file_manifestation.instances.each do |instance|
            xml.file(
                :id=>instance.path,
                :size=>signature.size,
                :datetime=>instance.datetime,
                :shelve=>'yes',
                :publish=>'yes',
                :preserve=>'yes') {
              fixity = signature.fixity
              xml.checksum(:type=>"MD5") {xml.text signature.md5 } if fixity[:md5]
              xml.checksum(:type=>"SHA-1") {xml.text signature.sha1} if fixity[:sha1]
              xml.checksum(:type=>"SHA-256") {xml.text signature.sha256} if fixity[:sha256]
            }
          end
        end
      }
    }
  end
  cm.to_xml
end

#generate_instance(node) ⇒ FileInstance

Returns The Moab::FileInstance object generated from the XML data.

Parameters:

  • node (Nokogiri::XML::Node) —

    The XML node containing file information

Returns:



85
86
87
88
89
90
 
# File 'lib/stanford/content_inventory.rb', line 85

def generate_instance(node)
  instance = FileInstance.new()
  instance.path = node.attributes['id'].content
  instance.datetime = node.attributes['datetime'].content rescue nil
  instance
end

#generate_signature(node) ⇒ FileSignature

Returns The Moab::FileSignature object generated from the XML data.

Parameters:

  • node (Nokogiri::XML::Node) —

    The XML node containing file information

Returns:



65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
 
# File 'lib/stanford/content_inventory.rb', line 65

def generate_signature(node)
  signature = FileSignature.new()
  signature.size = node.attributes['size'].content
  checksum_nodes = node.xpath('checksum')
  checksum_nodes.each do |checksum_node|
    case checksum_node.attributes['type'].content.upcase
      when 'MD5'
        signature.md5 = checksum_node.text
      when 'SHA1', 'SHA-1'
        signature.sha1 = checksum_node.text
      when 'SHA256', 'SHA-256'
        signature.sha256 = checksum_node.text
    end
  end
  signature
end

#group_from_cm(content_metadata, subset) ⇒ FileGroup

Returns The Moab::FileGroup object generated from a contentMetadata instance.

Examples:

Parameters:

  • content_metadata (String) —

    The contentMetadata as a string

  • subset (String) —

    Speciifes which subset of files to list (all|preserve|publish|shelve)

Returns:



38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
 
# File 'lib/stanford/content_inventory.rb', line 38

def group_from_cm(content_metadata, subset)
  ng_doc = Nokogiri::XML(content_metadata)
  validate_content_metadata(ng_doc)
  nodeset = case subset.to_s.downcase
    when 'preserve'
      ng_doc.xpath("//file[@preserve='yes']")
    when 'publish'
      ng_doc.xpath("//file[@publish='yes']")
    when 'shelve'
      ng_doc.xpath("//file[@shelve='yes']")
    when 'all'
      ng_doc.xpath("//file")
    else
        raise "Unknown disposition subset (#{subset})"
    end
  content_group = FileGroup.new(:group_id=>'content', :data_source => "contentMetadata-#{subset}")
  nodeset.each do |file_node|
    signature = generate_signature(file_node)
    instance = generate_instance(file_node)
    content_group.add_file_instance(signature, instance)
  end
  content_group
end

#inventory_from_cm(content_metadata, object_id, subset, version_id = nil) ⇒ FileInventory

Returns The versionInventory equivalent of the contentMetadata if the supplied content_metadata is blank or empty, then a skeletal FileInventory will be returned.

Parameters:

  • content_metadata (String) —

    The content metadata to be transformed into a versionInventory

  • object_id (String) —

    The identifier of the digital object

  • subset (String) —

    Speciifes which subset of files to list (all|preserve|publish|shelve)

  • version_id (Integer) (defaults to: nil) —

    The ID of the version whosen content metadata is to be transformed

Returns:

  • (FileInventory) —

    The versionInventory equivalent of the contentMetadata if the supplied content_metadata is blank or empty, then a skeletal FileInventory will be returned



22
23
24
25
26
27
28
29
30
31
 
# File 'lib/stanford/content_inventory.rb', line 22

def inventory_from_cm(content_metadata, object_id, subset, version_id=nil)
  # The contentMetadata datastream is not required for ingest, since some object types, such as collection or APO do not require one.
  # Many of these objects have contentMetadata with no child elements, such as this:
  #    <contentMetadata objectId="bd608mj3166" type="file"/>
  # but there are also objects that have no datasteam of this name at all
  cm_inventory = FileInventory.new(:type=>"version",:digital_object_id=>object_id, :version_id=>version_id)
  content_group = group_from_cm(content_metadata, subset)
  cm_inventory.groups << content_group
  cm_inventory
end

#remediate_checksum_nodes(file_node, signature) ⇒ void

This method returns an undefined value.

Returns update the file’s checksum elements if data missing, raise exception if inconsistent.

Parameters:

  • file_node (Nokogiri::XML::Element) —

    the File stanza being remediated

  • signature (FileSignature) —

    the fixity data for the file from the FileGroup



206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
 
# File 'lib/stanford/content_inventory.rb', line 206

def remediate_checksum_nodes(file_node, signature)
  # collect <checksum> elements for checksum types that are already present
  checksum_nodes = OrderedHash.new
  file_node.xpath('checksum').each do |checksum_node|
    type = @type_for_name[checksum_node['type']]
    checksum_nodes[type] = checksum_node
  end
  # add new <checksum> elements for the other checksum types that were missing
  @names_for_type.each do |type, names|
    unless checksum_nodes.has_key?(type)
      checksum_node = Nokogiri::XML::Element.new('checksum',file_node.document)
      checksum_node['type'] = names[0]
      file_node << checksum_node
      checksum_nodes[type] = checksum_node
    end
  end
  # make sure the <checksum> element has a content value
  checksum_nodes.each do |type,checksum_node|
    cm_checksum = checksum_node.content
    sig_checksum = signature.checksums[type]
    if cm_checksum.nil? or cm_checksum.empty?
      checksum_node.content = sig_checksum
    elsif cm_checksum != sig_checksum
      raise "Inconsistent #{type.to_s} for #{file_node['id']}: #{cm_checksum} != #{sig_checksum}"
    end
  end
end

#remediate_content_metadata(content_metadata, content_group) ⇒ String

Returns a remediated copy of the contentMetadata with fixity data filled in

Parameters:

  • content_metadata (String) —

    The contentMetadata as a string

  • content_group (FileGroup) —

    The Moab::FileGroup object used as the fixity data source

Returns:

  • (String) —

    Returns a remediated copy of the contentMetadata with fixity data filled in

See Also:



174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
 
# File 'lib/stanford/content_inventory.rb', line 174

def remediate_content_metadata(content_metadata, content_group)
  return nil if content_metadata.nil?
  return content_metadata if content_group.nil? or content_group.files.size < 1
  signature_for_path = content_group.path_hash
  @type_for_name = FileSignature.checksum_type_for_name
  @names_for_type = FileSignature.checksum_names_for_type
  ng_doc = Nokogiri::XML(content_metadata) { |x| x.noblanks }
  nodeset = ng_doc.xpath("//file")
  nodeset.each do |file_node|
    filepath = file_node['id']
    signature = signature_for_path[filepath]
    remediate_file_size(file_node, signature)
    remediate_checksum_nodes(file_node, signature)
  end
  ng_doc.to_xml(:indent => 2)
end

#remediate_file_size(file_node, signature) ⇒ void

This method returns an undefined value.

Returns update the file size attribute if missing, raise exception if inconsistent.

Parameters:

  • file_node (Nokogiri::XML::Element) —

    the File stanza being remediated

  • signature (FileSignature) —

    the fixity data for the file from the FileGroup



194
195
196
197
198
199
200
201
 
# File 'lib/stanford/content_inventory.rb', line 194

def remediate_file_size(file_node, signature)
  file_size = file_node['size']
  if file_size.nil? or file_size.empty?
    file_node['size'] = signature.size.to_s
  elsif file_size != signature.size.to_s
    raise "Inconsistent size for #{file_node['id']}: #{file_size} != #{signature.size.to_s}"
  end
end

#validate_content_metadata(content_metadata) ⇒ Boolean

Returns True if contentMetadata has essetial file attributes, else raise exception.

Parameters:

  • content_metadata (String, Nokogiri::XML::Document) —

    The contentMetadata as a string or XML doc

Returns:

  • (Boolean) —

    True if contentMetadata has essetial file attributes, else raise exception

Raises:



125
126
127
128
129
 
# File 'lib/stanford/content_inventory.rb', line 125

def validate_content_metadata(content_metadata)
  result = validate_content_metadata_details(content_metadata)
  raise Moab::InvalidMetadataException, result[0]+" ..." if result.size > 0
  true
end

#validate_content_metadata_details(content_metadata) ⇒ Array<String>

Returns List of problems found.

Parameters:

  • content_metadata (String, Nokogiri::XML::Document) —

    The contentMetadata as a string or XML doc

Returns:



133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
 
# File 'lib/stanford/content_inventory.rb', line 133

def validate_content_metadata_details(content_metadata)
  result = []
  content_metadata_doc =
    case content_metadata.class.name
      when "String"
        Nokogiri::XML(content_metadata)
      when "Pathname"
        Nokogiri::XML(content_metadata.read)
      when "Nokogiri::XML::Document"
        content_metadata
      else
        raise Moab::InvalidMetadataException, "Content Metadata is in unrecognized format"
      end
  nodeset = content_metadata_doc.xpath("//file")
  nodeset.each do |file_node|
    missing = ['id', 'size','md5','sha1']
    missing.delete('id') if file_node.has_attribute?('id')
    missing.delete('size') if file_node.has_attribute?('size')
    checksum_nodes = file_node.xpath('checksum')
    checksum_nodes.each do |checksum_node|
      case checksum_node.attributes['type'].content.upcase
        when 'MD5'
          missing.delete('md5')
        when 'SHA1', 'SHA-1'
          missing.delete('sha1')
      end
    end
    if missing.include?('id')
      result << "File node #{nodeset.index(file_node)} is missing #{missing.join(',')}"
    elsif missing.size > 0
      id = file_node['id']
      result << "File node having id='#{id}' is missing #{missing.join(',')}"
    end
  end
  result
end