Class: GDor::Indexer::SolrDocHash

Inherits:
SimpleDelegator
  • Object
show all
Defined in:
lib/gdor/indexer/solr_doc_hash.rb

Overview

Hash-like SolrDoc Object class

Instance Method Summary collapse

Constructor Details

#initialize(hash = {}) ⇒ SolrDocHash

Returns a new instance of SolrDocHash.



6
7
8
 
# File 'lib/gdor/indexer/solr_doc_hash.rb', line 6

def initialize(hash = {})
  super(hash)
end

Instance Method Details

#combine(new_hash) ⇒ GDor::Indexer::SolrDocHash

merge in field values from the new hash, with the following guarantees:

values for keys in new_hash will be a non-empty String or flat Array
keys will be removed from hash if all values are nil or empty

Parameters:

  • new_hash (Hash)

Returns:



34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
 
# File 'lib/gdor/indexer/solr_doc_hash.rb', line 34

def combine(new_hash)
  new_hash.select { |_key, value| Array(value).any? { |v| !v.blank? } }.each do |key, new_val|
    if field_present? key
      orig_val = self[key]
      case orig_val
      when Array
        self[key] += Array(new_val)
      else
        self[key] = Array(orig_val) + Array(new_val)
      end

      self[key] = self[key].reject(&:blank?).uniq
    else
      self[key] = new_val
    end
  end

  compact_blank_fields!
  self
end

#compact_blank_fields! ⇒ GDor::Indexer::SolrDocHash

Returns self.

Returns:



56
57
58
59
60
61
 
# File 'lib/gdor/indexer/solr_doc_hash.rb', line 56

def compact_blank_fields!
  keys.reject { |key| field_present? key }.each do |key|
    delete key
  end
  self
end

#druid ⇒ String

Returns:

  • (String) 


64
65
66
 
# File 'lib/gdor/indexer/solr_doc_hash.rb', line 64

def druid
  self[:druid]
end

#field_present?(field, exp_val = nil) ⇒ Boolean

when exp_val is nil, looks for non-empty existence of field when exp_val is a String, looks for matching value as a String or as a member of an Array when exp_val is a Regexp, looks for String value that matches, or Array with a String member that matches

Parameters:

  • exp_val (String, Regexp) (defaults to: nil)

Returns:

  • (Boolean) —

    true if the field is non-trivially present in the hash, false otherwise



15
16
17
18
19
20
21
22
23
24
25
26
27
 
# File 'lib/gdor/indexer/solr_doc_hash.rb', line 15

def field_present?(field, exp_val = nil)
  return false unless include?(field)
  return false unless Array(self[field]).any?(&:present?)

  case exp_val
  when nil
    true
  when Regexp
    Array(self[field]).any? { |s| exp_val.match(s) }
  else
    Array(self[field]).include? exp_val
  end
end

#validate_collection(config) ⇒ Object

validate fields that should be in hash for any collection object in SearchWorks Solr

See Also:



84
85
86
87
88
89
 
# File 'lib/gdor/indexer/solr_doc_hash.rb', line 84

def validate_collection(config)
  result = validate_gdor_fields(config)
  result << "#{druid} missing collection_type 'Digital Collection'\n" unless field_present?(:collection_type, 'Digital Collection')
  result << "#{druid} missing format_main_ssim 'Archive/Manuscript'\n" unless field_present?(:format_main_ssim, 'Archive/Manuscript')
  result
end

#validate_gdor_fields(config) ⇒ Object

validate fields that should be in hash for every gryphonDOR object in SearchWorks Solr

See Also:



93
94
95
96
97
98
99
100
101
 
# File 'lib/gdor/indexer/solr_doc_hash.rb', line 93

def validate_gdor_fields(config)
  result = []
  result << "#{druid} missing druid field\n" unless field_present?(:druid, druid)
  result << "#{druid} missing url_fulltext for purl\n" unless field_present?(:url_fulltext, "#{config.harvestdor.purl}/#{druid}")
  result << "#{druid} missing access_facet 'Online'\n" unless field_present?(:access_facet, 'Online')
  result << "#{druid} missing or bad display_type, possibly caused by unrecognized @type attribute on <contentMetadata>\n" unless field_present?(:display_type, /(file)|(image)|(media)|(book)/)
  result << "#{druid} missing building_facet 'Stanford Digital Repository'\n" unless field_present?(:building_facet, 'Stanford Digital Repository')
  result
end

#validate_item(config) ⇒ Array<String>

validate fields that should be in hash for any item object in SearchWorks Solr

Parameters:

  • config (Object) —

    Configuration object

Returns:

  • (Array<String>) —

    Array of messages suitable for notificaiton email and/or logs



71
72
73
74
75
76
77
78
79
80
 
# File 'lib/gdor/indexer/solr_doc_hash.rb', line 71

def validate_item(config)
  result = validate_gdor_fields(config)
  result << "#{druid} missing collection\n" unless field_present?(:collection)

  Array(self[:collection]).each do |collection_druid|
    result << "#{druid} missing collection_with_title (or collection #{collection_druid} is missing title)\n" unless field_present?(:collection_with_title, Regexp.new("#{collection_druid}-\\|-.+"))
  end
  result << "#{druid} missing file_id(s)\n" unless field_present?(:file_id)
  result
end

#validate_mods(_config) ⇒ Object

validate fields that should be in doc hash for every unmerged gryphonDOR object in SearchWorks Solr

See Also:



105
106
107
108
109
110
111
112
113
114
115
 
# File 'lib/gdor/indexer/solr_doc_hash.rb', line 105

def validate_mods(_config)
  result = []
  result << "#{druid} missing modsxml\n" unless field_present?(:modsxml)
  result << "#{druid} missing resource type\n" unless field_present?(:format_main_ssim)
  result << "#{druid} missing format\n" unless field_present?(:format) # for backwards compatibility
  result << "#{druid} missing title\n" unless field_present?(:title_display)
  result << "#{druid} missing pub year for date slider\n" unless field_present?(:pub_year_tisim)
  result << "#{druid} missing author\n" unless field_present?(:author_person_display)
  result << "#{druid} missing language\n" unless field_present?(:language)
  result
end