skala 0.3.0 → 1.1.0

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_edition.rb

@@ -0,0 +1,10 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetEdition < Transformator::Transformation::Step
+
+  def call
+    target.record.edition = transformation.read_source_values(".//display/edition").first
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_fulltext_available.rb

@@ -0,0 +1,12 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+# TODO: UBPB transformation: Move to custom adapter
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetFulltextAvailable < Transformator::Transformation::Step
+
+  def call
+    target.record.fulltext_available = transformation.read_source_values(".//delivery/fulltext").first != "no_fulltext"
+  end
+
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_id.rb

@@ -0,0 +1,11 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetId < Transformator::Transformation::Step
+
+  def call
+    target.record.id = source.at_xpath(".//control/recordid").text().gsub(/\ATN_/, "").split(":").last
+  end
+
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_identifier.rb

@@ -0,0 +1,46 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetIdentifier < Transformator::Transformation::Step
+
+  def call
+    sanitizer = Rails::Html::FullSanitizer.new
+
+    identifier_strings = transformation.read_source_values(".//display/identifier", split: ";")
+    identifier_strings.map!{|s| sanitizer.sanitize(s)}
+
+    isbns = extract_isbns(identifier_strings)
+    issns = extract_issns(identifier_strings)
+    dois  = extract_dois(identifier_strings)
+
+    target.record.isbn = isbns
+    target.record.issn = issns
+    target.record.identifier = dois.map{|i| {type: "doi", value: i}}
+  end
+
+private
+
+  def extract_isbns(identifier_strings)
+    extract(identifier_strings, /(isbn:|e-isbn:)\s+([a-z0-9\-\.\/]+)/i, match_index: 1)
+  end
+
+  def extract_issns(identifier_strings)
+    extract(identifier_strings, /(issn:|e-issn:)\s+([a-z0-9\-\.\/]+)/i, match_index: 1)
+  end
+
+  def extract_dois(identifier_strings)
+    extract(identifier_strings, /doi:\s+([a-z0-9\-\.\/]+)/i)
+  end
+
+  def extract(identifier_strings, regexp, match_index: 0)
+    values = []
+
+    identifier_strings.each do |s|
+      values += s.scan(regexp).map{|r| r[match_index]}
+    end
+
+    values.compact
+  end
+
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_is_part_of.rb

@@ -0,0 +1,14 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetIsPartOf < Transformator::Transformation::Step
+
+  def call
+    is_part_of = transformation.read_source_values(".//display/ispartof").map do |superorder|
+      { label: superorder }
+    end
+
+    target.record.is_part_of = is_part_of
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_language.rb

@@ -0,0 +1,10 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetLanguage < Transformator::Transformation::Step
+
+  def call
+    target.record.language = transformation.read_source_values(".//display/language", split: ";")
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_metadata.rb

@@ -0,0 +1,13 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetMetadata < Transformator::Transformation::Step
+
+  def call
+    target.id = source.attr("ID")
+    target.index = source.attr("SEARCH_ENGINE")
+    target.score = source.attr("RANK")
+    target.type = "record"
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_openurl.rb

@@ -0,0 +1,23 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetOpenurl < Transformator::Transformation::Step
+
+  def call
+    openurl = transformation.read_source_values(".//LINKS/openurl").first
+    if openurl.present?
+      openurl = remove_language_param(openurl)
+      target.record.openurl = openurl
+    end
+  end
+
+private
+
+  # Remove the language param to force the default language
+  # TODO: UBPB setting: Move to custom adapter
+  def remove_language_param(openurl)
+    openurl.split('&').map{|e| e.gsub(/req\.language=.+/, 'req.language=')}.join('&')
+  end
+
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_publisher.rb

@@ -0,0 +1,10 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetPublisher < Transformator::Transformation::Step
+
+  def call
+    target.record.publisher = transformation.read_source_values(".//display/publisher").first
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_source.rb

@@ -0,0 +1,10 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetSource < Transformator::Transformation::Step
+
+  def call
+    target.record.source = transformation.read_source_values(".//display/source").first
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_subject.rb

@@ -0,0 +1,10 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetSubject < Transformator::Transformation::Step
+
+  def call
+    target.record.subject = transformation.read_source_values(".//display/subject", split: ";")
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_title.rb

@@ -0,0 +1,10 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetTitle < Transformator::Transformation::Step
+
+  def call
+    target.record.title = transformation.read_source_values(".//display/title").first
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/doc_transformation/set_year_of_publication.rb

@@ -0,0 +1,10 @@
+require "transformator/transformation/step"
+require_relative "../doc_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::DocTransformation::
+  SetYearOfPublication < Transformator::Transformation::Step
+
+  def call
+    target.record.year_of_publication = transformation.read_source_values(".//display/creationdate").first
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/fix_primo_facets_bugs.rb

@@ -0,0 +1,108 @@
+require "transformator/transformation/step"
+require_relative "../result_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::
+  FixPrimoFacetsBugs < Transformator::Transformation::Step
+
+  def call
+    # please do not change order of the calls below, because they depend (in parts) on each other
+    fix_selected_facets_occur_slightly_different
+    fix_selected_facet_value_is_missing
+    fix_creationdate_facet_values_out_of_selected_range
+    # case 1 - there is no facet at all
+    #transformation.search_request.facets.each do |_requested_facet|
+    #  binding.pry
+    #end
+=begin
+    request[:body]["facets"].try do |_requested_facets|
+      _requested_facets.each do |_facet_name, _requested_facet|
+        unless target["facets"][_facet_name]
+          if _requested_facet.keys.include?("terms")
+            target["facets"][_facet_name] = {
+              "_type" => "terms",
+              "terms" => []
+            }
+          end
+        end
+      end
+    end
+=end
+
+    # case 2 - there is a facet, but the requested facet is not included
+=begin
+    deep_locate(
+      request,
+      -> (_key, _value, _) do
+        _key == "match" && _value.keys.any? do |__key|
+          __key.start_with?("facet")
+        end
+      end
+    )
+    .presence
+    .try do |_facet_queries|
+      _facet_queries.each do |_facet_query|
+        _facet_query["match"].try do |_match_query|
+          target["facets"][_match_query.keys.first]["terms"].try do |_target_facet_terms|
+            if _target_facet_terms.none? { |_term| _term["term"] == _match_query.values.first }
+              _target_facet_terms << {
+                "term" => _match_query.values.first,
+                "count" => 1
+              }
+            end
+          end
+        end
+      end
+    end
+=end
+  end
+
+  private
+
+  def fix_creationdate_facet_values_out_of_selected_range
+    creationdate_facet_query = transformation.search_request.facet_queries.find do |_facet_query| 
+      _facet_query.field == "facet_creationdate"
+    end
+
+    creationdate_facet = target.facets.find do |_target_facet|
+      _target_facet.field == "facet_creationdate"
+    end
+
+    if creationdate_facet_query
+      creationdate_facet.entries.select! do |_entry|
+        _entry.key.to_i >= creationdate_facet_query.gte.to_i &&
+        _entry.key.to_i <= creationdate_facet_query.lte.to_i
+      end
+    end
+  end
+
+  def fix_selected_facet_value_is_missing
+    transformation.search_request.facet_queries.each do |_facet_query|
+      target_facet = target.facets.find { |_facet| _facet.field == _facet_query.field }
+
+      if _facet_query.type.to_s == "match" && target_facet && target_facet.type.to_s == "terms"
+        if target_facet.terms.none? { |_term| _term.term == _facet_query.query }
+          term_class = target_facet.terms.first.class
+
+          target_facet.terms.push term_class.new(
+            count: target.total_hits,
+            term: _facet_query.query
+          )
+        end
+      end
+    end
+  end
+
+  def fix_selected_facets_occur_slightly_different
+    transformation.search_request.facet_queries.each do |_facet_query|
+      target_facet = target.facets.find { |_facet| _facet.field == _facet_query.field }
+
+      if _facet_query.type.to_s == "match" && target_facet && target_facet.type.to_s == "terms"
+        target_facet.terms.each do |_target_facet_term|
+          if _target_facet_term.term.downcase.strip == _facet_query.query.downcase
+            _target_facet_term.term = _facet_query.query
+          end
+        end
+      end
+    end
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/parse_search_brief_response.rb

@@ -0,0 +1,20 @@
+require "nokogiri"
+require "transformator/transformation/step"
+require_relative "../result_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::
+  ParseSearchBriefResponse < Transformator::Transformation::Step
+
+  def call
+    transformation.search_brief_response =
+    Nokogiri::XML(source)
+    .remove_namespaces!
+    .at_xpath("/Envelope/Body/searchBriefResponse/searchBriefReturn")
+    .try(:text)
+    .try do |_embedded_search_brief_response|
+      Nokogiri::XML(_embedded_search_brief_response) { |config| config.noblanks }
+      .remove_namespaces!
+      .at_xpath("//SEGMENTS/JAGROOT/RESULT")
+    end
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/set_facets.rb

@@ -0,0 +1,71 @@
+require "transformator/transformation/step"
+require_relative "../result_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::
+  SetFacets < Transformator::Transformation::Step
+
+  def call
+    requested_facets = transformation.search_request.facets
+
+    target.facets =
+    transformation.search_brief_response.xpath("//FACET")
+    .map do |_source_facet|
+      # in primo facet fields are named "facet_foo", while the name is "foo"
+      field = "facet_#{_source_facet.attr("NAME")}"
+
+      corresponding_search_request_facet = requested_facets.find do |_requested_facet|
+        _requested_facet.field == field
+      end
+
+      case corresponding_search_request_facet.try(:type)
+      when "histogram"
+        target.class::HistogramFacet.new(
+          field: field,
+          name: corresponding_search_request_facet.name,
+          entries: begin
+            _source_facet.xpath("./FACET_VALUES")
+            .select do |_source_facet_value|
+              _source_facet_value.attr("KEY")[/\d{4}/]
+            end
+            .each_with_object({}) do |_source_facet_value, _hash|
+              key = _source_facet_value.attr("KEY").to_i
+              count = _source_facet_value.attr("VALUE").to_i
+
+              _hash[key] = count
+            end
+            .tap do |_key_count_mapping|
+              keys = _key_count_mapping.keys.map(&:to_i)
+
+              (keys.min..keys.max).each do |_key_from_range|
+                _key_count_mapping[_key_from_range] ||= 0
+              end
+            end
+            .each_with_object([]) do |(_key, _count), _array|
+              _array.push(count: _count, key: _key)
+            end
+            .sort do |_entry, _other_entry|
+              _entry[:key] <=> _other_entry[:key]
+            end
+          end
+        )
+      when "terms"
+        target.class::TermsFacet.new(
+          field: field,
+          name: corresponding_search_request_facet.name,
+          terms: _source_facet.xpath("./FACET_VALUES").map do |_source_facet_value|
+            {
+              count: _source_facet_value.attr("VALUE").to_i,
+              term: begin
+                # https://github.com/ubpb/katalog_ubpb/issues/58
+                # https://github.com/sparklemotion/nokogiri/issues/214
+                _source_facet_value.to_s.match(/KEY="([^"]+)"/).captures.try(:first) ||
+                _source_facet_value.attr("KEY")
+              end
+            }
+          end
+        )
+      end
+    end
+    .compact
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/set_total_hits.rb

@@ -0,0 +1,13 @@
+require "transformator/transformation/step"
+require_relative "../result_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::
+  SetTotalHits < Transformator::Transformation::Step
+
+  def call
+    target.total_hits =
+    transformation.search_brief_response.at_xpath("//DOCSET").try do |_docset|
+      _docset.attr("TOTALHITS").to_i
+    end
+  end
+end

data/lib/skala/primo_adapter/search/result_transformation/sort_terms_facets_terms.rb

@@ -0,0 +1,14 @@
+require "transformator/transformation/step"
+require_relative "../result_transformation"
+
+class Skala::PrimoAdapter::Search::ResultTransformation::
+  SortTermsFacetsTerms < Transformator::Transformation::Step
+
+  def call
+    target.facets.each do |_target_facet|
+      if _target_facet.type.to_s == "terms"
+        _target_facet.terms.sort_by!(&:count).reverse!
+      end
+    end
+  end
+end

data/lib/skala/primo_adapter/soap_api.rb

@@ -0,0 +1,15 @@
+require_relative "../primo_adapter"
+
+class Skala::PrimoAdapter::SoapApi
+  require_relative "./soap_api/search_brief"
+
+  attr_accessor :adapter
+
+  def initialize(adapter)
+    self.adapter = adapter
+  end
+
+  def searchBrief(*args)
+    self.class::SearchBrief.new(adapter).call(*args)
+  end
+end