search_solr_tools 6.1.0 → 6.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cded5837c68e69c4dd22f9aea6b3efab1d4cc21570daba1b91b9561d7cde1216
-  data.tar.gz: 73a840e47c0cbc41a3fb13e62ca85735e20d87dc9f4b0acfd3d522ef1874ad34
+  metadata.gz: f9ced4643b8adbda2b5ef09192f036af86878e07243fe959448213762e0e5cc1
+  data.tar.gz: 0a5f27a7bc1d8c9c0c07a20b6fbf122d5a3b6163a5654db635d5c478ac4a21bc
 SHA512:
-  metadata.gz: 5056c01da21c54b3b4c86ddef264d2e0f9dbe83db472677ac1076a31a5e42e97a6ad0cd0a037b0431374a2d243bfc398dfcfa631863d3d4f3b3b334751b6e2c6
-  data.tar.gz: 39418c870ebada693b8ef2b75460d04e8ed0d1f9764e534d2195fa59fb5daaec4ae002ae4f016d05449a1a8a593b7e931261965b5756c664c6edfc7c94e6e3b1
+  metadata.gz: cc66f8b40c62e2640fd72ce05aa2ac01aa76c58c730b6c445976fc7cf6e43b88cff29ec088f73e5dff913c879f7a1a31016cb634d851cfb3adb7b8bb735614c8
+  data.tar.gz: f896f7b473f977f0e349d422f6568774342e6bdb66e1a7dad1cf4477d0ffa7e9b05b81184898ab2d4d69c44f8ca98a3aa6791234926190484820cca4b439dc7d

data/CHANGELOG.md

@@ -1,8 +1,17 @@
-## v6.1.0 (2022-07-14)
+## v6.3.0 (2023-07-24)
+
+  - Update Rubocop configuration to actually run against files, and make
+    necessary corrections to comply with Rubocop styling.
+
+## v6.2.0 (2023-07-18)
+
+  - Remove deprecated harvesters and associated tests, helpers, etc.
+
+## v6.1.0 (2023-07-14)
 
   - Updated a few other dependencies that weren't at the newest versions.
 
-## v6.0.0 (2022-07-14)
+## v6.0.0 (2023-07-14)
 
   - Updated Ruby to 3.2.2, updated gem dependencies to more recent versions.
 

data/bin/search_solr_tools

@@ -47,7 +47,7 @@ class SolrHarvestCLI < Thor
     end
 
     ping_status = SearchSolrTools::Helpers::HarvestStatus.new(
-      SearchSolrTools::Helpers::HarvestStatus::PING_SOLR => solr_success,
+      SearchSolrTools::Helpers::HarvestStatus::PING_SOLR   => solr_success,
       SearchSolrTools::Helpers::HarvestStatus::PING_SOURCE => source_success
     )
     raise SearchSolrTools::Errors::HarvestError, ping_status unless ping_status.ok?
@@ -64,9 +64,9 @@ class SolrHarvestCLI < Thor
       puts "Target: #{target}"
       begin
         harvest_class = get_harvester_class(target)
-        harvester = harvest_class.new(options[:environment], die_on_failure)
+        harvester = harvest_class.new(options[:environment], die_on_failure:)
         ping_status = SearchSolrTools::Helpers::HarvestStatus.new(
-          SearchSolrTools::Helpers::HarvestStatus::PING_SOLR => harvester.ping_solr,
+          SearchSolrTools::Helpers::HarvestStatus::PING_SOLR   => harvester.ping_solr,
           SearchSolrTools::Helpers::HarvestStatus::PING_SOURCE => harvester.ping_source
         )
         raise SearchSolrTools::Errors::HarvestError, ping_status unless ping_status.ok?
@@ -120,20 +120,8 @@ class SolrHarvestCLI < Thor
   no_tasks do
     def harvester_map
       {
-        'bco_dmo' => SearchSolrTools::Harvesters::BcoDmo,
-        'adc' => SearchSolrTools::Harvesters::Adc,
-        'echo' => SearchSolrTools::Harvesters::Echo,
-        'ices' => SearchSolrTools::Harvesters::Ices,
-        'nmi' => SearchSolrTools::Harvesters::Nmi,
-        'nodc' => SearchSolrTools::Harvesters::Nodc,
-        'r2r' => SearchSolrTools::Harvesters::R2R,
-        'rda' => SearchSolrTools::Harvesters::Rda,
-        'usgs' => SearchSolrTools::Harvesters::Usgs,
-        'tdar' => SearchSolrTools::Harvesters::Tdar,
-        'pdc' => SearchSolrTools::Harvesters::Pdc,
-        'nsidc' => SearchSolrTools::Harvesters::NsidcJson,
-        'nsidc_auto_suggest' => SearchSolrTools::Harvesters::NsidcAutoSuggest,
-        'ade_auto_suggest' => SearchSolrTools::Harvesters::AdeAutoSuggest
+        'nsidc'              => SearchSolrTools::Harvesters::NsidcJson,
+        'nsidc_auto_suggest' => SearchSolrTools::Harvesters::NsidcAutoSuggest
       }
     end
 

data/lib/search_solr_tools/config/environments.rb

@@ -1,9 +1,11 @@
+# frozen_string_literal: true
+
 require 'yaml'
 
 module SearchSolrTools
   # configuration to work with solr locally, or on integration/qa/staging/prod
   module SolrEnvironments
-    YAML_ENVS = YAML.load_file(File.expand_path('../environments.yaml', __FILE__))
+    YAML_ENVS = YAML.load_file(File.expand_path('environments.yaml', __dir__))
 
     def self.[](env = :development)
       YAML_ENVS[:common].merge(YAML_ENVS[env.to_sym])

data/lib/search_solr_tools/config/environments.yaml

@@ -3,38 +3,6 @@
   :collection_name: nsidc_oai
   :collection_path: solr
   :port: 8983
-  :bcodmo_url: http://www.bco-dmo.org/nsidc/arctic-deployments.json
-  :adc_url: https://arcticdata.io/metacat/d1/mn/v2/query/solr/select?q=northBoundCoord:%5B45.0%20TO%2090.0%5D
-  :data_one_url: https://cn.dataone.org/cn/v1/query/solr/select?q=northBoundCoord:%5B45.0%20TO%2090.0%5D
-  :echo_url: https://api.echo.nasa.gov/catalog-rest/echo_catalog/datasets.echo10?bounding_box=-180,45,180,90
-  :gtnp:
-    - http://www.gtnpdatabase.org/rest/boreholes/json
-    - http://www.gtnpdatabase.org/rest/activelayers/json
-  :ices_url: http://geo.ices.dk/geonetwork/srv/en/csw
-  :ncdc_paleo_url: https://gis.ncdc.noaa.gov/gptpaleo/csw
-  :nmi_url: http://arcticdata.met.no/metamod/oai
-  :nodc_url: https://data.nodc.noaa.gov/geoportal/csw
-  :pdc_url: http://www.polardata.ca/oai/provider
-  :rda_url: https://rda.ucar.edu/cgi-bin/oai
-  :tdar_url: http://core.tdar.org/search/rss
-  :usgs_url: https://www.sciencebase.gov/catalog/item/527cf4ede4b0850ea05182ee/csw
-  :eol:
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.SHEBA.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.SBI.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.PacMARS.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BASE.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.ATLAS.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.ARC_MIP.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.AMTS.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BOREAS.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BeringSea.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.ARCSS.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BEST.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BSIERP.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.BARROW.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.DBO.thredds.xml
-    - http://data.eol.ucar.edu/jedi/catalog/ucar.ncar.eol.project.ITEX.thredds.xml
-  :r2r_url: http://get.rvdata.us/services/cruise/
 
   # Not using DCS API v2 here because not all retired datasets have their "retired"
   # flag checked. For example, GLA01.033 is retired; GLA01.018 is not, but it

data/lib/search_solr_tools/errors/harvest_error.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module SearchSolrTools
   module Errors
     class HarvestError < StandardError
@@ -10,34 +12,47 @@ module SearchSolrTools
       ERRCODE_OTHER = 128
 
       ERRCODE_DESC = {
-          ERRCODE_SOLR_PING => 'Solr instance did not return a successful ping',
-          ERRCODE_SOURCE_PING => 'Source to be harvested did not return a successful ping',
-          ERRCODE_SOURCE_NO_RESULTS => 'Source to be harvested returned no documents matching query',
-          ERRCODE_SOURCE_HARVEST_ERROR => 'One or more source documents returned an error when trying to retrieve or translate',
-          ERRCODE_DOCUMENT_INVALID => 'One or more documents to be harvested was invalid (malformed)',
-          ERRCODE_INGEST_ERROR => 'Solr returned an error trying to ingest one or more harvested documents',
-          ERRCODE_OTHER => 'General error code for non-harvest related issues'
+        ERRCODE_SOLR_PING            => 'Solr instance did not return a successful ping',
+        ERRCODE_SOURCE_PING          => 'Source to be harvested did not return a successful ping',
+        ERRCODE_SOURCE_NO_RESULTS    => 'Source to be harvested returned no documents matching query',
+        ERRCODE_SOURCE_HARVEST_ERROR => 'One or more source documents returned an error when trying to retrieve or translate',
+        ERRCODE_DOCUMENT_INVALID     => 'One or more documents to be harvested was invalid (malformed)',
+        ERRCODE_INGEST_ERROR         => 'Solr returned an error trying to ingest one or more harvested documents',
+        ERRCODE_OTHER                => 'General error code for non-harvest related issues'
       }.freeze
 
       PING_ERRCODE_MAP = {
-        'ping_solr' => ERRCODE_SOLR_PING,
-        'ping_source' => ERRCODE_SOURCE_PING,
-      }
+        'ping_solr'   => ERRCODE_SOLR_PING,
+        'ping_source' => ERRCODE_SOURCE_PING
+      }.freeze
 
       STATUS_ERRCODE_MAP = {
-          Helpers::HarvestStatus::HARVEST_NO_DOCS => ERRCODE_SOURCE_NO_RESULTS,
-          Helpers::HarvestStatus::HARVEST_FAILURE => ERRCODE_SOURCE_HARVEST_ERROR,
-          Helpers::HarvestStatus::INGEST_ERR_INVALID_DOC => ERRCODE_DOCUMENT_INVALID,
-          Helpers::HarvestStatus::INGEST_ERR_SOLR_ERROR => ERRCODE_INGEST_ERROR,
-          Helpers::HarvestStatus::OTHER_ERROR => ERRCODE_OTHER
+        Helpers::HarvestStatus::HARVEST_NO_DOCS        => ERRCODE_SOURCE_NO_RESULTS,
+        Helpers::HarvestStatus::HARVEST_FAILURE        => ERRCODE_SOURCE_HARVEST_ERROR,
+        Helpers::HarvestStatus::INGEST_ERR_INVALID_DOC => ERRCODE_DOCUMENT_INVALID,
+        Helpers::HarvestStatus::INGEST_ERR_SOLR_ERROR  => ERRCODE_INGEST_ERROR,
+        Helpers::HarvestStatus::OTHER_ERROR            => ERRCODE_OTHER
       }.freeze
 
       # If code is -1, it means display all error codes
       def self.describe_exit_code(code = -1)
+        code_list = code_to_list(code)
+
+        codes = {}
+        code_list.each do |k|
+          next if code == -1 && !ERRCODE_DESC.keys.include?(k) # skip INVALID CODE if showing all codes
+
+          codes[k] = ERRCODE_DESC.keys.include?(k) ? ERRCODE_DESC[k] : 'INVALID CODE NUMBER'
+        end
+
+        codes
+      end
+
+      # Loop through all bit-flag values to produce a list of integers
+      def self.code_to_list(code)
         code = code.to_i
         code_list = []
 
-        # Loop through all bit-flag values
         [128, 64, 32, 16, 8, 4, 2, 1].each do |k|
           if code >= k || code == -1
             code_list.prepend k
@@ -45,20 +60,17 @@ module SearchSolrTools
           end
         end
 
-        codes = {}
-        code_list.each do |k|
-          next if code == -1 && !ERRCODE_DESC.keys.include?(k)  # skip INVALID CODE if showing all codes
-          codes[k] = ERRCODE_DESC.keys.include?(k) ? ERRCODE_DESC[k] : 'INVALID CODE NUMBER'
-        end
-
-        codes
+        code_list
       end
 
-      def initialize(status, message=nil)
+      def initialize(status, message = nil)
         @status_data = status
         @other_message = message
+
+        super message
       end
 
+      # rubocop:disable Metrics/AbcSize
       def exit_code
         if @status_data.nil?
           puts "OTHER ERROR REPORTED: #{@other_message}"
@@ -70,19 +82,20 @@ module SearchSolrTools
         code = 0
         code += ERRCODE_SOLR_PING unless @status_data.ping_solr
         code += ERRCODE_SOURCE_PING unless @status_data.ping_source
-        code += ERRCODE_SOURCE_NO_RESULTS if @status_data.status[Helpers::HarvestStatus::HARVEST_NO_DOCS] > 0
-        code += ERRCODE_SOURCE_HARVEST_ERROR if @status_data.status[Helpers::HarvestStatus::HARVEST_FAILURE] > 0
-        code += ERRCODE_DOCUMENT_INVALID if @status_data.status[Helpers::HarvestStatus::INGEST_ERR_INVALID_DOC] > 0
-        code += ERRCODE_INGEST_ERROR if @status_data.status[Helpers::HarvestStatus::INGEST_ERR_SOLR_ERROR] > 0
+        code += ERRCODE_SOURCE_NO_RESULTS if @status_data.status[Helpers::HarvestStatus::HARVEST_NO_DOCS].positive?
+        code += ERRCODE_SOURCE_HARVEST_ERROR if @status_data.status[Helpers::HarvestStatus::HARVEST_FAILURE].positive?
+        code += ERRCODE_DOCUMENT_INVALID if @status_data.status[Helpers::HarvestStatus::INGEST_ERR_INVALID_DOC].positive?
+        code += ERRCODE_INGEST_ERROR if @status_data.status[Helpers::HarvestStatus::INGEST_ERR_SOLR_ERROR].positive?
 
-        code = ERRCODE_OTHER if code == 0
+        code = ERRCODE_OTHER if code.zero?
 
         code
       end
+      # rubocop:enable Metrics/AbcSize
 
       def message
-        self.class.describe_exit_code(exit_code).map{|c,v| v}.join("\n")
+        self.class.describe_exit_code(exit_code).map { |_c, v| v }.join("\n")
       end
     end
   end
-end
+end

data/lib/search_solr_tools/harvesters/auto_suggest.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'base'
 require 'json'
 require 'rest-client'
@@ -6,8 +8,8 @@ module SearchSolrTools
   module Harvesters
     # Use the nsidc_oai core to populate the auto_suggest core
     class AutoSuggest < Base
-      def initialize(env = 'development', die_on_failure = false)
-        super env, die_on_failure
+      def initialize(env = 'development', die_on_failure: false)
+        super
         @env_settings = SolrEnvironments[@environment] # super sets @environment.
       end
 
@@ -50,7 +52,7 @@ module SearchSolrTools
 
         if status == Helpers::HarvestStatus::INGEST_OK
           puts "Added #{add_docs.size} auto suggest documents in one commit"
-          return Helpers::HarvestStatus.new(Helpers::HarvestStatus::INGEST_OK => add_docs)
+          Helpers::HarvestStatus.new(Helpers::HarvestStatus::INGEST_OK => add_docs)
         else
           puts "Failed adding #{add_docs.size} documents in single commit, retrying one by one"
           new_add_docs = []

data/lib/search_solr_tools/harvesters/base.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'multi_json'
 require 'nokogiri'
 require 'open-uri'
@@ -8,8 +10,6 @@ require 'time'
 require 'search_solr_tools'
 require_relative '../helpers/iso_namespaces'
 require_relative '../helpers/solr_format'
-require_relative '../helpers/iso_to_solr'
-
 
 module SearchSolrTools
   module Harvesters
@@ -21,7 +21,7 @@ module SearchSolrTools
       XML_CONTENT_TYPE = 'text/xml; charset=utf-8'
       JSON_CONTENT_TYPE = 'application/json; charset=utf-8'
 
-      def initialize(env = 'development', die_on_failure = false)
+      def initialize(env = 'development', die_on_failure: false)
         @environment = env
         @die_on_failure = die_on_failure
       end
@@ -52,7 +52,7 @@ module SearchSolrTools
             success = response.code == 200
             puts "Error in ping request: #{response.body}" unless success
           end
-        rescue => e
+        rescue StandardError => e
           puts "Rest exception while pinging Solr: #{e}"
         end
         success
@@ -62,7 +62,7 @@ module SearchSolrTools
       # to "ping" the data center.  Returns true if the ping is successful (or, as
       # in this default, no ping method was defined)
       def ping_source
-        puts "Harvester does not have ping method defined, assuming true"
+        puts 'Harvester does not have ping method defined, assuming true'
         true
       end
 
@@ -75,12 +75,12 @@ module SearchSolrTools
         harvest_status
       end
 
-      def delete_old_documents(timestamp, constraints, solr_core, force = false)
+      def delete_old_documents(timestamp, constraints, solr_core, force: false)
         constraints = sanitize_data_centers_constraints(constraints)
         delete_query = "last_update:[* TO #{timestamp}] AND #{constraints}"
         solr = RSolr.connect url: solr_url + "/#{solr_core}"
         unchanged_count = (solr.get 'select', params: { wt: :ruby, q: delete_query, rows: 0 })['response']['numFound'].to_i
-        if unchanged_count == 0
+        if unchanged_count.zero?
           puts "All documents were updated after #{timestamp}, nothing to delete"
         else
           puts "Begin removing documents older than #{timestamp}"
@@ -91,8 +91,8 @@ module SearchSolrTools
       def sanitize_data_centers_constraints(query_string)
         # Remove lucene special characters, preserve the query parameter and compress whitespace
         query_string.gsub!(/[:&|!~\-\(\)\{\}\[\]\^\*\?\+]+/, ' ')
-        query_string.gsub!(/data_centers /, 'data_centers:')
-        query_string.gsub!(/source /, 'source:')
+        query_string.gsub!('data_centers ', 'data_centers:')
+        query_string.gsub!('source ', 'source:')
         query_string.squeeze(' ').strip
       end
 
@@ -127,7 +127,7 @@ module SearchSolrTools
         status
       end
 
-      # TODO Need to return a specific type of failure:
+      # TODO: Need to return a specific type of failure:
       #   - Bad record content identified and no ingest attempted
       #   - Solr tries to ingest document and fails (bad content not detected prior to ingest)
       #   - Solr cannot insert document for reasons other than the document structure and content.
@@ -143,15 +143,15 @@ module SearchSolrTools
 
         # Some docs will cause solr to time out during the POST
         begin
-          RestClient.post(url, doc_serialized, content_type: content_type) do |response, _request, _result|
+          RestClient.post(url, doc_serialized, content_type:) do |response, _request, _result|
             success = response.code == 200
             unless success
               puts "Error for #{doc_serialized}\n\n response: #{response.body}"
               status = Helpers::HarvestStatus::INGEST_ERR_SOLR_ERROR
             end
           end
-        rescue => e
-          # TODO Need to provide more detail re: this failure so we know whether to
+        rescue StandardError => e
+          # TODO: Need to provide more detail re: this failure so we know whether to
           #  exit the job with a status != 0
           puts "Rest exception while POSTing to Solr: #{e}, for doc: #{doc_serialized}"
           status = Helpers::HarvestStatus::INGEST_ERR_SOLR_ERROR
@@ -161,11 +161,11 @@ module SearchSolrTools
 
       def get_serialized_doc(doc, content_type)
         if content_type.eql?(XML_CONTENT_TYPE)
-          return doc.respond_to?(:to_xml) ? doc.to_xml : doc
+          doc.respond_to?(:to_xml) ? doc.to_xml : doc
         elsif content_type.eql?(JSON_CONTENT_TYPE)
-          return MultiJson.dump(doc)
+          MultiJson.dump(doc)
         else
-          return doc
+          doc
         end
       end
 
@@ -178,17 +178,18 @@ module SearchSolrTools
 
         begin
           puts "Request: #{request_url}"
-          response = URI.open(request_url, read_timeout: timeout, 'Content-Type' => content_type)
+          response = URI.parse(request_url).open(read_timeout: timeout, 'Content-Type' => content_type)
         rescue OpenURI::HTTPError, Timeout::Error, Errno::ETIMEDOUT => e
           retries_left -= 1
           puts "## REQUEST FAILED ## #{e.class} ## Retrying #{retries_left} more times..."
 
-          retry if retries_left > 0
+          retry if retries_left.positive?
 
-          # TODO - Do we really need this "die_on_failure" anymore?  The empty return
+          # TODO: Do we really need this "die_on_failure" anymore?  The empty return
           #  will cause the "No Documents" error to be thrown in the harvester class
           #  now, so it will pretty much always "die on failure"
           raise e if @die_on_failure
+
           return
         end
         doc = Nokogiri.XML(response)
@@ -216,7 +217,7 @@ module SearchSolrTools
         spatial_coverages = doc.xpath(".//field[@name='spatial_coverages']").first
         return true if spatial_coverages.nil?
 
-        spatial_coverages = spatial_coverages.text.split(' ')
+        spatial_coverages = spatial_coverages.text.split
 
         # We've only seen the failure with 4 spatial coverage values
         return true if spatial_coverages.size < 4

data/lib/search_solr_tools/harvesters/nsidc_auto_suggest.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'auto_suggest'
 
 module SearchSolrTools
@@ -16,11 +18,11 @@ module SearchSolrTools
       def fields
         {
           'authoritative_id' => { weight: 1, source: 'NSIDC', creator: method(:standard_add_creator) },
-          'full_title' => { weight: 2, source: 'NSIDC', creator: method(:standard_add_creator) },
-          'copy_parameters' => { weight: 5, source: 'NSIDC', creator: method(:standard_add_creator) },
-          'full_platforms' => { weight: 2, source: 'NSIDC', creator: method(:short_full_split_add_creator) },
-          'full_sensors' => { weight: 2, source: 'NSIDC', creator: method(:short_full_split_add_creator) },
-          'full_authors' => { weight: 1, source: 'NSIDC', creator: method(:standard_add_creator) }
+          'full_title'       => { weight: 2, source: 'NSIDC', creator: method(:standard_add_creator) },
+          'copy_parameters'  => { weight: 5, source: 'NSIDC', creator: method(:standard_add_creator) },
+          'full_platforms'   => { weight: 2, source: 'NSIDC', creator: method(:short_full_split_add_creator) },
+          'full_sensors'     => { weight: 2, source: 'NSIDC', creator: method(:short_full_split_add_creator) },
+          'full_authors'     => { weight: 1, source: 'NSIDC', creator: method(:standard_add_creator) }
         }
       end
 

data/lib/search_solr_tools/harvesters/nsidc_json.rb

@@ -1,15 +1,16 @@
+# frozen_string_literal: true
+
 require 'json'
 require 'rest-client'
 
 require 'search_solr_tools'
 
-
 module SearchSolrTools
   module Harvesters
     # Harvests data from NSIDC OAI and inserts it into Solr after it has been translated
     class NsidcJson < Base
-      def initialize(env = 'development', die_on_failure = false)
-        super env, die_on_failure
+      def initialize(env = 'development', die_on_failure: false)
+        super
         @translator = Translators::NsidcJsonToSolr.new
         Helpers::FacetConfiguration.import_bin_configuration(env)
       end
@@ -19,7 +20,7 @@ module SearchSolrTools
           RestClient.options(nsidc_json_url) do |response, _request, _result|
             return response.code == 200
           end
-        rescue => e
+        rescue StandardError
           puts "Error trying to get options for #{nsidc_json_url} (ping)"
         end
         false
@@ -37,7 +38,7 @@ module SearchSolrTools
 
         status = insert_solr_docs result[:add_docs], Base::JSON_CONTENT_TYPE
 
-        status.record_status(Helpers::HarvestStatus::HARVEST_NO_DOCS) if result[:num_docs] == 0
+        status.record_status(Helpers::HarvestStatus::HARVEST_NO_DOCS) if (result[:num_docs]).zero?
 
         # Record the number of harvest failures; note that if this is 0, thats OK, the status will stay at 0
         status.record_status(Helpers::HarvestStatus::HARVEST_FAILURE, result[:failure_ids].length)
@@ -66,7 +67,7 @@ module SearchSolrTools
       # @param id [String] NSIDC authoritative ID for the dataset
       # @return [Hash] Parsed version of the JSON response
       def fetch_json_from_nsidc(id)
-        json_response = RestClient.get(nsidc_json_url + id + '.json')
+        json_response = RestClient.get("#{nsidc_json_url}#{id}.json")
         JSON.parse(json_response)
       end
 
@@ -81,13 +82,13 @@ module SearchSolrTools
           id = r.text.split('/').last
           begin
             docs << { 'add' => { 'doc' => @translator.translate(fetch_json_from_nsidc(id)) } }
-          rescue => e
+          rescue StandardError => e
             puts "Failed to fetch #{id} with error #{e}: #{e.backtrace}"
             failure_ids << id
           end
         end
 
-        { num_docs: all_docs.size, add_docs: docs, failure_ids: failure_ids }
+        { num_docs: all_docs.size, add_docs: docs, failure_ids: }
       end
     end
   end

data/lib/search_solr_tools/helpers/bounding_box_util.rb

@@ -1,4 +1,6 @@
-require_relative './iso_namespaces'
+# frozen_string_literal: true
+
+require_relative 'iso_namespaces'
 
 module SearchSolrTools
   module Helpers
@@ -8,12 +10,10 @@ module SearchSolrTools
       NORTHERN_GLOBAL_BOUNDARY = 85.0
 
       def self.bounding_box_hash_from_geo_json(geometry)
-        if geometry_is_point?(geometry)
-          return { west: geometry.x.to_s, south: geometry.y.to_s, east: geometry.x.to_s, north: geometry.y.to_s }
-        else
-          bbox = RGeo::Cartesian::BoundingBox.create_from_geometry(geometry)
-          return { west: bbox.min_x.to_s, south: bbox.min_y.to_s, east: bbox.max_x.to_s, north: bbox.max_y.to_s }
-        end
+        return { west: geometry.x.to_s, south: geometry.y.to_s, east: geometry.x.to_s, north: geometry.y.to_s } if geometry_is_point?(geometry)
+
+        bbox = RGeo::Cartesian::BoundingBox.create_from_geometry(geometry)
+        { west: bbox.min_x.to_s, south: bbox.min_y.to_s, east: bbox.max_x.to_s, north: bbox.max_y.to_s }
       end
 
       def self.geometry_is_point?(geometry)
@@ -30,7 +30,7 @@ module SearchSolrTools
       end
 
       def self.box_invalid?(box)
-        [:north, :south, :east, :west].any? { |d| box[d].to_s.empty? }
+        %i[north south east west].any? { |d| box[d].to_s.empty? }
       end
     end
   end

data/lib/search_solr_tools/helpers/facet_configuration.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'json'
 require 'rest_client'
 require 'singleton'
@@ -8,7 +10,7 @@ module SearchSolrTools
     class FacetConfiguration
       include Singleton
       def self.import_bin_configuration(env)
-        @bin_configuration = JSON.parse(RestClient.get(SolrEnvironments[env][:nsidc_dataset_metadata_url] + 'binConfiguration')) if @bin_configuration.nil?
+        @bin_configuration = JSON.parse(RestClient.get("#{SolrEnvironments[env][:nsidc_dataset_metadata_url]}binConfiguration")) if @bin_configuration.nil?
       end
 
       def self.get_facet_bin(facet_name)

data/lib/search_solr_tools/helpers/harvest_status.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module SearchSolrTools
   module Helpers
     class HarvestStatus
@@ -7,17 +9,17 @@ module SearchSolrTools
       INGEST_ERR_INVALID_DOC = :invalid
       INGEST_ERR_SOLR_ERROR = :solr_error
       OTHER_ERROR = :other
-      PING_SOLR = :ping_solr  # used for initialize only
-      PING_SOURCE = :ping_source  # used for initialize only
+      PING_SOLR = :ping_solr # used for initialize only
+      PING_SOURCE = :ping_source # used for initialize only
 
-      ERROR_STATUS = [HARVEST_NO_DOCS, HARVEST_FAILURE, INGEST_ERR_INVALID_DOC, INGEST_ERR_SOLR_ERROR, OTHER_ERROR]
+      ERROR_STATUS = [HARVEST_NO_DOCS, HARVEST_FAILURE, INGEST_ERR_INVALID_DOC, INGEST_ERR_SOLR_ERROR, OTHER_ERROR].freeze
 
-      attr_reader :status, :ping_solr, :ping_source
-      attr_writer :ping_solr, :ping_source
+      attr_accessor :ping_solr, :ping_source
+      attr_reader :status
 
       # init_info is an optional hash that contains the various status keys and the documents to
       # associate with them
-      def initialize(init_info={})
+      def initialize(init_info = {})
         @status = { INGEST_OK => 0 }
         @ping_solr = true
         @ping_source = true
@@ -36,9 +38,9 @@ module SearchSolrTools
       end
 
       def ok?
-        ERROR_STATUS.each { |s| return false unless @status[s] == 0 }
+        ERROR_STATUS.each { |s| return false unless (@status[s]).zero? }
         @ping_solr && @ping_source
       end
     end
   end
-end
+end

data/lib/search_solr_tools/helpers/iso_namespaces.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module SearchSolrTools
   module Helpers
     # Helper class to provide default namespaces for XML document parsing.
@@ -25,7 +27,7 @@ module SearchSolrTools
         'srv'    => 'http://www.isotc211.org/2005/srv',
         'xlink'  => 'http://www.w3.org/1999/xlink',
         'xsi'    => 'http://www.w3.org/2001/XMLSchema-instance'
-      }
+      }.freeze
     end
   end
 end