Merge pull request #2313 from broadinstitute/jb-azul-entity-update

Switching back to GET for Azul entity queries (SCP-6059)

Author: bistline

app/controllers/api/v1/bulk_download_controller.rb

@@ -162,8 +162,10 @@ def summary
         end
 
         @study_file_info = ::BulkDownloadService.get_download_info(valid_accessions)
-        hca_file_info = ::AzulSearchService.get_file_summary_info(hca_accessions)
-        @study_file_info += hca_file_info if hca_file_info.any?
+        if hca_accessions.present?
+          hca_file_info = ::AzulSearchService.get_file_summary_info(hca_accessions)
+          @study_file_info += hca_file_info if hca_file_info.any?
+        end
 
         render json: @study_file_info
       end

app/models/hca_azul_client.rb

@@ -15,7 +15,7 @@ class HcaAzulClient
   MANIFEST_FORMATS = %w[compact full terra.bdbag terra.pfb curl].freeze
 
   # maximum number of results to return
-  MAX_RESULTS = 80
+  MAX_RESULTS = 75
 
   # maximum length of query string (in characters) for requests
   MAX_QUERY_LENGTH = 8192
@@ -184,8 +184,9 @@ def catalogs
   #   - (ArgumentError) => if catalog is not in self.all_catalogs
   def projects(catalog: nil, query: {}, size: MAX_RESULTS)
     base_path = "#{api_root}/index/projects?size=#{size}"
+    base_path += "&filters=#{format_hash_as_query_string(query)}"
     path = append_catalog(base_path, catalog)
-    process_api_request(:post, path, payload: create_query_filters(query))
+    process_api_request(:get, path)
   end
 
   # simulate OR logic by splitting project queries on facet and joining results
@@ -280,10 +281,10 @@ def project_manifest_link(project_id, catalog: nil, format: 'compact')
   #   - (Hash) => List of files matching query
   def files(catalog: nil, query: {}, size: MAX_RESULTS)
     base_path = "#{api_root}/index/files?size=#{size}"
-    payload = create_query_filters(query)
+    base_path += "&filters=#{format_hash_as_query_string(query)}"
     path = append_catalog(base_path, catalog)
     # make API request, but fold in project information to each result so that this is preserved for later use
-    raw_results = process_api_request(:post, path, payload:)['hits']
+    raw_results = process_api_request(:get, path)['hits']
     results = []
     raw_results.each do |result|
       files = result['files']
@@ -404,6 +405,19 @@ def merge_query_objects(*query_objects)
     merged_query
   end
 
+  # take a Hash/JSON object and format as a query string parameter
+  #
+  # * *params*
+  #   - +query_params+ (Hash) => Hash of query parameters
+  #
+  # * *returns*
+  #   - (String) => URL-encoded string version of query parameters
+  def format_hash_as_query_string(query_params)
+    # replace Ruby => assignment operators with JSON standard colons (:)
+    sanitized_params = query_params.to_s.gsub(/=>/, ':')
+    CGI.escape(sanitized_params)
+  end
+
   # create a query filter object to use in a request body
   #
   # * *params*

test/integration/external/hca_azul_client_test.rb

@@ -163,9 +163,16 @@ def get_entries_from_response(response, key)
     assert_equal keys, file.keys.sort & keys
   end
 
-  # TODO: SCP-5564
-  #   - Fix "Global bulk download breaks in default use"
-  #   - Re-enable this test
+  test 'should format object for query string' do
+    query_object = { 'foo' => { 'bar' => 'baz' } }
+    expected_response = '%7B%22foo%22+%3A+%7B%22bar%22+%3A+%22baz%22%7D%7D'
+    formatted_query = @hca_azul_client.format_hash_as_query_string(query_object)
+    assert_equal expected_response, formatted_query
+    # assert we can get back to original object, but as JSON (removing whitespace that breaks comparison)
+    query_as_json = CGI.unescape(formatted_query).gsub(/\s/, '')
+    assert_equal query_object.to_json, query_as_json
+  end
+
   test 'should get HCA metadata tsv link' do
     skip_if_api_down
     manifest_info = @hca_azul_client.project_manifest_link(@project_id)