remove solr dependency. #50

Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (1.0.32)
+    bolognese (1.0.33)
       activesupport (>= 4.2.5, < 6)
       benchmark_methods (~> 0.7)
       bibtex-ruby (~> 4.1)

lib/bolognese/doi_utils.rb

@@ -17,9 +17,9 @@ def doi_resolver(doi, options = {})
       sandbox.present? || options[:sandbox] ? "https://handle.test.datacite.org/" : "https://doi.org/"
     end
 
-    def doi_search(doi, options = {})
+    def doi_api_url(doi, options = {})
       sandbox = Array(/handle.test.datacite.org/.match(doi)).last
-      sandbox.present? || options[:sandbox] ? "https://search.test.datacite.org/api" : "https://search.datacite.org/api"
+      sandbox.present? || options[:sandbox] ? "https://api.test.datacite.org/dois/" + doi_from_url(doi) : "https://api.datacite.org/dois/" + doi_from_url(doi)
     end
 
     def normalize_doi(doi, options = {})
@@ -46,10 +46,10 @@ def get_doi_ra(doi)
       prefix = validate_prefix(doi)
       return nil if prefix.blank?
 
-      url = "https://api.datacite.org/prefixes/#{prefix}"
+      url = "https://doi.org/ra/#{prefix}"
       result = Maremma.get(url)
 
-      result.body.fetch("data", {}).fetch('attributes', {}).fetch('registration-agency', nil)
+      result.body.dig("data", 0, "RA")
     end
   end
 end

lib/bolognese/metadata.rb

@@ -27,7 +27,8 @@ def initialize(input: nil, from: nil, **options)
       elsif input.present? && File.exist?(input)
         ext = File.extname(input)
         if %w(.bib .ris .xml .json).include?(ext)
-          hsh = { "url" => options[:url],
+          hsh = { 
+            "url" => options[:url],
             "state" => options[:state],
             "date_registered" => options[:date_registered],
             "date_updated" => options[:date_updated],
@@ -41,18 +42,19 @@ def initialize(input: nil, from: nil, **options)
           exit 1
         end
       else
-        hsh = { "url" => options[:url],
-                "state" => options[:state],
-                "date_registered" => options[:date_registered],
-                "date_updated" => options[:date_updated],
-                "provider_id" => options[:provider_id],
-                "client_id" => options[:client_id],
-                "content_url" => options[:content_url],
-                "creators" => options[:creators],
-                "contributors" => options[:contributors],
-                "titles" => options[:titles],
-                "publisher" => options[:publisher],
-                "publication_year" => options[:publication_year] }
+        hsh = { 
+          "url" => options[:url],
+          "state" => options[:state],
+          "date_registered" => options[:date_registered],
+          "date_updated" => options[:date_updated],
+          "provider_id" => options[:provider_id],
+          "client_id" => options[:client_id],
+          "content_url" => options[:content_url],
+          "creators" => options[:creators],
+          "contributors" => options[:contributors],
+          "titles" => options[:titles],
+          "publisher" => options[:publisher],
+          "publication_year" => options[:publication_year] }
         string = input
         @from = from || find_from_format(string: string)
       end

lib/bolognese/readers/datacite_reader.rb

@@ -6,15 +6,9 @@ module DataciteReader
       def get_datacite(id: nil, **options)
         return { "string" => nil, "state" => "not_found" } unless id.present?
 
-        search_url = doi_search(id, options)
-        doi = doi_from_url(id)
-        params = { q: "doi:#{doi}",
-                   fl: "doi,url,xml,state,allocator_symbol,datacentre_symbol,media,minted,updated",
-                   wt: "json" }
-        search_url += "?" + URI.encode_www_form(params)
-
-        response = Maremma.get search_url
-        attributes = response.body.dig("data", "response", "docs").first
+        api_url = doi_api_url(id, options)
+        response = Maremma.get api_url
+        attributes = response.body.dig("data", "attributes")
         return { "string" => nil, "state" => "not_found" } unless attributes.present?
 
         string = attributes.fetch('xml', nil)
@@ -34,17 +28,21 @@ def get_datacite(id: nil, **options)
           string = doc.to_xml(:indent => 2)
         end
 
-        content_url = Array.wrap(attributes.fetch("media", nil)).map do |media|
-          media.split(":", 2).last
+        client = Array.wrap(response.body.fetch("included", nil)).find { |m| m["type"] == "clients" }
+        client_id = client.to_h.fetch("id", nil)
+        provider_id = Array.wrap(client.to_h.fetch("relationships", nil)).find { |m| m["provider"].present? }.to_h.dig("provider", "data", "id")
+
+        content_url = attributes.fetch("contentUrl", nil) || Array.wrap(response.body.fetch("included", nil)).select { |m| m["type"] == "media" }.map do |m|
+          m.dig("attributes", "url")
         end.compact
 
         { "string" => string,
           "url" => attributes.fetch("url", nil),
           "state" => attributes.fetch("state", nil),
-          "date_registered" => attributes.fetch("minted", nil),
+          "date_registered" => attributes.fetch("registered", nil),
           "date_updated" => attributes.fetch("updated", nil),
-          "provider_id" => attributes.fetch("allocator_symbol", nil),
-          "client_id" => attributes.fetch("datacentre_symbol", nil),
+          "provider_id" => provider_id,
+          "client_id" => client_id,
           "content_url" => content_url }
       end
 

lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "1.0.32"
+  VERSION = "1.0.33"
 end

spec/doi_utils_spec.rb

@@ -51,47 +51,47 @@
     end
   end
 
-  context "doi search" do
+  context "doi_api_url" do
     it "doi" do
       doi = "10.5061/DRYAD.8515"
-      response = subject.doi_search(doi)
-      expect(response).to eq("https://search.datacite.org/api")
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.datacite.org/dois/10.5061/dryad.8515")
     end
 
     it "doi with protocol" do
       doi = "doi:10.5061/DRYAD.8515"
-      response = subject.doi_search(doi)
-      expect(response).to eq("https://search.datacite.org/api")
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.datacite.org/dois/10.5061/dryad.8515")
     end
 
     it "https url" do
       doi = "https://doi.org/10.5061/dryad.8515"
-      response = subject.doi_search(doi)
-      expect(response).to eq("https://search.datacite.org/api")
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.datacite.org/dois/10.5061/dryad.8515")
     end
 
     it "dx.doi.org url" do
       doi = "http://dx.doi.org/10.5061/dryad.8515"
-      response = subject.doi_search(doi)
-      expect(response).to eq("https://search.datacite.org/api")
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.datacite.org/dois/10.5061/dryad.8515")
     end
 
     it "test resolver" do
       doi = "https://handle.test.datacite.org/10.5061/dryad.8515"
-      response = subject.doi_search(doi)
-      expect(response).to eq("https://search.test.datacite.org/api")
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.test.datacite.org/dois/10.5061/dryad.8515")
     end
 
     it "test resolver http" do
       doi = "http://handle.test.datacite.org/10.5061/dryad.8515"
-      response = subject.doi_search(doi)
-      expect(response).to eq("https://search.test.datacite.org/api")
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.test.datacite.org/dois/10.5061/dryad.8515")
     end
 
     it "force test resolver" do
       doi = "https://doi.org/10.5061/dryad.8515"
-      response = subject.doi_search(doi, sandbox: true)
-      expect(response).to eq("https://search.test.datacite.org/api")
+      response = subject.doi_api_url(doi, sandbox: true)
+      expect(response).to eq("https://api.test.datacite.org/dois/10.5061/dryad.8515")
     end
   end