Merge pull request #162 from nulib/getty-relevancy-ordering

Fix Getty search relevancy issue

lib/authoritex/getty/aat.ex

@@ -27,11 +27,15 @@ defmodule Authoritex.Getty.AAT do
            skos:inScheme <http://vocab.getty.edu/aat/> ;
            gvp:prefLabelGVP [skosxl:literalForm ?name] .
         FILTER (#{sparql_search_filter(q)}) .
-      } LIMIT #{max_results}
+      } ORDER BY #{sparql_order_clause(q)} LIMIT #{max_results}
       """
     end
   end
 
+  defp sparql_order_clause(q) do
+    ~s{DESC(IF(REGEX(?name, "^#{q}$", "i"), 2, IF(REGEX(?name, "^#{q}", "i"), 1, 0)))}
+  end
+
   defp sparql_search_filter(q) do
     String.split(q)
     |> Enum.map_join(" && ", &~s{regex(?name, "#{&1}", "i")})

lib/authoritex/getty/base.ex

@@ -37,7 +37,7 @@ defmodule Authoritex.Getty.Base do
             {:error, 404}
 
           {:ok, [%{replaced_by: replaced_by}] = result} when replaced_by != "" ->
-            Logger.warn("#{id} is obsolete. Fetching replacement term #{replaced_by}.")
+            Logger.warning("#{id} is obsolete. Fetching replacement term #{replaced_by}.")
             fetch(replaced_by)
 
           {:ok, [result]} ->

lib/authoritex/getty/tgn.ex

@@ -29,11 +29,15 @@ defmodule Authoritex.Getty.TGN do
           gvp:prefLabelGVP [skosxl:literalForm ?name] ;
           gvp:parentString ?hint .
         FILTER (#{sparql_search_filter(q)}) .
-      } LIMIT #{max_results}
+      } ORDER BY #{sparql_order_clause(q)} LIMIT #{max_results}
       """
     end
   end
 
+  defp sparql_order_clause(q) do
+    ~s{DESC(IF(REGEX(?name, "^#{q}$", "i"), 2, IF(REGEX(?name, "^#{q}", "i"), 1, 0)))}
+  end
+
   defp sparql_search_filter(q) do
     String.split(q)
     |> Enum.map_join(" && ", &~s{regex(?name, "#{&1}", "i")})

lib/authoritex/getty/ulan.ex

@@ -31,11 +31,15 @@ defmodule Authoritex.Getty.ULAN do
             foaf:focus/gvp:biographyPreferred [schema:description ?hint] ;
             skos:altLabel ?alt .
         FILTER (#{sparql_search_filter(q)}) .
-      } LIMIT #{max_results}
+      } ORDER BY #{sparql_order_clause(q)} LIMIT #{max_results}
       """
     end
   end
 
+  defp sparql_order_clause(q) do
+    ~s{DESC(IF(REGEX(?name, "^#{q}$", "i"), 2, IF(REGEX(?name, "^#{q}", "i"), 1, 0)))}
+  end
+
   defp sparql_search_filter(q) do
     if String.contains?(q, " ") do
       String.split(q)

mix.exs

@@ -1,7 +1,7 @@
 defmodule Authoritex.MixProject do
   use Mix.Project
 
-  @version "1.0.3"
+  @version "1.1.0"
   @url "https://github.com/nulib/authoritex"
 
   def project do

test/authoritex/geonames_test.exs

@@ -6,14 +6,14 @@ defmodule Authoritex.GeoNamesTest do
     code: "geonames",
     description: "GeoNames geographical database",
     test_uris: [
-      "https://sws.geonames.org/4302561/"
+      "https://sws.geonames.org/4299276/"
     ],
-    bad_uri: "https://sws.geonames.org/43025619/",
+    bad_uri: "https://sws.geonames.org/42992769/",
     expected: [
       hint: "Kentucky, United States",
-      id: "https://sws.geonames.org/4302561/",
-      label: "Nicholasville",
-      qualified_label: "Nicholasville, Kentucky, United States",
+      id: "https://sws.geonames.org/4299276/",
+      label: "Louisville",
+      qualified_label: "Louisville, Kentucky, United States",
       variants: []
     ],
     search_result_term: "Kentucky",
@@ -28,9 +28,9 @@ defmodule Authoritex.GeoNamesTest do
     end
 
     test "proper URI form" do
-      assert Authoritex.fetch("https://sws.geonames.org/4302561") == {:error, 404}
-      assert Authoritex.fetch("http://sws.geonames.org/4302561") == {:error, :unknown_authority}
-      assert Authoritex.fetch("http://sws.geonames.org/4302561/") == {:error, :unknown_authority}
+      assert Authoritex.fetch("https://sws.geonames.org/4299276") == {:error, 404}
+      assert Authoritex.fetch("http://sws.geonames.org/4299276") == {:error, :unknown_authority}
+      assert Authoritex.fetch("http://sws.geonames.org/4299276/") == {:error, :unknown_authority}
     end
 
     test "search" do

test/authoritex/getty/aat_test.exs

@@ -15,7 +15,7 @@ defmodule Authoritex.Getty.AATTest do
       label: "dollars (paper money)",
       qualified_label: "dollars (paper money)",
       hint: nil,
-      variants: ["dollar (paper money)", "dollar bills", "dollar bill"]
+      variants: ["dollar (paper money)", "dollar bills", "dollar bill", "Dollars"]
     ],
     search_result_term: "dollars",
     search_count_term: "paint"

test/authoritex/getty_test.exs

@@ -27,7 +27,7 @@ defmodule Authoritex.GettyTest do
                     label: "dollars (paper money)",
                     qualified_label: "dollars (paper money)",
                     hint: nil,
-                    variants: ["dollar (paper money)", "dollar bills", "dollar bill"]
+                    variants: ["dollar (paper money)", "dollar bills", "dollar bill", "Dollars"]
                   }}
 
         assert Getty.fetch("http://vocab.getty.edu/tgn/2236134") ==

test/authoritex/homosaurus_test.exs

@@ -16,6 +16,6 @@ defmodule Authoritex.HomosaurusTest do
     ],
     search_result_term: "adop",
     search_count_term: "adop",
-    default_results: 10,
-    explicit_results: 10
+    default_results: 50,
+    explicit_results: 50
 end

test/authoritex_test.exs

@@ -47,16 +47,20 @@ defmodule AuthoritexTest do
   end
 
   describe "search/2" do
-    test "results" do
-      use_cassette "authoritex_search_results", match_requests_on: [:query] do
+    test "result count" do
+      use_cassette "authoritex_search_count", match_requests_on: [:query] do
         with {:ok, results} <- Authoritex.search("lcnaf", "smith") do
           assert length(results) == 30
         end
 
         with {:ok, results} <- Authoritex.search("lcnaf", "smith", 50) do
           assert length(results) == 50
         end
+      end
+    end
 
+    test "expected result" do
+      use_cassette "authoritex_search_results", match_requests_on: [:query] do
         with {:ok, results} <- Authoritex.search("lcnaf", "valim") do
           assert Enum.member?(results, %{
                    id: "http://id.loc.gov/authorities/names/no2011087251",

test/fixtures/vcr_cassettes/aat_fetch_failure.json

@@ -1,21 +1,22 @@
 [
   {
     "request": {
+      "options": [],
       "body": "",
+      "url": "http://vocab.getty.edu/sparql.xml?query=SELECT+DISTINCT+%3Fs+%3Fname+%3FreplacedBy+%28group_concat%28%3Falt%3B+separator%3D%22%7C%22%29+AS+%3Fvariants%29+%7B+BIND%28%3Chttp%3A%2F%2Fvocab.getty.edu%2Faat%2Fwrong-id%3E+as+%3Fs%29+OPTIONAL+%7B%3Fs+gvp%3AprefLabelGVP%2Fxl%3AliteralForm+%3Fname%7D+OPTIONAL+%7B%3Fs+dcterms%3AisReplacedBy+%3FreplacedBy%7D+OPTIONAL+%7B%3Fs+xl%3AaltLabel%2Fxl%3AliteralForm+%3Falt%7D+%7D+GROUP+BY+%3Fs+%3Fname+%3FreplacedBy+LIMIT+1",
       "headers": {
         "Accept": "application/sparql-results+xml;charset=UTF-8",
         "User-Agent": "Authoritex"
       },
       "method": "get",
-      "options": [],
-      "request_body": "",
-      "url": "http://vocab.getty.edu/sparql.xml?query=SELECT+DISTINCT+%3Fs+%3Fname+%3FreplacedBy+%28group_concat%28%3Falt%3B+separator%3D%22%7C%22%29+AS+%3Fvariants%29+%7B+BIND%28%3Chttp%3A%2F%2Fvocab.getty.edu%2Faat%2Fwrong-id%3E+as+%3Fs%29+OPTIONAL+%7B%3Fs+gvp%3AprefLabelGVP%2Fxl%3AliteralForm+%3Fname%7D+OPTIONAL+%7B%3Fs+dcterms%3AisReplacedBy+%3FreplacedBy%7D+OPTIONAL+%7B%3Fs+xl%3AaltLabel%2Fxl%3AliteralForm+%3Falt%7D+%7D+GROUP+BY+%3Fs+%3Fname+%3FreplacedBy+LIMIT+1"
+      "request_body": ""
     },
     "response": {
       "binary": false,
+      "type": "ok",
       "body": "<?xml version='1.0' encoding='UTF-8'?>\n<sparql xmlns='http://www.w3.org/2005/sparql-results#'>\n\t<head>\n\t\t<variable name='s'/>\n\t\t<variable name='name'/>\n\t\t<variable name='replacedBy'/>\n\t\t<variable name='variants'/>\n\t</head>\n\t<results>\n\t\t<result>\n\t\t\t<binding name='s'>\n\t\t\t\t<uri>http://vocab.getty.edu/aat/wrong-id</uri>\n\t\t\t</binding>\n\t\t\t<binding name='variants'>\n\t\t\t\t<literal></literal>\n\t\t\t</binding>\n\t\t</result>\n\t</results>\n</sparql>\n",
       "headers": {
-        "Date": "Tue, 18 Apr 2023 19:54:04 GMT",
+        "Date": "Thu, 02 May 2024 16:28:43 GMT",
         "Content-Type": "application/sparql-results+xml;charset=utf-8",
         "Transfer-Encoding": "chunked",
         "Connection": "keep-alive",
@@ -24,8 +25,7 @@
         "Content-Disposition": "attachment; filename=\"sparql.xml\"",
         "Content-Language": "en"
       },
-      "status_code": 200,
-      "type": "ok"
+      "status_code": 200
     }
   }
 ]

test/fixtures/vcr_cassettes/aat_fetch_success.json

@@ -1,21 +1,22 @@
 [
   {
     "request": {
+      "options": [],
       "body": "",
+      "url": "http://vocab.getty.edu/sparql.xml?query=SELECT+DISTINCT+%3Fs+%3Fname+%3FreplacedBy+%28group_concat%28%3Falt%3B+separator%3D%22%7C%22%29+AS+%3Fvariants%29+%7B+BIND%28%3Chttp%3A%2F%2Fvocab.getty.edu%2Faat%2F300265149%3E+as+%3Fs%29+OPTIONAL+%7B%3Fs+gvp%3AprefLabelGVP%2Fxl%3AliteralForm+%3Fname%7D+OPTIONAL+%7B%3Fs+dcterms%3AisReplacedBy+%3FreplacedBy%7D+OPTIONAL+%7B%3Fs+xl%3AaltLabel%2Fxl%3AliteralForm+%3Falt%7D+%7D+GROUP+BY+%3Fs+%3Fname+%3FreplacedBy+LIMIT+1",
       "headers": {
         "Accept": "application/sparql-results+xml;charset=UTF-8",
         "User-Agent": "Authoritex"
       },
       "method": "get",
-      "options": [],
-      "request_body": "",
-      "url": "http://vocab.getty.edu/sparql.xml?query=SELECT+DISTINCT+%3Fs+%3Fname+%3FreplacedBy+%28group_concat%28%3Falt%3B+separator%3D%22%7C%22%29+AS+%3Fvariants%29+%7B+BIND%28%3Chttp%3A%2F%2Fvocab.getty.edu%2Faat%2F300265149%3E+as+%3Fs%29+OPTIONAL+%7B%3Fs+gvp%3AprefLabelGVP%2Fxl%3AliteralForm+%3Fname%7D+OPTIONAL+%7B%3Fs+dcterms%3AisReplacedBy+%3FreplacedBy%7D+OPTIONAL+%7B%3Fs+xl%3AaltLabel%2Fxl%3AliteralForm+%3Falt%7D+%7D+GROUP+BY+%3Fs+%3Fname+%3FreplacedBy+LIMIT+1"
+      "request_body": ""
     },
     "response": {
       "binary": false,
-      "body": "<?xml version='1.0' encoding='UTF-8'?>\n<sparql xmlns='http://www.w3.org/2005/sparql-results#'>\n\t<head>\n\t\t<variable name='s'/>\n\t\t<variable name='name'/>\n\t\t<variable name='replacedBy'/>\n\t\t<variable name='variants'/>\n\t</head>\n\t<results>\n\t\t<result>\n\t\t\t<binding name='s'>\n\t\t\t\t<uri>http://vocab.getty.edu/aat/300265149</uri>\n\t\t\t</binding>\n\t\t\t<binding name='name'>\n\t\t\t\t<literal xml:lang='en'>dollars (paper money)</literal>\n\t\t\t</binding>\n\t\t\t<binding name='variants'>\n\t\t\t\t<literal>dollar (paper money)|dollar bills|dollar bill</literal>\n\t\t\t</binding>\n\t\t</result>\n\t</results>\n</sparql>\n",
+      "type": "ok",
+      "body": "<?xml version='1.0' encoding='UTF-8'?>\n<sparql xmlns='http://www.w3.org/2005/sparql-results#'>\n\t<head>\n\t\t<variable name='s'/>\n\t\t<variable name='name'/>\n\t\t<variable name='replacedBy'/>\n\t\t<variable name='variants'/>\n\t</head>\n\t<results>\n\t\t<result>\n\t\t\t<binding name='s'>\n\t\t\t\t<uri>http://vocab.getty.edu/aat/300265149</uri>\n\t\t\t</binding>\n\t\t\t<binding name='name'>\n\t\t\t\t<literal xml:lang='en'>dollars (paper money)</literal>\n\t\t\t</binding>\n\t\t\t<binding name='variants'>\n\t\t\t\t<literal>dollar (paper money)|dollar bills|dollar bill|Dollars</literal>\n\t\t\t</binding>\n\t\t</result>\n\t</results>\n</sparql>\n",
       "headers": {
-        "Date": "Tue, 18 Apr 2023 19:54:05 GMT",
+        "Date": "Thu, 02 May 2024 16:28:44 GMT",
         "Content-Type": "application/sparql-results+xml;charset=utf-8",
         "Transfer-Encoding": "chunked",
         "Connection": "keep-alive",
@@ -24,8 +25,7 @@
         "Content-Disposition": "attachment; filename=\"sparql.xml\"",
         "Content-Language": "en"
       },
-      "status_code": 200,
-      "type": "ok"
+      "status_code": 200
     }
   }
 ]