From 72b3e49cdccab3cfbafeabcc2bdefe87a47083a0 Mon Sep 17 00:00:00 2001
From: Mike Bryant <mikesname@gmail.com>
Date: Wed, 9 Oct 2024 12:08:08 +0100
Subject: [PATCH] In cases where multiple Wikipedia links exist, try and pick
 English

Otherwise, fall back to the first article
---
 helpers/TeiEditions_Helpers_DataFetcher.php | 15 ++++++++++++++-
 package-lock.json                           |  4 ++--
 package.json                                |  2 +-
 plugin.ini                                  |  2 +-
 test/unit/TeiEditions_TeiEnhancerTest.php   | 14 ++++++++++++--
 5 files changed, 30 insertions(+), 7 deletions(-)

diff --git a/helpers/TeiEditions_Helpers_DataFetcher.php b/helpers/TeiEditions_Helpers_DataFetcher.php
index 13f97cc..5ed9803 100644
--- a/helpers/TeiEditions_Helpers_DataFetcher.php
+++ b/helpers/TeiEditions_Helpers_DataFetcher.php
@@ -201,7 +201,20 @@ private function _getPlace($url, $lang = null)
 
         $lat = @(float)$xpath->query("/rdf:RDF/gn:Feature/wgs84_pos:lat")->item(0)->textContent;
         $lon = @(float)$xpath->query("/rdf:RDF/gn:Feature/wgs84_pos:long")->item(0)->textContent;
-        $wiki = @$xpath->query("/rdf:RDF/gn:Feature/gn:wikipediaArticle/@rdf:resource")->item(0)->textContent;
+        $wikis = $xpath->query("/rdf:RDF/gn:Feature/gn:wikipediaArticle/@rdf:resource");
+        // Sigh: try and prefer the English Wikipedia article if one exists, otherwise take the first:
+        $wiki = null;
+        if ($wikis->length) {
+            foreach ($wikis as $w) {
+                if (preg_match('/en.wikipedia.org\/wiki/', $w->textContent)) {
+                    $wiki = $w->textContent;
+                    break;
+                }
+            }
+            if (!$wiki) {
+                $wiki = $wikis->item(0)->textContent;
+            }
+        }
 
         $entity = TeiEditionsEntity::create(
            $this->_parseGeonamesPlaceName($xpath, $lang),
diff --git a/package-lock.json b/package-lock.json
index 377d72a..cee496e 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -1,12 +1,12 @@
 {
   "name": "tei-editions",
-  "version": "1.0.0-pre10",
+  "version": "1.0.0-pre11",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "tei-editions",
-      "version": "1.0.0-pre10",
+      "version": "1.0.0-pre11",
       "license": "EUPL-1.2",
       "devDependencies": {
         "grunt": "^1.5.3",
diff --git a/package.json b/package.json
index e79e31b..7e3bcd7 100644
--- a/package.json
+++ b/package.json
@@ -1,7 +1,7 @@
 {
   "name": "tei-editions",
   "description": "An Omeka plugin for importing and rendering structured TEI files for EHRI digital editions.",
-  "version": "1.0.0-pre10",
+  "version": "1.0.0-pre11",
   "devDependencies": {
     "grunt": "^1.5.3",
     "grunt-cli": "^1.4.3",
diff --git a/plugin.ini b/plugin.ini
index 3a3b646..2c4195a 100755
--- a/plugin.ini
+++ b/plugin.ini
@@ -7,4 +7,4 @@ link="https://github.com/EHRI/TeiEditions"
 support_link="https://github.com/EHRI/TeiEditions/issues"
 omeka_minimum_version="2.6"
 omeka_target_version="2.6"
-version="1.0.0-pre10"
+version="1.0.0-pre11"
diff --git a/test/unit/TeiEditions_TeiEnhancerTest.php b/test/unit/TeiEditions_TeiEnhancerTest.php
index 67f6971..6e162cc 100644
--- a/test/unit/TeiEditions_TeiEnhancerTest.php
+++ b/test/unit/TeiEditions_TeiEnhancerTest.php
@@ -24,7 +24,8 @@ public function setUp(): void
         $this->xpath->registerNamespace("t", TeiEditions_Helpers_DocumentProxy::TEI_NS);
     }
 
-    private function valueOf($path) {
+    private function valueOf($path)
+    {
         $n = $this->xpath->query($path);
         return $n->length ? $n->item(0)->textContent : "";
     }
@@ -56,7 +57,7 @@ public function test_addRefs()
             $this->valueOf("//t:fileDesc/t:sourceDesc/t:listPerson/t:person[1]/t:persName")
         );
         $this->assertRegExp(
-          "/11\/10\/1902\s+15\/10\/1980/",
+            "/11\/10\/1902\s+15\/10\/1980/",
             $this->valueOf("//t:fileDesc/t:sourceDesc/t:listPerson/t:person[1]/t:note/t:p[1]")
         );
         $this->assertEquals(
@@ -65,6 +66,15 @@ public function test_addRefs()
         );
     }
 
+    public function test_addCorrectWikipediaRefs()
+    {
+        $testdata = TeiEditions_Helpers_DocumentProxy::fromUriOrPath(
+            TEI_EDITIONS_TEST_DIR . "/resources/enhance-tei-2.xml");
+        $src = new TeiEditions_Helpers_DataFetcher([], "eng");
+        $num = (new TeiEditions_Helpers_TeiEnhancer($src))->addReferences($testdata);
+        $this->assertEquals(13, $num);
+    }
+
     public function test_addRefsWithLang()
     {
         // TODO: fix this so we can mock the data lookups!