[FINNA-623] QDC: Tweak check for detecting invalid language codes (#61)

NatLibFi · Feb 22, 2024 · deadb3b · deadb3b
1 parent 565e600
commit deadb3b
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 3 deletions.
diff --git a/src/RecordManager/Finna/Record/Qdc.php b/src/RecordManager/Finna/Record/Qdc.php
@@ -154,14 +154,21 @@ protected function getLanguages()
         $languages = [];
         foreach ($this->doc->language as $language) {
             foreach (explode(' ', trim((string)$language)) as $part) {
+                //Remove extra characters from start and end of a language
+                $check = trim($part, ', ');
                 $check = preg_replace(
                     '/^http:\/\/lexvo\.org\/id\/iso639-.\/(.*)/',
                     '$1',
-                    $part
+                    $check
                 );
+                // en_US
+                if (str_contains($check, '_')) {
+                    $check = explode('_', $check)[0];
+                }
                 // Check that the language given is in proper form
-                if (mb_strlen($check) > 9 || !ctype_lower($check)) {
-                    $this->storeWarning("unhandled language $check");
+                if (!$check || strlen($check) > 9 || !ctype_lower($check)) {
+                    $toLog = $part ?: 'EMPTY_VALUE';
+                    $this->storeWarning("unhandled language $toLog");
                     continue;
                 }
                 foreach (str_split($check, 3) as $code) {

diff --git a/tests/RecordManagerTest/Finna/Record/QdcTest.php b/tests/RecordManagerTest/Finna/Record/QdcTest.php
@@ -138,13 +138,20 @@ public function testQdcLanguageWarnings()
                 'unhandled language verylonglanguagehere',
                 'unhandled language EnGb',
                 'unhandled language caT',
+                'unhandled language po,tt',
+                'unhandled language ,',
+                'unhandled language EMPTY_VALUE',
             ],
             $record->getProcessingWarnings(),
             'getProcessingWarnings'
         );
         $this->compareArray(
             [
                 'fi',
+                'jp',
+                'sv',
+                'en',
+                'nr',
             ],
             $fields['language'],
             'LanguageCheckAfterWarnings'

diff --git a/tests/fixtures/Finna/record/qdc_language_warnings.xml b/tests/fixtures/Finna/record/qdc_language_warnings.xml
@@ -6,6 +6,11 @@
   <language type="iso">EnGb</language>
   <language type="iso">caT</language>
   <language type="iso">fi</language>
+  <language type="iso">jp_JP</language>
+  <language type="iso">sv, en, nr</language><!-- this should validate as a language -->
+  <language type="iso">po,tt</language><!--this should not validate as a language -->
+  <language type="iso">,</language>
+  <language type="iso"> </language>
   <rights>CC BY-NC-ND 4.0</rights>
   <publisher lang="fi">Sanitation Project, Research Institute for Humanity and Nature</publisher>
   <permaddress type="doi">http://dx.doi.org/https://doi.org/10.34416/svc.00029</permaddress>