[FINNA-1729] LIDO: Drop support for splitting titles (#160)

NatLibFi · Jun 26, 2024 · 828192b · 828192b
1 parent 04573c7
commit 828192b
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 81 deletions.
diff --git a/src/RecordManager/Base/Record/Lido.php b/src/RecordManager/Base/Record/Lido.php
@@ -146,29 +146,11 @@ public function toSolrArray(Database $db = null)
 
         $data['record_format'] = 'lido';
         $title = $this->getTitle(false);
-        if ($this->getDriverParam('splitTitles', false)) {
-            $titlePart = $this->metadataUtils->splitTitle($title);
-            if ($titlePart) {
-                $data['description'] = $title;
-                $title = $titlePart;
-            }
-        }
         $data['title'] = $data['title_short'] = $data['title_full'] = $title;
-        // Create sort title from the title that may have been split above:
         $data['title_sort'] = $this->metadataUtils->createSortTitle($title);
         $data['title_alt'] = $this->getAltTitles();
 
-        $description = $this->getDescription();
-        if ($description) {
-            if (
-                !empty($data['description'])
-                && !str_starts_with($description, $data['description'])
-            ) {
-                $data['description'] .= " -- $description";
-            } else {
-                $data['description'] = $description;
-            }
-        }
+        $data['description'] = $this->getDescription();
 
         $data['format'] = $this->getObjectWorkType();
         $data['identifier'] = $this->getIdentifier();

diff --git a/src/RecordManager/Base/Utils/MetadataUtils.php b/src/RecordManager/Base/Utils/MetadataUtils.php
@@ -935,68 +935,6 @@ function (&$val, $key, $chars) {
         return $array;
     }
 
-    /**
-     * Split title to main title and description. Tries to find the first sentence
-     * break where the title can be split.
-     *
-     * @param string $title Title to split
-     *
-     * @return null|string Null if title was not split, otherwise the initial
-     * title part
-     */
-    public function splitTitle($title)
-    {
-        $i = 0;
-        $parenLevel = 0;
-        $bracketLevel = 0;
-        // Make sure the title has single spaces for whitespace
-        $title = preg_replace('/\s+/', ' ', $title);
-        $titleWords = explode(' ', $title);
-        foreach ($titleWords as $word) {
-            ++$i;
-            $parenLevel += substr_count($word, '(');
-            $parenLevel -= substr_count($word, ')');
-            $bracketLevel += substr_count($word, '[');
-            $bracketLevel -= substr_count($word, ']');
-            if ($parenLevel == 0 && $bracketLevel == 0) {
-                // Try to avoid splitting at short words or the very beginning
-                if (
-                    substr($word, -1) == '.' && strlen($word) > 2
-                    && ($i > 1 || strlen($word) > 4)
-                ) {
-                    // Verify that the word is strippable (not abbreviation etc.)
-                    $leadStripped = $this->stripLeadingPunctuation(
-                        $word
-                    );
-                    $stripped = $this->stripTrailingPunctuation(
-                        $leadStripped
-                    );
-                    $nextFirst = isset($titleWords[$i])
-                        ? substr($titleWords[$i], 0, 1)
-                        : '';
-                    // 1.) There has to be something following this word.
-                    // 2.) The trailing period must be strippable or end with a year.
-                    // 3.) Next word has to start with a capital or digit
-                    // 4.) Not something like 12-p.
-                    // 5.) Not initials like A.N.
-                    if (
-                        $nextFirst
-                        && ($leadStripped != $stripped
-                        || preg_match('/^\d{4}\.$/', $word))
-                        && (is_numeric($nextFirst) || !ctype_lower($nextFirst))
-                        && !preg_match('/.+\-\w{1,2}\.$/', $word)
-                        && !preg_match('/^\w\.\w\.$/', $word) // initials
-                    ) {
-                        return $this->stripTrailingPunctuation(
-                            implode(' ', array_splice($titleWords, 0, $i))
-                        );
-                    }
-                }
-            }
-        }
-        return null;
-    }
-
     /**
      * Determine if a record is a hidden component part
      *

diff --git a/tests/RecordManagerTest/Base/Record/LidoTest.php b/tests/RecordManagerTest/Base/Record/LidoTest.php
@@ -64,6 +64,7 @@ public function testLido1()
             'title_sort' => 'luonnonsuojelusäädökset toimittanut raimo luhtanen'
                 . ' säädökset',
             'title_alt' => [],
+            'description' => '',
             'format' => 'Kirja',
             'identifier' => '26054',
             'institution' => 'Test Institution',
@@ -179,6 +180,7 @@ public function testLido1NonMergedTitle()
             'title_alt' => [
                 'Säädökset',
             ],
+            'description' => '',
             'format' => 'Kirja',
             'identifier' => '26054',
             'institution' => 'Test Institution',