From a68a03e628864ec42e503b9a859a5f7b32c43342 Mon Sep 17 00:00:00 2001
From: Joshua Ramon Enslin <joshua@museum-digital.de>
Date: Thu, 31 Aug 2023 16:09:21 +0200
Subject: [PATCH] Improve wikidata fetcher

---
 src/NodaWikidataFetcher.php       | 48 +++++++++++++++++++------------
 tests/NodaWikidataFetcherTest.php |  3 --
 2 files changed, 30 insertions(+), 21 deletions(-)

diff --git a/src/NodaWikidataFetcher.php b/src/NodaWikidataFetcher.php
index d90179e..f647107 100644
--- a/src/NodaWikidataFetcher.php
+++ b/src/NodaWikidataFetcher.php
@@ -134,7 +134,7 @@ final class NodaWikidataFetcher {
             $doc->loadXML('<section>' . trim($input) . '</section>');
         }
         catch (Exception $e) {
-            throw new Exception("Failed to load DOMDocument." . PHP_EOL . $e->getMessage() . PHP_EOL . PHP_EOL . $input);
+            throw new Exception("Failed to load DOMDocument." . PHP_EOL . $e->getMessage() . PHP_EOL . PHP_EOL . '---' . $input . '---');
         }
 
         $list = $doc->getElementsByTagName("style");
@@ -672,7 +672,6 @@ final class NodaWikidataFetcher {
                 ];
 
             }
-            # print_r($descs);
 
         }
 
@@ -1007,7 +1006,7 @@ final class NodaWikidataFetcher {
     }
 
     /**
-     * Gets the current description of a place.
+     * Returns the current description of a place.
      *
      * @param integer $onum Place ID.
      *
@@ -1029,6 +1028,29 @@ final class NodaWikidataFetcher {
 
     }
 
+    /**
+     * Returns the current description of a tag.
+     *
+     * @param integer $tag_id Tag ID.
+     *
+     * @return string
+     */
+    private function getTagDescription(int $tag_id):string {
+
+        $result = $this->_mysqli_noda->query_by_stmt("SELECT `tag_anmerkung`
+            FROM `tag`
+            WHERE `tag_id` = ?", "i", $tag_id);
+
+        if (!($cur = $result->fetch_row())) {
+            $result->close();
+            return '';
+        }
+        $result->close();
+
+        return $cur[0];
+
+    }
+
     /**
      * Function for entering base information about a place from wikidata.
      *
@@ -1239,14 +1261,14 @@ final class NodaWikidataFetcher {
         }
 
         if (isset($data['claims']['P625'])) {
-            $latitude_wd = $data['claims']['P625'][0]['mainsnak']['datavalue']['value']['latitude'];
-            $longitude_wd = $data['claims']['P625'][0]['mainsnak']['datavalue']['value']['longitude'];
+            $latitude_wd = \filter_var($data['claims']['P625'][0]['mainsnak']['datavalue']['value']['latitude'], FILTER_VALIDATE_FLOAT);
+            $longitude_wd = \filter_var($data['claims']['P625'][0]['mainsnak']['datavalue']['value']['longitude'], FILTER_VALIDATE_FLOAT);
         }
 
         NodaBatchInserter::linkNodaForPlace($this->_mysqli_noda, $onum, $nodaLinks, $erfasst_von);
 
         $this->_mysqli_noda->autocommit(false);
-        if (!empty($tgn_id)) {
+        if (!empty($tgn_id) and is_numeric($tgn_id)) {
 
             $updateStmt = $this->_mysqli_noda->do_prepare("UPDATE `orte`
                 SET `ort_land` = ?
@@ -1257,7 +1279,7 @@ final class NodaWikidataFetcher {
             unset($updateStmt);
 
         }
-        if (!empty($geonames_id)) {
+        if (!empty($geonames_id) and is_numeric($geonames_id)) {
 
             $updateStmt = $this->_mysqli_noda->do_prepare("UPDATE `orte`
                 SET `ort_geonames` = ?
@@ -1342,17 +1364,7 @@ final class NodaWikidataFetcher {
         $datafromwiki = '"' . $datafromwiki . '" - (Wikipedia (' . $lang . ') ' . date("d.m.Y") . ')';
         $datafromwiki = str_replace("'", "´", MD_STD::preg_replace_str("/\&\#91\;[0-9]\&\#93\;/", '', $datafromwiki));
 
-        $cergebnis = $this->_mysqli_noda->query_by_stmt("SELECT `tag_anmerkung`
-            FROM `tag`
-            WHERE `tag_id` = ?", "i", $tag_id);
-
-        if (!($cinfo = $cergebnis->fetch_row())) {
-            $cergebnis->close();
-            return $output;
-        }
-        $cergebnis->close();
-
-        $tag_anmerkung = $cinfo[0];
+        $tag_anmerkung = $this->getTagDescription($tag_id);
 
         $this->_mysqli_noda->autocommit(false);
 
diff --git a/tests/NodaWikidataFetcherTest.php b/tests/NodaWikidataFetcherTest.php
index 20c75b8..df6184f 100644
--- a/tests/NodaWikidataFetcherTest.php
+++ b/tests/NodaWikidataFetcherTest.php
@@ -255,9 +255,6 @@ Transclusion expansion time report (%,ms,calls,template)
             "Start of parsed Wikipedia text should be:" . PHP_EOL . PHP_EOL . $expected . PHP_EOL . PHP_EOL . 'Real start text is: ' . PHP_EOL . PHP_EOL . substr($output, 0, 250)
         );
 
-
-
-
     }
 
     /**