museum-digital/MDNodaHelpers
33
0
Fork 0
Code Issues 3 Pull Requests Releases Wiki Activity

Compare commits

base: museum-digital:12a7937218ea07d58f931bb949397921dba565a7
Branches Tags
museum-digital:master
..
compare: museum-digital:107a4cd640e362daf57df191f90fbe03ee993faa
Branches Tags
museum-digital:master

No commits in common. "12a7937218ea07d58f931bb949397921dba565a7" and "107a4cd640e362daf57df191f90fbe03ee993faa" have entirely different histories.
12a7937218 ... 107a4cd640

2 changed files with 24 additions and 32 deletions
Show Stats Expand all files Collapse all files

53
src/NodaWikidataFetcher.php
View File

@ -134,7 +134,7 @@ final class NodaWikidataFetcher {
$doc->loadXML('<section>' . trim($input) . '</section>'); $doc->loadXML('<section>' . trim($input) . '</section>');
} }
catch (Exception $e) { 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"); $list = $doc->getElementsByTagName("style");
@ -224,6 +224,7 @@ final class NodaWikidataFetcher {
$input = str_replace("\t", " ", $input); $input = str_replace("\t", " ", $input);
return $input;
// Remove newlines with ensuing spaces // Remove newlines with ensuing spaces
while (strpos($input, PHP_EOL . " ") !== false) { while (strpos($input, PHP_EOL . " ") !== false) {
$input = str_replace(PHP_EOL . " ", PHP_EOL, $input); $input = str_replace(PHP_EOL . " ", PHP_EOL, $input);
@ -628,7 +629,7 @@ final class NodaWikidataFetcher {
$output = []; $output = [];
# $descs = []; $descs = [];
foreach ($checkagainstLanguage as $lang) { foreach ($checkagainstLanguage as $lang) {
if (!empty($languagesToFetch[$lang]) && !empty($data['sitelinks'][$lang . 'wiki']) && !empty($wikilinks[$lang])) { if (!empty($languagesToFetch[$lang]) && !empty($data['sitelinks'][$lang . 'wiki']) && !empty($wikilinks[$lang])) {
@ -649,7 +650,7 @@ final class NodaWikidataFetcher {
} }
if ($tDescription !== '') { if ($tDescription !== '') {
# $descs[$lang] = $tDescription; $descs[$lang] = $tDescription;
$desc_cleaned = self::_cleanWikidataInput($tDescription); $desc_cleaned = self::_cleanWikidataInput($tDescription);
if ($desc_cleaned !== '') { if ($desc_cleaned !== '') {
$output[$lang] = [ $output[$lang] = [
@ -671,6 +672,7 @@ final class NodaWikidataFetcher {
]; ];
} }
# print_r($descs);
} }
@ -1005,7 +1007,7 @@ final class NodaWikidataFetcher {
} }
/** /**
* Returns the current description of a place. * Gets the current description of a place.
* *
* @param integer $onum Place ID. * @param integer $onum Place ID.
* *
@ -1027,29 +1029,6 @@ 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 $cur[0];
}
/** /**
* Function for entering base information about a place from wikidata. * Function for entering base information about a place from wikidata.
* *
@ -1260,14 +1239,14 @@ final class NodaWikidataFetcher {
} }
if (isset($data['claims']['P625'])) { if (isset($data['claims']['P625'])) {
$latitude_wd = \filter_var($data['claims']['P625'][0]['mainsnak']['datavalue']['value']['latitude'], FILTER_VALIDATE_FLOAT); $latitude_wd = $data['claims']['P625'][0]['mainsnak']['datavalue']['value']['latitude'];
$longitude_wd = \filter_var($data['claims']['P625'][0]['mainsnak']['datavalue']['value']['longitude'], FILTER_VALIDATE_FLOAT); $longitude_wd = $data['claims']['P625'][0]['mainsnak']['datavalue']['value']['longitude'];
} }
NodaBatchInserter::linkNodaForPlace($this->_mysqli_noda, $onum, $nodaLinks, $erfasst_von); NodaBatchInserter::linkNodaForPlace($this->_mysqli_noda, $onum, $nodaLinks, $erfasst_von);
$this->_mysqli_noda->autocommit(false); $this->_mysqli_noda->autocommit(false);
if (!empty($tgn_id) and is_numeric($tgn_id)) { if (!empty($tgn_id)) {
$updateStmt = $this->_mysqli_noda->do_prepare("UPDATE `orte` $updateStmt = $this->_mysqli_noda->do_prepare("UPDATE `orte`
SET `ort_land` = ? SET `ort_land` = ?
@ -1278,7 +1257,7 @@ final class NodaWikidataFetcher {
unset($updateStmt); unset($updateStmt);
} }
if (!empty($geonames_id) and is_numeric($geonames_id)) { if (!empty($geonames_id)) {
$updateStmt = $this->_mysqli_noda->do_prepare("UPDATE `orte` $updateStmt = $this->_mysqli_noda->do_prepare("UPDATE `orte`
SET `ort_geonames` = ? SET `ort_geonames` = ?
@ -1363,7 +1342,17 @@ final class NodaWikidataFetcher {
$datafromwiki = '"' . $datafromwiki . '" - (Wikipedia (' . $lang . ') ' . date("d.m.Y") . ')'; $datafromwiki = '"' . $datafromwiki . '" - (Wikipedia (' . $lang . ') ' . date("d.m.Y") . ')';
$datafromwiki = str_replace("'", "´", MD_STD::preg_replace_str("/\&\#91\;[0-9]\&\#93\;/", '', $datafromwiki)); $datafromwiki = str_replace("'", "´", MD_STD::preg_replace_str("/\&\#91\;[0-9]\&\#93\;/", '', $datafromwiki));
$tag_anmerkung = $this->getTagDescription($tag_id); $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];
$this->_mysqli_noda->autocommit(false); $this->_mysqli_noda->autocommit(false);

3
tests/NodaWikidataFetcherTest.php
View File

@ -255,6 +255,9 @@ 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) "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)
); );
} }
/** /**