[sysconfig] Update Wikipedia extraction logic

1 files changed, 24 insertions, 7 deletions

diff --git a/modules-available/sysconfig/addmodule_branding.inc.php b/modules-available/sysconfig/addmodule_branding.inc.php
index 4cbcb66f..54b2ad57 100644
--- a/modules-available/sysconfig/addmodule_branding.inc.php
+++ b/modules-available/sysconfig/addmodule_branding.inc.php
@@ -99,28 +99,44 @@ class Branding_ProcessFile extends AddModule_Base
 	private static function downloadSvg(string $svgName, string $url, &$title): bool
 	{
 		$title = false;
-		// [wikipedia] Did someone paste a link to a thumbnail of the svg? Let's fix that...
-		if (preg_match('#^(.*)/thumb/(.*\.svg)/.*\.svg#', $url, $out)) {
-			$url = $out[1] . '/' . $out[2];
-		}
 		for ($i = 0; $i < 5; ++$i) {
+			// [wikipedia] Did someone paste a link to a thumbnail of the svg? Let's fix that...
+			if (preg_match('#^(.*)/thumb/(.*\.svg)/.*\.svg#', $url, $out)) {
+				$url = $out[1] . '/' . $out[2];
+			}
 			$code = 400;
 			if (!Download::toFile($svgName, $url, 3, $code) || $code < 200 || $code > 299) {
 				Message::addError('remote-timeout', $url, $code);
 				return false;
 			}
-			$content = FileUtil::readFile($svgName, 25000);
+			$content = FileUtil::readFile($svgName, 250000);
 			// Is svg file?
 			if (strpos($content, '<svg') !== false)
 				return true; // Found an svg tag - don't try to find links to the actual image
 				
 			// [wikipedia] Try to be nice and detect links that might give a hint where the svg can be found
-			if (preg_match_all('#href="([^"]*upload.wikimedia.org/[^"]*/[^"]*/[^"]*\.svg|[^"]+/[^"]+:[^"]+\.svg[^"]*)"#', $content, $out, PREG_PATTERN_ORDER)) {
+			$out1 = $out2 = $out3 = null;
+			if (preg_match_all('#href="([^"]*upload.wikimedia.org/[^"]*/[^"]*/[^"]*\.svg)"#', $content, $out1, PREG_PATTERN_ORDER)
+					|| preg_match_all('#src="([^"]*upload.wikimedia.org/[^"]*/thumb/[^"]*\.svg/[^"]+\.svg[^"]*)"#', $content, $out2, PREG_PATTERN_ORDER)
+					|| preg_match_all('#href="([^"]+/[^"]+:[^"]+\.svg)"#', $content, $out3, PREG_PATTERN_ORDER)) {
 				if ($title === false && preg_match('#<title>([^<]*)</title>#i', $content, $tout)) {
 					$title = trim(preg_replace('/\W*Wikipedia.*/', '', $tout[1]));
 				}
 				$new = false;
-				foreach ($out[1] as $res) {
+				$out = [];
+				if (isset($out1[1])) {
+					$out += $out1[1];
+				}
+				if (isset($out2[1])) {
+					$out += $out2[1];
+				}
+				if (isset($out3[1])) {
+					$out += $out3[1];
+				}
+				foreach ($out as $res) {
+					error_log("Match '$res'");
+					if (!preg_match('/hochschule|univers|logo|siegel/i', $res))
+						continue;
 					if (strpos($res, 'action=edit') !== false)
 						continue;
 					$new = Branding_ProcessFile::internetCombineUrl($url, html_entity_decode($res, ENT_COMPAT, 'UTF-8'));
@@ -129,6 +145,7 @@ class Branding_ProcessFile extends AddModule_Base
 				}
 				if ($new === $url || $new === false)
 					break;
+				error_log("New: '$new'");
 				$url = $new;
 				continue;
 			}