fix: strip invalid XML characters from responses

src/BusProNet/Transport/XmlChannel.php

@@ -6,6 +6,7 @@ use App\BusProNet\Exception\ApiClientException;
 use App\BusProNet\Exception\ImmediateConnectionCloseException;
 use App\BusProNet\Exception\ResponseParserException;
 use App\BusProNet\MessageType;
+use App\BusProNet\Utility\XmlCharacterFilter;
 use App\BusProNet\XmlParser\ApiResponseParser;
 use App\Service\XmlDumpWriter;
 use Psr\Log\LoggerInterface;
@@ -208,8 +209,18 @@ class XmlChannel
      */
     private function parseResponse(MessageType $type, string $xml, array $parserArgs, string $requestId, bool $dumpOnFailure = true): mixed
     {
+        // The dump keeps the raw response; only the parser sees the filtered one.
+        $filtered = XmlCharacterFilter::strip($xml, $removed);
+        if ($removed > 0) {
+            $this->logger->warning('Removed characters not allowed in XML from API response', [
+                'request_id' => $requestId,
+                'type' => $type->value,
+                'removed' => $removed,
+            ]);
+        }
+
         try {
-            return $this->responseParser->parseXmlString($type, $xml, $parserArgs);
+            return $this->responseParser->parseXmlString($type, $filtered, $parserArgs);
         } catch (ResponseParserException $e) {
             // Keep the evidence even outside debug mode: without the raw response a parse
             // failure is not diagnosable after the fact. app:cleanup:xml-dumps prunes it.

src/BusProNet/Utility/XmlCharacterFilter.php

@@ -0,0 +1,60 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\BusProNet\Utility;
+
+/**
+ * Removes characters XML 1.0 does not allow from a BusPro response before it is parsed.
+ *
+ * BusPro escapes whatever a field holds into a character reference without checking it,
+ * so a NUL-padded value comes out as `&#x0;`. libxml rejects that as fatal and the whole
+ * response fails to parse, not just the field. The same applies to the raw control bytes.
+ */
+final class XmlCharacterFilter
+{
+    /**
+     * @param-out int $count the number of characters removed
+     */
+    public static function strip(string $xml, ?int &$count = null): string
+    {
+        $count = 0;
+
+        $xml = preg_replace_callback(
+            '/&#(x[0-9a-f]+|[0-9]+);/i',
+            static function (array $match) use (&$count): string {
+                $reference = $match[1];
+                $codePoint = 'x' === strtolower($reference[0]) ? hexdec(substr($reference, 1)) : (int) $reference;
+
+                if (self::isAllowed($codePoint)) {
+                    return $match[0];
+                }
+
+                ++$count;
+
+                return '';
+            },
+            $xml,
+        );
+
+        // Control bytes never occur inside a UTF-8 multibyte sequence, so a byte-wise match is safe.
+        $xml = preg_replace('/[\x00-\x08\x0B\x0C\x0E-\x1F]/', '', $xml, -1, $rawCount);
+
+        $count += $rawCount;
+
+        return $xml;
+    }
+
+    /**
+     * The Char production of XML 1.0.
+     */
+    private static function isAllowed(int|float $codePoint): bool
+    {
+        return 0x9 === $codePoint
+            || 0xA === $codePoint
+            || 0xD === $codePoint
+            || ($codePoint >= 0x20 && $codePoint <= 0xD7FF)
+            || ($codePoint >= 0xE000 && $codePoint <= 0xFFFD)
+            || ($codePoint >= 0x10000 && $codePoint <= 0x10FFFF);
+    }
+}

tests/BusProNet/Utility/XmlCharacterFilterTest.php

@@ -0,0 +1,75 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Tests\BusProNet\Utility;
+
+use App\BusProNet\Utility\XmlCharacterFilter;
+use App\BusProNet\XmlCrawlerFactory;
+use PHPUnit\Framework\Attributes\DataProvider;
+use PHPUnit\Framework\TestCase;
+
+class XmlCharacterFilterTest extends TestCase
+{
+    public function testNulPaddedFieldNoLongerBreaksTheResponse(): void
+    {
+        // Shape of a real KUNDENKONTO/Adressdaten response with a NUL-padded Geburtsort.
+        $xml = '<?xml version="1.0" encoding="utf-8"?><ergebnis><satz typ="KUNDENKONTO"></satz>'
+            .'<idperson>11297</idperson><adressdaten><name>Muster</name>'
+            .'<geburtsort>'.str_repeat('&#x0;', 40).'</geburtsort><beruf></beruf></adressdaten></ergebnis>';
+
+        $this->assertCount(0, XmlCrawlerFactory::create($xml));
+
+        $filtered = XmlCharacterFilter::strip($xml, $count);
+
+        $this->assertSame(40, $count);
+        $crawler = XmlCrawlerFactory::create($filtered);
+        $this->assertSame('Muster', $crawler->filterXPath('//adressdaten/name')->text());
+        $this->assertSame('', $crawler->filterXPath('//adressdaten/geburtsort')->text());
+    }
+
+    /** @return iterable<string, array{string}> */
+    public static function invalidProvider(): iterable
+    {
+        yield 'hex NUL' => ['&#x0;'];
+        yield 'hex NUL, padded and upper case' => ['&#X0000;'];
+        yield 'decimal NUL' => ['&#0;'];
+        yield 'vertical tab' => ['&#x0B;'];
+        yield 'decimal 31' => ['&#31;'];
+        yield 'surrogate' => ['&#xD800;'];
+        yield 'noncharacter FFFE' => ['&#xFFFE;'];
+        yield 'beyond Unicode' => ['&#x110000;'];
+        yield 'raw NUL byte' => ["\x00"];
+        yield 'raw escape byte' => ["\x1B"];
+    }
+
+    #[DataProvider('invalidProvider')]
+    public function testRemovesInvalidCharacters(string $invalid): void
+    {
+        $this->assertSame('<a>xy</a>', XmlCharacterFilter::strip('<a>x'.$invalid.'y</a>', $count));
+        $this->assertSame(1, $count);
+    }
+
+    /** @return iterable<string, array{string}> */
+    public static function validProvider(): iterable
+    {
+        yield 'tab' => ['&#x9;'];
+        yield 'line feed' => ['&#10;'];
+        yield 'carriage return' => ['&#xD;'];
+        yield 'space' => ['&#32;'];
+        yield 'umlaut' => ['&#xFC;'];
+        yield 'emoji' => ['&#x1F600;'];
+        yield 'named entity' => ['&amp;'];
+        yield 'raw tab and newline' => ["\t\r\n"];
+        yield 'raw UTF-8' => ['Grüße 😀'];
+    }
+
+    #[DataProvider('validProvider')]
+    public function testKeepsAllowedCharacters(string $valid): void
+    {
+        $xml = '<a>x'.$valid.'y</a>';
+
+        $this->assertSame($xml, XmlCharacterFilter::strip($xml, $count));
+        $this->assertSame(0, $count);
+    }
+}