XMLParser: fix premature stream termination on empty mid-stream reads

src/XMLParser.php

@@ -19,6 +19,8 @@ class XMLParser implements \Iterator
 
     private string $currentTagContent = '';
 
+    private bool $streamFinished = false;
+
     /**
      * The stack of the XML node names as go deeper into the tree.
      *
@@ -51,6 +53,7 @@ public function setUp(): void
         $this->currentTagName = null;
         $this->currentTagAttributes = [];
         $this->nodeNamesStack = [];
+        $this->streamFinished = false;
 
         $this->parser = \xml_parser_create();
 
@@ -179,25 +182,39 @@ public function rewind(): void
      *
      * Callbacks are called, nodes are pushed to the stack and iterator can go over them.
      *
+     * Loops until at least one node is queued or the stream is exhausted.
+     * This handles compressed or network streams that may return empty reads mid-stream
+     * (e.g. zlib flush markers or TCP packet boundaries) without signalling EOF.
+     *
      * @return void
      * @throws ParsingError
      */
     private function parseNextChunk(): void
     {
-        // the nodes stack has been iterated over, consume and parse the next piece of the XML stream
-        $data = stream_get_contents($this->stream, length: self::BATCH_READ_SIZE);
-        $isFinal = ($data === false);
+        if ($this->streamFinished) {
+            return;
+        }
 
-        $res = xml_parse($this->parser, $data, $isFinal);
+        while (empty($this->nodesQueue)) {
+            $data = stream_get_contents($this->stream, length: self::BATCH_READ_SIZE);
+            $isFinal = ($data === false || feof($this->stream));
 
-        if ($res === 0 /* returns 0 on failure */) {
-            // take more details from the parser instance and throw an exception
-            throw ParsingError::fromParserInstance($this->parser, is_string($data) ? $data : null);
-        }
+            if ($data === false) {
+                $data = '';
+            }
+
+            $res = xml_parse($this->parser, $data, $isFinal);
 
-        // we're done with reading and parsing the stream, close the XML parser instance
-        if ($isFinal) {
-            $this->close();
+            if ($res === 0 /* returns 0 on failure */) {
+                // take more details from the parser instance and throw an exception
+                throw ParsingError::fromParserInstance($this->parser, $data !== '' ? $data : null);
+            }
+
+            if ($isFinal) {
+                $this->streamFinished = true;
+                $this->close();
+                return;
+            }
         }
     }
 

tests/XMLParserNestedElementsTest.php

@@ -0,0 +1,139 @@
+<?php
+
+use Elecena\XmlIterator\Nodes\XMLNodeContent;
+use Elecena\XmlIterator\XMLParser;
+
+/**
+ * A PHP stream wrapper that inserts empty reads at regular intervals to simulate
+ * compressed or network streams (e.g. zlib flush markers, HTTP chunked boundaries).
+ *
+ * @see https://www.php.net/manual/en/stream.streamwrapper.example-1.php
+ */
+class IntermittentReadStreamWrapper
+{
+    public $context;
+
+    private static string $content = '';
+    private static int $emptyEveryN = 3;
+    private int $position = 0;
+    private int $readCount = 0;
+
+    public static function open(string $content, int $emptyEveryN = 3): string
+    {
+        self::$content = $content;
+        self::$emptyEveryN = $emptyEveryN;
+
+        if (in_array('intermittent', stream_get_wrappers())) {
+            stream_wrapper_unregister('intermittent');
+        }
+
+        stream_wrapper_register('intermittent', self::class);
+
+        return 'intermittent://stream';
+    }
+
+    public static function unregister(): void
+    {
+        if (in_array('intermittent', stream_get_wrappers())) {
+            stream_wrapper_unregister('intermittent');
+        }
+    }
+
+    public function stream_open(string $path, string $mode, int $options, ?string &$openedPath): bool
+    {
+        $this->position = 0;
+        $this->readCount = 0;
+        return true;
+    }
+
+    public function stream_read(int $count): string
+    {
+        $this->readCount++;
+
+        // Return empty string every N reads unless already at EOF
+        if ($this->readCount % self::$emptyEveryN === 0 && !$this->stream_eof()) {
+            return '';
+        }
+
+        $chunk = substr(self::$content, $this->position, $count);
+        $this->position += strlen($chunk);
+        return $chunk;
+    }
+
+    public function stream_eof(): bool
+    {
+        return $this->position >= strlen(self::$content);
+    }
+
+    public function stream_stat(): array
+    {
+        return [];
+    }
+}
+
+/**
+ * Tests XML parsing of nested elements that share the same tag name,
+ * as seen in the Discogs data dump (label > sublabels > label).
+ *
+ * @see https://github.com/elecena/xml-iterator/issues/13
+ */
+class XMLParserNestedElementsTest extends XMLParserTestCase
+{
+    private static string $xml = '';
+
+    public static function setUpBeforeClass(): void
+    {
+        $xml = '<?xml version="1.0" encoding="UTF-8"?><labels>';
+        for ($i = 0; $i < 300; $i++) {
+            $xml .= sprintf(
+                '<label id="%d"><name>Label %d</name><sublabels><label>Sub %d</label></sublabels></label>',
+                $i,
+                $i,
+                $i
+            );
+        }
+        $xml .= '</labels>';
+        self::$xml = $xml;
+    }
+
+    protected function tearDown(): void
+    {
+        IntermittentReadStreamWrapper::unregister();
+    }
+
+    protected function getParserStream()
+    {
+        return self::streamFromString(self::$xml);
+    }
+
+    public function testParsesNestedSameNameElements(): void
+    {
+        $labelCount = 0;
+
+        foreach ($this->getParser() as $node) {
+            if ($node instanceof XMLNodeContent && $node->name === 'label') {
+                $labelCount++;
+            }
+        }
+
+        // 300 outer labels + 300 inner labels inside sublabels
+        $this->assertSame(600, $labelCount);
+    }
+
+    public function testParsesStreamWithIntermittentEmptyReads(): void
+    {
+        $url = IntermittentReadStreamWrapper::open(self::$xml, emptyEveryN: 3);
+        $parser = new XMLParser(stream: fopen($url, 'r'));
+
+        $labelCount = 0;
+
+        foreach ($parser as $node) {
+            if ($node instanceof XMLNodeContent && $node->name === 'label') {
+                $labelCount++;
+            }
+        }
+
+        // 300 outer labels + 300 inner labels inside sublabels
+        $this->assertSame(600, $labelCount);
+    }
+}