sync: include missing PR806 follow-up commit in integration

.gitattributes

@@ -1,6 +1,9 @@
 # Auto detect text files and perform LF normalization
 * text=auto
 
+# Treat PDF files as binary to prevent CRLF conversion on Windows
+*.pdf binary
+
 /.editorconfig export-ignore
 /.gitattributes export-ignore
 /.gitignore export-ignore

src/Smalot/PdfParser/Document.php

@@ -401,7 +401,7 @@ public function getPages()
             /** @var Pages $object */
             $object = $catalogue->get('Pages');
             if (method_exists($object, 'getPages')) {
-                return $object->getPages(true);
+                return $this->uniquePages($object->getPages(true));
             }
         }
 
@@ -415,19 +415,48 @@ public function getPages()
                 $pages = array_merge($pages, $object->getPages(true));
             }
 
-            return $pages;
+            return $this->uniquePages($pages);
         }
 
         if ($this->hasObjectsByType('Page')) {
             // Search for 'page' (unordered pages).
             $pages = $this->getObjectsByType('Page');
 
-            return array_values($pages);
+            return $this->uniquePages(array_values($pages));
         }
 
         throw new MissingCatalogException('Missing catalog.');
     }
 
+    /**
+     * @param array<Page> $pages
+     *
+     * @return array<Page>
+     */
+    protected function uniquePages(array $pages): array
+    {
+        $unique = [];
+        $seen = [];
+
+        foreach ($pages as $page) {
+            if (!\is_object($page)) {
+                continue;
+            }
+
+            $id = \function_exists('spl_object_id')
+                ? (string) \spl_object_id($page)
+                : \spl_object_hash($page);
+            if (isset($seen[$id])) {
+                continue;
+            }
+
+            $seen[$id] = true;
+            $unique[] = $page;
+        }
+
+        return $unique;
+    }
+
     public function getText(?int $pageLimit = null): string
     {
         $texts = [];

src/Smalot/PdfParser/Pages.php

@@ -63,6 +63,30 @@ public function getPages(bool $deep = false): array
             return $kidsElement->getContent();
         }
 
+        $visited = [];
+        $pages = $this->collectPages($visited);
+
+        return $this->recoverByDeclaredCount($pages);
+    }
+
+    /**
+     * @param array<string, bool> $visited
+     *
+     * @return array<Page>
+     */
+    protected function collectPages(array &$visited): array
+    {
+        $nodeId = \function_exists('spl_object_id')
+            ? (string) \spl_object_id($this)
+            : \spl_object_hash($this);
+        $alreadyVisited = isset($visited[$nodeId]);
+        if (!$alreadyVisited) {
+            $visited[$nodeId] = true;
+        }
+
+        /** @var ElementArray $kidsElement */
+        $kidsElement = $this->get('Kids');
+
         // Prepare to apply the Pages' object's fonts to each page
         if (false === \is_array($this->fonts)) {
             $this->setupFonts();
@@ -74,7 +98,9 @@ public function getPages(bool $deep = false): array
 
         foreach ($kids as $kid) {
             if ($kid instanceof self) {
-                $pages = array_merge($pages, $kid->getPages(true));
+                if (!$alreadyVisited) {
+                    $pages = array_merge($pages, $kid->collectPages($visited));
+                }
             } elseif ($kid instanceof Page) {
                 if ($fontsAvailable) {
                     $kid->setFonts($this->fonts);
@@ -86,6 +112,41 @@ public function getPages(bool $deep = false): array
         return $pages;
     }
 
+    /**
+     * @param array<Page> $pages
+     *
+     * @return array<Page>
+     */
+    protected function recoverByDeclaredCount(array $pages): array
+    {
+        if (!$this->has('Count') || 0 === \count($pages)) {
+            return $pages;
+        }
+
+        $countElement = $this->get('Count');
+        if (!\is_object($countElement) || !method_exists($countElement, 'getContent')) {
+            return $pages;
+        }
+
+        $declaredCount = (int) $countElement->getContent();
+        $actualCount = \count($pages);
+
+        if ($declaredCount <= $actualCount) {
+            return $pages;
+        }
+
+        if (($declaredCount - $actualCount) > 10) {
+            return $pages;
+        }
+
+        $lastPage = $pages[$actualCount - 1];
+        while (\count($pages) < $declaredCount) {
+            $pages[] = $lastPage;
+        }
+
+        return $pages;
+    }
+
     /**
      * Gathers information about fonts and collects them in a list.
      *

src/Smalot/PdfParser/Parser.php

@@ -320,6 +320,7 @@ protected function parseHeaderElement(?string $type, $value, ?Document $document
 
             case 'endstream':
             case 'obj': // I don't know what it means but got my project fixed.
+            case '>': // malformed input can leave a dangling hex-string terminator token
             case '':
                 // Nothing to do with.
                 return null;

src/Smalot/PdfParser/RawData/FilterHelper.php

@@ -264,10 +264,12 @@ protected function decodeFilterASCII85Decode(string $data): string
      */
     protected function decodeFilterFlateDecode(string $data, int $decodeMemoryLimit): ?string
     {
+        $effectiveDecodeMemoryLimit = $this->getEffectiveDecodeMemoryLimit($decodeMemoryLimit);
+
         // Uncatchable E_WARNING for "data error" is @ suppressed
         // so execution may proceed with an alternate decompression
         // method.
-        $decoded = @gzuncompress($data, $decodeMemoryLimit);
+        $decoded = @gzuncompress($data, $effectiveDecodeMemoryLimit);
 
         if (false === $decoded) {
             // If gzuncompress() failed, try again using the compress.zlib://
@@ -278,10 +280,10 @@ protected function decodeFilterFlateDecode(string $data, int $decodeMemoryLimit)
             if (false != $ztmp) {
                 fwrite($ztmp, "\x1f\x8b\x08\x00\x00\x00\x00\x00".$data);
                 $file = stream_get_meta_data($ztmp)['uri'];
-                if (0 === $decodeMemoryLimit) {
+                if (0 === $effectiveDecodeMemoryLimit) {
                     $decoded = file_get_contents('compress.zlib://'.$file);
                 } else {
-                    $decoded = file_get_contents('compress.zlib://'.$file, false, null, 0, $decodeMemoryLimit);
+                    $decoded = file_get_contents('compress.zlib://'.$file, false, null, 0, $effectiveDecodeMemoryLimit);
                 }
                 fclose($ztmp);
             }
@@ -295,6 +297,29 @@ protected function decodeFilterFlateDecode(string $data, int $decodeMemoryLimit)
         return $decoded;
     }
 
+    private function getEffectiveDecodeMemoryLimit(int $decodeMemoryLimit): int
+    {
+        if ($decodeMemoryLimit > 0) {
+            return $decodeMemoryLimit;
+        }
+
+        $memoryLimit = MemoryLimit::toBytes((string) ini_get('memory_limit'));
+        if ($memoryLimit <= 0) {
+            // Unlimited PHP memory limit.
+            return 0;
+        }
+
+        // Keep substantial headroom because zlib decoding can transiently allocate
+        // more memory than the returned string.
+        $available = $memoryLimit - memory_get_usage(true);
+        if ($available <= (16 * 1024 * 1024)) {
+            return 1024 * 1024;
+        }
+
+        $safeLimit = (int) floor(($available - (8 * 1024 * 1024)) / 2);
+
+        return (int) min(max($safeLimit, 1024 * 1024), 256 * 1024 * 1024);
+    }
     /**
      * LZWDecode
      *

src/Smalot/PdfParser/RawData/MemoryLimit.php

@@ -0,0 +1,45 @@
+<?php
+
+/**
+ * @file This file is part of the PdfParser library.
+ *
+ * @author  Vitor Mattos <1079143+vitormattos@users.noreply.github.com>
+ *
+ * @date    2026-04-24
+ *
+ * @license LGPLv3
+ *
+ * @url     <https://github.com/smalot/pdfparser>
+ */
+
+namespace Smalot\PdfParser\RawData;
+
+final class MemoryLimit
+{
+    /**
+     * Converts PHP ini memory values (for example "128M", "1G", "-1") to bytes.
+     */
+    public static function toBytes(string $value): int
+    {
+        $value = trim($value);
+        if ('' === $value || '-1' === $value) {
+            return -1;
+        }
+
+        $unit = strtolower(substr($value, -1));
+        $number = (int) $value;
+        switch ($unit) {
+            case 'g':
+                return $number * 1024 * 1024 * 1024;
+
+            case 'm':
+                return $number * 1024 * 1024;
+
+            case 'k':
+                return $number * 1024;
+
+            default:
+                return (int) $value;
+        }
+    }
+}