Absorb spaces after 'stream' declarations (#642)

There looks to be some additional issues with fonts remaining, but the text content is now read by getText().
smalot · Sep 28, 2023 · 051ec84 · 051ec84
1 parent 5c48261
commit 051ec84
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 1 deletion.
diff --git a/src/Smalot/PdfParser/RawData/RawDataParser.php b/src/Smalot/PdfParser/RawData/RawDataParser.php
@@ -756,7 +756,7 @@ protected function getRawObject(string $pdfData, int $offset = 0, array $headerD
                     // start stream object
                     $objtype = 'stream';
                     $offset += 6;
-                    if (1 == preg_match('/^([\r]?[\n])/isU', substr($pdfData, $offset, 4), $matches)) {
+                    if (1 == preg_match('/^( *[\r]?[\n])/isU', substr($pdfData, $offset, 4), $matches)) {
                         $offset += \strlen($matches[0]);
 
                         // we get stream length here to later help preg_match test less data

diff --git a/tests/PHPUnit/Integration/RawData/RawDataParserTest.php b/tests/PHPUnit/Integration/RawData/RawDataParserTest.php
@@ -87,6 +87,25 @@ public function testGetRawObjectIssue372(): void
             ],
             $result
         );
+
+        // Test that spaces after a 'stream' declaration are absorbed
+        // See: https://github.com/smalot/pdfparser/issues/641
+        $data = 'stream '."\n";
+        $data .= 'streamdata'."\n";
+        $data .= 'endstream'."\n";
+        $data .= 'endobj';
+
+        $result = $this->fixture->exposeGetRawObject($data);
+
+        // Value 'streamdata'."\n" would be empty string without the fix
+        $this->assertEquals(
+            [
+                'stream',
+                'streamdata'."\n",
+                19,
+            ],
+            $result
+        );
     }
 
     /**