From 8d005080cc21dc1e476260f31b6e85681a307379 Mon Sep 17 00:00:00 2001
From: Brian Huisman <bhuisman@greywyvern.com>
Date: Mon, 25 Mar 2024 16:10:21 -0400
Subject: [PATCH] More robust check for BI within strings

`BI` "commands" within strings should not be parsed as the beginning of inline image blocks. Detect if the `BI` we found is inside a (string) and if it is, note the offset and move past it for the next match.
---
 src/Smalot/PdfParser/PDFObject.php          | 59 +++++++++++++++++----
 tests/PHPUnit/Integration/PDFObjectTest.php | 18 ++++++-
 2 files changed, 65 insertions(+), 12 deletions(-)

diff --git a/src/Smalot/PdfParser/PDFObject.php b/src/Smalot/PdfParser/PDFObject.php
index 20e92334..ab3b66e8 100644
--- a/src/Smalot/PdfParser/PDFObject.php
+++ b/src/Smalot/PdfParser/PDFObject.php
@@ -227,15 +227,54 @@ private function formatContent(?string $content): string
         // Find all inline image content and replace them so they aren't
         // affected by the next steps
         $pdfInlineImages = [];
-        while (preg_match('/\sBI\s(.+?)\sID\s(.+?)\sEI(?=\s|$)/s', $content, $text)) {
-            $id = uniqid('IMAGE_', true);
-            $pdfInlineImages[$id] = [$text[1], $text[2]];
-            $content = preg_replace(
-                '/'.preg_quote($text[0], '/').'/',
-                '^^^'.$id.'^^^',
-                $content,
-                1
-            );
+        $offsetBI = 0;
+        while (preg_match('/\sBI\s(\/.+?)\sID\s(.+?)\sEI(?=\s|$)/s', $content, $text, \PREG_OFFSET_CAPTURE, $offsetBI)) {
+            // Attempt to detemine if this instance of the 'BI' command
+            // actually occured within a (string) using the following
+            // steps:
+
+            // Remove any escaped parentheses from the alleged image
+            // characteristics data
+            $para = str_replace(['\\(', '\\)'], '', $text[1][0]);
+
+            // Remove all correctly ordered and balanced parentheses
+            // from (strings)
+            do {
+                $paraTest = $para;
+                $para = preg_replace('/\(([^)]*)\)/', '$1', $paraTest);
+            } while ($para != $paraTest);
+
+            $paraOpen = strpos($para, '(');
+            $paraClose = strpos($para, ')');
+
+            // If the remaining text contains a close parenthesis ')'
+            // AND it occurs before any open parenthesis, then we are
+            // almost certain to be inside a (string)
+            if (0 < $paraClose && (false === $paraOpen || $paraClose < $paraOpen)) {
+                // Bump the search offset forward and match again
+                $offsetBI = (int) $text[1][1];
+                continue;
+            }
+
+            // Double check that this is actually inline image data by
+            // parsing the alleged image characteristics as a dictionary
+            $dict = $this->parseDictionary('<<'.$text[1][0].'>>');
+
+            // Check if an image Width and Height are set in the dict
+            if ((isset($dict['W']) || isset($dict['Width']))
+                && (isset($dict['H']) || isset($dict['Height']))) {
+                $id = uniqid('IMAGE_', true);
+                $pdfInlineImages[$id] = [
+                    preg_replace(['/\r\n/', '/\r/', '/\n/'], ' ', $text[1][0]),
+                    preg_replace(['/\r\n/', '/\r/', '/\n/'], '', $text[2][0]),
+                ];
+                $content = preg_replace(
+                    '/'.preg_quote($text[0][0], '/').'/',
+                    '^^^'.$id.'^^^',
+                    $content,
+                    1
+                );
+            }
         }
 
         // Find all strings () and replace them so they aren't affected
@@ -338,7 +377,7 @@ private function formatContent(?string $content): string
         foreach ($pdfInlineImages as $id => $image) {
             $content = str_replace(
                 '^^^'.$id.'^^^',
-                "\r\nBI\r\n".$image[0]."\r\nID\r\n".$image[1]."\r\nEI\r\n",
+                "\r\nBI\r\n".$image[0]." ID\r\n".$image[1]." EI\r\n",
                 $content
             );
         }
diff --git a/tests/PHPUnit/Integration/PDFObjectTest.php b/tests/PHPUnit/Integration/PDFObjectTest.php
index a2ee699f..947a6944 100644
--- a/tests/PHPUnit/Integration/PDFObjectTest.php
+++ b/tests/PHPUnit/Integration/PDFObjectTest.php
@@ -298,15 +298,29 @@ public function testFormatContentInlineImages(): void
 
         $cleaned = $formatContent->invoke(
             $this->getPdfObjectInstance(new Document()),
-            'q 65.30 0 0 18.00 412 707 cm BI /W 544 /H 150
+            'BT (This BI /W 258 /H 51 /should not trigger /as a /PDF command) TD ET q 65.30 0 0 18.00 412 707 cm BI /W 544 /H 150
 /BPC 1 /IM true /F [/A85 /Fl] ID Gb"0F_$L6!$j/a\$:ma&h\'JnJJ9S?O_EA-W+%D^ClCH=FP3s5M-gStQm\'5/hc`C?<Q)riWgtEe:Po0dY_-er6$jM@#?n`E+#(sa"0Gk3&K>CqL(^pV$_-er6Ik`"-1]Q ;~> EI Q /F002 10.00 Tf 0.00 Tw 0 g'
         );
 
         // PdfParser should not be fooled by Q's in inline image data;
         // Only one 'Q' command should be found
         $commandQ = preg_match_all('/Q\r\n/', $cleaned);
-
         $this->assertEquals(1, $commandQ);
+
+        // The 'BI' inside a string should not be interpreted as the
+        // beginning of an inline image command
+        $this->assertStringContainsString('(This BI /W 258 /H 51 /should not trigger /as a /PDF command) TD', $cleaned);
+
+        $cleaned = $formatContent->invoke(
+            $this->getPdfObjectInstance(new Document()),
+            'BT (This BI /W 258 /H 51 /should not () \) trigger /as a /PDF command) TD (There is no ID inline image in this data) TD (Nothing but text EI should be found) TD ET'
+        );
+
+        $this->assertEquals('BT'."\r\n".
+'(This BI /W 258 /H 51 /should not () \) trigger /as a /PDF command) TD'."\r\n".
+'(There is no ID inline image in this data) TD'."\r\n".
+'(Nothing but text EI should be found) TD'."\r\n".
+'ET', $cleaned);
     }
 
     public function testGetSectionsText(): void