From ea15ab290c48c162fa2835ff24e493e8e6372e6d Mon Sep 17 00:00:00 2001
From: Markus Ast <m@rkusa.st>
Date: Mon, 5 Sep 2022 17:59:11 +0200
Subject: [PATCH] allow endstream position to be off by one

---
 lib/object/object.js | 45 +++++++++++++++++++++++++++++++++++---------
 1 file changed, 36 insertions(+), 9 deletions(-)

diff --git a/lib/object/object.js b/lib/object/object.js
index bd9ffcb..c2f2f43 100644
--- a/lib/object/object.js
+++ b/lib/object/object.js
@@ -103,18 +103,17 @@ class PDFObject {
           lexer.pos = pos
         }
 
-        const PDFStream = require('./stream') // lazy load, cause circular referecnes
+        const PDFStream = require('./stream') // lazy load, cause circular references
         const stream = new PDFStream(obj)
         stream.content = lexer.read(length)
-        lexer.skipEOL(1, true)
-
-        // not to be expected according to the PDF spec, but there are some PDF files that indent
-        // the stream
-        lexer.skipWhitespace(null, true)
 
-        if (lexer.readString(9) !== 'endstream') {
-          throw new Error('Invalid stream: `endstream` not found')
-        }
+        const orphan = parseEndStream(lexer)
+        if (orphan) {
+          const merged = new Uint8Array(stream.content.length + orphan.length);
+          merged.set(stream.content);
+          merged.set(orphan, stream.content.length);
+          stream.content = merged
+      }
 
         lexer.skipEOL(1, true)
       }
@@ -126,4 +125,32 @@ class PDFObject {
   }
 }
 
+/**
+ * Parse an `endstream` while being tolerant to different ways of how different producers understood
+ * the spec. If there ends up being any extra content left before the `endstream`, it is returned.
+ */
+function parseEndStream(lexer) {
+  // some producer put the endstream at the end of the last line, and some to the next line
+  // -> accept both
+  lexer.skipEOL(1, true)
+
+  // not to be expected according to the PDF spec, but there are some PDF files that indent
+  // the stream
+  lexer.skipWhitespace(null, true)
+
+  const end = lexer.readString(9)
+  if (end === 'endstream') {
+    return null
+  }
+
+  // allow off by one error for length here (#285)
+  if (end.endsWith("endstrea")) {
+    if (lexer.readString(1) === "m") {
+      return Uint8Array.of(end.charCodeAt(0))
+    }
+  }
+
+  throw new Error('Invalid stream: `endstream` not found')
+}
+
 module.exports = PDFObject