BUG: Wrong page inserted when PdfMerger.merge is done

Caused-by: #207

Why it wasn't detected by the tests: We don't have any tests that check
for the correct result of a merge. We just check for exceptions

How we prevent it in future: Unit test was added

Risk of the fix:
- We will have bigger file sizes again as #207 was effectively reverted
- We will need to adjust this test if we change the way we write PDFs

Closes: #1062

PyPDF2/_writer.py

@@ -913,16 +913,8 @@ def _sweep_indirect_references(
                 if newobj is None:
                     try:
                         newobj = data.pdf.get_object(data)
-                        hash_value = None
-                        if newobj is not None:
-                            hash_value = newobj.hash_value()
-                        # Check if object is already added to pdf.
-                        if hash_value in self._idnum_hash:
-                            return IndirectObject(self._idnum_hash[hash_value], 0, self)
                         self._objects.append(None)  # placeholder
                         idnum = len(self._objects)
-                        if hash_value is not None:
-                            self._idnum_hash[hash_value] = idnum
                         newobj_ido = IndirectObject(idnum, 0, self)
                         if data.pdf not in extern_map:
                             extern_map[data.pdf] = {}

tests/test_workflows.py

@@ -244,6 +244,9 @@ def test_merge_with_warning(url, name):
     ):
         merger.write("tmp.merged.pdf")
 
+    # Cleanup
+    os.remove("tmp.merged.pdf")
+
 
 @pytest.mark.parametrize(
     ("url", "name"),
@@ -261,6 +264,9 @@ def test_merge(url, name):
     merger.append(reader)
     merger.write("tmp.merged.pdf")
 
+    # Cleanup
+    os.remove("tmp.merged.pdf")
+
 
 @pytest.mark.parametrize(
     ("url", "name"),
@@ -335,3 +341,31 @@ def test_scale_rectangle_indirect_object():
 
     for page in reader.pages:
         page.scale(sx=2, sy=3)
+
+
+def test_merge_output():
+    # Arrange
+    base = os.path.join(RESOURCE_ROOT, "Seige_of_Vicksburg_Sample_OCR.pdf")
+    crazy = os.path.join(RESOURCE_ROOT, "crazyones.pdf")
+    expected = os.path.join(
+        RESOURCE_ROOT, "Seige_of_Vicksburg_Sample_OCR-crazyones-merged.pdf"
+    )
+
+    # Act
+    merger = PdfMerger(strict=True)
+    with pytest.warns(PdfReadWarning):
+        merger.append(base)
+    merger.merge(1, crazy)
+    stream = BytesIO()
+    merger.write(stream)
+
+    # Assert
+    stream.seek(0)
+    actual = stream.read()
+    with open(expected, "rb") as fp:
+        expected_data = fp.read()
+    if actual != expected_data:
+        # See https://github.com/pytest-dev/pytest/issues/9124
+        assert (
+            False
+        ), f"len(actual) = {len(actual):,} vs len(expected) = {len(expected_data):,}"