ROB: Raise proper error on infinite loop when reading objects (#3169)

Closes #3112.

pypdf/_reader.py

@@ -40,6 +40,7 @@
     Iterable,
     List,
     Optional,
+    Set,
     Tuple,
     Type,
     Union,
@@ -133,6 +134,7 @@ def __init__(
         self._validated_root: Optional[DictionaryObject] = None
 
         self._initialize_stream(stream)
+        self._known_objects: Set[Tuple[int, int]] = set()
 
         self._override_encryption = False
         self._encryption: Optional[Encryption] = None
@@ -447,7 +449,13 @@ def get_object(
                 )
             if self.strict:
                 assert generation == indirect_reference.generation
+
+            current_object = (indirect_reference.idnum, indirect_reference.generation)
+            if current_object in self._known_objects:
+                raise PdfReadError(f"Detected loop with self reference for {indirect_reference!r}.")
+            self._known_objects.add(current_object)
             retval = read_object(self.stream, self)  # type: ignore
+            self._known_objects.remove(current_object)
 
             # override encryption is used for the /Encrypt dictionary
             if not self._override_encryption and self._encryption is not None:

tests/test_reader.py

@@ -1806,3 +1806,15 @@ def test_issue2886(caplog):
 
     with pytest.raises(PdfReadError, match="Unexpected empty line in Xref table."):
         _ = PdfReader(BytesIO(get_data_from_url(url, name=name)))
+
+
+@pytest.mark.enable_socket
+def test_infinite_loop_for_length_value():
+    """Tests for #3112"""
+    url = "https://github.com/user-attachments/files/19106009/Special.n.15.du.jeudi.22.fevrier.2024.pdf"
+    name = "issue3112.pdf"
+
+    reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
+    writer = PdfWriter()
+    with pytest.raises(PdfReadError, match=r"^Detected loop with self reference for IndirectObject\(165, 0, \d+\)\.$"):
+        writer.add_page(reader.pages[0])