BUG: Handle IndirectObject in CCITTFaxDecode filter (#2965)

* BUG: Handle IndirectObject in CCITTFaxDecode filter * fix types * further typing fix * add test * add missing annotation * update parameter order to keep backwards compatibility
py-pdf · Dec 25, 2024 · 6d4e2ec · 6d4e2ec
1 parent 1951fdf
commit 6d4e2ec
Show file tree

Hide file tree

Showing 3 changed files with 32 additions and 16 deletions.
diff --git a/pypdf/filters.py b/pypdf/filters.py
@@ -38,6 +38,7 @@
 import struct
 import zlib
 from base64 import a85decode
+from dataclasses import dataclass
 from io import BytesIO
 from typing import Any, Dict, List, Optional, Tuple, Union, cast
 
@@ -477,17 +478,17 @@ def decode(
         return data
 
 
+@dataclass
 class CCITParameters:
     """§7.4.6, optional parameters for the CCITTFaxDecode filter."""
 
-    def __init__(self, K: int = 0, columns: int = 0, rows: int = 0) -> None:
-        self.K = K
-        self.EndOfBlock = None
-        self.EndOfLine = None
-        self.EncodedByteAlign = None
-        self.columns = columns  # width
-        self.rows = rows  # height
-        self.DamagedRowsBeforeError = None
+    K: int = 0
+    columns: int = 0
+    rows: int = 0
+    EndOfBlock: Union[int, None] = None
+    EndOfLine: Union[int, None] = None
+    EncodedByteAlign: Union[int, None] = None
+    DamagedRowsBeforeError: Union[int, None] = None
 
     @property
     def group(self) -> int:
@@ -513,7 +514,7 @@ class CCITTFaxDecode:
     @staticmethod
     def _get_parameters(
         parameters: Union[None, ArrayObject, DictionaryObject, IndirectObject],
-        rows: int,
+        rows: Union[int, IndirectObject],
     ) -> CCITParameters:
         # §7.4.6, optional parameters for the CCITTFaxDecode filter
         k = 0
@@ -525,16 +526,16 @@ def _get_parameters(
             if isinstance(parameters_unwrapped, ArrayObject):
                 for decode_parm in parameters_unwrapped:
                     if CCITT.COLUMNS in decode_parm:
-                        columns = decode_parm[CCITT.COLUMNS]
+                        columns = decode_parm[CCITT.COLUMNS].get_object()
                     if CCITT.K in decode_parm:
-                        k = decode_parm[CCITT.K]
+                        k = decode_parm[CCITT.K].get_object()
             else:
                 if CCITT.COLUMNS in parameters_unwrapped:
-                    columns = parameters_unwrapped[CCITT.COLUMNS]  # type: ignore
+                    columns = parameters_unwrapped[CCITT.COLUMNS].get_object()  # type: ignore
                 if CCITT.K in parameters_unwrapped:
-                    k = parameters_unwrapped[CCITT.K]  # type: ignore
+                    k = parameters_unwrapped[CCITT.K].get_object()  # type: ignore
 
-        return CCITParameters(k, columns, rows)
+        return CCITParameters(K=k, columns=columns, rows=int(rows))
 
     @staticmethod
     def decode(

diff --git a/pypdf/generic/_base.py b/pypdf/generic/_base.py
@@ -396,6 +396,10 @@ def __float__(self) -> str:
         # in this case we are looking for the pointed data
         return self.get_object().__float__()  # type: ignore
 
+    def __int__(self) -> int:
+        # in this case we are looking for the pointed data
+        return self.get_object().__int__()  # type: ignore
+
     def __str__(self) -> str:
         # in this case we are looking for the pointed data
         return self.get_object().__str__()

diff --git a/tests/test_filters.py b/tests/test_filters.py
@@ -18,7 +18,7 @@
     CCITTFaxDecode,
     FlateDecode,
 )
-from pypdf.generic import ArrayObject, DictionaryObject, NameObject, NumberObject
+from pypdf.generic import ArrayObject, DictionaryObject, IndirectObject, NameObject, NumberObject
 
 from . import PILContext, get_data_from_url
 from .test_encryption import HAS_AES
@@ -196,14 +196,25 @@ def test_ccitparameters():
     ("parameters", "expected_k"),
     [
         (None, 0),
-        (ArrayObject([{"/K": 1}, {"/Columns": 13}]), 1),
+        (ArrayObject([{"/K": NumberObject(1)}, {"/Columns": NumberObject(13)}]), 1),
     ],
 )
 def test_ccitt_get_parameters(parameters, expected_k):
     parameters = CCITTFaxDecode._get_parameters(parameters=parameters, rows=0)
     assert parameters.K == expected_k  # noqa: SIM300
 
 
+def test_ccitt_get_parameters__indirect_object():
+    class Pdf:
+        def get_object(self, reference) -> NumberObject:
+            return NumberObject(42)
+
+    parameters = CCITTFaxDecode._get_parameters(
+        parameters=None, rows=IndirectObject(13, 1, Pdf())
+    )
+    assert parameters.rows == 42
+
+
 def test_ccitt_fax_decode():
     data = b""
     parameters = DictionaryObject(