ROB: Deal with content streams not containing streams (#3005)

Closes #2995.

pypdf/_page.py

@@ -2101,15 +2101,16 @@ def process_operation(operator: bytes, operands: List[Any]) -> None:
             elif operator == b"TJ":
                 # The space width may be smaller than the font width, so the width should be 95%.
                 _confirm_space_width = _space_width * 0.95
-                for op in operands[0]:
-                    if isinstance(op, (str, bytes)):
-                        process_operation(b"Tj", [op])
-                    if isinstance(op, (int, float, NumberObject, FloatObject)) and (
-                        (abs(float(op)) >= _confirm_space_width)
-                        and (len(text) > 0)
-                        and (text[-1] != " ")
-                    ):
-                        process_operation(b"Tj", [" "])
+                if operands:
+                    for op in operands[0]:
+                        if isinstance(op, (str, bytes)):
+                            process_operation(b"Tj", [op])
+                        if isinstance(op, (int, float, NumberObject, FloatObject)) and (
+                            (abs(float(op)) >= _confirm_space_width)
+                            and (len(text) > 0)
+                            and (text[-1] != " ")
+                        ):
+                            process_operation(b"Tj", [" "])
             elif operator == b"Do":
                 output += text
                 if visitor_text is not None:

pypdf/generic/_data_structures.py

@@ -1178,7 +1178,18 @@ def __init__(
             if isinstance(stream, ArrayObject):
                 data = b""
                 for s in stream:
-                    data += s.get_object().get_data()
+                    s_resolved = s.get_object()
+                    if isinstance(s_resolved, NullObject):
+                        continue
+                    if not isinstance(s_resolved, StreamObject):
+                        # No need to emit an exception here for now - the PDF structure
+                        # seems to already be broken beforehand in these cases.
+                        logger_warning(
+                            f"Expected StreamObject, got {type(s_resolved).__name__} instead. Data might be wrong.",
+                            __name__
+                        )
+                    else:
+                        data += s_resolved.get_data()
                     if len(data) == 0 or data[-1] != b"\n":
                         data += b"\n"
                 super().set_data(bytes(data))

tests/test_generic.py

@@ -1209,3 +1209,13 @@ def test_coverage_streamobject():
     co[NameObject("/testkey")] = NameObject("/test")
     co.decoded_self = DecodedStreamObject()
     assert "/testkey" in co.replicate(writer)
+
+
+def test_contentstream_arrayobject_containing_nullobject(caplog):
+    stream_object = DecodedStreamObject()
+    stream_object.set_data(b"Hello World!")
+
+    input_stream = ArrayObject([NullObject(), stream_object])
+    content_stream = ContentStream(stream=input_stream, pdf=None)
+    assert content_stream.get_data() == b"Hello World!\n"
+    assert caplog.text == ""

tests/test_text_extraction.py

@@ -11,7 +11,7 @@
 
 from pypdf import PdfReader, mult
 from pypdf._text_extraction import set_custom_rtl
-from pypdf.errors import ParseError
+from pypdf.errors import ParseError, PdfReadError
 
 from . import get_data_from_url
 
@@ -165,7 +165,7 @@ def test_layout_mode_indirect_sequence_font_widths():
     # Cover the situation where the sequence for font widths is an IndirectObject
     # ref https://github.com/py-pdf/pypdf/pull/2788
     url = "https://github.com/user-attachments/files/16491621/2788_example.pdf"
-    name ="2788_example.pdf"
+    name = "2788_example.pdf"
     reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
     assert reader.pages[0].extract_text(extraction_mode="layout") == ""
     url = "https://github.com/user-attachments/files/16491619/2788_example_malformed.pdf"
@@ -175,9 +175,11 @@ def test_layout_mode_indirect_sequence_font_widths():
         reader.pages[0].extract_text(extraction_mode="layout")
         assert str(exc.value).startswith("Invalid font width definition")
 
+
 def dummy_visitor_text(text, ctm, tm, fd, fs):
     pass
 
+
 @patch("pypdf._page.logger_warning")
 def test_layout_mode_warnings(mock_logger_warning):
     # Check that a warning is issued when an argument is ignored
@@ -274,6 +276,29 @@ def test_infinite_loop_arrays():
     assert "RNA structure comparison" in extracted
 
 
+@pytest.mark.enable_socket
+def test_content_stream_is_dictionary_object(caplog):
+    """Tests for #2995."""
+    url = "https://github.com/user-attachments/files/18049322/6fa5fd46-5f98-4a67-800d-5e2362b0164f.pdf"
+    name = "iss2995.pdf"
+    data = get_data_from_url(url, name=name)
+
+    reader = PdfReader(BytesIO(data))
+    page = reader.pages[0]
+    assert "\nYours faithfully   \n" in page.extract_text()
+    assert "Expected StreamObject, got DictionaryObject instead. Data might be wrong." in caplog.text
+    caplog.clear()
+
+    reader = PdfReader(BytesIO(data), strict=True)
+    page = reader.pages[0]
+    with pytest.raises(PdfReadError) as exception:
+        page.extract_text()
+    assert (
+        "Invalid Elementary Object starting with b\\'\\\\x18\\' @3557: b\\'ateDecode/Length 629\\\\x18ck["
+        in exception.value.args[0]
+    )
+
+
 @pytest.mark.enable_socket
 def test_tz_with_no_operands():
     """Tests for #2975"""