ENH: Add Cloning

PyPDF2/_merger.py

@@ -702,6 +702,7 @@ def add_outline_item(
             title,
             page_number,
             parent,
+            None,
             color,
             bold,
             italic,

PyPDF2/_page.py

@@ -46,6 +46,7 @@
 )
 
 from ._cmap import build_char_map, unknown_char_map
+from ._protocols import PdfReaderProtocol
 from ._utils import (
     CompressedTransformationMatrix,
     File,
@@ -288,16 +289,17 @@ class PageObject(DictionaryObject):
         this object in its source PDF
     """
 
+    original_page: "PageObject"  # very local use in writer when appending
+
     def __init__(
         self,
-        pdf: Optional[Any] = None,  # PdfReader
+        pdf: Optional[PdfReaderProtocol] = None,
         indirect_reference: Optional[IndirectObject] = None,
         indirect_ref: Optional[IndirectObject] = None,
     ) -> None:
-        from ._reader import PdfReader
 
         DictionaryObject.__init__(self)
-        self.pdf: Optional[PdfReader] = pdf
+        self.pdf: Optional[PdfReaderProtocol] = pdf
         if indirect_ref is not None:  # deprecated
             warnings.warn(
                 "Use indirect_reference instead of indirect_ref.", DeprecationWarning

PyPDF2/_protocols.py

@@ -0,0 +1,65 @@
+"""Helpers for working with PDF types."""
+
+from io import BufferedReader, BufferedWriter, BytesIO, FileIO
+from pathlib import Path
+from typing import Any, Dict, List, Optional, Tuple, Union
+
+try:
+    # Python 3.8+: https://peps.python.org/pep-0586
+    from typing import Protocol  # type: ignore[attr-defined]
+except ImportError:
+    from typing_extensions import Protocol  # type: ignore[misc]
+
+from ._utils import StrByteType
+
+
+class PdfObjectProtocol(Protocol):
+    indirect_reference: Any
+
+    def clone(
+        self,
+        pdf_dest: Any,
+        force_duplicate: bool = False,
+        ignore_fields: Union[Tuple[str, ...], List[str], None] = (),
+    ) -> Any:
+        ...
+
+    def _reference_clone(self, clone: Any, pdf_dest: Any) -> Any:
+        ...
+
+    def get_object(self) -> Optional["PdfObjectProtocol"]:
+        ...
+
+
+class PdfReaderProtocol(Protocol):  # pragma: no cover
+    @property
+    def pdf_header(self) -> str:
+        ...
+
+    @property
+    def strict(self) -> bool:
+        ...
+
+    @property
+    def xref(self) -> Dict[int, Dict[int, Any]]:
+        ...
+
+    @property
+    def pages(self) -> List[Any]:
+        ...
+
+    def get_object(self, indirect_reference: Any) -> Optional[PdfObjectProtocol]:
+        ...
+
+
+class PdfWriterProtocol(Protocol):  # pragma: no cover
+    _objects: List[Any]
+    _id_translated: Dict[int, Dict[int, int]]
+
+    def get_object(self, indirect_reference: Any) -> Optional[PdfObjectProtocol]:
+        ...
+
+    def write(
+        self, stream: Union[Path, StrByteType]
+    ) -> Tuple[bool, Union[FileIO, BytesIO, BufferedReader, BufferedWriter]]:
+        ...

PyPDF2/_reader.py

@@ -972,6 +972,7 @@ def _build_outline_item(self, node: DictionaryObject) -> Optional[Destination]:
                 # absolute value = num. visible children
                 # positive = open/unfolded, negative = closed/folded
                 outline_item[NameObject("/Count")] = node["/Count"]
+        outline_item.node = node
         return outline_item
 
     @property
@@ -1389,6 +1390,8 @@ def cache_indirect_object(
                 raise PdfReadError(msg)
             logger_warning(msg, __name__)
         self.resolved_objects[(generation, idnum)] = obj
+        if obj is not None:
+            obj.indirect_reference = IndirectObject(idnum, generation, self)
         return obj
 
     def cacheIndirectObject(