ENH: Add page deletion feature to PdfWriter (#1843)

Note: This does not delete the objects from the PDF, just the page entry

pypdf/_page.py

@@ -2451,6 +2451,47 @@ def __getitem__(
  raise IndexError("sequence index out of range")
  return self.get_function(index)
 
+ def __delitem__(self, index: Union[int, slice]) -> None:
+ if isinstance(index, slice):
+ r = list(range(*index.indices(len(self))))
+ # pages have to be deleted from last to first
+ r.sort()
+ r.reverse()
+ for p in r:
+ del self[p]
+ return
+ if not isinstance(index, int):
+ raise TypeError("index must be integers")
+ len_self = len(self)
+ if index < 0:
+ # support negative indexes
+ index = len_self + index
+ if index < 0 or index >= len_self:
+ raise IndexError("index out of range")
+ ind = self[index].indirect_reference
+ assert ind is not None
+ parent = cast(DictionaryObject, ind.get_object()).get("/Parent", None)
+ while parent is not None:
+ parent = cast(DictionaryObject, parent.get_object())
+ try:
+ i = parent["/Kids"].index(ind)
+ del parent["/Kids"][i]
+ try:
+ assert ind is not None
+ del ind.pdf.flattened_pages[index] # case of page in a Reader
+ except AttributeError:
+ pass
+ if "/Count" in parent:
+ parent[NameObject("/Count")] = NumberObject(parent["/Count"] - 1)
+ if len(parent["/Kids"]) == 0:
+ # No more objects in this part of this sub tree
+ ind = parent.indirect_reference
+ parent = cast(DictionaryObject, parent.get("/Parent", None))
+ else:
+ parent = None
+ except ValueError: # from index
+ raise PdfReadError(f"Page Not Found in Page Tree {ind}")
+
  def __iter__(self) -> Iterator[PageObject]:
  for i in range(len(self)):
  yield self[i]

pypdf/_writer.py

@@ -487,7 +487,18 @@ def getNumPages(self) -> int: # deprecated
 
  @property
  def pages(self) -> List[PageObject]:
- """Property that emulates a list of :class:`PageObject<pypdf._page.PageObject>`."""
+ """
+ Property that emulates a list of :class:`PageObject<pypdf._page.PageObject>`.
+ this property allows to get a page or a range of pages.
+
+ It provides also capability to remove a page/range of page from the list
+ (through del operator)
+ Note: only the page entry is removed. As the objects beneath can be used
+ somewhere else.
+ a solution to completely remove them - if they are not used somewhere -
+ is to write to a buffer/temporary and to then load it into a new PdfWriter
+ object.
+ """
  return _VirtualList(self._get_num_pages, self.get_page) # type: ignore
 
  def add_blank_page(

tests/test_page.py

@@ -11,7 +11,7 @@
 from pypdf import PdfReader, PdfWriter, Transformation
 from pypdf._page import PageObject
 from pypdf.constants import PageAttributes as PG
-from pypdf.errors import DeprecationError, PdfReadWarning
+from pypdf.errors import DeprecationError, PdfReadError, PdfReadWarning
 from pypdf.generic import (
  ArrayObject,
  ContentStream,
@@ -1143,6 +1143,46 @@ def test_pages_printing():
  reader.pages[0].images["~1~"]
 
 
+@pytest.mark.enable_socket()
+def test_del_pages():
+ url = "https://corpora.tika.apache.org/base/docs/govdocs1/941/941536.pdf"
+ name = "tika-941536.pdf"
+ writer = PdfWriter(clone_from=BytesIO(get_pdf_from_url(url, name=name)))
+ ll = len(writer.pages)
+ pp = writer.pages[1].indirect_reference
+ del writer.pages[1]
+ assert len(writer.pages) == ll - 1
+ pages = writer._pages.get_object()
+ assert pages["/Count"] == ll - 1
+ assert len(pages["/Kids"]) == ll - 1
+ assert pp not in pages["/Kids"]
+ del writer.pages[-2]
+ with pytest.raises(TypeError):
+ del writer.pages["aa"]
+ with pytest.raises(IndexError):
+ del writer.pages[9999]
+ pp = tuple(p.indirect_reference for p in writer.pages[3:5])
+ ll = len(writer.pages)
+ del writer.pages[3:5]
+ assert len(writer.pages) == ll - 2
+ for p in pp:
+ assert p not in pages["/Kids"]
+ # del whole arborescence
+ reader = PdfReader(BytesIO(get_pdf_from_url(url, name=name)))
+ # error case
+ pp = reader.pages[2]
+ i = pp["/Parent"].get_object()["/Kids"].index(pp.indirect_reference)
+ del pp["/Parent"].get_object()["/Kids"][i]
+ with pytest.raises(PdfReadError):
+ del reader.pages[2]
+ # reader is corrupted we have to reload it
+ reader = PdfReader(BytesIO(get_pdf_from_url(url, name=name)))
+ del reader.pages[:]
+ assert len(reader.pages) == 0
+ assert len(reader.trailer["/Root"]["/Pages"]["/Kids"]) == 0
+ assert len(reader.flattened_pages) == 0
+
+
 def test_pdf_pages_missing_type():
  pdf_path = RESOURCE_ROOT / "crazyones.pdf"
  reader = PdfReader(pdf_path)