BUG: Make reader.get_fields also return dropdowns with options

PyPDF2/_reader.py

@@ -65,7 +65,11 @@
 from .constants import CatalogDictionary as CD
 from .constants import Core as CO
 from .constants import DocumentInformationAttributes as DI
-from .constants import FieldDictionaryAttributes, GoToActionArguments
+from .constants import (
+ FieldDictionaryAttributes,
+ GoToActionArguments,
+ CheckboxRadioButtonAttributes,
+)
 from .constants import PageAttributes as PG
 from .constants import PagesAttributes as PA
 from .constants import TrailerKeys as TK
@@ -478,6 +482,7 @@ def get_fields(
  ``None`` if form data could not be located.
  """
  field_attributes = FieldDictionaryAttributes.attributes_dict()
+ field_attributes.update(CheckboxRadioButtonAttributes.attributes_dict())
  if retval is None:
  retval = {}
  catalog = cast(DictionaryObject, self.trailer[TK.ROOT])
@@ -488,7 +493,6 @@ def get_fields(
  return None
  if tree is None:
  return retval
-
  self._check_kids(tree, retval, fileobj)
  for attr in field_attributes:
  if attr in tree:
@@ -548,7 +552,12 @@ def _check_kids(
  self.get_fields(kid.get_object(), retval, fileobj)
 
  def _write_field(self, fileobj: Any, field: Any, field_attributes: Any) -> None:
- for attr in FieldDictionaryAttributes.attributes():
+ field_attributes_tuple = FieldDictionaryAttributes.attributes()
+ field_attributes_tuple = (
+ field_attributes_tuple + CheckboxRadioButtonAttributes.attributes()
+ )
+
+ for attr in field_attributes_tuple:
  if attr in (
  FieldDictionaryAttributes.Kids,
  FieldDictionaryAttributes.AA,

PyPDF2/constants.py

@@ -332,6 +332,22 @@ def attributes_dict(cls) -> Dict[str, str]:
  }
 
 
+class CheckboxRadioButtonAttributes:
+ """TABLE 8.76 Field flags common to all field types"""
+
+ Opt = "/Opt" # Options, Optional
+
+ @classmethod
+ def attributes(cls) -> Tuple[str, ...]:
+ return (cls.Opt,)
+
+ @classmethod
+ def attributes_dict(cls) -> Dict[str, str]:
+ return {
+ cls.Opt: "Options",
+ }
+
+
 class FieldFlag(IntFlag):
  """TABLE 8.70 Field flags common to all field types"""
 
@@ -411,6 +427,7 @@ class CatalogDictionary:
  CatalogAttributes,
  CatalogDictionary,
  CcittFaxDecodeParameters,
+ CheckboxRadioButtonAttributes,
  ColorSpaces,
  Core,
  DocumentInformationAttributes,

PyPDF2/generic.py

@@ -68,7 +68,7 @@
  skip_over_comment,
  str_,
 )
-from .constants import FieldDictionaryAttributes
+from .constants import CheckboxRadioButtonAttributes, FieldDictionaryAttributes
 from .constants import FilterTypes as FT
 from .constants import StreamAttributes as SA
 from .constants import TypArguments as TA
@@ -1618,8 +1618,9 @@ class Field(TreeObject):
 
  def __init__(self, data: Dict[str, Any]) -> None:
  DictionaryObject.__init__(self)
-
- for attr in FieldDictionaryAttributes.attributes():
+ Field_attributes = FieldDictionaryAttributes.attributes()
+ Field_attributes = Field_attributes + CheckboxRadioButtonAttributes.attributes()
+ for attr in Field_attributes:
  try:
  self[NameObject(attr)] = data[attr]
  except KeyError:

tests/test_generic.py

@@ -11,6 +11,7 @@
  Bookmark,
  BooleanObject,
  ByteStringObject,
+ CheckboxRadioButtonAttributes,
  Destination,
  DictionaryObject,
  FloatObject,
@@ -491,3 +492,7 @@ def test_issue_997():
 
  # cleanup
  os.remove(merged_filename)
+
+
+def test_CheckboxRadioButtonAttributes_opt():
+ assert "/Opt" in CheckboxRadioButtonAttributes.attributes_dict()

tests/test_workflows.py

@@ -22,6 +22,13 @@
 sys.path.append(PROJECT_ROOT)
 
 
+def test_dropdown_items():
+ with open(os.path.join(RESOURCE_ROOT, "libreoffice-form.pdf"), "rb") as inputfile:
+ reader = PdfReader(inputfile)
+ fields = reader.get_fields()
+ assert "/Opt" in fields["Nationality"].keys()
+
+
 def test_PdfReaderFileLoad():
  """
  Test loading and parsing of a file. Extract text of the file and compare to expected