BUG: Generate static appearance for fields

pypdf/_reader.py

@@ -69,11 +69,11 @@
 from .constants import CatalogDictionary as CD
 from .constants import (
  CheckboxRadioButtonAttributes,
- FieldDictionaryAttributes,
  GoToActionArguments,
 )
 from .constants import Core as CO
 from .constants import DocumentInformationAttributes as DI
+from .constants import FieldDictionaryAttributes as FA
 from .constants import PageAttributes as PG
 from .constants import PagesAttributes as PA
 from .constants import TrailerKeys as TK
@@ -542,7 +542,7 @@ def get_fields(
  default, the mapping name is used for keys.
  ``None`` if form data could not be located.
  """
- field_attributes = FieldDictionaryAttributes.attributes_dict()
+ field_attributes = FA.attributes_dict()
  field_attributes.update(CheckboxRadioButtonAttributes.attributes_dict())
  if retval is None:
  retval = {}
@@ -626,6 +626,26 @@ def _build_field(
  self._write_field(fileobj, field, field_attributes)
  fileobj.write("\n")
  retval[key] = Field(field)
+ obj = retval[key].indirect_reference.get_object() # to get the full object
+ if obj.get(FA.FT, "") == "/Ch":
+ retval[key][NameObject("/_States_")] = obj[NameObject(FA.Opt)]
+ if obj.get(FA.FT, "") == "/Btn" and "/AP" in obj:
+ # Checkbox
+ retval[key][NameObject("/_States_")] = ArrayObject(
+ list(obj["/AP"]["/N"].keys())
+ )
+ if "/Off" not in retval[key]["/_States_"]:
+ retval[key][NameObject("/_States_")].append(NameObject("/Off"))
+ elif obj.get(FA.FT, "") == "/Btn" and obj.get(FA.Ff, 0) & FA.FfBits.Radio != 0:
+ states = []
+ for k in obj.get(FA.Kids, {}):
+ k = k.get_object()
+ for s in list(k["/AP"]["/N"].keys()):
+ if s not in states:
+ states.append(s)
+ retval[key][NameObject("/_States_")] = ArrayObject(states)
+ if obj.get(FA.Ff, 0) & FA.FfBits.NoToggleToOff != 0:
+ del retval[key]["/_States_"][retval[key]["/_States_"].index("/Off")]
 
  def _check_kids(
  self, tree: Union[TreeObject, DictionaryObject], retval: Any, fileobj: Any
@@ -636,20 +656,20 @@ def _check_kids(
  self.get_fields(kid.get_object(), retval, fileobj)
 
  def _write_field(self, fileobj: Any, field: Any, field_attributes: Any) -> None:
- field_attributes_tuple = FieldDictionaryAttributes.attributes()
+ field_attributes_tuple = FA.attributes()
  field_attributes_tuple = (
  field_attributes_tuple + CheckboxRadioButtonAttributes.attributes()
  )
 
  for attr in field_attributes_tuple:
  if attr in (
- FieldDictionaryAttributes.Kids,
- FieldDictionaryAttributes.AA,
+ FA.Kids,
+ FA.AA,
  ):
  continue
  attr_name = field_attributes[attr]
  try:
- if attr == FieldDictionaryAttributes.FT:
+ if attr == FA.FT:
  # Make the field type value more clear
  types = {
  "/Btn": "Button",
@@ -659,12 +679,12 @@ def _write_field(self, fileobj: Any, field: Any, field_attributes: Any) -> None:
  }
  if field[attr] in types:
  fileobj.write(f"{attr_name}: {types[field[attr]]}\n")
- elif attr == FieldDictionaryAttributes.Parent:
+ elif attr == FA.Parent:
  # Let's just write the name of the parent
  try:
- name = field[attr][FieldDictionaryAttributes.TM]
+ name = field[attr][FA.TM]
  except KeyError:
- name = field[attr][FieldDictionaryAttributes.T]
+ name = field[attr][FA.T]
  fileobj.write(f"{attr_name}: {name}\n")
  else:
  fileobj.write(f"{attr_name}: {field[attr]}\n")

pypdf/_writer.py

@@ -32,7 +32,6 @@
 import decimal
 import enum
 import hashlib
-import logging
 import re
 import struct
 import uuid
@@ -72,10 +71,10 @@
  deprecation_with_replacement,
  logger_warning,
 )
+from .constants import AnnotationDictionaryAttributes as AA
+from .constants import CatalogAttributes as CA
 from .constants import (
- AnnotationDictionaryAttributes,
  CatalogDictionary,
- FieldDictionaryAttributes,
  FieldFlag,
  FileSpecificationDictionaryEntries,
  GoToActionArguments,
@@ -84,9 +83,11 @@
  TypFitArguments,
  UserAccessPermissions,
 )
-from .constants import CatalogAttributes as CA
 from .constants import Core as CO
 from .constants import EncryptionDictAttributes as ED
+from .constants import (
+ FieldDictionaryAttributes as FA,
+)
 from .constants import PageAttributes as PG
 from .constants import PagesAttributes as PA
 from .constants import StreamAttributes as SA
@@ -127,9 +128,6 @@
  ZoomArgType,
 )
 
-logger = logging.getLogger(__name__)
-
-
 OPTIONAL_READ_WRITE_FIELD = FieldFlag(0)
 ALL_DOCUMENT_PERMISSIONS = UserAccessPermissions((2**31 - 1) - 3)
 
@@ -332,7 +330,20 @@ def _add_page(
  pages[NameObject(PA.COUNT)] = NumberObject(page_count + 1)
  return page
 
- def set_need_appearances_writer(self) -> None:
+ def set_need_appearances_writer(self, state: bool = True) -> None:
+ """
+ Sets the "NeedAppearances" flag in the PDF writer.
+
+ The "NeedAppearances" flag indicates whether the appearance dictionary
+ for form fields should be automatically generated by the PDF viewer or
+ if the embedded appearence should be used.
+
+ Args:
+ state: The actual value of the NeedAppearances flag.
+
+ Returns:
+ None
+ """
  # See 12.7.2 and 7.7.2 for more information:
  # http://www.adobe.com/content/dam/acom/en/devnet/acrobat/pdfs/PDF32000_2008.pdf
  try:
@@ -343,9 +354,13 @@ def set_need_appearances_writer(self) -> None:
  ] = self._add_object(DictionaryObject())
 
  need_appearances = NameObject(InteractiveFormDictEntries.NeedAppearances)
- self._root_object[CatalogDictionary.ACRO_FORM][need_appearances] = BooleanObject(True) # type: ignore
- except Exception as exc:
- logger.error("set_need_appearances_writer() catch : %s", repr(exc))
+ cast(DictionaryObject, self._root_object[CatalogDictionary.ACRO_FORM])[
+ need_appearances
+ ] = BooleanObject(state)
+ except Exception as exc: # pragma: no cover
+ logger_warning(
+ f"set_need_appearances_writer({state}) catch : {exc}", __name__
+ )
 
  def add_page(
  self,
@@ -801,11 +816,96 @@ def _get_qualified_field_name(self, parent: DictionaryObject) -> Optional[str]:
  return qualified_parent + "." + cast(str, parent["/T"])
  return cast(str, parent["/T"])
 
+ def _update_text_field(self, field: DictionaryObject) -> None:
+ # Calculate rectangle dimensions
+ _rct = cast(RectangleObject, field[AA.Rect])
+ rct = RectangleObject((0, 0, _rct[2] - _rct[0], _rct[3] - _rct[1]))
+
+ # Extract font information
+ font_properties: Any = (
+ cast(str, field[AA.DA]).replace("\n", " ").replace("\r", " ").split(" ")
+ )
+ font_name = font_properties[font_properties.index("Tf") - 2]
+ font_height = float(font_properties[font_properties.index("Tf") - 1])
+ y_offset = rct.height - 1 - font_height
+
+ # Retrieve field text and selected values
+ field_flags = field.get(FA.Ff, 0)
+ if field.get(FA.FT, "/Tx") == "/Ch" and field_flags & FA.FfBits.Combo == 0:
+ txt = "\n".join(field.get(FA.Opt, {}))
+ sel = field.get("/V", [])
+ if not isinstance(sel, list):
+ sel = [sel]
+ else: # /Tx
+ txt = field.get("/V", "")
+ sel = []
+
+ # Generate appearance stream
+ ap_stream = f"q\n/Tx BMC \nq\n1 1 {rct.width - 1} {rct.height - 1} re\nW\nBT\n{field[AA.DA]}\n".encode()
+ for line_number, line in enumerate(txt.replace("\n", "\r").split("\r")):
+ if line in sel:
+ # may be improved but can not find how get fill working => replaced with lined box
+ ap_stream += (
+ f"1 {y_offset - (line_number * font_height * 1.4) - 1} {rct.width - 2} {font_height + 2} re\n"
+ f"0.5 0.5 0.5 rg s\n{field[AA.DA]}\n"
+ ).encode()
+ if line_number == 0:
+ ap_stream += f"2 {y_offset} Td\n".encode()
+ else:
+ # Td is a relative translation
+ ap_stream += f"0 {- font_height * 1.4} Td\n".encode()
+ ap_stream += b"(" + str(line).encode("UTF-8") + b") Tj\n"
+ ap_stream += b"ET\nQ\nEMC\nQ\n"
+
+ # Create appearance dictionary
+ dct = DecodedStreamObject.initialize_from_dictionary(
+ {
+ NameObject("/Type"): NameObject("/XObject"),
+ NameObject("/Subtype"): NameObject("/Form"),
+ NameObject("/BBox"): rct,
+ "__streamdata__": ByteStringObject(ap_stream),
+ "/Length": 0,
+ }
+ )
+
+ # Retrieve font information from AcroForm dictionary
+ dr: Any = cast(
+ dict, cast(DictionaryObject, self._root_object["/AcroForm"]).get("/DR", {})
+ )
+ if isinstance(dr, IndirectObject):
+ dr = dr.get_object()
+ dr = dr.get("/Font", {})
+ if isinstance(dr, IndirectObject):
+ dr = dr.get_object()
+
+ # Update Resources with font information if necessary
+ if font_name in dr:
+ dct[NameObject("/Resources")] = DictionaryObject(
+ {
+ NameObject("/Font"): DictionaryObject(
+ {NameObject(font_name): dr[font_name].indirect_reference}
+ )
+ }
+ )
+ if AA.AP not in field:
+ field[NameObject(AA.AP)] = DictionaryObject(
+ {NameObject("/N"): self._add_object(dct)}
+ )
+ elif "/N" not in cast(DictionaryObject, field[AA.AP]):
+ cast(DictionaryObject, field[NameObject(AA.AP)])[
+ NameObject("/N")
+ ] = self._add_object(dct)
+ else: # [/AP][/N] exists
+ n = field[AA.AP]["/N"].indirect_reference.idnum # type: ignore
+ self._objects[n - 1] = dct
+ dct.indirect_reference = IndirectObject(n, 0, self)
+
  def update_page_form_field_values(
  self,
  page: PageObject,
  fields: Dict[str, Any],
  flags: FieldFlag = OPTIONAL_READ_WRITE_FIELD,
+ auto_regen: Optional[bool] = True,
  ) -> None:
  """
  Update the form field values for a given page from a fields dictionary.
@@ -821,57 +921,57 @@ def update_page_form_field_values(
  flags: An integer (0 to 7). The first bit sets ReadOnly, the
  second bit sets Required, the third bit sets NoExport. See
  PDF Reference Table 8.70 for details.
+ auto_regen: set/unset the need_appearances flag ;
+ the flag is unchanged if auto_regen is None
  """
- self.set_need_appearances_writer()
+ if isinstance(auto_regen, bool):
+ self.set_need_appearances_writer(auto_regen)
  # Iterate through pages, update field values
  if PG.ANNOTS not in page:
  logger_warning("No fields to update on this page", __name__)
  return
- for j in range(len(page[PG.ANNOTS])): # type: ignore
- writer_annot = page[PG.ANNOTS][j].get_object() # type: ignore
+ for writer_annot in page[PG.ANNOTS]: # type: ignore
+ writer_annot = cast(DictionaryObject, writer_annot.get_object())
  # retrieve parent field values, if present
- writer_parent_annot = DictionaryObject() # fallback if it's not there
- if PG.PARENT in writer_annot:
-  writer_parent_annot = writer_annot[PG.PARENT]
- for field in fields:
+ writer_parent_annot = writer_annot.get(
+  PG.PARENT, DictionaryObject()
+ ).get_object()
+ for field, value in fields.items():
  if (
- writer_annot.get(FieldDictionaryAttributes.T) == field
+ writer_annot.get(FA.T) == field
  or self._get_qualified_field_name(writer_annot) == field
  ):
- if writer_annot.get(FieldDictionaryAttributes.FT) == "/Btn":
- writer_annot.update(
- {
- NameObject(
- AnnotationDictionaryAttributes.AS
- ): NameObject(fields[field])
- }
- )
- writer_annot.update(
- {
- NameObject(FieldDictionaryAttributes.V): TextStringObject(
- fields[field]
- )
- }
- )
+ if isinstance(value, list):
+ lst = ArrayObject()
+ for v in value:
+ lst.append(TextStringObject(v))
+ writer_annot[NameObject(FA.V)] = lst
+ else:
+ writer_annot[NameObject(FA.V)] = TextStringObject(value)
+ if writer_annot.get(FA.FT) in ("/Btn"):
+ # case of Checkbox button (no /FT found in Radio widgets
+ writer_annot[NameObject(AA.AS)] = NameObject(value)
+ elif (
+ writer_annot.get(FA.FT) == "/Tx"
+ or writer_annot.get(FA.FT) == "/Ch"
+ ):
+ # textbox
+ self._update_text_field(writer_annot)
+ elif writer_annot.get(FA.FT) == "/Sig":
+ # signature
+ logger_warning("Signature forms not implemented yet", __name__)
  if flags:
- writer_annot.update(
- {
- NameObject(FieldDictionaryAttributes.Ff): NumberObject(
- flags
- )
- }
- )
+ writer_annot[NameObject(FA.Ff)] = NumberObject(flags)
  elif (
- writer_parent_annot.get(FieldDictionaryAttributes.T) == field
+ writer_parent_annot.get(FA.T) == field
  or self._get_qualified_field_name(writer_parent_annot) == field
  ):
- writer_parent_annot.update(
- {
- NameObject(FieldDictionaryAttributes.V): TextStringObject(
- fields[field]
- )
- }
- )
+ writer_parent_annot[NameObject(FA.V)] = TextStringObject(value)
+ for k in writer_parent_annot[NameObject(FA.Kids)]:
+ k = k.get_object()
+ k[NameObject(AA.AS)] = NameObject(
+ value if value in k[AA.AP]["/N"] else "/Off"
+ )
 
  def updatePageFormFieldValues(
  self,
@@ -2166,14 +2266,12 @@ def add_uri(
  lnk = DictionaryObject()
  lnk.update(
  {
- NameObject(AnnotationDictionaryAttributes.Type): NameObject(PG.ANNOTS),
- NameObject(AnnotationDictionaryAttributes.Subtype): NameObject("/Link"),
- NameObject(AnnotationDictionaryAttributes.P): page_link,
- NameObject(AnnotationDictionaryAttributes.Rect): rect,
+ NameObject(AA.Type): NameObject(PG.ANNOTS),
+ NameObject(AA.Subtype): NameObject("/Link"),
+ NameObject(AA.P): page_link,
+ NameObject(AA.Rect): rect,
  NameObject("/H"): NameObject("/I"),
- NameObject(AnnotationDictionaryAttributes.Border): ArrayObject(
- border_arr
- ),
+ NameObject(AA.Border): ArrayObject(border_arr),
  NameObject("/A"): lnk2,
  }
  )