Fix checkbox marks disappearing during PDF flattening

commonforms/form_creator.py

@@ -1,11 +1,12 @@
-from pypdf import PdfWriter, PdfReader
+from pypdf import PdfReader, PdfWriter
 from pypdf.annotations import AnnotationDictionary
 from pypdf.generic import (
-    NameObject,
     ArrayObject,
+    DecodedStreamObject,
+    DictionaryObject,
+    NameObject,
     NumberObject,
     TextStringObject,
-    DictionaryObject,
 )
 
 from commonforms.utils import BoundingBox
@@ -83,6 +84,44 @@ def __init__(
         super().__init__()
         pdf_value = NameObject("/Off") if not value else NameObject("/Yes")
 
+        width = max(float(rect[2]) - float(rect[0]), 1)
+        height = max(float(rect[3]) - float(rect[1]), 1)
+
+        def appearance(checked: bool) -> DecodedStreamObject:
+            stream = DecodedStreamObject()
+            stream.update(
+                {
+                    NameObject("/Type"): NameObject("/XObject"),
+                    NameObject("/Subtype"): NameObject("/Form"),
+                    NameObject("/BBox"): ArrayObject(
+                        [
+                            NumberObject(0),
+                            NumberObject(0),
+                            NumberObject(width),
+                            NumberObject(height),
+                        ]
+                    ),
+                    NameObject("/Resources"): DictionaryObject(),
+                }
+            )
+            if checked:
+                stream.set_data(
+                    f"q 0 0 0 RG {min(width, height) * 0.12:.4f} w "
+                    f"{width * 0.18:.4f} {height * 0.50:.4f} m "
+                    f"{width * 0.43:.4f} {height * 0.23:.4f} l "
+                    f"{width * 0.84:.4f} {height * 0.78:.4f} l S Q".encode()
+                )
+            else:
+                stream.set_data(b"")
+            return stream
+
+        normal_appearance = DictionaryObject(
+            {
+                NameObject("/Off"): appearance(False),
+                NameObject("/Yes"): appearance(True),
+            }
+        )
+
         self.update(
             {
                 NameObject("/Type"): NameObject("/Annot"),
@@ -93,6 +132,9 @@ def __init__(
                 NameObject("/Rect"): rect,
                 NameObject("/V"): pdf_value,
                 NameObject("/AS"): pdf_value,
+                NameObject("/AP"): DictionaryObject(
+                    {NameObject("/N"): normal_appearance}
+                ),
                 NameObject("/T"): TextStringObject(name),
             }
         )
@@ -162,6 +204,18 @@ def add_text_box(
     def add_checkbox(self, name: str, page: int, bounding_box: BoundingBox) -> None:
         rect = rect_for(bounding_box, self.writer.pages[page])
         checkbox = Checkbox(name=name, rect=rect)
+        appearance = checkbox[NameObject("/AP")]
+        if not isinstance(appearance, DictionaryObject):
+            raise TypeError(
+                "Generated checkbox has an invalid /AP appearance dictionary"
+            )
+        normal_appearance = appearance[NameObject("/N")]
+        if not isinstance(normal_appearance, DictionaryObject):
+            raise TypeError(
+                "Generated checkbox has an invalid /AP /N appearance dictionary"
+            )
+        for state in (NameObject("/Off"), NameObject("/Yes")):
+            normal_appearance[state] = self.writer._add_object(normal_appearance[state])
         self.writer.add_annotation(page_number=page, annotation=checkbox)
 
     def add_signature(self, name: str, page: int, bounding_box: BoundingBox) -> None:

tests/form_creator_test.py

@@ -0,0 +1,73 @@
+from pypdf import PdfReader, PdfWriter
+from pypdf.generic import ContentStream, IndirectObject, NameObject
+
+from commonforms.form_creator import PyPdfFormCreator
+from commonforms.utils import BoundingBox
+
+
+def test_checkbox_appearance_streams_are_indirect_and_flatten_checked_state(tmp_path):
+    input_path = tmp_path / "blank.pdf"
+    checkbox_path = tmp_path / "checkbox.pdf"
+    flattened_path = tmp_path / "flattened.pdf"
+
+    blank_writer = PdfWriter()
+    blank_writer.add_blank_page(width=612, height=792)
+    with input_path.open("wb") as output:
+        blank_writer.write(output)
+    blank_writer.close()
+
+    creator = PyPdfFormCreator(input_path)
+    creator.add_checkbox("agree", 0, BoundingBox(x0=0.1, y0=0.1, x1=0.2, y1=0.2))
+    creator.save(checkbox_path)
+    creator.close()
+
+    checkbox_reader = PdfReader(checkbox_path)
+    annotation = checkbox_reader.pages[0]["/Annots"][0].get_object()
+    normal_appearances = annotation["/AP"]["/N"]
+
+    for state in ("/Off", "/Yes"):
+        appearance_reference = normal_appearances.raw_get(NameObject(state))
+        assert isinstance(appearance_reference, IndirectObject)
+
+        appearance = appearance_reference.get_object()
+        assert appearance["/Type"] == "/XObject"
+        assert appearance["/Subtype"] == "/Form"
+        assert "/BBox" in appearance
+        assert "/Resources" in appearance
+
+    # pypdf versions can preserve insignificant PDF whitespace in empty streams.
+    assert normal_appearances["/Off"].get_data().strip() == b""
+    checked_appearance = normal_appearances["/Yes"]
+    checked_bbox = list(checked_appearance["/BBox"])
+    checked_resources = dict(checked_appearance["/Resources"])
+    checked_data = checked_appearance.get_data()
+    assert checked_data
+
+    flatten_writer = PdfWriter(clone_from=checkbox_reader)
+    flatten_writer.update_page_form_field_values(
+        None, {"agree": "/Yes"}, auto_regenerate=False, flatten=True
+    )
+    with flattened_path.open("wb") as output:
+        flatten_writer.write(output)
+    flatten_writer.close()
+    checkbox_reader.close()
+
+    flattened_reader = PdfReader(flattened_path)
+    flattened_page = flattened_reader.pages[0]
+    operations = ContentStream(
+        flattened_page.get_contents(), flattened_reader
+    ).operations
+    do_operations = [operands for operands, operator in operations if operator == b"Do"]
+
+    assert len(do_operations) == 1
+    xobject_name = do_operations[0][0]
+    flattened_appearance = (
+        flattened_page["/Resources"]["/XObject"].raw_get(xobject_name).get_object()
+    )
+
+    assert flattened_appearance["/Type"] == "/XObject"
+    assert flattened_appearance["/Subtype"] == "/Form"
+    assert list(flattened_appearance["/BBox"]) == checked_bbox
+    assert dict(flattened_appearance["/Resources"]) == checked_resources
+    assert flattened_appearance.get_data() == checked_data
+    flattened_reader.close()