PyPDFForm 5.2.4__tar.gz → 5.2.5__tar.gz

{pypdfform-5.2.4 → pypdfform-5.2.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: PyPDFForm
-Version: 5.2.4
+Version: 5.2.5
 Summary: The Python library & CLI for PDF forms.
 Author: Jinge Li
 License-Expression: MIT

{pypdfform-5.2.4 → pypdfform-5.2.5}/PyPDFForm/__init__.py

@@ -18,7 +18,7 @@ from Python code or from the command line.
 
 import logging
 
-__version__ = "5.2.4"
+__version__ = "5.2.5"
 
 from .lib.annotations import Annotations
 from .lib.assets.blank import BlankPage

{pypdfform-5.2.4 → pypdfform-5.2.5}/PyPDFForm/lib/egress.py

@@ -28,6 +28,7 @@ from .constants import (
     Fields,
     JavaScript,
     OpenAction,
+    Parent,
     Root,
     S,
     Title,
@@ -140,10 +141,10 @@ def rebuild_acroform_fields(
 
     The existing `/Fields` array is replaced, creating an AcroForm dictionary
     when necessary. Each page annotation is resolved to a widget key, and only
-    annotations whose keys are present in `widget_keys` are added to the new
-    array. Page annotation arrays are left unchanged. When no matching page
-    annotations are found, the original PDF stream is returned unchanged to
-    avoid an unnecessary rewrite.
+    annotations whose keys are present in `widget_keys` contribute their
+    top-level field object to the new array. Page annotation arrays are left
+    unchanged. When no matching page annotations are found, the original PDF
+    stream is returned unchanged to avoid an unnecessary rewrite.
 
     Args:
         pdf (bytes): The PDF stream whose AcroForm fields should be rebuilt.
@@ -155,25 +156,110 @@ def rebuild_acroform_fields(
         bytes: The PDF stream with a rebuilt AcroForm `/Fields` array, or the
             original stream when there are no matching widgets to rebuild.
     """
+    if not widget_keys:
+        return pdf
+
     writer = PdfWriter(BytesIO(pdf))
     root = writer._root_object  # type: ignore # noqa: SLF001 # # pylint: disable=W0212
 
-    if AcroForm not in root:
-        root[NameObject(AcroForm)] = DictionaryObject({})
-    root[AcroForm][NameObject(Fields)] = ArrayObject([])
-
-    needs_update = False
+    fields = ArrayObject([])
+    seen_fields = set()
     for page in writer.pages:
         for annot in page.get(Annots, []):
             key = get_widget_key(annot.get_object(), use_full_widget_name)
             if key in widget_keys:
-                root[AcroForm][Fields].append(annot)
-                needs_update = True
+                field_ref = _get_root_field_reference(writer, annot)
+                field_key = _field_reference_key(field_ref)
+                if field_key not in seen_fields:
+                    fields.append(field_ref)
+                    seen_fields.add(field_key)
 
-    if not needs_update:
+    if not seen_fields:
         return pdf
 
+    if AcroForm not in root:
+        root[NameObject(AcroForm)] = DictionaryObject({})
+    root[AcroForm][NameObject(Fields)] = fields
+
     with BytesIO() as f:
         writer.write(f)
         f.seek(0)
         return f.read()
+
+
+def _get_root_field_reference(writer: PdfWriter, annot):
+    """
+    Returns the top-level AcroForm field reference for an annotation.
+
+    Widget annotations can be leaf nodes under a parent field dictionary. The
+    catalog `/AcroForm/Fields` array must point at root fields rather than
+    child widgets, otherwise readers such as pypdf will not expose hierarchical
+    fields like radio groups via `get_fields`.
+
+    Args:
+        writer (PdfWriter): The PDF writer that owns the annotation objects.
+        annot: The widget annotation reference or object.
+
+    Returns:
+        IndirectObject: The top-level AcroForm field reference.
+    """
+    field = annot
+    field_object = field.get_object()
+    visited = set()
+
+    while Parent in field_object and id(field_object) not in visited:
+        visited.add(id(field_object))
+        field = field_object[Parent]
+        field_object = field.get_object()
+
+    return _ensure_indirect_reference(writer, field)
+
+
+def _ensure_indirect_reference(writer: PdfWriter, field):
+    """
+    Returns an indirect reference for a field object.
+
+    pypdf expects entries in `/AcroForm/Fields` to have an indirect reference
+    when it builds field objects. Parent fields cloned from page annotations can
+    appear as direct dictionaries that still know their original indirect
+    reference, so prefer that before adding a new indirect object.
+
+    Args:
+        writer (PdfWriter): The PDF writer that will own newly added objects.
+        field: The field reference or direct field object to normalize.
+
+    Returns:
+        IndirectObject: An indirect reference to the field object.
+    """
+    field_object = field.get_object()
+    indirect_reference = getattr(field_object, "indirect_reference", None)
+    if indirect_reference is not None:
+        return indirect_reference
+
+    result = field
+    if not hasattr(field, "idnum"):
+        result = writer._add_object(field_object)  # type: ignore[attr-defined] # noqa: SLF001 # pylint: disable=W0212
+
+    return result
+
+
+def _field_reference_key(field_ref) -> tuple:
+    """
+    Returns a stable key for deduplicating root fields in one writer.
+
+    Args:
+        field_ref: The field reference to identify.
+
+    Returns:
+        tuple: A stable key for the field reference.
+    """
+    field_object = field_ref.get_object()
+    indirect_reference = getattr(field_object, "indirect_reference", field_ref)
+    idnum = getattr(indirect_reference, "idnum", None)
+    generation = getattr(indirect_reference, "generation", None)
+
+    result = (id(field_object),)
+    if idnum is not None:
+        result = idnum, generation
+
+    return result

{pypdfform-5.2.4 → pypdfform-5.2.5}/PyPDFForm/lib/utils.py

@@ -354,16 +354,15 @@ def get_version(pdf: bytes) -> str | None:
             start with a known PDF version identifier.
     """
 
-    result = None
-    for each in VERSION_IDENTIFIERS:
-        if pdf.startswith(each):
-            result = each.replace(VERSION_IDENTIFIER_PREFIX, b"").decode()
-            break
+    version_identifier_length = len(VERSION_IDENTIFIERS[0])
+    version_identifier = pdf[:version_identifier_length]
+    if version_identifier not in VERSION_IDENTIFIERS:
+        return None
 
-    return result
+    return version_identifier[len(VERSION_IDENTIFIER_PREFIX) :].decode()
 
 
-def set_version(pdf: bytes, old: str, new: str) -> bytes:
+def set_version(pdf: bytes, old: str | None, new: str | None) -> bytes:
     """
     Replaces the first PDF header version marker in a byte stream.
 
@@ -374,15 +373,34 @@ def set_version(pdf: bytes, old: str, new: str) -> bytes:
 
     Args:
         pdf (bytes): The PDF stream to update.
-        old (str): The currently present PDF version string.
-        new (str): The PDF version string to write into the header.
+        old (str | None): The currently present PDF version string. When
+            None, the header token after `%PDF-` is replaced if one exists.
+        new (str | None): The PDF version string to write into the header. When
+            None, the PDF stream is returned unchanged.
 
     Returns:
         bytes: The PDF stream with the first matching version marker replaced.
     """
 
-    return pdf.replace(
-        VERSION_IDENTIFIER_PREFIX + bytes(old, "utf-8"),
-        VERSION_IDENTIFIER_PREFIX + bytes(new, "utf-8"),
-        1,
-    )
+    if new is None or old == new:
+        return pdf
+
+    new_header = VERSION_IDENTIFIER_PREFIX + new.encode()
+
+    if old is None:
+        if not pdf.startswith(VERSION_IDENTIFIER_PREFIX):
+            return pdf
+
+        header_end = len(pdf)
+        for separator in (b"\r", b"\n"):
+            separator_index = pdf.find(separator, len(VERSION_IDENTIFIER_PREFIX))
+            if separator_index != -1:
+                header_end = min(header_end, separator_index)
+
+        return new_header + pdf[header_end:]
+
+    old_header = VERSION_IDENTIFIER_PREFIX + old.encode()
+    if pdf.startswith(old_header):
+        return new_header + pdf[len(old_header) :]
+
+    return pdf.replace(old_header, new_header, 1)

{pypdfform-5.2.4 → pypdfform-5.2.5}/PyPDFForm/lib/wrapper.py

@@ -585,8 +585,10 @@ class PdfWrapper:
             PdfWrapper: The `PdfWrapper` object, allowing for method chaining.
         """
 
-        if self.version:
-            self._stream = set_version(self._read(), self.version, version)
+        current_stream = self._read()
+        updated_stream = set_version(current_stream, self.version, version)
+        if updated_stream != current_stream:
+            self._stream = updated_stream
             self._version = version
 
         return self

{pypdfform-5.2.4 → pypdfform-5.2.5}/PyPDFForm.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: PyPDFForm
-Version: 5.2.4
+Version: 5.2.5
 Summary: The Python library & CLI for PDF forms.
 Author: Jinge Li
 License-Expression: MIT

{pypdfform-5.2.4 → pypdfform-5.2.5}/tests/test_functional.py

@@ -21,6 +21,7 @@ from PyPDFForm.lib.constants import (
 from PyPDFForm.lib.deprecation import deprecation_notice
 from PyPDFForm.lib.middleware.base import Widget
 from PyPDFForm.lib.template import get_widget_key, get_widgets_by_page
+from PyPDFForm.lib.utils import get_version, set_version
 
 
 def test_deprecation_warning():
@@ -113,10 +114,11 @@ def test_write(template_stream, tmp_path):
 
 def test_write_io(template_stream):
     buff = BytesIO()
-    PdfWrapper(template_stream).write(buff)
+    obj = PdfWrapper(template_stream)
+    obj.write(buff)
     buff.seek(0)
 
-    assert buff.read()
+    assert PdfWrapper(buff.read()).widgets.keys() == obj.widgets.keys()
 
 
 def test_fill_flatten_then_unflatten(template_stream, pdf_samples, data_dict, request):
@@ -507,7 +509,14 @@ def test_version(pdf_samples):
 
     obj = PdfWrapper(os.path.join(pdf_samples, "versions", "unknown.pdf"))
     assert obj.version is None
-    assert obj.read()
+    stream = obj.read()
+    assert stream.startswith(b"%PDF-2.1")
+    assert get_version(b"not a pdf") is None
+    assert set_version(stream, obj.version, None) == stream
+    assert set_version(b"not a pdf", None, "2.0") == b"not a pdf"
+    assert set_version(b"prefix %PDF-1.7\n", "1.7", "2.0") == b"prefix %PDF-2.0\n"
+    assert obj.change_version("2.0").version == "2.0"
+    assert obj.read().startswith(b"%PDF-2.0")
 
 
 @pytest.mark.requires_zlib_over_zlib_ng
@@ -855,7 +864,10 @@ def test_remove_fields_update_widgets(template_stream):
 
 
 def test_remove_fields_no_keys_specified(template_stream):
-    assert PdfWrapper(template_stream).remove_fields([]).read()
+    assert (
+        PdfWrapper(template_stream).remove_fields([]).widgets.keys()
+        == PdfWrapper(template_stream).widgets.keys()
+    )
 
 
 def test_merge(template_stream):