PyPDFForm 4.3.0__tar.gz → 4.4.0__tar.gz

{pypdfform-4.3.0 → pypdfform-4.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: PyPDFForm
-Version: 4.3.0
+Version: 4.4.0
 Summary: The Python library for PDF forms.
 Author: Jinge Li
 License-Expression: MIT

{pypdfform-4.3.0 → pypdfform-4.4.0}/PyPDFForm/__init__.py

@@ -20,7 +20,7 @@ The library supports various PDF form features, including:
 PyPDFForm aims to simplify PDF form manipulation, making it accessible to developers of all skill levels.
 """
 
-__version__ = "4.3.0"
+__version__ = "4.4.0"
 
 from .assets.blank import BlankPage
 from .middleware import Widgets

{pypdfform-4.3.0 → pypdfform-4.4.0}/PyPDFForm/template.py

@@ -28,6 +28,45 @@ from .patterns import (DROPDOWN_CHOICE_PATTERNS, WIDGET_DESCRIPTION_PATTERNS,
 from .utils import extract_widget_property, find_pattern_match, stream_to_io
 
 
+@lru_cache
+def get_metadata(pdf: bytes) -> dict:
+    """
+    Retrieves the metadata of a PDF.
+
+    Args:
+        pdf (bytes): The PDF stream to extract metadata from.
+
+    Returns:
+        dict: A dictionary containing the PDF's metadata.
+    """
+    if not pdf:
+        return {}
+
+    reader = PdfReader(stream_to_io(pdf))
+    return reader.metadata or {}
+
+
+def set_metadata(pdf: bytes, metadata: dict) -> bytes:
+    """
+    Sets the metadata of a PDF.
+
+    Args:
+        pdf (bytes): The PDF stream to set metadata for.
+        metadata (dict): A dictionary containing the metadata to be set.
+
+    Returns:
+        bytes: The updated PDF stream with the new metadata.
+    """
+    reader = PdfReader(stream_to_io(pdf))
+    writer = PdfWriter(clone_from=reader)
+    writer.add_metadata(metadata)
+
+    with BytesIO() as f:
+        writer.write(f)
+        f.seek(0)
+        return f.read()
+
+
 def build_widgets(
     pdf_stream: bytes,
     use_full_widget_name: bool,

{pypdfform-4.3.0 → pypdfform-4.4.0}/PyPDFForm/wrapper.py

@@ -38,7 +38,8 @@ from .middleware.dropdown import Dropdown
 from .middleware.signature import Signature
 from .middleware.text import Text
 from .raw import RawText, RawTypes
-from .template import build_widgets, update_widget_keys
+from .template import (build_widgets, get_metadata, set_metadata,
+                       update_widget_keys)
 from .types import PdfArray
 from .utils import (generate_unique_suffix, get_page_streams, merge_pdfs,
                     remove_all_widgets)
@@ -68,6 +69,7 @@ class PdfWrapper:
                 - `use_full_widget_name` (bool): Whether to use the full widget name when filling the form.
                 - `need_appearances` (bool): Whether to set the `NeedAppearances` flag in the PDF's AcroForm dictionary.
                 - `generate_appearance_streams` (bool): Whether to explicitly generate appearance streams for all form fields using pikepdf.
+                - `preserve_metadata` (bool): Whether to preserve the original metadata of the PDF.
                 - `title` (str): The title of the PDF document.
 
     """
@@ -76,6 +78,7 @@ class PdfWrapper:
         ("use_full_widget_name", False),
         ("need_appearances", False),
         ("generate_appearance_streams", False),
+        ("preserve_metadata", False),
         ("title", None),
     ]
 
@@ -105,6 +108,9 @@ class PdfWrapper:
         self._stream = fp_or_f_obj_or_stream_to_stream(template)
         self.widgets = {}
         self.title: str = None
+        self._metadata = (
+            get_metadata(self._read()) if kwargs.get("preserve_metadata") else {}
+        )
         self._on_open_javascript = None
         self._available_fonts = {}  # for setting /F1
         self._font_register_events = []  # for reregister
@@ -356,7 +362,8 @@ class PdfWrapper:
         2. If `need_appearances` is enabled, it handles appearance streams and the
            `/NeedAppearances` flag, which may include removing XFA and explicitly
            generating appearance streams.
-        3. If a title or on-open JavaScript is set, it updates the PDF properties
+        3. If `preserve_metadata` is enabled, it preserves the original metadata of the PDF.
+        4. If a title or on-open JavaScript is set, it updates the PDF properties
            accordingly.
 
         Returns:
@@ -369,6 +376,10 @@ class PdfWrapper:
                 result, getattr(self, "generate_appearance_streams")
             )  # cached
 
+        if getattr(self, "preserve_metadata"):
+            # TODO: refactor with preserve_pdf_properties
+            result = set_metadata(result, self._metadata)
+
         if any([self.title, self.on_open_javascript]):
             result = preserve_pdf_properties(
                 result,

{pypdfform-4.3.0 → pypdfform-4.4.0}/PyPDFForm.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: PyPDFForm
-Version: 4.3.0
+Version: 4.4.0
 Summary: The Python library for PDF forms.
 Author: Jinge Li
 License-Expression: MIT

{pypdfform-4.3.0 → pypdfform-4.4.0}/PyPDFForm.egg-info/SOURCES.txt

@@ -54,7 +54,6 @@ tests/test_draw_elements.py
 tests/test_dropdown.py
 tests/test_extract_values.py
 tests/test_fill_max_length_text_field.py
-tests/test_fill_method.py
 tests/test_font_widths.py
 tests/test_functional.py
 tests/test_generate_appearance_streams.py

pypdfform-4.4.0/tests/test_bulk_create_fields.py

@@ -0,0 +1,146 @@
+# -*- coding: utf-8 -*-
+
+import os
+
+import pytest
+
+from PyPDFForm import Fields, PdfWrapper
+
+
+@pytest.mark.posix_only
+def test_bulk_create_fields_stress_max(pdf_samples, request):
+    expected_path = os.path.join(
+        pdf_samples, "bulk_create_fields", "test_bulk_create_fields_stress_max.pdf"
+    )
+
+    fields = []
+    x = 0
+    y = 0
+    margin = 25
+    while x <= 575:
+        while y <= 625:
+            fields.append(
+                Fields.TextField(f"text_{x}_{y}", 1, x, y, width=margin, height=margin)
+            )
+            fields.append(Fields.CheckBoxField(f"check_{x}_{y}", 2, x, y, size=margin))
+            fields.append(
+                Fields.DropdownField(
+                    f"dropdown_{x}_{y}",
+                    3,
+                    x,
+                    y,
+                    width=margin,
+                    height=margin,
+                    options=["foo", "bar"],
+                )
+            )
+            fields.append(
+                Fields.RadioGroup(
+                    f"radio_{x}_{y}", 4, [x, x + 12.5], [y, y + 12.5], size=12.5
+                )
+            )
+            fields.append(
+                Fields.ImageField(
+                    f"image_{x}_{y}", 5, x, y, width=margin, height=margin
+                )
+            )
+            fields.append(
+                Fields.SignatureField(
+                    f"signature_{x}_{y}", 6, x, y, width=margin, height=margin
+                )
+            )
+            y += margin
+        y = 0
+        x += margin
+
+    with open(expected_path, "rb+") as f:
+        obj = PdfWrapper()
+        for _ in range(6):
+            obj += PdfWrapper(os.path.join(pdf_samples, "dummy.pdf"))
+        obj.bulk_create_fields(fields)
+
+        request.config.results["expected_path"] = expected_path
+        request.config.results["stream"] = obj.read()
+
+        expected = f.read()
+
+        assert len(obj.read()) == len(expected)
+        assert obj.read() == expected
+
+
+@pytest.mark.posix_only
+def test_bulk_create_fields_stress_max_mixed(pdf_samples, request):
+    expected_path = os.path.join(
+        pdf_samples,
+        "bulk_create_fields",
+        "test_bulk_create_fields_stress_max_mixed.pdf",
+    )
+
+    fields = []
+    x = 0
+    y = 0
+    margin = 25
+    while x <= 450:
+        while y <= 625:
+            fields.append(
+                Fields.TextField(f"text_{x}_{y}", 1, x, y, width=margin, height=margin)
+            )
+            fields.append(
+                Fields.CheckBoxField(f"check_{x}_{y}", 1, x + margin, y, size=margin)
+            )
+            fields.append(
+                Fields.DropdownField(
+                    f"dropdown_{x}_{y}",
+                    1,
+                    x,
+                    y + margin,
+                    width=margin,
+                    height=margin,
+                    options=["foo", "bar"],
+                )
+            )
+            fields.append(
+                Fields.RadioGroup(
+                    f"radio_{x}_{y}",
+                    1,
+                    [x + margin, x + margin + 12.5],
+                    [y + margin, y + margin + 12.5],
+                    size=12.5,
+                )
+            )
+            fields.append(
+                Fields.ImageField(
+                    f"image_{x}_{y}",
+                    1,
+                    x + margin + margin,
+                    y,
+                    width=margin,
+                    height=margin,
+                )
+            )
+            fields.append(
+                Fields.SignatureField(
+                    f"signature_{x}_{y}",
+                    1,
+                    x + margin + margin,
+                    y + margin,
+                    width=margin,
+                    height=margin,
+                )
+            )
+            y += 50
+        y = 0
+        x += 75
+
+    with open(expected_path, "rb+") as f:
+        obj = PdfWrapper(os.path.join(pdf_samples, "dummy.pdf")).bulk_create_fields(
+            fields
+        )
+
+        request.config.results["expected_path"] = expected_path
+        request.config.results["stream"] = obj.read()
+
+        expected = f.read()
+
+        assert len(obj.read()) == len(expected)
+        assert obj.read() == expected