Merge pull request #2143 from okkays/kk/radio-buttons

Include radio buttons when processing forms

tests/test_api.py

@@ -510,6 +510,13 @@ def test_partial_pdf_custom_metadata():
     ('<input>', ['/Tx', '/V ()']),
     ('<input value="">', ['/Tx', '/V ()']),
     ('<input type="checkbox">', ['/Btn']),
+    ('<input type="radio">',
+     ['/Btn', '/V /Off', '/AS /Off', '/Ff 49152']),
+    ('<input checked type="radio" name="foo" value="value">',
+     ['/Btn', '/T (1)', '/V /dmFsdWU=', '/AS /dmFsdWU=']),
+    ('<form><input type="radio" name="foo" value="v0"></form>'
+     '<form><input checked type="radio" name="foo" value="v1"></form>',
+     ['/Btn', '/AS /djE=', '/V /djE=', '/AS /Off', '/V /Off']),
     ('<textarea></textarea>', ['/Tx', '/V ()']),
     ('<select><option value="a">A</option></select>', ['/Ch', '/Opt']),
     ('<select>'
@@ -525,7 +532,8 @@ def test_partial_pdf_custom_metadata():
 def test_pdf_inputs(html, fields):
     stdout = _run('--pdf-forms --uncompressed-pdf - -', html.encode())
     assert b'AcroForm' in stdout
-    assert all(field.encode() in stdout for field in fields)
+    for field in fields:
+        assert field.encode() in stdout
     stdout = _run('--uncompressed-pdf - -', html.encode())
     assert b'AcroForm' not in stdout
 

weasyprint/anchors.py

@@ -27,10 +27,11 @@ def rectangle_aabb(matrix, pos_x, pos_y, width, height):
     return box_x1, box_y1, box_x2, box_y2
 
 
-def gather_anchors(box, anchors, links, bookmarks, inputs, parent_matrix=None):
+def gather_anchors(box, anchors, links, bookmarks, forms, parent_matrix=None,
+                   parent_form=None):
     """Gather anchors and other data related to specific positions in PDF.
 
-    Currently finds anchors, links, bookmarks and inputs.
+    Currently finds anchors, links, bookmarks and forms.
 
     """
     # Get box transformation matrix.
@@ -89,6 +90,11 @@ def gather_anchors(box, anchors, links, bookmarks, inputs, parent_matrix=None):
     has_anchor = anchor_name and anchor_name not in anchors
     is_input = box.is_input()
 
+    if box.is_form():
+        parent_form = box.element
+        if parent_form not in forms:
+            forms[parent_form] = []
+
     if has_bookmark or has_link or has_anchor or is_input:
         if is_input:
             pos_x, pos_y = box.content_box_x(), box.content_box_y()
@@ -106,7 +112,7 @@ def gather_anchors(box, anchors, links, bookmarks, inputs, parent_matrix=None):
                 link_type = 'attachment'
             links.append((link_type, target, rectangle, box))
         if is_input:
-            inputs.append((box.element, box.style, rectangle))
+            forms[parent_form].append((box.element, box.style, rectangle))
         if matrix and (has_bookmark or has_anchor):
             pos_x, pos_y = matrix.transform_point(pos_x, pos_y)
         if has_bookmark:
@@ -116,7 +122,7 @@ def gather_anchors(box, anchors, links, bookmarks, inputs, parent_matrix=None):
             anchors[anchor_name] = pos_x, pos_y
 
     for child in box.all_children():
-        gather_anchors(child, anchors, links, bookmarks, inputs, matrix)
+        gather_anchors(child, anchors, links, bookmarks, forms, matrix, parent_form)
 
 
 def make_page_bookmark_tree(page, skipped_levels, last_by_depth,

weasyprint/css/html5_ua.css

@@ -187,8 +187,10 @@ input[value]::before {
   overflow: hidden;
 }
 input::before,
-input[value=""]::before {
-  content: " ";
+input[value=""]::before,
+input[type="checkbox"]::before,
+input[type="radio"]::before {
+  content: "";
 }
 select {
   background: lightgrey;

weasyprint/document.py

@@ -28,6 +28,7 @@ class Page:
     instantiated directly.
 
     """
+
     def __init__(self, page_box):
         #: The page width, including margins, in CSS pixels.
         self.width = page_box.margin_width()
@@ -67,14 +68,15 @@ def __init__(self, page_box):
         #: ``(x, y)`` point in CSS pixels from the top-left of the page.
         self.anchors = {}
 
-        #: The :obj:`list` of ``(element, attributes, rectangle)`` :obj:`tuples
-        #: <tuple>`. A ``rectangle`` is ``(x, y, width, height)``, in CSS
+        #: The :obj:`dict` mapping form elements to a list
+        #: of ``(element, attributes, rectangle)`` :obj:`tuples <tuple>`.
+        #: A ``rectangle`` is ``(x, y, width, height)``, in CSS
         #: pixels from the top-left of the page. ``atributes`` is a
         #: :obj:`dict` of HTML tag attributes and values.
-        self.inputs = []
+        #: The key ``None`` will contain inputs that are not part of a form.
+        self.forms = {None: []}
 
-        gather_anchors(
-            page_box, self.anchors, self.links, self.bookmarks, self.inputs)
+        gather_anchors(page_box, self.anchors, self.links, self.bookmarks, self.forms)
         self._page_box = page_box
 
     def paint(self, stream, scale=1):
@@ -105,6 +107,7 @@ class DocumentMetadata:
     New attributes may be added in future versions of WeasyPrint.
 
     """
+
     def __init__(self, title=None, authors=None, description=None,
                  keywords=None, generator=None, created=None, modified=None,
                  attachments=None, lang=None, custom=None):
@@ -162,6 +165,7 @@ class DiskCache:
     (i.e. RasterImage instances) are still stored in memory.
 
     """
+
     def __init__(self, folder):
         self._path = Path(folder)
         self._path.mkdir(parents=True, exist_ok=True)

weasyprint/formatting_structure/boxes.py

@@ -336,6 +336,12 @@ def is_input(self):
                 return not isinstance(self, (LineBox, TextBox))
         return False
 
+    def is_form(self):
+        """Return whether this box is a form element."""
+        if self.element is None:
+            return False
+        return self.element.tag == 'form'
+
 
 class ParentBox(Box):
     """A box that has children."""

weasyprint/pdf/__init__.py

@@ -11,7 +11,7 @@
 from .stream import Stream
 
 from .anchors import (  # isort:skip
-    add_annotations, add_inputs, add_links, add_outlines, resolve_links,
+    add_annotations, add_forms, add_links, add_outlines, resolve_links,
     write_pdf_attachment)
 
 VARIANTS = {
@@ -184,8 +184,8 @@ def generate_pdf(document, target, zoom, **options):
         add_annotations(
             links_and_anchors[0], matrix, document, pdf, pdf_page, annot_files,
             compress)
-        add_inputs(
-            page.inputs, matrix, pdf, pdf_page, resources, stream,
+        add_forms(
+            page.forms, matrix, pdf, pdf_page, resources, stream,
             document.font_config.font_map, compress)
         page.paint(stream, scale)
 

weasyprint/pdf/anchors.py

@@ -1,7 +1,9 @@
 """Insert anchors, links, bookmarks and inputs in PDFs."""
 
+import collections
 import io
 import mimetypes
+from base64 import b64encode
 from hashlib import md5
 from os.path import basename
 from urllib.parse import unquote, urlsplit
@@ -91,10 +93,10 @@ def add_outlines(pdf, bookmarks, parent=None):
     return outlines, count
 
 
-def add_inputs(inputs, matrix, pdf, page, resources, stream, font_map,
-               compress):
+def add_forms(forms, matrix, pdf, page, resources, stream, font_map,
+              compress):
     """Include form inputs in PDF."""
-    if not inputs:
+    if not forms or not any(forms.values()):
         return
 
     if 'Annots' not in page:
@@ -109,20 +111,43 @@ def add_inputs(inputs, matrix, pdf, page, resources, stream, font_map,
     context = ffi.gc(
         pango.pango_font_map_create_context(font_map),
         gobject.g_object_unref)
-    for i, (element, style, rectangle) in enumerate(inputs):
+    inputs_with_forms = [
+        (form, element, style, rectangle)
+        for form, inputs in forms.items()
+        for element, style, rectangle in inputs
+    ]
+    radio_groups = collections.defaultdict(dict)
+    for i, (form, element, style, rectangle) in enumerate(inputs_with_forms):
         rectangle = (
             *matrix.transform_point(*rectangle[:2]),
             *matrix.transform_point(*rectangle[2:]))
 
         input_type = element.attrib.get('type')
+        input_value = element.attrib.get('value', 'Yes')
         default_name = f'unknown-{page_reference.decode()}-{i}'
         input_name = element.attrib.get('name', default_name)
         # TODO: where does this 0.75 scale come from?
         font_size = style['font_size'] * 0.75
         field_stream = pydyf.Stream(compress=compress)
         field_stream.set_color_rgb(*style['color'][:3])
-        if input_type == 'checkbox':
-            # Checkboxes
+        if input_type in ('radio', 'checkbox'):
+            if input_type == 'radio':
+                if input_name not in radio_groups[form]:
+                    radio_groups[form][input_name] = group = pydyf.Dictionary({
+                        'FT': '/Btn',
+                        'Ff': (1 << (15 - 1)) + (1 << (16 - 1)),  # NoToggle & Radio
+                        'T': pydyf.String(f'{len(radio_groups)}'),
+                        'V': '/Off',
+                        'Kids': pydyf.Array(),
+                    })
+                    pdf.add_object(group)
+                    pdf.catalog['AcroForm']['Fields'].append(group.reference)
+                group = radio_groups[form][input_name]
+                character = 'l'  # Disc character in Dingbats
+            else:
+                character = '4'  # Check character in Dingbats
+
+            # Create stream when input is checked
             width = rectangle[2] - rectangle[0]
             height = rectangle[1] - rectangle[3]
             checked_stream = pydyf.Stream(extra={
@@ -135,32 +160,40 @@ def add_inputs(inputs, matrix, pdf, page, resources, stream, font_map,
             checked_stream.begin_text()
             checked_stream.set_color_rgb(*style['color'][:3])
             checked_stream.set_font_size('ZaDb', font_size)
-            # Center (let’s assume that Dingbat’s check has a 0.8em size)
-            x = (width - font_size * 0.8) / 2
-            y = (height - font_size * 0.8) / 2
+            # Center (let’s assume that Dingbat’s characters have a 0.75em size)
+            x = (width - font_size * 0.75) / 2
+            y = (height - font_size * 0.75) / 2
             checked_stream.move_text_to(x, y)
-            checked_stream.show_text_string('4')
+            checked_stream.show_text_string(character)
             checked_stream.end_text()
             checked_stream.pop_state()
             pdf.add_object(checked_stream)
 
             checked = 'checked' in element.attrib
             field_stream.set_font_size('ZaDb', font_size)
+            key = b64encode(input_value.encode(), altchars=b"+-").decode()
             field = pydyf.Dictionary({
                 'Type': '/Annot',
                 'Subtype': '/Widget',
                 'Rect': pydyf.Array(rectangle),
                 'FT': '/Btn',
                 'F': 1 << (3 - 1),  # Print flag
                 'P': page.reference,
-                'T': pydyf.String(input_name),
-                'V': '/Yes' if checked else '/Off',
+                'AS': f'/{key}' if checked else '/Off',
                 'AP': pydyf.Dictionary({'N': pydyf.Dictionary({
-                    'Yes': checked_stream.reference,
-                })}),
-                'AS': '/Yes' if checked else '/Off',
+                    key: checked_stream.reference})}),
+                'MK': pydyf.Dictionary({'CA': pydyf.String(character)}),
                 'DA': pydyf.String(b' '.join(field_stream.stream)),
             })
+            pdf.add_object(field)
+            if input_type == 'radio':
+                field['Parent'] = group.reference
+                if checked:
+                    group['V'] = f'/{key}'
+                group['Kids'].append(field.reference)
+            else:
+                field['T'] = pydyf.String(input_name)
+                field['V'] = field['AS']
         elif element.tag == 'select':
             # Select fields
             font_description = get_font_description(style)
@@ -174,7 +207,7 @@ def add_inputs(inputs, matrix, pdf, page, resources, stream, font_map,
             selected_values = []
             for option in element:
                 value = pydyf.String(option.attrib.get('value', ''))
-                text = pydyf.String(option.text or "")
+                text = pydyf.String(option.text or '')
                 options.append(pydyf.Array([value, text]))
                 if 'selected' in option.attrib:
                     selected_values.append(value)
@@ -198,6 +231,7 @@ def add_inputs(inputs, matrix, pdf, page, resources, stream, font_map,
                 field['V'] = (
                     selected_values[-1] if selected_values
                     else pydyf.String(''))
+            pdf.add_object(field)
         else:
             # Text, password, textarea, files, and unknown
             font_description = get_font_description(style)
@@ -231,8 +265,8 @@ def add_inputs(inputs, matrix, pdf, page, resources, stream, font_map,
             maxlength = element.get('maxlength')
             if maxlength and maxlength.isdigit():
                 field['MaxLen'] = element.get('maxlength')
+            pdf.add_object(field)
 
-        pdf.add_object(field)
         page['Annots'].append(field.reference)
         pdf.catalog['AcroForm']['Fields'].append(field.reference)