pdf: add support for combo fields (#75373)

passerelle/apps/pdf/forms.py

@@ -15,6 +15,8 @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
 from django import forms
+from django.urls import reverse
+from django.utils.html import mark_safe
 from django.utils.translation import gettext_lazy as _
 
 from passerelle.utils.forms import ConditionField, TemplateField
@@ -44,11 +46,30 @@ class FieldsMappingEditForm(forms.ModelForm):
                     values = ', '.join('"%s"' % value for value in field.radio_possible_values)
                     help_text = _('text template, possibles values %s') % values
                     field_class = TemplateField
+                elif field.widget_type in ('list', 'combo'):
+                    ds_url = (
+                        reverse(
+                            'generic-endpoint',
+                            kwargs={
+                                'connector': 'pdf',
+                                'endpoint': 'field-values',
+                                'slug': self.instance.slug,
+                            },
+                        )
+                        + '?digest_id='
+                        + field.digest_id
+                    )
+                    help_text = mark_safe(
+                        _('text template, possibles values <a href="%s">data source</a>') % ds_url
+                    )
+                    field_class = TemplateField
                 else:
                     continue
-                label = _('field {number} ({help_text})').format(number=i + 1, help_text=help_text)
+                label = _('field {number}').format(number=i + 1)
                 initial = fields_mapping.get(name, '')
-                self.fields[name] = field_class(label=label, required=False, initial=initial)
+                self.fields[name] = field_class(
+                    label=label, required=False, initial=initial, help_text=help_text
+                )
                 self.fields[name].page_number = page.page_number
                 self.fields[name].widget.attrs['tabindex'] = '0'
                 self.fields[name].widget.attrs['class'] = '0'

passerelle/apps/pdf/models.py

@@ -133,6 +133,7 @@ class Resource(BaseResource):
         description=_('Returns the assembly of received PDF files'),
         perm='can_access',
         methods=['post'],
+        display_order=0,
         post={
             'request_body': {'schema': {'application/json': ASSEMBLE_SCHEMA}},
             'input_example': {
@@ -207,6 +208,7 @@ class Resource(BaseResource):
         description=_('Fills the input PDF form with fields applying mappings to the received payload'),
         perm='can_access',
         methods=['post'],
+        display_order=1,
         parameters={
             'filename': {'description': _('file name')},
             'flatten': {'description': _('remove PDF fields, keep only the drawed values')},
@@ -252,6 +254,8 @@ class Resource(BaseResource):
                         value = evaluate_template(mapping_template, post_data)
                     elif field.widget_type == 'radio':
                         value = evaluate_template(mapping_template, post_data)
+                    elif field.widget_type in ('combo', 'list'):
+                        value = evaluate_template(mapping_template, post_data)
                         self.logger.info('field=%r value=%r', field, value)
                     else:
                         raise NotImplementedError
@@ -261,3 +265,28 @@ class Resource(BaseResource):
         response['Content-Disposition'] = 'attachment; filename="%s"' % filename
         pdf.write(response, flatten=flatten_pdf)
         return response
+
+    @endpoint(
+        name='field-values',
+        description=_('Return possible values for PDF\'s combo or list form fields'),
+        perm='can_access',
+        parameters={
+            'digest_id': {'description': _('Identifier of the field')},
+        },
+    )
+    def field_values(self, request, digest_id):
+        if not self.fill_form_file:
+            raise APIError('not PDF file configured')
+
+        with self.fill_form_file.open() as fd:
+            pdf_content = fd.read()
+
+        pdf = PDF(pdf_content)
+        fields = [field for page in pdf.pages for field in page.fields if field.digest_id == digest_id]
+        if not fields:
+            raise APIError(f'unknown digest-id {digest_id!r}')
+        field = fields[0]
+        if field.widget_type not in ('list', 'combo'):
+            raise APIError(f'wrong field type for digest-id {digest_id!r}: {field.widget_type}')
+
+        return {'data': [{'id': value, 'text': value} for _, value in field.combo_possible_values]}

passerelle/utils/pdf.py

@@ -25,8 +25,9 @@ import typing
 
 import pdfrw
 
-RADIO_FLAG = 2**15
-PUSH_BUTTON_FLAG = 2**16
+RADIO_FLAG = 1 << 15  # bit 16
+PUSH_BUTTON_FLAG = 1 << 16  # bit 17
+LIST_FLAG = 1 << 17  # bit 18
 
 
 class Rect(typing.NamedTuple):
@@ -65,6 +66,10 @@ class FieldFlags(int):
     def is_push_button(self):
         return self & PUSH_BUTTON_FLAG
 
+    @property
+    def is_list(self):
+        return self & LIST_FLAG
+
 
 @dataclasses.dataclass(frozen=True)
 class Widget:
@@ -104,6 +109,11 @@ class Widget:
             return 'checkbox'
         elif self.field_type == pdfrw.PdfName.Tx:
             return 'text'
+        elif self.field_type == pdfrw.PdfName.Ch:
+            if self.field_flags.is_list:
+                return 'list'
+            else:
+                return 'combo'
         else:
             raise NotImplementedError
 
@@ -156,6 +166,11 @@ class Widget:
         assert self.widget_type == 'radio'
         return list(list(kid.AP.N.keys())[0][1:] for kid in self.kids_ordered_by_rect if kid.AP and kid.AP.N)
 
+    @property
+    def combo_possible_values(self):
+        assert self.widget_type in ('list', 'combo')
+        return [(option[0].decode(), option[1].decode()) for option in self.annotation.Opt]
+
     @property
     def value(self):
         if self.widget_type == 'text':
@@ -166,6 +181,8 @@ class Widget:
             return self.annotation.V == self.checkbox_true_value
         elif self.widget_type == 'radio':
             return self.annotation.V.lstrip('/') if self.annotation.V else None
+        elif self.widget_type in ('list', 'combo'):
+            return self.annotation.V.decode() if self.annotation.V is not None else None
 
     def set(self, value):
         # allow rendering of values in Acrobat Reader
@@ -186,6 +203,15 @@ class Widget:
                 else:
                     kid.update(pdfrw.PdfDict(AS=pdfrw.PdfName.Off))
             self.annotation.update(pdfrw.PdfDict(V=radio_value))
+        elif self.widget_type in ('list', 'combo'):
+            for export, combo_value in self.combo_possible_values:
+                if combo_value == value:
+                    self.annotation.update(
+                        pdfrw.PdfDict(
+                            V=pdfrw.PdfString.from_unicode(export), AS=pdfrw.PdfString.from_unicode(export)
+                        )
+                    )
+                    break
 
     @classmethod
     def from_pdf_widget(cls, page, pdf_widget):

tests/test_utils_pdf.py

@@ -104,3 +104,16 @@ def test_radio_button():
     assert radio.radio_possible_values == ['H', 'F']
     radio.set('H')
     assert radio.value == 'H'
+
+
+def test_combo_box():
+    with open('tests/data/cerfa_14011-02.pdf', 'rb') as fd:
+        pdf = PDF(content=fd)
+    combo = [field for field in pdf.page(0).fields if field.name == 'topmostSubform[0].Page1[0].Pays[0]']
+    assert len(combo) == 1
+    combo = combo[0]
+    assert len(combo.combo_possible_values) == 235
+    combo.set('X')
+    assert combo.value is None
+    combo.set('FRANCE')
+    assert combo.value == 'FRANCE'