pdf: add xfdf template possibiity in fill-form endpoint (#74797)

passerelle/apps/pdf/migrations/0003_resource_xfdf_template.py

@@ -0,0 +1,28 @@
+# Generated by Django 2.2.26 on 2023-02-23 15:02
+
+from django.db import migrations, models
+
+import passerelle.utils.models
+import passerelle.utils.templates
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('pdf', '0002_resource_fill_form_file'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='resource',
+            name='xfdf_template',
+            field=models.FileField(
+                blank=True,
+                help_text='Django template, used to create a XFDF for fill-form, rendered with payload',
+                null=True,
+                upload_to=passerelle.utils.models.resource_file_upload_to,
+                validators=[passerelle.utils.templates.validate_template],
+                verbose_name='XFDF Template',
+            ),
+        ),
+    ]

passerelle/apps/pdf/models.py

@@ -25,12 +25,14 @@ from django.conf import settings
 from django.core.exceptions import ValidationError
 from django.db import models
 from django.http.response import HttpResponse
+from django.template.base import VariableDoesNotExist
 from django.utils.translation import gettext_lazy as _
 
 from passerelle.base.models import BaseResource
 from passerelle.utils.api import endpoint
 from passerelle.utils.jsonresponse import APIError
 from passerelle.utils.models import resource_file_upload_to
+from passerelle.utils.templates import render_to_string, validate_template
 
 PDF_FILE_OBJECT = {
     'type': 'object',
@@ -85,7 +87,7 @@ FILL_FORM_SCHEMA = {
     'title': '',
     'description': '',
     'type': 'object',
-    'required': ['filename', 'fields'],
+    'required': ['filename'],
     'unflatten': True,
     'properties': OrderedDict(
         {
@@ -94,7 +96,7 @@ FILL_FORM_SCHEMA = {
                 'type': 'string',
             },
             'input-form': PDF_FILE_OBJECT,
-            'fields': {
+            'xfdf': {
                 'description': _('hierarchical dictionary of fields'),
                 'type': 'object',
             },
@@ -123,6 +125,14 @@ class Resource(BaseResource):
         null=True,
         blank=True,
     )
+    xfdf_template = models.FileField(
+        _('XFDF Template'),
+        upload_to=resource_file_upload_to,
+        help_text=_('Django template, used to create a XFDF for fill-form, rendered with payload'),
+        validators=[validate_template],
+        null=True,
+        blank=True,
+    )
 
     class Meta:
         verbose_name = _('PDF')
@@ -191,34 +201,40 @@ class Resource(BaseResource):
             'request_body': {'schema': {'application/json': FILL_FORM_SCHEMA}},
             'input_example': {
                 'filename': 'filled.pdf',
-                'fields/Page1[0]/FirstName[0]': 'John',
+                'xfdf/Page1[0]/FirstName[0]': 'John',
-                'fields/Page1[0]/LastName[0]': 'Doe',
+                'xfdf/Page1[0]/LastName[0]': 'Doe',
-                'fields/Page2[0]/Checkbox[0]': '0',
+                'xfdf/Page2[0]/Checkbox[0]': '0',
-                'fields/Page2[0]/Checkbox[1]': '1',
+                'xfdf/Page2[0]/Checkbox[1]': '1',
             },
         },
     )
     def fill_form(self, request, post_data):
-        filename = post_data.pop('filename')
+        filename = post_data['filename']
-        fields = post_data.pop('fields')
+        if 'xfdf' in post_data:
+            fields = post_data.pop('xfdf')
+        elif self.xfdf_template:
+            fields = None
+        else:
+            raise APIError("missing 'xfdf' property (no XFDF template)", http_status=400)
 
-        xfdf_root = ET.Element('xfdf')
+        if fields is not None:
-        xfdf_root.attrib['xmlns'] = 'http://ns.adobe.com/xfdf/'
+            xfdf_root = ET.Element('xfdf')
-        xfdf_root.attrib['xml:space'] = 'preserve'
+            xfdf_root.attrib['xmlns'] = 'http://ns.adobe.com/xfdf/'
-        xfdf_f = ET.SubElement(xfdf_root, 'f')
+            xfdf_root.attrib['xml:space'] = 'preserve'
-        xfdf_fields = ET.SubElement(xfdf_root, 'fields')
+            xfdf_f = ET.SubElement(xfdf_root, 'f')
+            xfdf_fields = ET.SubElement(xfdf_root, 'fields')
 
-        def add_fields(element, fields):
+            def add_fields(element, fields):
-            if isinstance(fields, dict):
+                if isinstance(fields, dict):
-                for key in fields:
+                    for key in fields:
-                    field = ET.SubElement(element, 'field')
+                        field = ET.SubElement(element, 'field')
-                    field.attrib['name'] = key
+                        field.attrib['name'] = key
-                    add_fields(field, fields[key])
+                        add_fields(field, fields[key])
-            else:
+                else:
-                value = ET.SubElement(element, 'value')
+                    value = ET.SubElement(element, 'value')
-                value.text = str(fields)
+                    value.text = str(fields)
 
-        add_fields(xfdf_fields, fields)
+            add_fields(xfdf_fields, fields)
 
         with tempfile.TemporaryDirectory(prefix='passerelle-pdftk-%s-fill-form-' % self.id) as tmpdir:
             if isinstance(post_data.get('input-form'), dict) and post_data['input-form'].get('content'):
@@ -228,13 +244,25 @@ class Resource(BaseResource):
             elif self.fill_form_file:
                 input_filename = self.fill_form_file.path
             else:
-                raise APIError("missing or bad 'input-form' property", http_status=400)
+                raise APIError(
+                    "missing or bad 'input-form' property (no default input file)", http_status=400
+                )
             # create xfdf
             xfdf_filename = os.path.join(tmpdir, 'fields.xfdf')
-            xfdf_f.attrib['href'] = input_filename
+            if fields is not None:
-            with open(xfdf_filename, mode='wb') as fd:
+                xfdf_f.attrib['href'] = input_filename
-                ET.indent(xfdf_root)
+                with open(xfdf_filename, mode='wb') as fd:
-                ET.ElementTree(xfdf_root).write(fd, encoding='UTF-8', xml_declaration=True)
+                    ET.indent(xfdf_root)
+                    ET.ElementTree(xfdf_root).write(fd, encoding='UTF-8', xml_declaration=True)
+            else:
+                self.xfdf_template.seek(0)
+                xfdf_template = self.xfdf_template.read().decode()
+                try:
+                    xfdf_content = render_to_string(xfdf_template, post_data)
+                except VariableDoesNotExist as exc:
+                    raise APIError("cannot render XFDF template: %s" % exc, http_status=400)
+                with open(xfdf_filename, mode='w') as fd:
+                    fd.write(xfdf_content)
 
             # call pdftk fill_form
             pdf_content = self.run_pdftk(args=[input_filename, 'fill_form', xfdf_filename])
@@ -254,5 +282,5 @@ class Resource(BaseResource):
         for line in dump.splitlines():
             unflatten_separated += '<br>%s' % line
             if line.startswith('FieldName: '):
-                unflatten_separated += ' → <b>fields/%s</b>' % line[11:].replace('.', '/')
+                unflatten_separated += ' → <b>xfdf/%s</b>' % line[11:].replace('.', '/')
         return unflatten_separated

tests/test_pdf.py

@@ -154,7 +154,7 @@ def test_pdf_fill_form(mocked_check_output, app, pdf):
 
     payload = {
         'filename': 'foo.pdf',
-        'fields/fname': 'John',
+        'xfdf/fname': 'John',
         'input-form': {'content': acroform_b64content},
     }
     mocked_check_output.side_effect = check_xml
@@ -176,7 +176,7 @@ def test_pdf_fill_form(mocked_check_output, app, pdf):
     pdf.save()
     payload = {
         'filename': 'bar.pdf',
-        'fields/fname': 'John',
+        'xfdf/fname': 'John',
     }
     mocked_check_output.reset_mock()
     resp = app.post_json(endpoint, params=payload, status=200)
@@ -196,7 +196,7 @@ def test_pdf_fill_form(mocked_check_output, app, pdf):
     # pdftk errors (faked)
     payload = {
         'filename': 'foo.pdf',
-        'fields/fname': 'Bill',
+        'xfdf/fname': 'Bill',
         'input-form': {'content': acroform_b64content},
     }
     mocked_check_output.reset_mock()
@@ -228,22 +228,22 @@ def test_pdf_fill_form(mocked_check_output, app, pdf):
     payload = {'filename': 'out.pdf'}
     resp = app.post_json(endpoint, params=payload, status=400)
     assert resp.json['err'] == 1
-    assert resp.json['err_desc'] == "'fields' is a required property"
+    assert resp.json['err_desc'] == "missing 'xfdf' property (no XFDF template)"
 
-    payload = {'filename': 'out.pdf', 'fields': 'not-a-dict'}
+    payload = {'filename': 'out.pdf', 'xfdf': 'not-a-dict'}
     resp = app.post_json(endpoint, params=payload, status=400)
     assert resp.json['err'] == 1
-    assert resp.json['err_desc'] == "fields: 'not-a-dict' is not of type 'object'"
+    assert resp.json['err_desc'] == "xfdf: 'not-a-dict' is not of type 'object'"
 
     pdf.fill_form_file = None  # no default PDF form
     pdf.save()
     payload = {
         'filename': 'bar.pdf',
-        'fields/fname': 'Alice',
+        'xfdf/fname': 'Alice',
     }
     resp = app.post_json(endpoint, params=payload, status=400)
     assert resp.json['err'] == 1
-    assert resp.json['err_desc'] == "missing or bad 'input-form' property"
+    assert resp.json['err_desc'] == "missing or bad 'input-form' property (no default input file)"
 
     resp = app.get(endpoint, status=405)
 
@@ -255,7 +255,7 @@ def test_pdf_real_pdftk_fillform(admin_user, app, pdf, settings):
     endpoint = generic_endpoint_url('pdf', 'fill-form', slug=pdf.slug)
     payload = {
         'filename': 'filled.pdf',
-        'fields/fname': 'ThisIsMyFirstName',
+        'xfdf/fname': 'ThisIsMyFirstName',
         'input-form': {'content': acroform_b64content},
     }
     resp = app.post_json(endpoint, params=payload, status=200)
@@ -271,11 +271,11 @@ def test_pdf_real_pdftk_fillform(admin_user, app, pdf, settings):
     manage_url = reverse('view-connector', kwargs={'connector': 'pdf', 'slug': pdf.slug})
     resp = app.get(manage_url)
     assert 'panel-dumpfields' not in resp.text
-    assert '<b>fields/fname</b>' not in resp.text
+    assert '<b>xfdf/fname</b>' not in resp.text
     app = login(app)
     resp = app.get(manage_url)
     assert 'panel-dumpfields' in resp.text
-    assert '<b>fields/fname</b>' in resp.text
+    assert '<b>xfdf/fname</b>' in resp.text
 
 
 def test_pdf_validator(pdf):