Fix PDF relabel rewrite for nested fields

Author: LemmaLegal

formfyxer/lit_explorer.py

@@ -26,6 +26,7 @@
     unlock_pdf_in_place,
     is_tagged,
     get_original_text_with_fields,
+    rename_pdf_fields,
 )
 
 import math
@@ -2081,16 +2082,13 @@ def _looks_reasonable(candidate: str) -> bool:
         stats["debug fields"] = debug_fields
     if rewrite:
         try:
-            my_pdf = pikepdf.Pdf.open(in_file, allow_overwriting_input=True)
-            fields_too = (
-                my_pdf.Root.AcroForm.Fields
-            )  # [0]["/Kids"][0]["/Kids"][0]["/Kids"][0]["/Kids"]
-            # print(repr(fields_too))
-            for k, field_name in enumerate(new_names):
-                # print(k,field)
-                fields_too[k].T = re.sub(r"^\*", "", field_name)
-            my_pdf.save(in_file)
-            my_pdf.close()
+            rename_mapping = {
+                old_name: re.sub(r"^\*", "", new_name)
+                for old_name, new_name in zip(field_names, new_names)
+                if old_name != new_name
+            }
+            if rename_mapping:
+                rename_pdf_fields(in_file, in_file, rename_mapping)
         except Exception as ex:
             stats["error"] = f"could not change form fields: {ex}"
     return stats

formfyxer/tests/test_lit_explorer_pdf_labeling.py

@@ -73,6 +73,86 @@ def test_parse_form_accepts_model_and_openai_base_url(self):
         self.assertIn("model", signature.parameters)
         self.assertIn("openai_base_url", signature.parameters)
 
+    @patch("formfyxer.lit_explorer.rename_pdf_fields")
+    @patch("formfyxer.lit_explorer.needs_calculations", return_value=False)
+    @patch("formfyxer.lit_explorer.is_tagged", return_value=False)
+    @patch("formfyxer.lit_explorer.get_sensitive_data_types", return_value=[])
+    @patch("formfyxer.lit_explorer.field_types_and_sizes", return_value=[])
+    @patch("formfyxer.lit_explorer.all_caps_words", return_value=0)
+    @patch("formfyxer.lit_explorer.transformed_sentences", return_value=[])
+    @patch("formfyxer.lit_explorer.get_citations", return_value=[])
+    @patch("formfyxer.lit_explorer.get_passive_sentences", return_value=[])
+    @patch("formfyxer.lit_explorer.split_sentences", return_value=["Sentence one."])
+    @patch("formfyxer.lit_explorer.spot", return_value=[])
+    @patch("formfyxer.lit_explorer.guess_form_name", return_value="")
+    @patch("formfyxer.lit_explorer.describe_form", return_value="")
+    @patch("formfyxer.lit_explorer.cleanup_text", side_effect=lambda text: text)
+    @patch("formfyxer.lit_explorer.extract_text", return_value="Sample form text")
+    @patch(
+        "formfyxer.lit_explorer.get_existing_pdf_fields",
+        return_value=[
+            [
+                SimpleNamespace(name="field_a"),
+                SimpleNamespace(name="field_b"),
+            ]
+        ],
+    )
+    @patch("formfyxer.lit_explorer.unlock_pdf_in_place")
+    @patch("formfyxer.lit_explorer.pikepdf.open")
+    @patch("formfyxer.lit_explorer.get_openai_api_key_from_sources", return_value="sk-test")
+    @patch(
+        "formfyxer.lit_explorer.rename_pdf_fields_with_context",
+        return_value={
+            "field_a": "*users.name.first",
+            "field_b": "users.name.last",
+        },
+    )
+    def test_parse_form_rewrite_uses_rename_helper_for_nested_fields(
+        self,
+        _mock_rename_with_context,
+        _mock_get_api_key,
+        mock_pikepdf_open,
+        _mock_unlock,
+        _mock_get_existing_fields,
+        _mock_extract_text,
+        _mock_cleanup_text,
+        _mock_describe_form,
+        _mock_guess_form_name,
+        _mock_spot,
+        _mock_split_sentences,
+        _mock_get_passive_sentences,
+        _mock_get_citations,
+        _mock_transformed_sentences,
+        _mock_all_caps_words,
+        _mock_field_types_and_sizes,
+        _mock_get_sensitive_data_types,
+        _mock_is_tagged,
+        _mock_needs_calculations,
+        mock_rename_pdf_fields,
+    ):
+        fake_pdf = SimpleNamespace(pages=[object()], docinfo=SimpleNamespace(Title="Form"))
+        mock_pikepdf_open.return_value = fake_pdf
+
+        with patch(
+            "formfyxer.lit_explorer.textstat.text_standard", return_value=6.0
+        ), patch(
+            "formfyxer.lit_explorer.textstat.difficult_words_list", return_value=[]
+        ), patch(
+            "formfyxer.lit_explorer.time_to_answer_form", return_value=[-1, -1]
+        ):
+            result = parse_form("fake.pdf", rewrite=True)
+
+        mock_rename_pdf_fields.assert_called_once_with(
+            "fake.pdf",
+            "fake.pdf",
+            {
+                "field_a": "users.name.first",
+                "field_b": "users.name.last",
+            },
+        )
+        self.assertEqual(result["fields_old"], ["field_a", "field_b"])
+        self.assertEqual(result["fields"], ["*users.name.first", "users.name.last"])
+
     def test_get_original_text_with_fields_handles_pdfminer_state(self):
         fixture = Path(__file__).parent / "affidavit_supplement.pdf"
         with tempfile.NamedTemporaryFile(suffix=".txt", delete=False) as temp_file: