get_data_fields with integration test

codeforamerica · Mar 12, 2016 · 01ddaec · 01ddaec
1 parent 992ba3a
commit 01ddaec
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 18 deletions.
diff --git a/src/pdftk_wrapper.py b/src/pdftk_wrapper.py
@@ -50,14 +50,14 @@ def _clean_up_tmp_files(self):
             path = self._tmp_files.pop()
             os.remove(path)
 
-    def _get_file_contents(self, path, decode=False):
+    def _get_file_contents(self, path, decode=False, encoding=None):
         """given a file path, return the contents of the file
         if decode is True, the contents will be decoded using the default
         encoding
         """
         bytestring = open(path, 'rb').read()
         if decode:
-            return bytestring.decode(self.encoding)
+            return bytestring.decode(encoding or self.encoding)
         return bytestring
 
     def get_fdf(self, fp):
@@ -68,14 +68,18 @@ def get_fdf(self, fp):
         tmp_outfile = self._write_tmp_file()
         self.run_command([fp, 'generate_fdf',
             'output', tmp_outfile])
-        return self._get_file_contents(
+        contents = self._get_file_contents(
             tmp_outfile, decode=True)
-
-    def get_xfdf(self, fp):
-        return None
+        return contents
 
     def get_data_fields(self, fp):
-        return None
+        fp = self._coerce_to_file_path(fp)
+        tmp_outfile = self._write_tmp_file()
+        self.run_command([fp, 'dump_data_fields_utf8',
+            'output', tmp_outfile])
+        contents = self._get_file_contents(
+            tmp_outfile, decode=True, encoding='utf-8')
+        return contents
 
     def parse_fdf_fields(self, fdf_str):
         yield None

diff --git a/tests/integration/test_pdftk.py b/tests/integration/test_pdftk.py
@@ -17,6 +17,7 @@ def setUp(self):
         self.tmp_dir = 'data'
         self.sample_form_path = 'data/sample_pdfs/sample_form.pdf'
         self.fdf_str_sample = '%FDF-1.2\n%âãÏÓ\n1 0 obj \n<<\n/FDF \n<<\n/Fields [\n<<\n/V /Yes\n/T (Language 2 Check Box)\n>> \n<<\n/V ()\n/T (Address 2 Text Box)\n>> \n<<\n/V /Off\n/T (Language 3 Check Box)\n>> \n<<\n/V ()\n/T (City Text Box)\n>> \n<<\n/V /Off\n/T (Language 1 Check Box)\n>> \n<<\n/V /Off\n/T (Driving License Check Box)\n>> \n<<\n/V ()\n/T (Given Name Text Box)\n>> \n<<\n/V /Off\n/T (Language 5 Check Box)\n>> \n<<\n/V ()\n/T (House nr Text Box)\n>> \n<<\n/V (150)\n/T (Height Formatted Field)\n>> \n<<\n/V ()\n/T (Family Name Text Box)\n>> \n<<\n/V ()\n/T (Address 1 Text Box)\n>> \n<<\n/V /Off\n/T (Language 4 Check Box)\n>> \n<<\n/V ()\n/T (Postcode Text Box)\n>>]\n>>\n>>\nendobj \ntrailer\n\n<<\n/Root 1 0 R\n>>\n%%EOF\n'
+        self.data_fields_str_sample = '---\nFieldType: Text\nFieldName: Given Name Text Box\nFieldNameAlt: First name\nFieldFlags: 0\nFieldValue: \nFieldJustification: Left\nFieldMaxLength: 40\n---\nFieldType: Text\nFieldName: Family Name Text Box\nFieldNameAlt: Last name\nFieldFlags: 0\nFieldValue: \nFieldJustification: Left\nFieldMaxLength: 40\n---\nFieldType: Text\nFieldName: Address 1 Text Box\nFieldFlags: 0\nFieldValue: \nFieldJustification: Left\nFieldMaxLength: 40\n---\nFieldType: Text\nFieldName: House nr Text Box\nFieldNameAlt: House and floor\nFieldFlags: 0\nFieldValue: \nFieldJustification: Left\nFieldMaxLength: 20\n---\nFieldType: Text\nFieldName: Address 2 Text Box\nFieldFlags: 0\nFieldValue: \nFieldJustification: Left\nFieldMaxLength: 40\n---\nFieldType: Text\nFieldName: Postcode Text Box\nFieldFlags: 0\nFieldValue: \nFieldJustification: Left\nFieldMaxLength: 20\n---\nFieldType: Text\nFieldName: City Text Box\nFieldFlags: 0\nFieldValue: \nFieldJustification: Left\nFieldMaxLength: 40\n---\nFieldType: Text\nFieldName: Height Formatted Field\nFieldNameAlt: Value from 40 to 250 cm\nFieldFlags: 0\nFieldValue: 150\nFieldValueDefault: 150\nFieldJustification: Left\nFieldMaxLength: 20\n---\nFieldType: Button\nFieldName: Driving License Check Box\nFieldNameAlt: Car driving license\nFieldFlags: 0\nFieldValue: Off\nFieldJustification: Left\nFieldStateOption: Off\nFieldStateOption: Yes\n---\nFieldType: Button\nFieldName: Language 1 Check Box\nFieldFlags: 0\nFieldValue: Off\nFieldJustification: Left\nFieldStateOption: Off\nFieldStateOption: Yes\n---\nFieldType: Button\nFieldName: Language 2 Check Box\nFieldFlags: 0\nFieldValue: Yes\nFieldJustification: Left\nFieldStateOption: Off\nFieldStateOption: Yes\n---\nFieldType: Button\nFieldName: Language 3 Check Box\nFieldFlags: 0\nFieldValue: Off\nFieldJustification: Left\nFieldStateOption: Off\nFieldStateOption: Yes\n---\nFieldType: Button\nFieldName: Language 4 Check Box\nFieldFlags: 0\nFieldValue: Off\nFieldJustification: Left\nFieldStateOption: Off\nFieldStateOption: Yes\n---\nFieldType: Button\nFieldName: Language 5 Check Box\nFieldFlags: 0\nFieldValue: Off\nFieldJustification: Left\nFieldStateOption: Off\nFieldStateOption: Yes\n'
 
     def test_pdftk_errors(self):
         pdftk = PDFTKWrapper()
@@ -32,11 +33,18 @@ def test_pdftk_errors(self):
         with self.assertRaises(PdftkError):
             pdftk.run_command(args)
 
-    def test_generate_fdf(self):
+    def test_get_fdf(self):
         pdftk = PDFTKWrapper()
         results = pdftk.get_fdf(self.sample_form_path)
         self.assertEqual(results, self.fdf_str_sample)
 
+    def test_get_data_fields(self):
+        pdftk = PDFTKWrapper()
+        results = pdftk.get_data_fields(self.sample_form_path)
+        self.assertEqual(results, self.data_fields_str_sample)
+
+
+
 
 
 
diff --git a/tests/unit/test_pdftk.py b/tests/unit/test_pdftk.py
@@ -33,16 +33,6 @@ def test_get_fdf(self):
         contents_getter.assert_called_once_with(
             'tmp_path.fdf', decode=True)
 
-    def test_get_xfdf(self):
-        pdftk = PDFTKWrapper()
-        fp = 'something.pdf'
-        results = pdftk.get_xfdf(fp)
-
-    def test_get_data_fields(self):
-        pdftk = PDFTKWrapper()
-        fp = 'something.pdf'
-        results = pdftk.get_data_fields(fp)
-
     def test_parse_fdf_fields(self):
         pdftk = PDFTKWrapper()
         fdf_sample = """