Merge pull request #239 from transifex/docx_allow_space_removal

Allow space control on docx files

Author: igavriil

openformats/formats/office_open_xml/parser.py

@@ -168,7 +168,7 @@ def compile_paragraph(cls, paragraph, rels_soup, stringset):
         ).find_all(text=True)
 
         leading_spaces = 0
-
+        empty_text_element = None
         added_hl_text_elements = defaultdict(list)
         deleted_hl_text_elements = defaultdict(list)
 
@@ -178,6 +178,7 @@ def compile_paragraph(cls, paragraph, rels_soup, stringset):
             # and remove leading whitespace from the next string
             if not text.strip():
                 leading_spaces = len(text) - len(text.strip())
+                empty_text_element = text_element
                 continue
             else:
                 hyperlink_url = cls.get_hyperlink_url(
@@ -193,6 +194,10 @@ def compile_paragraph(cls, paragraph, rels_soup, stringset):
                 translation = six.text_type(translation_part)
                 if not translation[:leading_spaces].strip():
                     translation = translation[leading_spaces:]
+                else:
+                    if empty_text_element:
+                        cls.remove_text_element(empty_text_element)
+
                 leading_spaces = 0
 
             # the text parts of the translation are more that the

openformats/tests/formats/docx/test_docx.py

@@ -122,6 +122,60 @@ def test_simple_file(self):
         self.assertEqual(openstring.string, translation)
         self.assertEqual(openstring.string, openstring.key)
 
+    def test_space_control(self):
+        path = '{}/special_cases_2.docx'.format(self.TESTFILE_BASE)
+        with open(path, 'rb') as f:
+            content = f.read()
+
+        handler = DocxHandler()
+        template, stringset = handler.parse(content)
+
+        self.assertEqual(len(stringset), 1)
+
+        openstring = stringset[0]
+        self.assertEqual(openstring.order, 0)
+        self.assertEqual(
+            openstring.string,
+            u'one two <tx href="https://www.transifex.com/">three </tx><tx> four </tx>five'  # noqa
+        )
+        self.assertEqual(openstring.string, openstring.key)
+
+        translation = u'ενα δύο <tx href="https://www.transifex.com/">τρία </tx><tx> τέσσερα </tx>πέντε'  # noqa
+        stringset = [
+            OpenString(openstring.key, translation, order=1)
+        ]
+
+        content = handler.compile(template, stringset)
+        template, stringset = handler.parse(content)
+
+        self.assertEqual(len(stringset), 1)
+
+        openstring = stringset[0]
+        self.assertEqual(openstring.order, 0)
+        self.assertEqual(
+            openstring.string,
+            u'ενα δύο <tx href="https://www.transifex.com/">τρία </tx><tx> τέσσερα </tx>πέντε'
+        )
+        self.assertEqual(openstring.string, openstring.key)
+
+        translation = u'ενα δύο<tx href="https://www.transifex.com/">τρία</tx><tx>τέσσερα</tx>πέντε'  # noqa
+        stringset = [
+            OpenString(openstring.key, translation, order=1)
+        ]
+
+        content = handler.compile(template, stringset)
+        template, stringset = handler.parse(content)
+
+        self.assertEqual(len(stringset), 1)
+
+        openstring = stringset[0]
+        self.assertEqual(openstring.order, 0)
+        self.assertEqual(
+            openstring.string,
+            u'ενα δύο<tx href="https://www.transifex.com/">τρία</tx><tx>τέσσερα</tx>πέντε'
+        )
+        self.assertEqual(openstring.string, openstring.key)
+
     def test_hyperlink_reorder(self):
         path = '{}/special_cases_2.docx'.format(self.TESTFILE_BASE)
         with open(path, 'rb') as f:
@@ -166,8 +220,6 @@ def test_hyperlink_reorder(self):
         self.assertEqual(text_elements[1].parent.rPr.color, None)
         self.assertEqual(text_elements[1].parent.rPr.u, None)
 
-
-
     def test_complex_file(self):
         path = '{}/complex.docx'.format(self.TESTFILE_BASE)
         with open(path, 'rb') as f: