Merge pull request #326 from mathjazz/bug-1193860

Fix bug 1193860: escape quotes in some DTD files

pontoon/sync/formats/silme.py

@@ -4,10 +4,11 @@
 """
 import codecs
 import os
+import silme
+
 from collections import OrderedDict
 from copy import copy
 
-import silme
 from silme.format.dtd import FormatParser as DTDParser
 from silme.format.ini import FormatParser as IniParser
 from silme.format.inc import FormatParser as IncParser
@@ -78,6 +79,9 @@ def __init__(self, parser, path, source_resource=None):
         self.source_resource = source_resource
         self.entities = OrderedDict()  # Preserve entity order.
 
+        # Bug 1193860: unescape quotes in some files
+        self.escape_quotes_on = 'mobile/android/base' in path and parser is DTDParser
+
         # Copy entities from the source_resource if it's available.
         if source_resource:
             for key, entity in source_resource.entities.items():
@@ -103,6 +107,10 @@ def __init__(self, parser, path, source_resource=None):
         current_order = 0
         for obj in self.structure:
             if isinstance(obj, silme.core.entity.Entity):
+
+                if self.escape_quotes_on:
+                    obj.value = self.unescape_quotes(obj.value)
+
                 entity = SilmeEntity(obj, comments, current_order)
                 self.entities[entity.key] = entity
                 current_order += 1
@@ -118,6 +126,22 @@ def __init__(self, parser, path, source_resource=None):
     def translations(self):
         return self.entities.values()
 
+    def escape_quotes(self, value):
+        """
+        DTD files can use single or double quotes for identifying strings,
+        so " and ' are the safe bet that will work in both cases.
+        """
+        value = value.replace('"', '\\"')
+        value = value.replace("'", '\\'')
+        return value
+
+    def unescape_quotes(self, value):
+        value = value.replace('\\"', '"')
+        value = value.replace('\\"', '"')
+        value = value.replace('\\'', "'")
+        value = value.replace("\\'", "'")
+        return value
+
     def save(self, locale):
         """
         Load the source resource, modify it with changes made to this
@@ -144,7 +168,12 @@ def save(self, locale):
 
             translated_entity = self.entities.get(key)
             if translated_entity and None in translated_entity.strings:
-                new_structure.modify_entity(key, translated_entity.strings[None])
+                translation = translated_entity.strings[None]
+
+                if self.escape_quotes_on:
+                    translation = self.escape_quotes(translation)
+
+                new_structure.modify_entity(key, translation)
             else:
                 # Remove untranslated entity and following newline
                 pos = new_structure.entity_pos(key)

pontoon/sync/tests/formats/test_silme.py

@@ -185,6 +185,47 @@ def test_save_translation_identical(self):
 
         self.run_save_translation_identical(source_string, input_string, expected_string)
 
+    def test_quotes(self):
+        input_strings = dedent("""
+            <!ENTITY SingleQuote "\'">
+            <!ENTITY SingleQuoteEntity "\&apos;">
+            <!ENTITY DoubleQuote '\"'>
+            <!ENTITY DoubleQuoteEntity "\&quot;">
+        """)
+
+        # Make sure path contains 'mobile/android/base'
+        dirname = os.path.join(tempfile.mkdtemp(), 'mobile', 'android', 'base')
+        os.makedirs(dirname)
+
+        fd, path = tempfile.mkstemp(dir=dirname)
+        with os.fdopen(fd, 'w') as f:
+            f.write(input_strings)
+
+        resource = self.parse(path)
+
+        # Unescape quotes when parsing
+        assert_attributes_equal(resource.translations[0], strings={None: "'"})
+        assert_attributes_equal(resource.translations[1], strings={None: "'"})
+        assert_attributes_equal(resource.translations[2], strings={None: '"'})
+        assert_attributes_equal(resource.translations[3], strings={None: '"'})
+
+        # Escape quotes when saving
+        translated_resource = silme.SilmeResource(
+            DTDParser, path, source_resource=resource
+        )
+        translated_resource.translations[0].strings[None] = "Single Quote '"
+        translated_resource.translations[1].strings[None] = 'Double Quote "'
+
+        translated_resource.save(self.locale)
+
+        expected_string = dedent("""
+            <!ENTITY SingleQuote "Single Quote \&apos;">
+            <!ENTITY SingleQuoteEntity "Double Quote \&quot;">
+            <!ENTITY DoubleQuote '\&quot;'>
+            <!ENTITY DoubleQuoteEntity "\&quot;">
+        """)
+        self.assert_file_content(path, expected_string)
+
 
 BASE_PROPERTIES_FILE = """
 # Sample comment