Add test for soft-dotted characters.

The test is added to run_exhaustive_tests.py, and checks all
combinations of soft-dotted characters with combining marks above.

It's disabled for now, since the fonts fail it.

2 files changed, 46 insertions, 0 deletions

diff --git a/scripts/layout.py b/scripts/layout.py
index f65cf01..dc85ed2 100644
--- a/scripts/layout.py
+++ b/scripts/layout.py
@@ -32,3 +32,16 @@ def get_advances(text, font):
     _advance_cache[(text, font)] = advances
     return advances
 
+_shape_cache = {}
+def get_shapes(text, font):
+    """Get a list of shaped glyphs for text rendered in a font."""
+    try:
+        return _shape_cache[(text, font)]
+    except KeyError:
+        pass
+
+    hb_output = render.run_harfbuzz_on_text(text, font, None)
+    hb_output = json.loads(hb_output)
+    shapes = [glyph['g'] for glyph in hb_output]
+    _shape_cache[(text, font)] = shapes
+    return shapes
diff --git a/scripts/run_exhaustive_tests.py b/scripts/run_exhaustive_tests.py
index 02d1583..5cda50c 100755
--- a/scripts/run_exhaustive_tests.py
+++ b/scripts/run_exhaustive_tests.py
@@ -76,5 +76,38 @@ class TestSpacingMarks(unittest.TestCase):
                         'but it should not' % (ord(base_letter), ord(mark)))
 
 
+class TestSoftDottedChars(unittest.TestCase):
+    """Tests that soft-dotted characters lose their dots."""
+
+    def setUp(self):
+        self.font_files, _ = load_fonts()
+        charset = coverage.character_set(self.font_files[0])
+        self.marks_to_test = [char for char in charset
+                              if unicode_data.combining(char) == 230]
+        self.advance_cache = {}
+
+    def test_combinations(self):
+        """Tests that soft-dotted characters lose their dots when combined."""
+
+        return  # FIXME: Test is currently disabled, since the fonts fail it
+
+        for font in self.font_files:
+            print 'Testing %s for soft-dotted combinations...' % font
+
+            # TODO: replace the following list with actual derivation based on
+            # Unicode's soft-dotted property
+            for base_letter in (u'ij\u012F\u0249\u0268\u029D\u02B2\u03F3\u0456'
+                                u'\u0458\u1D62\u1D96\u1DA4\u1DA8\u1E2D\u1ECB'
+                                u'\u2071\u2C7C'):
+                print 'Testing %s combinations' % base_letter.encode('UTF-8')
+                for mark in self.marks_to_test:
+                    mark = unichr(mark)
+                    letter_only = layout.get_advances(base_letter, font)
+                    combination = layout.get_advances(base_letter + mark, font)
+                    self.assertNotEqual(combination[0], letter_only[0],
+                        "The sequence <%04X, %04X> doesn't lose its dot, "
+                        "but it should" % (ord(base_letter), ord(mark)))
+
+
 if __name__ == '__main__':
     unittest.main()