Fix a bug causing … is converted to "", "", "…" (#121)

* Change a variable name to modifield_to_original, to make it align with the original Java implementation

* Fix a bug causing … is converted to "", "", "…"

* Fix tests for according to the new replace method

* Fix comment format

Author: Sorami Hisamoto

sudachipy/utf8inputtextbuilder.py

@@ -24,7 +24,7 @@ def __init__(self, text, grammar):
         self.grammar = grammar
         self.original_text = text
         self.modified_text = text
-        self.text_offsets = list(range(len(self.original_text) + 1))
+        self.modified_to_original = list(range(len(self.original_text) + 1))
         # 注: サロゲートペア文字は考慮していない
 
     def replace(self, begin, end, str_):
@@ -42,15 +42,17 @@ def replace(self, begin, end, str_):
 
         self.modified_text = str_.join([self.modified_text[:begin], self.modified_text[end:]])
 
-        offset = self.text_offsets[begin]
+        modified_begin = self.modified_to_original[begin]
+        modified_end = self.modified_to_original[end]
         length = len(str_)
         if end - begin > length:
-            del self.text_offsets[begin + length:end]
-        for i in range(length):
+            del self.modified_to_original[begin + length:end]
+        self.modified_to_original[begin] = modified_begin
+        for i in range(1, length):
             if begin + i < end:
-                self.text_offsets[begin + i] = offset
+                self.modified_to_original[begin + i] = modified_end
             else:
-                self.text_offsets.insert(begin + i, offset)
+                self.modified_to_original.insert(begin + i, modified_end)
 
     def get_original_text(self):
         return self.original_text
@@ -70,10 +72,10 @@ def build(self):
             # 注: サロゲートペア文字は考慮していない
             for _ in range(self.utf8_byte_length(ord(self.modified_text[i]))):
                 byte_indexes[j] = i
-                offsets[j] = self.text_offsets[i]
+                offsets[j] = self.modified_to_original[i]
                 j += 1
         byte_indexes[length] = len(modified_string_text)
-        offsets[length] = self.text_offsets[-1]
+        offsets[length] = self.modified_to_original[-1]
 
         char_categories = self.get_char_category_types(modified_string_text)
         char_category_continuities = self.get_char_category_continuities(modified_string_text, length, char_categories)

tests/plugin/test_default_input_text_plugin.py

@@ -60,7 +60,7 @@ def test_before_rewrite(self):
         self.assertEqual(9, text.get_original_index(24))
         self.assertEqual(9, text.get_original_index(26))
 
-    def test_after_write(self):
+    def test_after_rewrite(self):
         self.assertEqual(self.original_text, self.builder.get_original_text())
         self.assertEqual(self.original_text, self.builder.get_text())
         self.plugin.rewrite(self.builder)
@@ -76,7 +76,8 @@ def test_after_write(self):
         self.assertEqual(1, text.get_original_index(2))
         self.assertEqual(2, text.get_original_index(3))
         self.assertEqual(4, text.get_original_index(7))
-        self.assertEqual(4, text.get_original_index(11))
+        self.assertEqual(5, text.get_original_index(8))
+        self.assertEqual(5, text.get_original_index(11))
         self.assertEqual(7, text.get_original_index(15))
         self.assertEqual(7, text.get_original_index(17))
 

tests/test_tokenizer.py

@@ -70,6 +70,16 @@ def test_tokenize_kanji_alphabet_word(self):
         self.assertEqual(len(self.tokenizer_obj.tokenize('ab')), 1)
         self.assertEqual(len(self.tokenizer_obj.tokenize('特ab')), 2)
 
+    def test_tokenizer_with_dots(self):
+        ms = self.tokenizer_obj.tokenize('京都…')
+        self.assertEqual(4, ms.size())
+        self.assertEqual(ms[1].surface(), '…')
+        self.assertEqual(ms[1].normalized_form(), '.')
+        self.assertEqual(ms[2].surface(), '')
+        self.assertEqual(ms[2].normalized_form(), '.')
+        self.assertEqual(ms[3].surface(), '')
+        self.assertEqual(ms[3].normalized_form(), '.')
+
 
 if __name__ == '__main__':
     unittest.main()

tests/test_utf8inputtext.py

@@ -117,8 +117,8 @@ def test_replace_with_same_length(self):
         self.assertEqual(input_.get_original_index(12), 7)
         self.assertEqual(input_.get_original_index(13), 8)
         self.assertEqual(input_.get_original_index(15), 8)
-        self.assertEqual(input_.get_original_index(16), 8)
-        self.assertEqual(input_.get_original_index(18), 8)
+        self.assertEqual(input_.get_original_index(16), 10)
+        self.assertEqual(input_.get_original_index(18), 10)
         self.assertEqual(input_.get_original_index(19), 10)
         self.assertEqual(input_.get_original_index(22), 10)
         self.assertEqual(input_.get_original_index(31), 13)
@@ -147,13 +147,13 @@ def test_replaceWithInsertion(self):
         self.assertEqual(input_.get_original_text(), self.TEXT)
         self.assertEqual(input_.get_text(), "âｂC1あ234あああ𡈽アｺﾞ")
         self.assertEqual(len(input_.get_byte_text()), 35)
-        self.assertEqual(input_.get_original_index(0), 0)
-        self.assertEqual(input_.get_original_index(12), 7)
-        self.assertEqual(input_.get_original_index(13), 8)
-        self.assertEqual(input_.get_original_index(21), 8)
-        self.assertEqual(input_.get_original_index(22), 10)
-        self.assertEqual(input_.get_original_index(25), 10)
-        self.assertEqual(input_.get_original_index(35), 14)
+        self.assertEqual(input_.get_original_index(0), 0)  # â
+        self.assertEqual(input_.get_original_index(12), 7)  # 4
+        self.assertEqual(input_.get_original_index(13), 8)  # >あ< ああ
+        self.assertEqual(input_.get_original_index(21), 10)  # ああ >あ<
+        self.assertEqual(input_.get_original_index(22), 10)  # 𡈽
+        self.assertEqual(input_.get_original_index(25), 10)  # 𡈽
+        self.assertEqual(input_.get_original_index(35), 14)  #  ﾞ
 
     def test_replaceMultiTimes(self):
         self.builder.replace(0, 1, "a")