Issue #9974: When untokenizing, use row info to insert backslash+newline.

terryjreedy · terryjreedy · commit 9dc3a36c849c · 2014-02-23T23:33:08.000-05:00
Original patches by A. Kuchling and G. Rees (#12691).
diff --git a/Lib/test/test_tokenize.py b/Lib/test/test_tokenize.py
@@ -2,7 +2,7 @@
 Tests for the tokenize module.
 
 The tests can be really simple. Given a small fragment of source
-code, print out a table with tokens. The ENDMARK is omitted for
+code, print out a table with tokens. The ENDMARKER is omitted for
 brevity.
 
     >>> dump_tokens("1 + 1")
@@ -1180,15 +1180,30 @@ def test_pathological_trailing_whitespace(self):
 class UntokenizeTest(TestCase):
 
     def test_bad_input_order(self):
+        # raise if previous row
         u = Untokenizer()
         u.prev_row = 2
         u.prev_col = 2
         with self.assertRaises(ValueError) as cm:
             u.add_whitespace((1,3))
         self.assertEqual(cm.exception.args[0],
                 'start (1,3) precedes previous end (2,2)')
+        # raise if previous column in row
         self.assertRaises(ValueError, u.add_whitespace, (2,1))
 
+    def test_backslash_continuation(self):
+        # The problem is that <whitespace>\<newline> leaves no token
+        u = Untokenizer()
+        u.prev_row = 1
+        u.prev_col =  1
+        u.tokens = []
+        u.add_whitespace((2, 0))
+        self.assertEqual(u.tokens, ['\\\n'])
+        u.prev_row = 2
+        u.add_whitespace((4, 4))
+        self.assertEqual(u.tokens, ['\\\n', '\\\n\\\n', '    '])
+        self.assertTrue(roundtrip('a\n  b\n    c\n  \\\n  c\n'))
+
     def test_iter_compat(self):
         u = Untokenizer()
         token = (NAME, 'Hello')
diff --git a/Lib/tokenize.py b/Lib/tokenize.py
@@ -234,6 +234,10 @@ def add_whitespace(self, start):
         if row < self.prev_row or row == self.prev_row and col < self.prev_col:
             raise ValueError("start ({},{}) precedes previous end ({},{})"
                              .format(row, col, self.prev_row, self.prev_col))
+        row_offset = row - self.prev_row
+        if row_offset:  
+            self.tokens.append("\\\n" * row_offset)
+            self.prev_col = 0
         col_offset = col - self.prev_col
         if col_offset:
             self.tokens.append(" " * col_offset)
@@ -248,6 +252,8 @@ def untokenize(self, iterable):
             if tok_type == ENCODING:
                 self.encoding = token
                 continue
+            if tok_type == ENDMARKER:
+                break
             self.add_whitespace(start)
             self.tokens.append(token)
             self.prev_row, self.prev_col = end
