Diff parser: Fix BOM with indentation issues

parso/python/diff.py

@@ -427,7 +427,8 @@ class DiffParser(object):
         tokens = self._tokenizer(
             lines,
             start_pos=(line_offset + 1, 0),
-            indents=indents
+            indents=indents,
+            is_first_token=line_offset == 0,
         )
         stack = self._active_parser.stack
         self._replace_tos_indent = None

parso/python/tokenize.py

@@ -389,7 +389,7 @@ def _print_tokens(func):
 
 
 # @_print_tokens
-def tokenize_lines(lines, version_info, start_pos=(1, 0), indents=None):
+def tokenize_lines(lines, version_info, start_pos=(1, 0), indents=None, is_first_token=True):
     """
     A heavily modified Python standard library tokenizer.
 
@@ -423,14 +423,13 @@ def tokenize_lines(lines, version_info, start_pos=(1, 0), indents=None):
     new_line = True
     prefix = ''  # Should never be required, but here for safety
     additional_prefix = ''
-    first = True
     lnum = start_pos[0] - 1
     fstring_stack = []
     for line in lines:  # loop over lines in stream
         lnum += 1
         pos = 0
         max_ = len(line)
-        if first:
+        if is_first_token:
             if line.startswith(BOM_UTF8_STRING):
                 additional_prefix = BOM_UTF8_STRING
                 line = line[1:]
@@ -441,7 +440,7 @@ def tokenize_lines(lines, version_info, start_pos=(1, 0), indents=None):
             pos = start_pos[1]
             max_ += start_pos[1]
 
-            first = False
+            is_first_token = False
 
         if contstr:                                         # continued string
             endmatch = endprog.match(line)

test/test_diff_parser.py

@@ -1565,3 +1565,23 @@ class Grammar:
         ''')
     differ.initialize(code1)
     differ.parse(code2, parsers=3, copies=1, expect_error_leaves=True)
+
+
+def test_byte_order_mark(differ):
+    code2 = dedent('''\
+
+        x
+        \ufeff
+                 else :
+        ''')
+    differ.initialize('\n')
+    differ.parse(code2, parsers=2, expect_error_leaves=True)
+
+    code3 = dedent('''\
+        \ufeff
+                 if:
+
+        x
+        ''')
+    differ.initialize('\n')
+    differ.parse(code3, parsers=2, expect_error_leaves=True)