1 r"""Fixer for unicode. 2 3 * Changes unicode to str and unichr to chr. 4 5 * If "...\u..." is not unicode literal change it into "...\\u...". 6 7 * Change u"..." into "...". 8 9 """ 10 11 from ..pgen2 import token 12 from .. import fixer_base 13 14 _mapping = {u"unichr" : u"chr", u"unicode" : u"str"} 15 16 class FixUnicode(fixer_base.BaseFix): 17 BM_compatible = True 18 PATTERN = "STRING | 'unicode' | 'unichr'" 19 20 def start_tree(self, tree, filename): 21 super(FixUnicode, self).start_tree(tree, filename) 22 self.unicode_literals = 'unicode_literals' in tree.future_features 23 24 def transform(self, node, results): 25 if node.type == token.NAME: 26 new = node.clone() 27 new.value = _mapping[node.value] 28 return new 29 elif node.type == token.STRING: 30 val = node.value 31 if not self.unicode_literals and val[0] in u'\'"' and u'\\' in val: 32 val = ur'\\'.join([ 33 v.replace(u'\\u', ur'\\u').replace(u'\\U', ur'\\U') 34 for v in val.split(ur'\\') 35 ]) 36 if val[0] in u'uU': 37 val = val[1:] 38 if val == node.value: 39 return node 40 new = node.clone() 41 new.value = val 42 return new 43
