contrib/byteify-strings.py
changeset 38384 1d9c97db465f
parent 36617 5246f940a48e
child 38385 a2976c27dac4
equal deleted inserted replaced
38383:23dc901cdf13 38384:1d9c97db465f
       
     1 # byteify-strings.py - transform string literals to be Python 3 safe
       
     2 #
       
     3 # Copyright 2015 Gregory Szorc <gregory.szorc@gmail.com>
       
     4 #
       
     5 # This software may be used and distributed according to the terms of the
       
     6 # GNU General Public License version 2 or any later version.
       
     7 
       
     8 from __future__ import absolute_import
       
     9 
       
    10 import io
       
    11 import token
       
    12 import tokenize
       
    13 
       
    14 if True:
       
    15     def replacetokens(tokens, fullname):
       
    16         """Transform a stream of tokens from raw to Python 3.
       
    17 
       
    18         Returns a generator of possibly rewritten tokens.
       
    19 
       
    20         The input token list may be mutated as part of processing. However,
       
    21         its changes do not necessarily match the output token stream.
       
    22         """
       
    23         futureimpline = False
       
    24 
       
    25         # The following utility functions access the tokens list and i index of
       
    26         # the for i, t enumerate(tokens) loop below
       
    27         def _isop(j, *o):
       
    28             """Assert that tokens[j] is an OP with one of the given values"""
       
    29             try:
       
    30                 return tokens[j].type == token.OP and tokens[j].string in o
       
    31             except IndexError:
       
    32                 return False
       
    33 
       
    34         def _findargnofcall(n):
       
    35             """Find arg n of a call expression (start at 0)
       
    36 
       
    37             Returns index of the first token of that argument, or None if
       
    38             there is not that many arguments.
       
    39 
       
    40             Assumes that token[i + 1] is '('.
       
    41 
       
    42             """
       
    43             nested = 0
       
    44             for j in range(i + 2, len(tokens)):
       
    45                 if _isop(j, ')', ']', '}'):
       
    46                     # end of call, tuple, subscription or dict / set
       
    47                     nested -= 1
       
    48                     if nested < 0:
       
    49                         return None
       
    50                 elif n == 0:
       
    51                     # this is the starting position of arg
       
    52                     return j
       
    53                 elif _isop(j, '(', '[', '{'):
       
    54                     nested += 1
       
    55                 elif _isop(j, ',') and nested == 0:
       
    56                     n -= 1
       
    57 
       
    58             return None
       
    59 
       
    60         def _ensureunicode(j):
       
    61             """Make sure the token at j is a unicode string
       
    62 
       
    63             This rewrites a string token to include the unicode literal prefix
       
    64             so the string transformer won't add the byte prefix.
       
    65 
       
    66             Ignores tokens that are not strings. Assumes bounds checking has
       
    67             already been done.
       
    68 
       
    69             """
       
    70             st = tokens[j]
       
    71             if st.type == token.STRING and st.string.startswith(("'", '"')):
       
    72                 tokens[j] = st._replace(string='u%s' % st.string)
       
    73 
       
    74         for i, t in enumerate(tokens):
       
    75             # Convert most string literals to byte literals. String literals
       
    76             # in Python 2 are bytes. String literals in Python 3 are unicode.
       
    77             # Most strings in Mercurial are bytes and unicode strings are rare.
       
    78             # Rather than rewrite all string literals to use ``b''`` to indicate
       
    79             # byte strings, we apply this token transformer to insert the ``b``
       
    80             # prefix nearly everywhere.
       
    81             if t.type == token.STRING:
       
    82                 s = t.string
       
    83 
       
    84                 # Preserve docstrings as string literals. This is inconsistent
       
    85                 # with regular unprefixed strings. However, the
       
    86                 # "from __future__" parsing (which allows a module docstring to
       
    87                 # exist before it) doesn't properly handle the docstring if it
       
    88                 # is b''' prefixed, leading to a SyntaxError. We leave all
       
    89                 # docstrings as unprefixed to avoid this. This means Mercurial
       
    90                 # components touching docstrings need to handle unicode,
       
    91                 # unfortunately.
       
    92                 if s[0:3] in ("'''", '"""'):
       
    93                     yield t
       
    94                     continue
       
    95 
       
    96                 # If the first character isn't a quote, it is likely a string
       
    97                 # prefixing character (such as 'b', 'u', or 'r'. Ignore.
       
    98                 if s[0] not in ("'", '"'):
       
    99                     yield t
       
   100                     continue
       
   101 
       
   102                 # String literal. Prefix to make a b'' string.
       
   103                 yield t._replace(string='b%s' % t.string)
       
   104                 continue
       
   105 
       
   106             # Insert compatibility imports at "from __future__ import" line.
       
   107             # No '\n' should be added to preserve line numbers.
       
   108             if (t.type == token.NAME and t.string == 'import' and
       
   109                 all(u.type == token.NAME for u in tokens[i - 2:i]) and
       
   110                 [u.string for u in tokens[i - 2:i]] == ['from', '__future__']):
       
   111                 futureimpline = True
       
   112             if t.type == token.NEWLINE and futureimpline:
       
   113                 futureimpline = False
       
   114                 if fullname == 'mercurial.pycompat':
       
   115                     yield t
       
   116                     continue
       
   117                 r, c = t.start
       
   118                 l = (b'; from mercurial.pycompat import '
       
   119                      b'delattr, getattr, hasattr, setattr, xrange, '
       
   120                      b'open, unicode\n')
       
   121                 for u in tokenize.tokenize(io.BytesIO(l).readline):
       
   122                     if u.type in (tokenize.ENCODING, token.ENDMARKER):
       
   123                         continue
       
   124                     yield u._replace(
       
   125                         start=(r, c + u.start[1]), end=(r, c + u.end[1]))
       
   126                 continue
       
   127 
       
   128             # This looks like a function call.
       
   129             if t.type == token.NAME and _isop(i + 1, '('):
       
   130                 fn = t.string
       
   131 
       
   132                 # *attr() builtins don't accept byte strings to 2nd argument.
       
   133                 if (fn in ('getattr', 'setattr', 'hasattr', 'safehasattr') and
       
   134                         not _isop(i - 1, '.')):
       
   135                     arg1idx = _findargnofcall(1)
       
   136                     if arg1idx is not None:
       
   137                         _ensureunicode(arg1idx)
       
   138 
       
   139                 # .encode() and .decode() on str/bytes/unicode don't accept
       
   140                 # byte strings on Python 3.
       
   141                 elif fn in ('encode', 'decode') and _isop(i - 1, '.'):
       
   142                     for argn in range(2):
       
   143                         argidx = _findargnofcall(argn)
       
   144                         if argidx is not None:
       
   145                             _ensureunicode(argidx)
       
   146 
       
   147                 # It changes iteritems/values to items/values as they are not
       
   148                 # present in Python 3 world.
       
   149                 elif fn in ('iteritems', 'itervalues'):
       
   150                     yield t._replace(string=fn[4:])
       
   151                     continue
       
   152 
       
   153             # Emit unmodified token.
       
   154             yield t