contrib/hgfixes/fix_bytes.py
author Mads Kiilerich <mads@kiilerich.com>
Tue, 31 Jul 2012 03:30:42 +0200
branchstable
changeset 17299 e51d4aedace9
parent 11748 37a70a784397
child 21637 48ef68004ec9
permissions -rw-r--r--
check-code: indent 4 spaces in py files
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
11747
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
     1
"""Fixer that changes plain strings to bytes strings."""
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
     2
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
     3
import re
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
     4
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
     5
from lib2to3 import fixer_base
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
     6
from lib2to3.pgen2 import token
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
     7
from lib2to3.fixer_util import Name
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
     8
from lib2to3.pygram import python_symbols as syms
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
     9
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    10
_re = re.compile(r'[rR]?[\'\"]')
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    11
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    12
# XXX: Implementing a blacklist in 2to3 turned out to be more troublesome than
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    13
# blacklisting some modules inside the fixers. So, this is what I came with.
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    14
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    15
blacklist = ['mercurial/demandimport.py',
11748
37a70a784397 py3kcompat: added a "compatibility layer" for py3k
Renato Cunha <renatoc@gmail.com>
parents: 11747
diff changeset
    16
             'mercurial/py3kcompat.py', # valid python 3 already
11747
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    17
             'mercurial/i18n.py',
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    18
            ]
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    19
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    20
def isdocstring(node):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    21
    def isclassorfunction(ancestor):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    22
        symbols = (syms.funcdef, syms.classdef)
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    23
        # if the current node is a child of a function definition, a class
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    24
        # definition or a file, then it is a docstring
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    25
        if ancestor.type == syms.simple_stmt:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    26
            try:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    27
                while True:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    28
                    if ancestor.type in symbols:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    29
                        return True
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    30
                    ancestor = ancestor.parent
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    31
            except AttributeError:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    32
                return False
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    33
        return False
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    34
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    35
    def ismodule(ancestor):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    36
        # Our child is a docstring if we are a simple statement, and our
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    37
        # ancestor is file_input. In other words, our child is a lone string in
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    38
        # the source file.
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    39
        try:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    40
            if (ancestor.type == syms.simple_stmt and
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    41
                ancestor.parent.type == syms.file_input):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    42
                    return True
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    43
        except AttributeError:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    44
            return False
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    45
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    46
    def isdocassignment(ancestor):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    47
        # Assigning to __doc__, definitely a string
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    48
        try:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    49
            while True:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    50
                if (ancestor.type == syms.expr_stmt and
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    51
                    Name('__doc__') in ancestor.children):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    52
                        return True
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    53
                ancestor = ancestor.parent
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    54
        except AttributeError:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    55
            return False
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    56
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    57
    if ismodule(node.parent) or \
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    58
       isdocassignment(node.parent) or \
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    59
       isclassorfunction(node.parent):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    60
        return True
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    61
    return False
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    62
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    63
def shouldtransform(node):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    64
    specialnames = ['__main__']
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    65
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    66
    if node.value in specialnames:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    67
        return False
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    68
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    69
    ggparent = node.parent.parent.parent
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    70
    sggparent = str(ggparent)
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    71
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    72
    if 'getattr' in sggparent or \
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    73
       'hasattr' in sggparent or \
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    74
       'setattr' in sggparent or \
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    75
       'encode' in sggparent or \
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    76
       'decode' in sggparent:
17299
e51d4aedace9 check-code: indent 4 spaces in py files
Mads Kiilerich <mads@kiilerich.com>
parents: 11748
diff changeset
    77
        return False
11747
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    78
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    79
    return True
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    80
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    81
class FixBytes(fixer_base.BaseFix):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    82
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    83
    PATTERN = 'STRING'
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    84
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    85
    def transform(self, node, results):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    86
        if self.filename in blacklist:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    87
            return
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    88
        if node.type == token.STRING:
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    89
            if _re.match(node.value):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    90
                if isdocstring(node):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    91
                    return
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    92
                if not shouldtransform(node):
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    93
                    return
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    94
                new = node.clone()
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    95
                new.value = 'b' + new.value
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    96
                return new
40d5633889bb hgfixes: add a fixer to convert plain strings to bytestrings
Renato Cunha <renatoc@gmail.com>
parents:
diff changeset
    97