1 #!/usr/bin/env python

     2 #

     3 # hggettext - carefully extract docstrings for Mercurial

     4 #

     5 # Copyright 2009 Matt Mackall <mpm@selenic.com> and others

     6 #

     7 # This software may be used and distributed according to the terms of the

     8 # GNU General Public License version 2, incorporated herein by reference.

     9 

    10 # The normalize function is taken from pygettext which is distributed

    11 # with Python under the Python License, which is GPL compatible.

    12 

    13 """Extract docstrings from Mercurial commands.

    14 

    15 Compared to pygettext, this script knows about the cmdtable and table

    16 dictionaries used by Mercurial, and will only extract docstrings from

    17 functions mentioned therein.

    18 

    19 Use xgettext like normal to extract strings marked as translatable and

    20 join the message cataloges to get the final catalog.

    21 """

    22 

    23 from mercurial import demandimport; demandimport.enable()

    24 import sys, inspect

    25 

    26 

    27 def escape(s):

    28     # The order is important, the backslash must be escaped first

    29     # since the other replacements introduce new backslashes

    30     # themselves.

    31     s = s.replace('\\', '\\\\')

    32     s = s.replace('\n', '\\n')

    33     s = s.replace('\r', '\\r')

    34     s = s.replace('\t', '\\t')

    35     s = s.replace('"', '\\"')

    36     return s

    37 

    38 

    39 def normalize(s):

    40     # This converts the various Python string types into a format that

    41     # is appropriate for .po files, namely much closer to C style.

    42     lines = s.split('\n')

    43     if len(lines) == 1:

    44         s = '"' + escape(s) + '"'

    45     else:

    46         if not lines[-1]:

    47             del lines[-1]

    48             lines[-1] = lines[-1] + '\n'

    49         lines = map(escape, lines)

    50         lineterm = '\\n"\n"'

    51         s = '""\n"' + lineterm.join(lines) + '"'

    52     return s

    53 

    54 

    55 def poentry(path, lineno, s):

    56     return ('#: %s:%d\n' % (path, lineno) +

    57             'msgid %s\n' % normalize(s) +

    58             'msgstr ""\n')

    59 

    60 

    61 def offset(src, doc, name, default):

    62     """Compute offset or issue a warning on stdout."""

    63     # Backslashes in doc appear doubled in src.

    64     end = src.find(doc.replace('\\', '\\\\'))

    65     if end == -1:

    66         # This can happen if the docstring contains unnecessary escape

    67         # sequences such as \" in a triple-quoted string. The problem

    68         # is that \" is turned into " and so doc wont appear in src.

    69         sys.stderr.write("warning: unknown offset in %s, assuming %d lines\n"

    70                          % (name, default))

    71         return default

    72     else:

    73         return src.count('\n', 0, end)

    74 

    75 

    76 def importpath(path):

    77     """Import a path like foo/bar/baz.py and return the baz module."""

    78     if path.endswith('.py'):

    79         path = path[:-3]

    80     if path.endswith('/__init__'):

    81         path = path[:-9]

    82     path = path.replace('/', '.')

    83     mod = __import__(path)

    84     for comp in path.split('.')[1:]:

    85         mod = getattr(mod, comp)

    86     return mod

    87 

    88 

    89 def docstrings(path):

    90     """Extract docstrings from path.

    91 

    92     This respects the Mercurial cmdtable/table convention and will

    93     only extract docstrings from functions mentioned in these tables.

    94     """

    95     mod = importpath(path)

    96     if mod.__doc__:

    97         src = open(path).read()

    98         lineno = 1 + offset(src, mod.__doc__, path, 7)

    99         print poentry(path, lineno, mod.__doc__)

   100 

   101     cmdtable = getattr(mod, 'cmdtable', {})

   102     if not cmdtable:

   103         # Maybe we are processing mercurial.commands?

   104         cmdtable = getattr(mod, 'table', {})

   105 

   106     for entry in cmdtable.itervalues():

   107         func = entry[0]

   108         if func.__doc__:

   109             src = inspect.getsource(func)

   110             name = "%s.%s" % (path, func.__name__)

   111             lineno = func.func_code.co_firstlineno

   112             lineno += offset(src, func.__doc__, name, 1)

   113             print poentry(path, lineno, func.__doc__)

   114 

   115 

   116 if __name__ == "__main__":

   117     for path in sys.argv[1:]:

   118         docstrings(path)