Finding untranslated Python strings

string-checker.py

#! /usr/bin/env python3

import ast
import gettext
from gettext import gettext as _
import sys


def get_func_name(node):
    cls = node.__class__.__name__

    if cls == 'Call':
        return get_func_name(node.func)
    elif cls == 'Attribute':
        return '{}.{}'.format(
            get_func_name(node.value),
            node.attr)
    elif cls == 'Name':
        return get_func_name(node.id)
    elif cls == 'str':
        return node
    elif cls == 'Str':
        return "<String literal>"
    elif cls == 'Subscript':
        return '{}[{}]'.format(get_func_name(node.value),
                               get_func_name(node.slice))
    elif cls == 'Index':
        return get_func_name(node.value)
    else:
        print('ERROR: Unknown class: {}'.format(cls))


class ShowStrings(ast.NodeVisitor):
    TRANSLATION_FUNCTIONS = [
        '_',  # gettext.gettext is often imported under this name
        'gettext',
        'gettext.gettext',
        # FIXME: this list is pretty much incomplete
    ]
    UNTRANSLATED = 'untranslated 9'

    def __init__(self, filename=None):
        super(ShowStrings, self).__init__()

        self.in_call = []
        self.filename = filename or '<parsed string>'

    def visit_with_trace(self, node, func):
        self.in_call.append((func, node.lineno, node.col_offset))
        self.visit(node)
        self.in_call.pop()

    def visit_Str(self, node):
        # TODO: make it possible to ignore untranslated strings
        # TODO: make this ignore docstrings

        # if we are not in a translator function, issue a warning
        if not self.in_call or \
           self.in_call[-1][0] not in self.TRANSLATION_FUNCTIONS:
            try:
                funcname = self.in_call[-1][0]
            except IndexError:
                funcname = None

            funcall_msg = "outside a function call" if funcname is None \
                          else "inside a call to {funcname}".format(
                                  funcname=funcname)

            print("WARNING: Untranslated string found at "
                  "{filename}:{line}:{col} {funcall_msg}".format(
                      filename=self.filename,
                      line=node.lineno,
                      col=node.col_offset,
                      funcall_msg=funcall_msg))

    def visit_Call(self, node):
        # if we are in a translator function, issue a warninc
        if self.in_call and self.in_call[-1][0] in self.TRANSLATION_FUNCTIONS:
            print("WARNING: function call within a translation function at "
                  "{filename}:{line}:{col}".format(filename=self.filename,
                                                   line=node.lineno,
                                                   col=node.col_offset))
        funcname = get_func_name(node)

        for arg in node.args:
            self.visit_with_trace(arg, funcname)

        for kwarg in node.keywords:
            self.visit_with_trace(kwarg.value, funcname)

    def generic_visit(self, node):
        # if we are inside a translator function, issue a warning
        if self.in_call and self.in_call[-1][0] in self.TRANSLATION_FUNCTIONS:
            # Some ast nodes, like Add don’t have position information
            if hasattr(node, 'lineno'):
                print("WARNING: something not a string ({klass}) found in a "
                      "translation function at {filename}:{line}:{col}".format(
                          filename=self.filename,
                          klass=node.__class__.__name__,
                          line=node.lineno,
                          col=node.col_offset))
            else:
                print("WARNING: something not a string ({klass}) found in a "
                      "translation function.  Position unknown; function call "
                      "is at {filename}:{line}:{col}".format(
                          filename=self.filename,
                          klass=node.__class__.__name__,
                          line=self.in_call[-1][1],
                          col=self.in_call[-1][2]))

        super(ShowStrings, self).generic_visit(node)


def tst(*args, **kwargs):
    pass


def actual_tests():
    _('translated 1')
    tst(_('translated 2'))
    tst(gettext.gettext('translated 3'))
    tst(_('translated 4') + 'native 1')
    tst('native 2'
        'native 3')
    tst(_('native 4' + 'native 5'))
    tst('native 6', b='native 7')
    tst(_(tst('hello!')))


if __name__ == '__main__':
    try:
        filename = sys.argv[1]
    except IndexError:
        filename = __file__
        print("INFO:    No filename specified, checking myself.")

    with open(filename, 'r') as f:
        code = f.read()

    root = ast.parse(code)

    show_strings = ShowStrings(filename=filename)
    show_strings.visit(root)