gecko-dev/config/Expression.py

210 строки
6.2 KiB
Python
Исходник Обычный вид История

# ***** BEGIN LICENSE BLOCK *****
# Version: MPL 1.1/GPL 2.0/LGPL 2.1
#
# The contents of this file are subject to the Mozilla Public License Version
# 1.1 (the "License"); you may not use this file except in compliance with
# the License. You may obtain a copy of the License at
# http://www.mozilla.org/MPL/
#
# Software distributed under the License is distributed on an "AS IS" basis,
# WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
# for the specific language governing rights and limitations under the
# License.
#
# The Original Code is Mozilla build system.
#
# The Initial Developer of the Original Code is
# Mozilla Foundation.
# Portions created by the Initial Developer are Copyright (C) 2007
# the Initial Developer. All Rights Reserved.
#
# Contributor(s):
# Axel Hecht <axel@pike.org>
#
# Alternatively, the contents of this file may be used under the terms of
# either the GNU General Public License Version 2 or later (the "GPL"), or
# the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
# in which case the provisions of the GPL or the LGPL are applicable instead
# of those above. If you wish to allow use of your version of this file only
# under the terms of either the GPL or the LGPL, and not to allow others to
# use your version of this file under the terms of the MPL, indicate your
# decision by deleting the provisions above and replace them with the notice
# and other provisions required by the GPL or the LGPL. If you do not delete
# the provisions above, a recipient may use your version of this file under
# the terms of any one of the MPL, the GPL or the LGPL.
#
# ***** END LICENSE BLOCK *****
"""
Parses and evaluates simple statements for Preprocessor:
Expression currently supports the following grammar, whitespace is ignored:
expression :
unary ( ( '==' | '!=' ) unary ) ? ;
unary :
'!'? value ;
value :
[0-9]+ # integer
| \w+ # string identifier or value;
"""
import re
class Expression:
def __init__(self, expression_string):
"""
Create a new expression with this string.
The expression will already be parsed into an Abstract Syntax Tree.
"""
self.content = expression_string
self.offset = 0
self.__ignore_whitespace()
self.e = self.__get_equality()
if self.content:
raise Expression.ParseError, self
def __get_equality(self):
"""
Production: unary ( ( '==' | '!=' ) unary ) ?
"""
if not len(self.content):
return None
rv = Expression.__AST("equality")
# unary
rv.append(self.__get_unary())
self.__ignore_whitespace()
if not re.match('[=!]=', self.content):
# no equality needed, short cut to our prime unary
return rv[0]
# append operator
rv.append(Expression.__ASTLeaf('op', self.content[:2]))
self.__strip(2)
self.__ignore_whitespace()
rv.append(self.__get_unary())
self.__ignore_whitespace()
return rv
def __get_unary(self):
"""
Production: '!'? value
"""
# eat whitespace right away, too
not_ws = re.match('!\s*', self.content)
if not not_ws:
return self.__get_value()
rv = Expression.__AST('not')
self.__strip(not_ws.end())
rv.append(self.__get_value())
self.__ignore_whitespace()
return rv
def __get_value(self):
"""
Production: ( [0-9]+ | \w+)
Note that the order is important, and the expression is kind-of
ambiguous as \w includes 0-9. One could make it unambiguous by
removing 0-9 from the first char of a string literal.
"""
rv = None
word_len = re.match('[0-9]*', self.content).end()
if word_len:
value = int(self.content[:word_len])
rv = Expression.__ASTLeaf('int', value)
else:
word_len = re.match('\w*', self.content).end()
if word_len:
rv = Expression.__ASTLeaf('string', self.content[:word_len])
else:
raise Expression.ParseError, self
self.__strip(word_len)
self.__ignore_whitespace()
return rv
def __ignore_whitespace(self):
ws_len = re.match('\s*', self.content).end()
self.__strip(ws_len)
return
def __strip(self, length):
"""
Remove a given amount of chars from the input and update
the offset.
"""
self.content = self.content[length:]
self.offset += length
def evaluate(self, context):
"""
Evaluate the expression with the given context
"""
# Helper function to evaluate __get_equality results
def eval_equality(tok):
left = opmap[tok[0].type](tok[0])
right = opmap[tok[2].type](tok[2])
rv = left == right
if tok[1].value == '!=':
rv = not rv
return rv
# Mapping from token types to evaluator functions
# Apart from (non-)equality, all these can be simple lambda forms.
opmap = {
'equality': eval_equality,
'not': lambda tok: not opmap[tok[0].type](tok[0]),
'string': lambda tok: context[tok.value],
'int': lambda tok: tok.value}
return opmap[self.e.type](self.e);
class __AST(list):
"""
Internal class implementing Abstract Syntax Tree nodes
"""
def __init__(self, type):
self.type = type
super(self.__class__, self).__init__(self)
class __ASTLeaf:
"""
Internal class implementing Abstract Syntax Tree leafs
"""
def __init__(self, type, value):
self.value = value
self.type = type
def __str__(self):
return self.value.__str__()
def __repr__(self):
return self.value.__repr__()
class ParseError(StandardError):
"""
Error raised when parsing fails.
It has two members, offset and content, which give the offset of the
error and the offending content.
"""
def __init__(self, expression):
self.offset = expression.offset
self.content = expression.content[:3]
def __str__(self):
return 'Unexpected content at offset %i, "%s"'%(self.offset, self.content)
class Context(dict):
"""
This class holds variable values by subclassing dict, and while it
truthfully reports True and False on
name in context
it returns the variable name itself on
context["name"]
to reflect the ambiguity between string literals and preprocessor
variables.
"""
def __getitem__(self, key):
if key in self:
return super(self.__class__, self).__getitem__(key)
return key