views:

46

answers:

3

I am using a re.sub callback to replace substrings with random values, but I would like the random values to be the same across different strings. Since the re.sub callback does not allow arguments, I am not sure how to do this.

Here is a simplified version of what I'm doing:

def evaluate(match):
    mappings = {'A': 1, 'B': 2}
    return str(eval(match.group(0)[2:-1], mappings))

# variables = {'A':[1,2,3,4,5], 'B':[1,2,3,4,5]}    
# mappings2 = {k:v[random.randint(0,len(v)-1)] for k, v in variables.items()}
string_one: "#{A} + #{B}"
string_two: "#{A+B}"
newstring_one = sub(r'\#\{([^#]+)\}', evaluate, string_one)
newstring_two = sub(r'\#\{([^#]+)\}', evaluate, string_two)

Now, as it stands, the strings will be properly evaluated: newstring_one is "1 + 2" and newstring_two is "3". But I want to be able to pick the values randomly, and still have them replaced in both strings. This would involve deleting the 'mappings' line in 'evaluate', and using something like the two commented lines. How, though, can I get my randomly chosen mappings2 to be used when eval-ing both strings, if I cannot pass it as an argument in the re.sub callback function?

Many thanks.

+1  A: 

You could use a function object.

 class A(object):
  def __init__(self, mappings):
    self.mappings = mappings
  def __call__(self, match):
    return str(eval(match.group(0)[2:-1], self.mappings))

 evaluate = A({'A': 1, 'B': 2})
KennyTM
+3  A: 

The easiest way I guess is to make use of functools.partial, which allows you create a "partially evaluated" function:

from functools import partial

def evaluate(match, mappings):
    return str(eval(match.group(0)[2:-1], mappings))

mappings = {'A': 1, 'B': 2}  # Or whatever ...

newstring = sub(r'\#\{([^#]+)\}', partial(evaluate, mappings=mappings), string)
Ferdinand Beyer
Just a note: if you reverse the order of the `evaluate` arguments, you can just say `partial(evaluate, mappings)`.
ΤΖΩΤΖΙΟΥ
Yes, but I used the argument order on purpose since to me it feels more natural to define `evaluate()` this way and I wanted to demonstrate the flexibility of `partial()`.
Ferdinand Beyer
+2  A: 

You could create a closure.

def evaluator(mappings):
  def f(match):
    return str(eval(match.group(0)[2:-1], mappings))
  return f

evaluate = evaluator({'A': 1, 'B': 2})

Since f is just a single statement, you could simply use lambda:

def evaluator(mappings):
  return lambda match: str(eval(match.group(0)[2:-1], mappings))
KennyTM