How to convert an user-input into a function in python 2.7

Question

I am pretty new to programming but I want to define a function like the following with user input:

function = raw_input('What function shall be used? ') #user puts in "(1+1/x)^x"
def f(x): #using this to define
 y = (1+1/x)**x 
 return y

I hope it is clear what I try to do. Could you use input and define a priori that sin is used as math.sin? But how do I define the function afterwards?

Edited to add specification from the comments

Let's say the valid input is python code such as "(1+1/x)**x" like in the question. Could you now use eval to define a function f(x) which can be later used?

• juanpa.arrivillaga: Give then way the question is worded, it sounds like you want to define your own mini-language, or computer algebra system.
• Paulo Scardine: You can use eval, but then you have to trust the user input will not blow up the server.
• hiro protagonist: sympy.org may help...
• juanpa.arrivillaga: @PauloScardine no, you could not use eval because eval evaluates a python expression, but (1+1/x)^x does not mean (1+1/x)**x in python.
• zwer: You can use eval() but don't... It's very unsafe and your users would need to know at least a bit of Python (e.g. something to the power of something else is defined as ** not ^) - and if they do, why not let them define their own modules that your app will load instead of requiring them to manually type math formulas?

Answer 1

As I commented, this can be done with something like eval, but is unsafe because you should never trust random user input.

>>> def expr_to_fn(expr):
... expr = expr.replace('^', '**')
... symbols = set(re.findall(r'([A-Z,a-z]+)', expr))
... fn = eval("lambda {}: {}".format(",".join(symbols), expr))
... return fn
>>> f = expr_to_fn("(1+1/x)^x")
>>> f(2)
2.25

The safe way is to implenet a DSL and a parser, or use a CAS (computer algebra system) like sympy, as suggested by Hiro.

• juanpa.arrivillaga: I think this is almost there, but it still doesn't deal with the issue of sin, cos and the like. However, this does make me understand what you meant by using a context dictionary, which could probably take this the rest of the way there.
• Chiray: Totally overlooked Hiro's comment but sympy might be what I was looking for.
• Chiray: I will try what you did later tonight and see how far I can get with it, thank you!
• Paulo Scardine: @juanpa.arrivillaga I accept your apology but I will have to left that as an exercise for the OP because I lack the time to post a complete work solution right now.