Resolved

That's exactly why I hate illogical usage of global variables. C logic with headers is so much better in cases like this.

Random tests ask for 0 position, although the description says "Strings are 1-indexes (sic)" and expect strange results, e.g.:

INPUT: "Eu%s", pos1: 3, pos2: 0
None should equal 'Eus%Euss'
 
INPUT: "^r/p", pos1: 0, pos2: 1
None should equal 'pr/^^r/p'
 
INPUT: "3/", pos1: 0, pos2: 1
None should equal '/33/'
 
INPUT: "q", pos1: 0, pos2: 0
None should equal 'qqqq'

symbols is used inside a lambda/function, and code inside those is executed only when you actually call the function. And as the first call is done after teh definition of symbols, there's no problem.

still the expected result to compute before.

anyway...

Too much bother :)

right on

is that you who worte them? If so, see message above... :/

random tests there are, now, ok, but well... what a "piece of code"... x/

• no need to make a copy of the string: strings are immutable in python...
• don't build the input string using concatenation => append the chars to a list, then join it at the end.
• use the describe/it blocks to separate the different parts of the tests (read this)
• not a problem (with the current runner), but that would be better that you compute the expected value before the assertion (this way, if the code of te user raises an error, the stacktrace won't show the name of the reference solution).

Resolved

Renamed it to swap_letters

Renamed parameter to word

Added to description that positions starts from 1. Also added one more example.

It is not clear that positions are 1-indexed.

That's rather an issue.