possible bug in while loop test

[Peter Otten]

Charles Hixson wrote:

> python3 --version
> Python 3.5.3
> 
> Running on Debian stretch
> 
> In this code s is a string parameter
> 
> while (j < k   and \
>           (s[j].isalnum())      or   \
>            (s[j] in seps and s[j+1].isalnum()) ):
>     j   =   j + 1
>     print ("i = {0}, j = {1}, k = {2}, len[s] = {3}".   \
>           format(i, j, k, len(s) ) )
> 
> Yields the result:
> 
> i = 25, j = 27, k = 31, len[s] = 32
> i = 25, j = 28, k = 31, len[s] = 32
> i = 25, j = 29, k = 31, len[s] = 32
> i = 25, j = 30, k = 31, len[s] = 32
> i = 25, j = 31, k = 31, len[s] = 32
> Traceback (most recent call last):
>    File "parse1.py", line 40, in <module>
>      print (parse1("The gostack distimms the doshes."))
>    File "parse1.py", line 21, in parse1
>      (s[j] in seps and s[j+1].isalnum()) ):
> IndexError: string index out of range
> 
> I hesitate to report this, because I've been wrong so often, but it
> looks to me like the last iteration should not have happened since j is
> not less than k.

You have made a bit of a mess of the while condition. Removing random space 
and newlines:

(j < k and (s[j].isalnum()) or (s[j] in seps and s[j+1].isalnum()))

The structure is

a and b or (c and d)  # a: j<k
                      # b: (s[j].isalnum())
                      # c: s[j] in seps
                      # d: s[j+1].isalnum()
which is evalued as

(a and b) or (c and d)

a and b --> False as you suspect, but then (c and d) is evalued, and if c 
succeeds also the expression d aka s[j+1].isalnum() which fails on accessing 
s[j+1].