In Python 3.9.2, parsing 'z' (small letter) as '%z' (time zone offset) using datetime.strptime emits an IndexError.

>>> from datetime import datetime
>>> datetime.strptime('z', '%z')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.9/_strptime.py", line 568, in _strptime_datetime
    tt, fraction, gmtoff_fraction = _strptime(data_string, format)
  File "/usr/local/lib/python3.9/_strptime.py", line 453, in _strptime
    if z[3] == ':':
IndexError: string index out of range

I expect ValueError (or some another useful error) as follows.
ValueError: time data 'z' does not match format '%z'

This is caused by compiling '%z' to a pattern containing 'Z' (for UTC) with the IGNORECASE flag and accessing z[3] without noticing 'z' is accepted by the regexp.

I noticed another unexpectedeffect of the IGNORECASE flag. It enables some non-ascii characters to match against the alphabets.

>>> from datetime import datetime
>>> datetime.strptime("Apr\u0130l", "%B")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.9/_strptime.py", line 568, in _strptime_datetime
    tt, fraction, gmtoff_fraction = _strptime(data_string, format)
  File "/usr/local/lib/python3.9/_strptime.py", line 391, in _strptime
    month = locale_time.f_month.index(found_dict['B'].lower())
ValueError: 'apri̇l' is not in list

I expect time data does not match error. The ASCII flag will disable matching unexpected unicode characters.

I will address the original issue regarding '%z', but the second issue actually has to do with the Unicode representation of Turkish characters. In Turkish, the letter I ('\u0049') is a capital ı ('\u0131') and the letter İ ('\u0130') is a capital i ('\u0069'). In Python however, the lowercase of I is i, as in English.

>>> '\u0049'.lower()
'i'
>>> '\u0130'.lower()
'i̇'

We see that the lowercase forms of both I and İ are i, consistent with English in one case and Turkish in the other.

@noormichael Thank you for submitting a patch, I confirmed the original issue is fixed. I'm ok this ticket is closed. Regarding the second issue, I learned it is a Turkish character (thanks!), but the error is same type so will not cause such a critical issue.