[ 2441975 ] Naming convention for Testsuites Management

[HolQue]

Hi Son,

derived from the regular expression and naming proposals in

#206 (comment)
#206 (comment)

and also based on

#366

I created this regular expression as naming convention for the Testsuites Management:

"^[\p{L}_][\p{L}0-9_\+\-\*/]*$"

It seems to work properly in pure Python:

search_pattern = "^[\p{L}_][\p{L}0-9_\+\-\*/]*$"
regex_pattern  = re.compile(search_pattern)

key_name = "123" # not allowed
key_name = "a123" # allowed
key_name = "_a123" # allowed
key_name = "𠼭_a123" # allowed
key_name = "𠼭_a€123" # not allowed
key_name = "𠼭_a#123" # not allowed
key_name = "𠼭_a-123" # allowed
key_name = "𠼭_a+123" # allowed
key_name = "𠼭_a*123" # allowed
key_name = "𠼭_a/123" # allowed
key_name = "$-𠼭_a/123" # not allowed
key_name = "[$-??_a/123]" # not allowed
key_name = "{$-??_a/123}" # not allowed

In CJasonPreprocessor.py

I exchanged

import re

by

import regex as re

But I get other (invalid) results than above in pure Python (names are accepted instead of being declined).

OK, maybe 'import regex as re' only is too simple and further adaptions in the JsonPreprocessor code are requiered.

But maybe you can check this and give "^[\p{L}_][\p{L}0-9_\+\-\*/]*$" a chance.

[test-fullautomation]

Hi Holger,

I wonder if we really want "*","+" and "/" as apart of a variable name.
Sorry, but I forgot the background for this after all the long discussions...

Can you short explain.

Thank you,
Thomas

[HolQue]

Hi Thomas,

You are right. As I mentioned above, the regular expression I have derived from already existing proposal. But they are wrong (belong to Python, but not to Robot Framework).

The Robot Framework is more restrictive than Python. A name must not start with an underline and must not contain special characters.

Therefore new proposal

search_pattern = "^[\p{L}][\p{L}0-9_]*$"

[test-fullautomation]

Hi @HolQue ,
Are you sure?
n underscores at the beginning of a variable works for me with robot.
Even one undercore alone works as variable name.

Proposal:
search_pattern = "^[\p{L}][\p{L}0-9]*$"

Please check.
Thank you,
Thomas

[HolQue]

Hi Thomas,

*** Variables ***

${ABC}    A
${__ABC}    B


*** Test Cases ***

QuickTest

    Log    ABC: ${ABC}    console=yes
    Log    __ABC: ${__ABC}    console=yes

Result:

INFO - ABC: A
INFO - __ABC: A

Same behavior when these variables are read in by JsonPreprocessor.

[HolQue]

Robot Framework main documentation tells:

"Robot Framework variables, similarly as keywords, are case-insensitive, and also spaces and underscores are ignored."

I several times stumbled about this "underscores are ignored" statement. Therefore I would be very careful when using them nevertheless.

[test-fullautomation]

I agree, but this means also that leading underscore(s) is/are allowed...

Puh.... then jsonpreprocessor would treat it properly, but after import into robot the last asigned value would survive. Not good...

O.K. we will allow underscores, but the document must have a fat warning that same variable names only different because of underscores will be treated as the same variables. Together with your example from above.