Unquoted String Envvar is Incorrectly Parsed as Integer

[kupmanyu]

Observed behavior

An unquoted String envvar that matches the regex [0-9]+[kKmMgGtTpPeE][0-9].* (e.g. "0K1abc" or "8m4dwr") is incorrectly parsed as an integer and not as a string.

The below error is present in the logs:

nats-server: variable reference for 'TEST_KEY' on line 10 could not be parsed: Parse error on line 1: 'Expected a top-level value to end with a new line, comment or EOF, but got '4' instead.'

Related to #3431 and its fix in #3434 and #3436.

The issue is probably related to the unicode.isDigit(r) check in the lexConvenientNumber function here

Expected behavior

String envvar is parsed as a string.

Server and client version

Tested on both 2.9.16-scratch and latest

Host environment

No response

Steps to reproduce

Here is a unit test that shows this issue. It can be part of the TestConvenientIntegerValues test in conf/lex_test.go.

expectedItems := []item{
	{itemKey, "foo", 1, 0},
	{itemString, "0K1abc", 1, 6},
	{itemEOF, "", 1, 0},
}
lx := lex("foo = 0K1abc")
expect(t, lx, expectedItems)

Expected result:

Test case passes

Actual Result:

Test case fails with error:

Testing: 'foo = 0K1abc'
        Expected "(String, '0K1abc', 1, 6)", received "(Integer, '0K', 1, 6)"

[kupmanyu]

The regex [0-9][kKmMgGtTpPeE][0-9].* should actually be [0-9]+[kKmMgGtTpPeE][0-9].* since the string "80m3hfe" also raises an error. I will update the issue with the correct regex.