Fix issue https://github.com/nunit/nunit-console/issues/1680 #1681
Conversation
Changes: - If the filter is "tests" and the operator is regex, take the value without normalization. - If the filter is "tests" and the operator is comparison, take the value using "GetTestName". - GetTestName takes the value to the end, rather than truncating at the first closing parenthesis. - GetTestName removes extra spaces everywhere except in quoted arguments. - GetTestName ignores escaped quotes.
|
@dotnet-policy-service agree |
| rhs = GetTestName(); | ||
| rhs = op == MATCH_OP || op == NOMATCH_OP | ||
| ? Expect(TokenKind.String, TokenKind.Word) | ||
| : GetTestName(); |
There was a problem hiding this comment.
All tests pass without this change. Could you either remove it or add a test that shows it's required?
There was a problem hiding this comment.
Regular expressions lack predictability in string formatting, which can lead to mismatched quotes, potentially resulting in the removal of spaces in undesirable places.
Examples:
- A regular expression can be an arbitrary substring of test names.
tests:
a.b.c("1111", "a b")
a.b.c("2222", "b c")
a.b.c("3333", "a b")
a.b.c("3333", "a b с")
a.b.c(Enum.X, "a b с")
command:
nunit "test.dll" -where 'test =~ /a b c/' --explore
expected:
a.b.c("3333", "a b с")
a.b.c(Enum.X, "a b с")
actual:
<nothing>, because /a b c/ => /abc/
- A regular expression can shorten the string using special constructs.
before: a\.b\.c\(\"[^"]+", "a b"\)
after: a\.b\.c\(\"[^"]+", "ab"\)
^
before: .1111., .a b.
after: .1111.,.ab.
^ ^
before: .*", "a b c"
after: .*", "abc"
^ ^
before: (X|"), "a b c"
after: (X|"), "abc"
^ ^
before: 3"|"b c"
after: 3"|"bc"
^
I can remove the 'if' statement or add tests for that cases. Which option is better?
There was a problem hiding this comment.
Ah, of course! A regex may be an actual test name, but is only required to match a test name. So calling TestTestName really doesn't make sense. This means that the user has to be especially precise in formulating regular expressions and also has to be aware of how best to enter them at the command-line for each OS. But it has always been so with regex. :-)
No change needed. I'm going to merge this.
| continue; | ||
| sb.Append(ch); | ||
| inQuotes = (!inQuotes && ch == '"') || (inQuotes && ch != '"') || (inQuotes && ch == '"' && inEscape); | ||
| inEscape = inQuotes && !inEscape && ch == '\\'; |
Fixes #1680
Changes: