Replace non-printable characters in display names

[sbrannen]

Background

While working on #4698, I created tests using blank strings like the following.

@Retention(RetentionPolicy.RUNTIME)
@ParameterizedTest
@NullSource
@ValueSource(strings = { "", " ", "   ", "\t", "\r", "\r\n", "\n" })
@interface EmptyReasonsTest {
}

That produces display names similar to the following, which are not very useful since the last 7 are identical.

If I wrap the values in single quotes as follows...

@ParameterizedTest(name = "[{index}] reason=''{0}''")

... the results are better.

Proposal

But... I think we can do even better.

Namely if we replace non-printable characters, then configuration like this...

@ParameterizedTest(name = "[{index}] reason=''{0}''")
@NullSource
@ValueSource(strings = { "", " ", "   ", "\t", "\r", "\r\n", "\n", "\u200B" })

... could have display names like this:

Proof of Concept

Invoking the following from ParameterizedInvocationNameFormatter.MessageFormatPartialFormatter's makeReadable() method resulted in the last screenshot in the Proposal section.

private String replaceNonPrintableCharacters(String string) {
	return string
			.replace("\n", "\\n")
			.replace("\r", "\\r")
			.replace("\t", "\\t")
			.replaceAll("[\\p{Cc}\\p{Cf}\\p{Co}\\p{Cn}]", "?");
}

Related Issues

• Sanitize display names before they are output via the ConsoleLauncher #1713
• Quote text-based arguments in display names for parameterized tests #4716
• CSV header names do not work with {argumentsWithNames} display name pattern #4783
• Consistently use name = value formatting for arguments with names in parameterized tests #4786

[sbrannen]

Side Note

Prior art: StringUtils.replaceIsoControlCharacters(...)

junit-framework/junit-platform-commons/src/main/java/org/junit/platform/commons/util/StringUtils.java

Lines 244 to 247 in 6b7f17d

	public static @Nullable String replaceIsoControlCharacters(@Nullable String str, String replacement) {
	Preconditions.notNull(replacement, "replacement must not be null");
	return str == null ? null : ISO_CONTROL_PATTERN.matcher(str).replaceAll(replacement);
	}

[sbrannen]

As an alternative to displaying CR and LF as \r and \n, we could consider using <CR> / <LF> or ␍ / ␤.

The two Unicode symbols are &#x240D and &#x2424, and they are rather small.

Though... perhaps small "symbols" like that would actually work well in display names.

---

Options

• CR
  • \r
  • <CR>
  • ␍
• LF:
  • \n
  • <LF>
  • ␤

---

Thoughts?

[marcphilipp]

I think using \r and \n is most clear. I'm wondering, though, if anyone would pass "\\n" and be confused. However, that applies to all options.