[Bug]: Playwright does not recognize Pattern quotation escape sequence (\Q ... \E)

[jhyot]

Version

1.47.0

Steps to reproduce

Using Playwright Java, navigate to https://playwright.dev

Run the following assertions:

working:

assertThat(page.getByRole(AriaRole.HEADING)
  .filter(new Locator.FilterOptions().setHasText(Pattern.compile("Playwright")))
).isVisible();

not working:

assertThat(page.getByRole(AriaRole.HEADING)
  .filter(new Locator.FilterOptions().setHasText(Pattern.compile(Pattern.quote("Playwright"))))
).isVisible();

Expected behavior

Both patterns should work, as they match exactly the same strings.

Actual behavior

The second pattern fails to find the DOM element.

Additional context

Java has the Regex special escapes \Q and \E, which means all characters are quoted as is betwen those two escape sequences. This is exactly what Pattern.quote does.

In other words, these are equivalent (as you can see if you print the result of Pattern.quote):
Pattern.compile("\\Qsometext\\E")
Pattern.compile(Pattern.quote("sometext"))

This is documented here (search for "Quotation"): https://docs.oracle.com/javase/8/docs/api/java/util/regex/Pattern.html

This is speculation, but why it might fail is if the regex is just transferred to the Javascript engine and executed there. Javascript does not seem to have the \Q and \E special escapes, which leads to the meaning of the passed regex to change.

You can also check here, switch between Java and ECMAScript engines; it works for Java but not for Javascript: https://regex101.com/r/lJ1XNH/1

Environment

macOS Sequoia, arm64
Java 17
Chromium browser (others not tested)

[yury-s]

This is speculation, but why it might fail is if the regex is just transferred to the Javascript engine and executed there. Javascript does not seem to have the \Q and \E special escapes, which leads to the meaning of the passed regex to change.

Yes, this is correct, Playwright shares the implementation between all language ports and it is JavaScript regex syntax that we support.

[ilonachan]

This is speculation, but why it might fail is if the regex is just transferred to the Javascript engine and executed there. Javascript does not seem to have the \Q and \E special escapes, which leads to the meaning of the passed regex to change.

Yes, this is correct, Playwright shares the implementation between all language ports and it is JavaScript regex syntax that we support.

That fact should be PROMINENTLY displayed somewhere. It's very confusing because you integrate with the existing Pattern API of Java, and so anyone would reasonably expect that Java pattern matching is used. Ideally you'd even convert Java patterns into JS patterns, but it'd also be fine to make your own JSRegEx marker class to avoid confusion.