Repository navigation
matchesRegex placeholder - regex with comma #306
Description
Using a matchesRegex placeholder with a regex containing a comma (,) requires configuring a different placeholderArgsSeparatorRegex so the regex is not splitted in two arguments.
For example
<a>${xmlunit.matchesRegex(.{300,400})}</a>
would results in just .{300 used as a regex (which would fail to be parsed as a regex Pattern in this case)
This might not be obvious without inspecting the library code,
It also seems to me that it's mixing internal parsing of the placeholder depending on the argument value used.
Suggestion:
- require quotes around arguments? would not be backwards compatible
- make the arguments parsing / splitting placeholder-specific? e.g. matchesRegex would take the whole string and not split it?
- document the behavior in https://github.com/xmlunit/user-guide/wiki/Placeholders / unit tests linked from there
Note I didn't test if there is a similar problem with ) inside the regex.
Related: wiremock/wiremock#3094
Please feel free to close if this is expected and the library used should use the proper constructor and configure the placeholderArgsSeparatorRegex
Activity
bodewig commentedon Jul 2, 2025
Your second suggestion would probably require a new method on the
PlaceholderHandlerinterface that signalled who was responsible for parsing the argument list. So this would be backwards incompatible as well (xmlunit-placeholders like most of XMLUnit still targets Java 1.7 so we can't use a default method on the interface).Right now documenting the behavior explicitly seems the best was forward.
DidierLoiseau commentedon Sep 2, 2025
The situation is somewhat similar with
(, because the defaultargsRegexwill look for the last opening parenthesis. It then fails to match thekeywordand thus fails to match.A simple workaround for this could be to just look for the first parenthesis, by starting the regex with
(.*?)\s*instead (note the?which makes it ungreedy). This might well cover most parenthesis issues without breaking anything.However for comas, I guess you could try to match opening and closing parentheses and [square] brackets, but it would probably be complicated and would still not cover the case of someone who would want to match just an opening or closing parenthesis.
Maybe some kind of quoting/escaping would be possible?