Colon identified as a spam character

Could you please try enabling Special Characters language script and see if the problem still persists?

Is Special Characters language script available in 2.0.2? (That is the version I'm using, and I can't see an option for Special Characters. Do I need to be testing this with 2.0.x-dev?)

On 2.0.2 I have the following language scripts enabled, and it doesn't allow a colon as in "5:30pm":
- Currency Symbols
- Latin
- Miscellaneous Symbols

Miscellaneous Symbols it is. Please enable that one and give another try as semicolon and similar symbols are not part of Latin language scripts.

I can confirm that even with Miscellaneous Symbols enabled, a colon is still being flagged as spam, using the default reject patterns which include a series of spaces with a colon.

When I use Latin as the only language script (WITHOUT Miscellaneous Symbols), and remove " :" from the reject patterns or put quotation marks around it, colons are not flagged as spam. To me this suggests that colons are part of Latin language script, unrelated to Miscellaneous Symbols, and the default reject pattern is the issue.

If " :" is intended to be a pre-installed reject pattern, could it be enclosed in quotes to avoid over-blocking?

First a (unit?) test should be added to ensure only expected words are blocked and no others. colon should be one example.

Then we can see the status and fix it :)

@Grevil will check this in the context of ✨ Needs schema check and tests for all functionality Needs work as soon as we have the time.

@lrwebks: Could you check this please and add a test?

I have definitely run into this issue before, where not enclosing strings in YAML was leading to unexpected behavior, so I will certainly take a look. I am also fairly confident that the colon is part of the Latin language character set, since it is a part of ASCII, which is almost entirely contained within the Latin set. I'll check it out!

The culprit of the problem is the following, as far as I can see:
In line 280 of protected_forms.module there is a helper function to turn the comma separated string from the settings page into an array. Sensibly enough, it also strips away the whitespace from all array items, since some users put a whitespace after each comma and others don't:

// Trim white spaces of array values in php.
$array = array_map('trim', $array);

This of course also deletes the placed whitespace in front of the colon, that is then flagged as spam.

A simple but not very gentle solution would of course be to not trim white space and force the user to not put spaces after their comma…
What do you think about this regarding a solution?

textarea separated by newlines makes a lot of sense to me and then remove the trimming and keep expected whitespaces.
This will need an update hook to replace the comma by "\n" and trim the values.

Please ensure that the default values are preceded by a space by default: :, :,. Anyway I'd even vote to remove them, because I think they are super risky in general?

@AltaGrade: What was the idea behind these signs?

I guess @lrwebks proposed solution will also solve ✨ Possibility to set words as reject pattern Needs review ?

Everything should be working now. I have also tested the update hook manually and the conversion from the comma separated list to the line separated list works absolutely fine.