dev/core#2146 - Long unicode contact names get truncated badly causing a crash

[demeritcowboy]

Overview

https://lab.civicrm.org/dev/core/-/issues/2146

1. Go to create a contact. It's slightly easier to see with Organization.
2. Enter a pretty long name using unicode characters into the organization name field, around 128 characters.
3. Click Save.
4. DB Error: Unknown error.

Before

Error

After

No error. Names that are too long get truncated.

Technical Details

substr() isn't multibyte-aware. But also if you look a couple lines up, it already calls copyValues(), which in turn calls CRM_Utils_String::ellipsify() which is multibyte-aware. So the string is already truncated correctly before it tries to then mangle it.
We can't call ellipsify though because then that breaks the way it currently works for Individual.

Comments

Has test. Without the patch the last 3 org tests fail with something like: [nativecode=1366 ** Incorrect string value: '\xD1' for column civicrm_contact.sort_name at row 1]. Some of the individual tests fail for similar or incorrect string length truncation reasons.

[civibot]

(Standard links)

• If this is your first pull-request for CiviCRM, please browse CONTRIBUTING.md for information about the development and testing processes.
• If you are reviewing this pull-request, you may wish to consult the test sites and the Review Standards (long template, short template).

[demeritcowboy]

D'oh the call to Individual::format() puts it back again if it's Individual. Will update...

[seamuslee001]

@demeritcowboy test fails relate

[demeritcowboy]

Ok yeah I guess that's something that would need concept approval, whether names should be silently truncated (current), or have some indication it's been truncated (the change), or should it flat out fail to create the contact (other). I was keeping it consistent with copyValues but that's obviously not what's currently tested.

Maybe keeping same as current is best for now.

Relatedly, I did notice api v3 and v4 enforce Org name length differently. v3 fails, v4 truncates.

[demeritcowboy]

Have updated so it leaves it as silent truncation for individual, same as before patch, but without db failure for both individual and org for long unicode.

[seamuslee001]

This looks fine to me now merging

[demeritcowboy]

Thanks!