[uac_registrant] Match reply Contact with binding_URI by parameter.

[vasilevalex]

Added new integer module parameter 'match_contact' to choose, what
parts of URI from Contact header should match binding_URI when
OpenSIPS chooses expiration time for registration.

Summary
Solving #2557

Details
It simplifies using uac_registrant module in cluster environment, when there is one DB per cluster, and each host uses its' own IP-address for SIP.

Solution
For example, setting modparam("uac_registrant", "match_contact", 3) allows matching only by user, ignoring host. As binding_URI contains specific host, and every server sends REGISTER with its' own IP-address.

Compatibility
Default behaviour is not changed.

Closing issues
closes #2557

[stale]

Any updates here? No progress has been made in the last 15 days, marking as stale. Will close this issue if no further updates are made in the next 30 days.

[github-actions]

Any updates here? No progress has been made in the last 30 days, marking as stale.

[github-actions]

Any updates here? No progress has been made in the last 30 days, marking as stale.

[stale]

Any updates here? No progress has been made in the last 15 days, marking as stale. Will close this issue if no further updates are made in the next 30 days.

[stale]

Any updates here? No progress has been made in the last 15 days, marking as stale. Will close this issue if no further updates are made in the next 30 days.

[liviuchircu]

I like the idea behind this PR, as it brings a lot more flexibility to uac_registrant when trying to achieve interoperability with various UAS and their reply Contacts.

Still, I don't like that we are encouraging opensips.cfg code which uses these hardcore Contact matching algorithm bitmasks such as 3, 11, 19, 67, 273, 1374, etc., which increase the script reviewing difficulty, as well as forcing script developers to constantly consult the parser/parse_uri.h C program file, which I think should be avoided at all costs.

Suggestion to fix the above: add a "string CSV options" converter function in parser/parse_uri.h to reverse human-readable input such as "user|passwd|userparam|transport" into the same enum uri_match_flags as you are using now. The function will be called once at startup (during fixup phase) anyway, so it will come with zero runtime overhead, but with a big QoL improvement.

Finally, let's also add some documentation in the modules/uac_registrant/doc/uac_registrant_admin.xml file before we can merge this new feature (modparam).

[vasilevalex]

@liviuchircu can you please review, when you have time?

[jes]

I just had a look at this and noticed a couple of potential issues:

1. the "maybe we're lucky and straight-forward comparison succeeds" compares with strncasecmp which is not correct as detailed in uac_registrant: compare Contact URI case-insensitively #3263 (review)
2. it allocates fixed-size buffers on the stack for char unescaped_a[UNESCAPED_BUF_LEN], unescaped_b[UNESCAPED_BUF_LEN], but then writes into them without any bounds-checking (EDIT: ah, I see there is bounds-checking in unescape_user)

[bogdan-iancu]

@vasilevalex , could you detail/describe the scenario where the host part should be ignored (in a clustered setup) ?

[vasilevalex]

Hi, @jes @bogdan-iancu. I described why we need in some cases to compare Contact header from 200 Ok only with partial match in #2557 .
I can add here. If database with credentials is used by several OpenSIPS server with clusterer setup. All the servers use the same "binding_URI". But all the servers have different IP addresses. So we distribute registrations through cluster. In 200 Ok we will receive different Contact URI, depending of the OpenSIPS server which sent the REGISTER. It means that we must also always change binding_URI in database to have Contact matching. And in case of failover of one of the servers we must not just move shared tag to another server, but also fix all the records in DB changing IP-address in the binding_URI. That's the case, when we want to compare URI from Contact header only partially, excluding host part.
P.S. @jes I'll try to check the issues you mentioned.