Regression: XML checks fail if less than character < is inside a string value

[mstovenour]

I have existing scenario files that are modeled directly out of the documentation: https://sipp.readthedocs.io/en/latest/scenarios/variables.html?highlight=search_in#variables

<recv request="INVITE">
  <action>
    <ereg regexp="<sip:([^;@]*)" search_in="hdr" header="To:" assign_to="dummy,uri" />
  </action>
</recv>
<Reference variables="dummy" />

The less than character < inside the quotes of the regexp tag (regexp="<sip:([^;@]*)") causes the following failure when running with any release after v3.6.0: 2024-04-20 12:55:46.344095 1713635746.344095: Invalid XML in scenario

Existing scenario files work fine up until v3.5.3. They will not pass the XML check logic starting with v3.6.0.

I'm pretty sure this commit introduced the regression: https://github.com/SIPp/sipp/commit/411ccf40ab8fe77293e4a0f3ac8ad7739569d293 @wdoekes

It might be the deletion of this function call as the xp_find_start_tag_end() function is checking for embedded strings in quotes "".

[mstovenour]

This looks like a duplicate of #670

[wdoekes]

Which in turn looks like a duplicate of #414

But if the documentation wrongly advises ="<..." then that should be fixed.

[mstovenour]

I see the conversation in #414 and I'll concede that SIPp previously supported invalid XML syntax, but it did so for two decades; yes I'm old. I think this would have taken me a lot less time if SIPp were to implement the following changes:

• Change documentation so it doesn't suggest incorrect language (pull request coming)
• Change error message so it clearly calls out the regression to give future me a hint (pull request coming)
• Add line number counting while parsing XML so this syntax check can explicitly state the line number (can you meet half way and help with this one?)

[mstovenour]

I just submitted a pull request for my suggestion to yet more notes for future me.