Failed to parse XHTML content from Jira Macro Repair

Issue Summary

When updating pages to fix broken macros (such as in Jira Macro Repair), user generated content needs to be parsed, updated, and then saved. Due to errors in the format of the XHTML, it's possible some pages cannot be parsed or saved, causing the content fixing to fail. 

This is reproducible on Data Center: with suitable samples of XHTML

Steps to Reproduce

N/A

Expected Results

Pages should be able to be parsed and saved. 

Actual Results

Some pages fail when being parsed or being updated. 

 
Troubleshooting

When parsing fails, this log entry is logged from Jira Macro Repair: 

Failed to parse content on page

or

Skipping page update due to parsing error.

Use these steps to find these failed update entries

• First run this Splunk query with the cloud ID of the tenant:

"<cloudId>" `connie-prod` "ValidatorMessageRunner"

• Use the trace ID from one of the logs (ext.traceId) with the following Splunk query:

 "<traceId>" `connie-prod` level=ERROR | stats count by message

• There can be several traceId values to work through

The error stack trace exception will show

com.atlassian.confluence.content.render.xhtml.XhtmlException: RuntimeException occurred while performing an XHTML storage transformation (Cannot invoke "javax.xml.stream.events.Attribute.getValue()" because "macroNameAttribute" is null)
...
Caused by: java.lang.NullPointerException: Cannot invoke "javax.xml.stream.events.Attribute.getValue()" because "macroNameAttribute" is null

Workaround

Manually fix those pages using the steps detailed in this KB

https://hello.atlassian.net/wiki/spaces/MCENTRAL/pages/1039987637/Fix+Jira+macros+on+Confluence+pages+after+server-to-cloud+migration