This is my proposed fix to #368 that builds on the work of #376 but adds a small separation of code and a unit-test for the case that the Scribe content is empty.
I decided not to use a mock browser for the document substitution so this still isn't very testable for genuine tree traversal but it should address people's immediate problems.
This is my proposed fix to #368 that builds on the work of #376 but adds a small separation of code and a unit-test for the case that the Scribe content is empty.
I decided not to use a mock browser for the document substitution so this still isn't very testable for genuine tree traversal but it should address people's immediate problems.