How can I use square brackets in a mapping dataflow path?

Question

I'm trying read a file from a directory that contains square brackets in the path using a mapping dataflow in Azure Synapse, like this:

/path/to/[a].[b]/some/file/*.csv

If I enter the path exactly as shown above, the dataflow will "import" the schema from the file, but no data will be displayed. If I rename the directory and remove the square brackets, the file(s) can be read in as expected.

/path/to/a.b/some/file/*.csv

I've tried using a backtick with the brackets and escaping them with a single or double backslash \, but nothing has worked. I know Spark can read in a directory like this if the path looks like /path/to/\[a\].\[b\]/some/file/*.csv, but if I enter a string that looks like that, Synapse complains about a token recognition error. If I enter double backslashes, like /path/to/\\[a\\].\\[b\\]/some/file/*.csv, the I get the same behavior as not including them at all.

I found this and it helps explain the issue, but there isn't a solution.