Deal with em dashes, en dashes, etc. about ssn-redaction HOT 2 OPEN

I tried dealing with this within regular expressions, but it went badly. We're using a backreference for the second appearance of a hyphen, but that assumes that the same character is being used for each hyphen. If the first one is a hyphen and the second one is an em dash, then the SSN won't be found.

What we really need is to match a second hyphen-shaped thing if a first hyphen-shaped thing is found.

Or, of course, we can just convert all em dashes, en dashes, and minus signs to hyphens before running this regex, as proposed in the initial issue.

from ssn-redaction.

I think the proper solution is to do a conversion on the input stream. Replace −, –, —, ~, and ‐ with -. (There are other characters that look the same, but OCR software is vanishingly unlikely to use them, in my experience.)

I'm assigning this to @jazzido, since this is now out of my bailiwick.

from ssn-redaction.