I wrote a script (attached to this bug report) with the intent to automate replacing some fancy Unicode characters with their simple ASCII equivalents in txt and html documents. The script works fine on small files, (no problems encountered up to 100kb), but some larger books are simply blanked or replaced with 1-2 characters. (Seen quotation marks ("), apostrophes ('), and dashes (--) so far.) And occasionally I will find a chunk of it missing.
The issue is reproducible with those books, if I undo and try again.
But if I fill a file with megabytes of randomness the replace function still works on that.
However, if I add a couple of the characters I am looking to replace to that random document, it suddenly gets a chunk bitten out of it by the plugin.