That happened to me once. It can be quite a project to recover from this, and there can be no guarantee of success. I don't think any of the standard file unerase/data-recovery programs are going to help because you already have a zero-byte file of the same name, so there's nothing to search for. Are you technically inclined? I had to go in with a hex editor, hunt for a large block of random data (the lost file), manually locate the file's endpoints, select the block and save it as a new file. The greatest difficulty was in finding and confirming the endpoints. In my case I was able to visually identify what appeared to be the endpoints because there was a perceptible transition between the lost file's fully random data and the non-random data that happened to surround it, but that was sheer luck, and it's not always going to happen. And even then, I had to test the (assumed) headers to ensure that I was in the right place. The headers look just like the rest of the random data, so I grabbed likely samples, saved them as files and tested them one by one until I found the exact spot. Are you up for all that?