To determine the format of the input, anytopnm uses the file program (possibly assisted by the magic numbers file fragment included with Netpbm). If that fails (very few image formats have magic numbers), anytopnm looks at the filename extension. If that fails, anytopnm punts.
The type of the output file depends on the input image.
If file indicates that the input file is compressed (either via Unix compress, gzip, or bzip compression), anytopnm uncompresses it and proceeds as above with the uncompressed result.
If file indicates that the input file is encoded by uuencode or btoa, anytopnm decodes it and proceeds as above with the decoded result.
If file is - or not given, anytopnm takes its input from Standard Input.