replace ToASCII with mime base64

[Alexfilus]

it fixes non ascii attach names such as cyrillic

[jhillyerd]

I fixed the static check failure in #198 -- that should go away if you rebase.

We'll need to update the golden files, please run the unit tests, inspect the output and run go test -update if it looks correct to you.

[Alexfilus]

I hope how it's right order of commits

[jhillyerd]

Looks like it picked up my PR, but no changes to the golden files:

--- FAIL: TestEncodePartQuotedHeaders (0.00s)
    encode_test.go:83: Test output did not match testdata/encode/part-quoted-headers.golden
        To update golden file, run: go test -update
    encode_test.go:83: diff -want +got:
        |-Content-Disposition: attachment; filename="arvizturo \"x\"
        |- tukorfurogep.zip"; modification-date="01 Feb 03 04:05 GMT"
        |+Content-Disposition: attachment;
        |+ filename="=?utf-8?b?w6FydsOtenTFsXLFkSAieCIgdMO8a8O2cmbDunLDs2fDqXAuemlw?=";
        |+ modification-date="01 Feb 03 04:05 GMT"
        | Content-Id: <mycontentid>
        | Content-Transfer-Encoding: base64
        | Content-Type: application/zip; boundary=enmime-abcdefg0123456789;
        |- charset=binary; name="arvizturo \"x\" tukorfurogep.zip";
        |+ charset=binary;
        |+ name="=?utf-8?b?w6FydsOtenTFsXLFkSAieCIgdMO8a8O2cmbDunLDs2fDqXAuemlw?=";
        |  param1=myparameter1; param2=myparameter2
        | 
        | WklQWklQWklQ
        | 
        
--- FAIL: TestEncodePartQuotedPrintableHeaders (0.00s)
    encode_test.go:104: Test output did not match testdata/encode/part-quoted-printable-headers.golden
        To update golden file, run: go test -update
    encode_test.go:104: diff -want +got:
        |-Content-Disposition: attachment; filename="arvizturo \"x\"
        |- tukorfurogep.zip"; modification-date="01 Feb 03 04:05 GMT"
        |+Content-Disposition: attachment;
        |+ filename="=?utf-8?b?w6FydsOtenTFsXLFkSAieCIgdMO8a8O2cmbDunLDs2fDqXAuemlw?=";
        |+ modification-date="01 Feb 03 04:05 GMT"
        | Content-Id: <mycontentid>
        | Content-Transfer-Encoding: base64
        | Content-Type: application/zip; boundary=enmime-abcdefg0123456789;
        |- charset=binary; name="arvizturo \"x\" tukorfurogep.zip";
        |+ charset=binary;
        |+ name="=?utf-8?b?w6FydsOtenTFsXLFkSAieCIgdMO8a8O2cmbDunLDs2fDqXAuemlw?=";
        |  param1=myparameter1; param2=myparameter2
        | X-Qp-Header: =?utf-8?q?Just_enough_to_need_qp_=E2=98=86?=
        | 
        | WklQWklQWklQ
        | 

[Alexfilus]

I see, those tests checks file names got from ToAscii function witch replace some symbols to similar, but it is not keep original file names.
mime.BEncoding.Encode allows to keep original file names, it is especially important in case of cyrillic file names, witch was looked like _________.pdf

Maybe I should replace golden files?

[jhillyerd]

Yes, we should update the golden files if the new output looks correct.

Something that I didn't realize when initially reviewing this PR, is that BEncode will always to base64. That makes it difficult to read an ASCII filename when looking at the raw text. I don't think we should do this.

Instead, please use selectTransferEncoding in encode.go to pick the encoder. That will allow us to pass through pure ASCII unchanged, use q-encoding for filenames with a few special characters, or b-encoding for ones with a majority.

[jhillyerd]

Do you plan to keep working on this? Otherwise I'll take a crack at it.

[Alexfilus]

I'm sorry for delay. I made changes in golden files and add switch case for file name encodings.

[jhillyerd]

Looks good, thank you!