Support decoding of binary QR codes

[KlaasJelmer]

When the ZBar decoder encounters binary data, it guesses the data encoding while converting it to text. This behavior destroys other types of data.

Version 0.23.1 of the ZBar library supports decoding binary QR codes using configuration option ZBAR_CFG_BINARY. This PR allows settings this flag from the decode() function

[coveralls]

Coverage decreased (-0.2%) to 97.877% when pulling 2b12b686079bee0500e438783d103a17b2a5bdf3 on KlaasJelmer:master into 833b375c0e84077943b7100cc9dc22a7bd48754b on NaturalHistoryMuseum:master.

[maglub]

Is there a good way to get attention to this pull request?

This would be quite a nice and useful workaround for issues in Europe, where the base library zbar mis-interprets the encoding:

• https://github.com/mchehab/zbar/blob/9f0cbc70aba627afe02cb0afdb8fbc39eb4f229d/zbar/qrcode/qrdectxt.c#L212
• Wrong text encoding assumption: https://github.com/mchehab/zbar/issues/212
• Related to Decoding umlauts: https://github.com/NaturalHistoryMuseum/pyzbar/issues/14

My workaround for now is the following, which could be avoided if one could just get the binary representation of the QR codes:

  try:
    logging.info("  - Trying the ugly workaround to re-encode utf-8 to big5, then back to utf-8")
    logging.info(res[0].data.decode('utf-8').encode('big5'))
    return res[0].data.decode('utf-8').encode('big5').decode('utf-8')
  except:
    logging.info("  - Re-encoding failed, so the qrcode is probably ok already in utf-8")
    logging.info(res[0].data.decode('utf-8'))
    return res[0].data.decode('utf-8')

[pauuser]

Getting the same issue. Hope this pull request gets more attention as soon as possible!