Investigate character encoding in Ambit devices

openambitproject / openambit

openambit

280 stars 82 forks source link

Investigate character encoding in Ambit devices #70

Closed svenstorp closed 9 years ago

svenstorp commented 9 years ago

The character encoding of the freetext strings in the devices needs to be investigated, today ISO-8859-1 is assumed, but is not necessarily correct. Evaluation is required for both Ambit/Ambit2 and Ambit3. Issue related to #63.

HELPING OUT:

Start PCAP dump (howto: http://openambit.org/wireshark-dissector)
Create new Sport mode with lots of nasty special characters on movescount.com
Sync device with Moveslink
Record an exercise (10 seconds duration is sufficient)
Sync new exercise with Moveslink
Stop PCAP dump
Repeat the above for all available languages (device setting on movescount.com)
Send PCAP-dumps to emil at openambit.org

svenstorp commented 9 years ago

ISO-8859-15 for all available languages (Ambit2)!

paddy-hack commented 9 years ago

Thanks for checking but I just want to make positively sure that it's really neither ISO-8859-1 nor Windows-1252. Did you check the following characters: €, Š, š, Ž, ž, Œ, œ, and Ÿ? If ISO-8859-15 is indeed used, they should correspond to 0xA4, 0xA6, 0xA8, 0xB4, 0xB8, 0xBC, 0xBD, and 0xBE, respectively.

ISO-8859-1 does not support these characters and defines different ones for those hex-values (¤, ¦, ¨, ´, ¸, ¼, ½, and ¾, respectively). Windows-1252 follows ISO-8859-1 for these hex-values but does support the characters listed but in the 0x80-0x9F range.

svenstorp commented 9 years ago

Yes I did (although the only characters movescount allowed me to use were Œ, œ, and Ÿ)

paddy-hack commented 9 years ago

So French is completely supported by MovesCount but Finnish isn't. Odd ;-) Anyway, Š, š, Ž, and ž are only used in loanwords in Finnish if one may rely on Wikipedia.

paddy-hack commented 9 years ago

I mentioned in private communication with @svenstorp that we could/should link the clock encoding for the activity name to the clock's language setting. 22a545f simply hardcodes it to ISO-8859-15 for all languages and all clock types. When other encodings are encountered in the wild I'll be more than happy to address that.

paddy-hack commented 9 years ago

When other encodings are encountered in the wild I'll be more than happy to address that.

As part of another issue :wink: