I've thought alot about this today, in particular. I'll save you the various efforts and leave you with my current best-bet. On the iPhone (I realize this may not help all smartphone users), click help->diagnostics->send diagnostic information, then delete firstname.lastname@example.org and type in your own email.
When you get your sqllite file, you can open it with (you guessed it) sql lite.
I have a variety of challenges down to how to convert the datetimes, and the data since the servers went down seems to be partly encoded in a blob.
There's a file above in this thread for the zeo data decoder library, and the documentation inside has a lot of good info that I'm guessing as of now is more or less directly compatible with the above mentioned fields, I'll check tonight but my guess is that the numbers I was thinking were seconds could be counts of 30 second epochs. I'll clarify this more if and when I can correlate the various fields to the doucmentation.
I also don't mean to imply that these files will ever work with the sd or web api utilities. My goal is more to be able to retrieve and use the data myself.
I have yet to determine if it's easier to guestimate the missing fields (risetime for example), or try to parse the blob.
I will continue to look to see if someone else has already done this to avoid having to do it myself, but at least this seems to answer the question how to get the data off the smartphone.
I do find it nice that it provides the 5 minute state data.
zsleepgraph, and zbasehypnogram in particular.
zbasehynogram starts off 00000000000000000022211222221111...
I am a little bothered that I can't quite correlate that to the image I see in the app. So if anyone can volunteer what the differerent numbers mean it'll help.
Those two fields alone are probably what most people want. Those and the summary columns that deal with time in each mode of sleep.
I dabbled with select datetime(zbedtime ,'unixepoch','utc') from zzosleeprecord and it gives me the best date time so far, so last night it says 1982-06-04 05:14:32
I'm assuming the year is just not specified, so I will add an offset to bring that up to 2013, and the 5:14am adjusted for PST sounds about right, again close enough for me.
The only other thing I'm leaning towards right now is timeinrem/60 gives me 3.65, but in the app it says 3:39, so I guess that's 65 100ths, thats the same way the timeclock at work does it so I guess it's not that uncommon, so I'll have to adjust for that too
Not sure this helps anyone, but I thought I'd throw this into the ether.
There's a link above to Zeo Data Decoder Library and the documentation in that file has a section 3.2 Sleep Information which I'm hopeful will help me make more sense of the data, and I will follow up to this after a day or two (I just found this thread today).
If I get sidetracked indefinitely it looks like there's a lot of good data in there that may (or may not) be the same as what's in the sql lite data.
I also don't mean to imply that any of this will make your iphone bluetooth data compatible with the sd card utilities. My objective is more to get it into either excel or access where I can continue to ferret away my sleep data, and hopefully produce some kind of meaningful report of trends over time.