Enhance openambit2gpx to add heart rate and improve commandline-handling

[vascotenner]

Added more advanced argument parsing. Make output file optional. Allows for averaging heart rate data.

[vascotenner]

@myadzel: I have been thinking about averaging. My first thought was indeed that is should be not in this script. However, after some more thoughts about what information is really important for the average user I decided that averaging is very usefull. For that reason I have added a switch to switch off the averaging behaviour (--no-avg-hr). The converter does not do an 1-to-1 conversion to gpx anyway; The original log file contains more data: the sensor data is only written at the location of a GPS point.

I am not aware of any normalizers in use. Do you have a reference to such normalizers? In point of usability of the script, the inclusion of a normalizer is nice.

[myadzel]

@vascotenner

We are talking about different things. I mean that you code filtering the pulse values less than 40 and more than 220 (my initial comment for this line in the code) and that it does not need to be done in this function. Its purpose is just to convert IBI data to HR and all:

HR = 60000 / IBI

The script does not have to filter values that go beyond the threshold. This is the work of "normalizers", but in this case it is not necessary - openambit receives data, and normalization is already their processing, which should be in another program (if someone needs it).

[vascotenner]

The issue there is that in the conversion process, HR is dropped or extrapolated. In that case noise in the sensor data can be amplified. Reasonable and traceble assumptions on the data can reduce the noise in the data. But I agree, this a certain choice that might be overriden by a power-user. The question is than if we need more customable "features", that is good for a power-user, or a more simpler (easier to use and to maintain) script that is working for 90% of the users?

I hope to push and update to this PR that makes it possible to turn of this behaviour.

[centic9]

If there is still interest in these changes, it would be good to split them into separate changes so they can be applied and tested separately.

[vascotenner]

I am using this script every week, and hence it would be nice of it ends up in master. However, I do not have time to rework this part in separated changes. It is quite a rebuild of the previous scripts, because during development I encountered some additional bugs. If interested, I can resolve the conflict with master.

[centic9]

I would be interested in improving the GPX creation some more, especially now that it becomes harder to let Suunto Movescount do the GPX creation.

However there were a few bugfixes applied to the version on master as well, how can we keep these included?

[centic9]

PR #211 contains only the changes to how HR are computed.

[vascotenner]

When #271 is merged, I will work on integrating this one.

Are there currently any tests with input log and output gpx files available? I then could extend/update them to match this new additions.

[centic9]

There is a little bit of testing as part of test.sh, but it currently only performs the transformation of a single log to gpx and verifies it against a "template".

I just saw that this is actually broken by the IBI changes as we do not export HR "0" any longer (which is useful anyway), but the test is easy to adjust by updating the stored .gpx file.

[vascotenner]

We actually did HR=0, but not in the old version. I have update the testscript, and changed to IBI to HR code to only report HR when there is something. This solves the hr=0 issues

[centic9]

Thanks for the work!