Libre2Link-Nightscout: Extract L2 scan data from LL database, upload to NS

Prerequisites:

• Android platform tools for your OS (for adb, to be installed in $PATH or tools/)
• Android Backup Extractor https://sourceforge.net/projects/adbextractor/ (abe.jar)
• Rename abe.sh in tools/ to abe, install abe.jar next to it
• Install sqlite3, required to dump the database
• Install ruby, needed to convert date/time representations in an OS-independent way

How to do it:

• Connect your data collector smartphone

• Run adb devices to verify you've got access; confirm on Android device if necessary

• Apparently, opening the LibreLink app in the foreground may keep it from being stopped.

• Run db-from-app to backup the com.freestylelibre.app.de data, decompress and unpack the tree containing the database - intermediate files are removed on success

• VERY IMPORTANT: RESTART THE LIBRELINK APP TO AVOID DATA LOSS if it's no longer running!

• Run db-to-dump to extract sensor serials and glucose readings from the database

• Create .settings (using the template) with your Nightscout server data and credentials

• Run query-last-sensor to get the last "Sensor Change" treatment known to Nightscout

• Run query-last-sgv to get the timestamp of the last sgv reading submitted to Nightscout

• Run dump-to-sensors-json to convert the sensor list into an uploadable json file, selecting only events newer than the already known ones

• Run dump-to-sgv-json to convert both the 15-minute and manual-scan readings into an uploadable json file, selecting only events newer than the already known ones (there are no 1-minute readings if no bluetooth connection has been established!)

• Use json-upload to upload the (non-empty only) result files to Nightscout; logs are kept in logs/

• Archive: create a tarball in SAVE/ of tmp/, data/ and log/

• Clean up: remove whole tmp/ subtree, possibly do the same with data/

• This can also be done by running make all after connecting your smartphone.

Get only sensor record(s):

• Run make serial. This will backup the database, fetch the latest sensor records from Nightscout, upload the yet unknown ones, and finally clean up.

Volume of data collected:

• The database storing the glucose values, sas.db, grows by about 10MB per month, or 5MB per sensor lifetime. Each (uncompressed) tarball will have about the same size!

Tested on:

• MacOSX 10.11.6, HomeBrew, Ruby 2.3.1