Scala Common Enrich: add ua-parser enrichment

[alexanderdean]

The current useragent library takes an algorithm-based approach - there are other libraries out there which are based on a regularly updated 'database' (YAML) of useragent strings. We should move to one of these.

[yalisassoon]

I think this is the library we should use: https://github.com/tobie/ua-parser
Master YAML is here: https://github.com/tobie/ua-parser/blob/master/regexes.yaml

[rcs]

+1. Just wanted to pull a quick "Mobile vs non-mobile" device thing, and this was pretty sad with the exclusion of iOS.

What can be done to help move this forward?

[alexanderdean]

To move this forward, someone needs to:

1. Have a play with ua-parser
2. Grok which fields ua-parser will generate
3. Figure out how they map onto our existing fields - any renames, deletes, additions
4. Add ua-parser as a dependency
5. Remove the old library in hadoop-etl/lib
6. Replace the old library code with ua-parser code
7. Update the Redshift table definition
8. Provide a migration script from old Redshift table definition to new one
9. Update the Postgres table definition
10. Provide a migration script from old Postgres table definition to new one
11. Update the Analytics Cookbook recipes as appropriate

[kingo55]

For the interim, should we just move to the latest version of Useragent Utils (this is the one we're using right?)? They're up to 1.11 now and look to have fixed a number of things.

https://github.com/HaraldWalker/user-agent-utils

[alexanderdean]

Hey @kingo55 - agreed, we'll do that in the next release: #416

[alexanderdean]

@pkallos had a play around with this - see #576 for details. Unfortunately ua-parser takes away things (e.g. browserType), even as it gives us enhanced accuracy.

So rather than a direct replacement, maybe we leave till we have pluggable enrichments, and then we offer both different UA parsing options

[alexanderdean]

Ticket depends on: snowplow/iglu-central#96, snowplow/iglu-central#94