Referer parser library.
To use Ref Inspector with your projects, edit your mix.exs
file and add the
required dependencies:
defp deps do
[{ :ref_inspector, "~> 0.19" }]
end
Probably the easiest way to manage startup is by simply
adding :ref_inspector
to the list of applications:
def application do
[ applications: [ :ref_inspector ]]
end
A second possible approach is to take care of supervision yourself. This
means you should add :ref_inspector
to your included applications instead:
def application do
[ included_applications: [ :ref_inspector ]]
end
And also add the appropriate RefInspector.Supervisor
to your hierarchy:
# in your application/supervisor
children = [
# ...
supervisor(RefInspector.Supervisor, [])
# ..
]
Using mix ref_inspector.download
you can store a local copy of the
regexes database in the configured path. This database is taken from the
referer-parser project.
The local path of the downloaded file will be shown to you upon command invocation.
If you want to download the database files using your application you can
directly call RefInspector.Downloader.download/0
.
When using both the mix task and a default remote configuration an informational README is placed next to the downloaded file(s). This behaviour can be deactivated by configuration:
config :ref_inspector,
skip_download_readme: true
Add the referer database you want to use to your project configuration:
use Mix.Config
# static configuration
config :ref_inspector,
database_files: [ "referers_search.yml", "referers_social.yml" ],
database_path: Path.join(Mix.Utils.mix_home, "ref_inspector") ]
# dynamic configuration
# { mod, fun } tuple without arguments
# called upon supervisor (re-) start
config :ref_inspector
init: { MyInitModule, :my_init_fun }
The remote urls of your database file are configurable:
# default configuration
config :ref_inspector,
remote_urls: [{
"referers.yml",
"https://s3-eu-west-1.amazonaws.com/snowplow-hosted-assets/third-party/referer-parser/referers-latest.yml"
}]
# custom configuration
config :ref_inspector,
remote_urls: [
"http://some.host/custom.yml",
{ "local_filename.yml", "http://some.host/remote_filename.yml" }
]
All files configured will be saved as under the configured database path. If you have not configure a custom file path the filename extracted from the url will be taken.
When reloading the old database is deleted with a configurable delay. The delay
is defined in milliseconds with a default of 30_000
.
config :ref_inspector,
ets_cleanup_delay: 30_000
The database is downloaded using
:hackney
. To pass custom configuration
values to hackney you can use the key :http_opts
in your config:
config :ref_inspector,
http_opts: [ proxy: "http://mycompanyproxy.com" ]
Please see
:hackney.request/5
for a complete list of available options.
All parsing requests are internally done using a :poolboy
worker pool. The
behaviour of this pool can be configured:
config :ref_inspector,
pool: [ max_overflow: 10, size: 5 ]
As these options are passed unmodified please look at the official poolboy documentation for details.
Defaults are defined in the module RefInspector.Pool
.
Sometimes (for example after downloading a new database set) it is required to reload the internal database. This can be done asynchronously:
RefInspector.reload()
This process is handled in the background, so for some time the old data will be used for lookups.
If you need to check if the database is still empty or (at least partially!)
loaded, you can use RefInspector.ready?/0
. Please be aware that this method
checks the current state and not what will happen after an (potentially running)
reload is finished.
iex(1)> RefInspector.parse("http://www.google.com/search?q=ref_inspector")
%RefInspector.Result{
referer: "http://www.google.com/search?q=ref_inspector",
medium: :search,
source: "google",
term: "ref_inspector"
}
Medium will be one of :unknown
, :email
, :paid
, :search
or :social
(always an atom). If configured to do so it might also be :internal
.
Source will be :unknown
(as atom) if nothing was matched, otherwise a string
with the detected provider.
Term will be :none
(as atom) if no query parameters were given to parse or the
provider does not send any terms to detect (mostly social or email referers).
Otherwise it will be an unencoded string will the term passed (can be empty).
Referer will return the passed referer unmodified.
To exclude some domains from parsing you can mark them as :internal
using
your configuration:
config :ref_inspector,
internal: [ "www.example.com", "www.example.org" ]
If a referer matches (== ends with) at least one of the configured domains
(paths ignored!), it will return a result with the medium :internal
.
Both :source
and :term
will be left at the initial/unknown state not
intended for further processing.
Referers.yml taken from the referer-parser project. See there for detailed license information about the data contained.