Serialize and deserialize ORGANIZER with params

[ryanwinchester]

Closes #46

This serializes and desializes ORGANIZER similar to how ATTENDEE works.

The only problem is that this is a breaking change for deserializing since it's now a map (like attendee) and not a string.

One way we could make deserializing not a breaking change would be to pass in some "optional options" opts \\ [], but they'd have to be passed all the way through from:
ICalendar.Deserialize.from_ics/1
-> ICalendar.Util.Deserialize.build_event/1
-> ICalendar.Util.Deserialize.parse_attrs/2

What do you think we should do?

[lpil]

Why is this skipped?

[ryanwinchester]

This is to show that it is a breaking change. It no longer deserializes to a string. It deserializes to a map. This is the original test, I added a new test just below this one for the new functionality.

assert event.organizer == "mailto:[email protected]"

Will fail, now.

I left this in (but skipped so we have passing tests) so you can compare the old functionality with the new.

[lpil]

Let's update this test to pass now

[ryanwinchester]

Let's update this test to pass now

I duplicated in the test below with the new functionality.

icalendar/test/icalendar/util/deserialize_test.exs

Lines 107 to 134 in 902df35

	test "Include ORGANIZER w/ params and ATTENDEEs in event" do
	event =
	"""
	BEGIN:VEVENT
	DTSTART:20190711T130000Z
	DTEND:20190711T150000Z
	DTSTAMP:20190719T195201Z
	ORGANIZER;[email protected]:mailto:[email protected]
	UID:7E212264-C604-4071-892B-E0A28048F1BA
	ATTENDEE;CUTYPE=INDIVIDUAL;ROLE=REQ-PARTICIPANT;PARTSTAT=ACCEPTED;[email protected];X-NUM-GUESTS=0:mailto:[email protected]
	ATTENDEE;CUTYPE=INDIVIDUAL;ROLE=REQ-PARTICIPANT;PARTSTAT=ACCEPTED;[email protected];X-NUM-GUESTS=0:mailto:[email protected]
	ATTENDEE;CUTYPE=INDIVIDUAL;ROLE=REQ-PARTICIPANT;PARTSTAT=ACCEPTED;CN=James SM;X-NUM-GUESTS=0:mailto:[email protected]
	CREATED:20190709T192336Z
	DESCRIPTION:
	LAST-MODIFIED:20190711T130843Z
	LOCATION:In-person at Volta and https://zoom.us/j/12345678
	SEQUENCE:0
	STATUS:CONFIRMED
	SUMMARY:Design session
	END:VEVENT
	"""
	|> String.trim()
	|> String.split("\n")
	|> Deserialize.build_event()
	assert %Event{organizer: organizer} = event
	assert %{:original_value => "mailto:[email protected]", "CN" => "[email protected]"} = organizer
	end

So we can just delete this one.

[ryanwinchester]

@lpil removed

[jarrodmoldrich]

Hi @lpil and @ryanwinchester 👋 I see this PR has stagnated, and I'm realising it's something I need. I think the delta is pretty small for a feature change like this, so I think with a small tweak we can maintain backwards compatibility and push this through.

One way we could make deserializing not a breaking change would be to pass in some "optional options"

The small tweak I'd suggest is simply using Elixir style function pattern matching

def build("ORGANIZER", %{} = organizer) do
    params =
      for {key, val} <- organizer, key != :original_value, into: "" do
        ";#{key}=#{sanitize(val)}"
      end

    "ORGANIZER#{params}:#{organizer.original_value}\n"
end 

def build("ORGANIZER" = key, value) when is_binary(value) do
   "#{key}:#{Value.to_ics(value)}\n"  
end

So just copying the vanilla def build(key, value) fallback implementation. We could even extract out the implementation out into an internal function like defp build_raw(key, value) and call it from both. This way it won't break backwards compatibility. I also added a sanitize for the parameter values, as injected commas and semi-colons could break the file.

Does all that make sense? Or have I misjudged? I'm happy to make these changes and update the PR. At the second I'm lacking the time, but I could contribute to maintenance, if you're wanting to share some of the responsibility @lpil

In the meantime I'll make these changes on my own fork. Let me know.

[ryanwinchester]

@jarrodmoldrich if you add this test back in, with your changes, does it pass?

79b9159#diff-38283140e12894618fc3e09cccbe6d5aa4bfc35942eb9da277360ede3077e77dL78-L105

[jarrodmoldrich]

I just ran mix test on my branch from main, which includes the old test (but not your new one) and it passed. The code is here:

jarrodmoldrich@4cbeaff