Prune islands with mode-less stop vertices

[bas-hbt]

Summary

For island pruning islands with stops without modes are not considered as only ferries anymore. Currently we have issues with stops that exist in our data that have no service but are linked to small islands.

In the following screenshot you can see the problem

[codecov]

Codecov Report

Attention: Patch coverage is 66.66667% with 4 lines in your changes are missing coverage. Please review.

Project coverage is 67.87%. Comparing base (c1a9196) to head (8230308).
Report is 30 commits behind head on dev-2.x.

Files	Patch %	Lines
...aph_builder/module/islandpruning/PruneIslands.java	0.00%	2 Missing and 1 partial ⚠️
...r/graph_builder/module/islandpruning/Subgraph.java	88.88%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@              Coverage Diff              @@
##             dev-2.x    #5782      +/-   ##
=============================================
+ Coverage      67.81%   67.87%   +0.06%     
- Complexity     16532    16552      +20     
=============================================
  Files           1906     1908       +2     
  Lines          72275    72337      +62     
  Branches        7443     7441       -2     
=============================================
+ Hits           49015    49102      +87     
+ Misses         20740    20714      -26     
- Partials        2520     2521       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[leonardehrenfried]

Please move this code into Subgraph and write documentation about all the cases, in particular the case of the empty modes.

[leonardehrenfried]

That would also make writing a test for it easier, since you only need to test Subgraph not the entire module..

[leonardehrenfried]

Also please coordinate with @vesameskanen if the handling of islands with stops with an empty mode need to have a configuration switch. We believe the check for the empty modes was introduced because there are actual islands in Finland that don't have any service in the winter and those should not be pruned.

Another option would be to fix the data in OSM.

[vesameskanen]

The reason for considering a stop with no modes as a potential ferry stop was indeed winter time traffic. I think this is quite a specific assumption, and therefore can be changed as proposed in this PR. Removing small islands which have no wintertime connection hardly causes any issues in our routing.

[leonardehrenfried]

Your latest change is a great improvement.

[leonardehrenfried]

Suggested change

	for (Iterator<TransitStopVertex> vIter = this.stopIterator(); vIter.hasNext();) {
	TransitStopVertex v = vIter.next();
	for (var v : stopsVertexSet) {

[leonardehrenfried]

Suggested change

	private static RegularStop regularStop1;
	private static RegularStop regularStop2;
	@BeforeAll
	static void setUp() {
	regularStop1 =
	RegularStop
	.of(new FeedScopedId("HH-GTFS", "TEST1"), () -> 0)
	.withCoordinate(53.54948, 9.98455)
	.build();
	regularStop2 =
	RegularStop
	.of(new FeedScopedId("HH-GTFS", "TEST2"), () -> 0)
	.withCoordinate(53.55272, 9.99480)
	.build();
	}
	private static final TEST_MODEL = TransitDataForTest.of();
	private static final RegularStop regularStop1 = TEST_MODEL.stop("TEST-1").build();
	private static final RegularStop regularStop2 = TEST_MODEL.stop("TEST-2").build();

[leonardehrenfried]

Suggested change

	for (Iterator<TransitStopVertex> vIter = stopIterator(); vIter.hasNext();) {
	Vertex vx = vIter.next();
	envelope.expandToInclude(vx.getCoordinate());
	}
	for (var vx : stopsVertexSet) {
	Vertex vx = vIter.next();
	envelope.expandToInclude(vx.getCoordinate());
	}

[leonardehrenfried]

Great and extensive test! ❤️

[leonardehrenfried]

Suggested change

	* @return true if only ferries stop at the subgraph and false if other or no other modes are
	* @return true if only ferries stop at the subgraph and false if other or no modes are

[leonardehrenfried]

Use a pattern variable here, please.