Allow filtering out of timeshifted itineraries with same routes and stops

[leonardehrenfried]

Summary

Adds a filter that removes itineraries from the result set if they are found to be using the same stops/stations and routes but at different times. This is useful if you don't care so much about "best" results but want to show a greater variety.

I'm a little unsure about the name of the config param: removeTimeShiftedDuplicates. Does anyone have a better idea?

Configuration example

router-config.json

{
  "routingDefaults": {
    "itineraryFilters" : {
      "removeItinerariesWithSameRoutesAndStops" : true
    }
  }
}

Issue

Closes #4274

Unit tests

Unit and module tests added.

Documentation

Added.

[leonardehrenfried]

cc @ivanwimt @FrancoisG-WIMT

[codecov-commenter]

Codecov Report

Merging #4298 (c95e835) into dev-2.x (f1b43ce) will increase coverage by 0.32%.
The diff coverage is 94.28%.

@@              Coverage Diff              @@
##             dev-2.x    #4298      +/-   ##
=============================================
+ Coverage      57.23%   57.55%   +0.32%     
- Complexity     10792    10941     +149     
=============================================
  Files           1429     1442      +13     
  Lines          58013    58285     +272     
  Branches        6729     6773      +44     
=============================================
+ Hits           33202    33548     +346     
+ Misses         22799    22714      -85     
- Partials        2012     2023      +11     

Impacted Files	Coverage Δ
...nner/standalone/config/ItineraryFiltersMapper.java	11.11% <0.00%> (-0.43%)	⬇️
...m/filterchain/ItineraryListFilterChainBuilder.java	91.50% <100.00%> (+0.69%)	⬆️
...m/filterchain/groupids/GroupByAllSameStations.java	73.33% <100.00%> (-4.45%)	⬇️
...ilterchain/groupids/GroupBySameRoutesAndStops.java	100.00% <100.00%> (ø)
...thm/mapping/RoutingRequestToFilterChainMapper.java	77.77% <100.00%> (+0.85%)	⬆️
...routing/api/request/ItineraryFilterParameters.java	100.00% <100.00%> (ø)
...entripplanner/transit/model/site/StopLocation.java	62.50% <100.00%> (+8.65%)	⬆️
...ain/java/org/opentripplanner/model/StreetNote.java	56.25% <0.00%> (-43.75%)	⬇️
...r/routing/services/notes/MatcherAndStreetNote.java	63.63% <0.00%> (-36.37%)	⬇️
...va/org/opentripplanner/ext/flex/trip/FlexTrip.java	71.42% <0.00%> (-11.91%)	⬇️
... and 129 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[leonardehrenfried]

At today's meeting we decided that we don't want to overload the term "duplicate" and will use a very wordy, but descriptive config param: removeTimeshiftedItinerariesWithSameRoutesAndStops.

[t2gran]

Do you want to keep the lowest cost or the earliest departure?

[leonardehrenfried]

I want to keep the earliest departure.

[t2gran]

I think this should be the earliest itinerary for a depart-after search, and the latest for a arrive-by search?

[leonardehrenfried]

Could you give me a hint how to achieve that?

[leonardehrenfried]

Should I be passing the arriveBy flag through to the chain and then the the ListSection to head or tail respectively?

[t2gran]

I this this would work: new GroupByFilter<>(, new SortingFilter(SortOrderComparator.comparator(sortOrder)) .... Strictly, this will sort street itineraries by them selves - but I guess we do not care in this case, so it is ok.

Can you move the new GroupByFilter<>(...) into a builder method as well, like the buildGroupByTripIdAndDistanceFilters()?

[leonardehrenfried]

Done for both.

[hannesj]

This is still sorting on generalized cost.

[leonardehrenfried]

Sorry, now it's really fixed.

[miklcct]

Is it possible to expose this parameter in the GraphQL APIs?