Fix TPoint generation for real-time generated patterns

[hannesj]

Summary

Currently the generation crashes if the TripTransferPoint#getStopPositionInPattern() is greater than RoutingTripPattern#numberOfStopsInPattern, or it would create a transfer between the wrong stop, if the pattern changed. This changes the handling so that we check the index and compare the stop at that point, and generate a constrained transfer only if it is the same stop, or its sibling, at the specified stop point in pattern.

[codecov]

Codecov Report

Base: 61.87% // Head: 61.97% // Increases project coverage by +0.09% 🎉

Coverage data is based on head (d9542e3) compared to base (ea3a29b).
Patch coverage: 56.80% of modified lines in pull request are covered.

Additional details and impacted files

@@              Coverage Diff              @@
##             dev-2.x    #4787      +/-   ##
=============================================
+ Coverage      61.87%   61.97%   +0.09%     
- Complexity     12831    12915      +84     
=============================================
  Files           1619     1644      +25     
  Lines          64958    65527     +569     
  Branches        7078     7131      +53     
=============================================
+ Hits           40196    40612     +416     
- Misses         22481    22603     +122     
- Partials        2281     2312      +31     

Impacted Files	Coverage Δ
...ipplanner/ext/datastore/gs/GsDataSourceModule.java	0.00% <ø> (ø)
...aphqlapi/datafetchers/LegacyGraphQLAgencyImpl.java	26.66% <0.00%> (ø)
...graphqlapi/datafetchers/LegacyGraphQLFeedImpl.java	11.11% <0.00%> (ø)
...phqlapi/datafetchers/LegacyGraphQLPatternImpl.java	13.51% <0.00%> (+0.23%)	⬆️
...qlapi/datafetchers/LegacyGraphQLQueryTypeImpl.java	7.78% <0.00%> (+0.05%)	⬆️
...raphqlapi/datafetchers/LegacyGraphQLRouteImpl.java	19.00% <0.00%> (ø)
...graphqlapi/datafetchers/LegacyGraphQLStopImpl.java	11.11% <0.00%> (ø)
...graphqlapi/datafetchers/LegacyGraphQLTripImpl.java	13.40% <0.00%> (ø)
...a/org/opentripplanner/ext/siri/EntityResolver.java	0.00% <0.00%> (ø)
...opentripplanner/ext/siri/SiriFuzzyTripMatcher.java	12.88% <0.00%> (ø)
... and 231 more

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

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[leonardehrenfried]

I checked this out now and it doesn't fix the issue I'm having when using the same transfers from two feeds with the same feed id.

You can find the reproduction case here: https://leonard.io/ibi/hopelink-duplicate-feed-ids.zip (it's all input sources including config in a single zip).

However, I think I made it clear that I'm only too happy to tell IBI that duplicate feed entities from different feeds aren't supported.

Could you look at the example and form an opinion if you still want to fix that case? (Not asking you to do it just if we should support it.)

[jtorin]

Unfortunately this PR doesn't fix our bug at Skånetrafiken either. The AIOOB exception moves to the processing of the non-realtime RoutingTripPattern, to around the code: scheduledPattern.stopIndex(stopPosInPattern)

Nevertheless: I like the direction the PR takes the code, unless the processing of the entries in two separate batches is adding complexity unnecessarily.

[hannesj]

Unfortunately this PR doesn't fix our bug at Skånetrafiken either. The AIOOB exception moves to the processing of the non-realtime RoutingTripPattern, to around the code: scheduledPattern.stopIndex(stopPosInPattern)

If that is producing the issue, then there is something wrong with the org.opentripplanner.model.transfer.TripTransferPoint#stopPositionInPattern. Does the trip come from a GTFS or NeTEx data source?

[jtorin]

If that is producing the issue, then there is something wrong with the org.opentripplanner.model.transfer.TripTransferPoint#stopPositionInPattern. Does the trip come from a GTFS or NeTEx data source?

The trips I see failing is from NeTEx.

[hannesj]

Can you add what the fields of the TripTransferPoint are, and how the trip pattern looks like? And the source interchange from NeTEx

[jtorin]

The trips I see failing is from NeTEx.

Just wanted to add that the problems we're seeing seems to be unrelated to this PR.