Fix bug with reading Set values from Lua scripts.

[SiarheiFedartsou]

Issue

Extracted from #6279

Tasklist

• CHANGELOG.md entry (How to write a changelog entry)
• update relevant Wiki pages
• add tests (see testing documentation)
• review
• adjust for comments
• cherry pick to release branch

Requirements / Relations

Link any requirements here. Other pull requests this PR is based on?

[SiarheiFedartsou]

https://github.com/Project-OSRM/osrm-backend/pull/6279/files#r920389235

[SiarheiFedartsou]

Note that this test will start failing if we just change Set to Sequence here (without changes from this PR)

osrm-backend/profiles/foot.lua

Line 67 in 06b1b98

suffix_list = Set {

[mjjbell]

Thanks for separating this out.
I'll do some digging to fully understand what's going on with the tests.

[mjjbell]

It turns out that the suffixes listed in the profile are used for matching suffixes and prefixes

osrm-backend/include/util/guidance/name_announcements.hpp

Lines 132 to 144 in 1a3d0f7

	const auto checkForPrefixOrSuffixChange =
	[](const StringView &first, const StringView &second, const SuffixTable &suffix_table) {
	std::string first_prefix, first_suffix, second_prefix, second_suffix;
	std::tie(first_prefix, first_suffix, second_prefix, second_suffix) =
	decompose(first, second);
	const auto checkTable = [&](const std::string &str) {
	return str.empty() || suffix_table.isSuffix(str);
	};
	return checkTable(first_prefix) && checkTable(first_suffix) &&
	checkTable(second_prefix) && checkTable(second_suffix);
	};

In the test we have roads sj nj ej and wj. It filters out the common substring j and then decides the names of inbound and outbound roads to a turn are "the same" if the remaining prefix and suffix values are in the profile set. This is the case here (n, s, e, w), and wasn't before (true?).

In the case where the turn is straight ahead and the names are "the same", it will remove the announcement from the guidance, hence the change in the test route output.

I don't think it's worth changing any of this logic. This test road names are not representative of real data, so it shouldn't be replicated in OSM. I will though add a test that shows how the bugfix produces the expected turn collapse behaviour in guidance output.