Fixes DISCO-2637: Replace SuggestionQuery options with providers. [firefox-android: linabutler/DISCO-2637] [firefox-ios: linabutler/DISCO-2637]

[linabutler]

This commit removes the include_sponsored and include_non_sponsored options, and replaces them with a list of providers. This lets consumers request specific suggestion types: AMP and Wikipedia only on mobile; and AMP, Wikipedia, and active experiments on Desktop.

This is a backward-incompatible API change.

Pull Request checklist

• Breaking changes: This PR follows our breaking change policy
  • This PR follows the breaking change policy:
    • This PR has no breaking API changes, or
    • There are corresponding PRs for our consumer applications that resolve the breaking changes and have been approved
• Quality: This PR builds and tests run cleanly
  • Note:
    • For changes that need extra cross-platform testing, consider adding [ci full] to the PR title.
    • If this pull request includes a breaking change, consider cutting a new release after merging.
• Tests: This PR includes thorough tests or an explanation of why it does not
• Changelog: This PR includes a changelog entry in CHANGELOG.md or an explanation of why it does not need one
  • Any breaking changes to Swift or Kotlin binding APIs are noted explicitly
• Dependencies: This PR follows our dependency management guidelines
  • Any new dependencies are accompanied by a summary of the due dilligence applied in selecting them.

Branch builds: add [firefox-android: branch-name] to the PR title.

[codecov-commenter]

Codecov Report

Attention: 32 lines in your changes are missing coverage. Please review.

Comparison is base (fe621c9) 36.72% compared to head (c22e000) 36.71%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #5867      +/-   ##
==========================================
- Coverage   36.72%   36.71%   -0.01%     
==========================================
  Files         348      348              
  Lines       33532    33541       +9     
==========================================
  Hits        12313    12313              
- Misses      21219    21228       +9     

Files	Coverage Δ
components/suggest/src/suggestion.rs	0.00% <ø> (ø)
components/suggest/src/lib.rs	0.00% <0.00%> (ø)
components/suggest/src/store.rs	0.00% <0.00%> (ø)
components/suggest/src/db.rs	0.00% <0.00%> (ø)

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

[linabutler]

The SuggestionProvider variants are all integers, so, as with providers_to_sql_list above, we can safely interpolate them directly into the query string, without worrying about SQL injection. This is a trick we use in Places, too! ✨

[linabutler]

Pocket keywords use a separate query, so, if we aren't matching Pocket suggestions, we can skip querying pocket_keywords entirely.

That also means we end up with a dynamically-sized list of params—two for Pocket; one for non-Pocket—so we have to use a Vec<(&str, &dyn ToSql)> to hold them.

[linabutler]

This transformation filters out the Nones from our query_and_then(...) iterator, then accumulates the remaining suggestions into a vector. It leans on a few iterator helpers to do this:

• transpose turns a Result<Option<Suggestion>> into an Option<Result<Suggestion>>.
• An Option is an iterator—it yields zero elements if it's None, and one if it's Some—so we can use it with flat_map to unwrap optionals. This turns our iterator of Option<Result<Suggestion>> into an iterator of Result<Suggestion>.
• collect::<Result<_>> accumulates the iterator of Result<Suggestion> into a Result<Vec<Suggestion>>.

[ncloudioj]

LGTM. Thanks!

I wonder if we could use a "cheaper" construct for SuggestionQuery.providers. Creating a new vector (despite the providers mostly remaining the same) for each query seems a bit wasteful to me. Would a bitflag fit the bill? I am thinking of this:

use bitflags::bitflags;

bitflags! {
    pub struct Providers: u32 {
        const Amp = 1;
        const Wikipedia = 1 << 1;
        const Pocket = 1 << 2;
        const Amo = 1 << 3;
    }
}

pub struct SuggestionQuery {
    pub keyword: String,
    pub providers: Providers, 
}

// For mobile
let query = SuggestionQuery {
    keyword: "foo".to_own(),
    providers = Providers::Amp | Providers::Wikipedia,
};

if query.providers.contains(Providers::Pocket) {
} else ...

WDYT?

[linabutler]

Thanks for the review!

I thought bitflags—or even a set—would be perfect here, too! Unfortunately, they're tricky to do with UniFFI. Neither UDL nor UniFFI's serialization format supports custom values for fieldless enums, so I don't think it would be possible today for Swift, Kotlin, or JS to pass a bitmask, and have Rust see the same value.

We could use a custom type to work around that, but then we couldn't use the enum syntax in UDL, because we can't declare a custom value for each variant.

[ncloudioj]

Bummer. Yep, the FFI limitations! If we use an opaque u32, then we'd sacrifice the type safety 🤷‍♂️. let's leave it as is then.

[linabutler]

PRs resolving the breaking changes for our consumers:

• Android: No bug - Update uses of SuggestionQuery in Firefox Suggest. mozilla-mobile/firefox-android#4073
• iOS: Bump FXIOS-7566 DISCO-2637 [v120] Update uses of SuggestionQuery. mozilla-mobile/firefox-ios#16818

The process for landing breaking changes like this is roughly:

• Land the Application Services PR first.
• Wait for the automated a-s version bumps in Firefox for Android and iOS to fail. (I'll drop links to those failing PRs here when that happens, so that folks can follow the process along).
• Fold in the Android and iOS PRs that fix the breakages into the version bump PRs, push, and land that as a squashed commit. This keeps the version bump, and the fix for the breakages, in a single commit.