This will allow us to distinguish between method and constructors, and
get better kind information for locals.

Ported from mozsearch/semanticdb-kotlinc@6affaf7

This changes the symbols of getters and setters to be `x.get().` and
`x.set().` instead of `getX().` and `setX().`.

Ported from mozsearch/semanticdb-kotlinc@7b4bd70

This will allow us to get the parent symbol for locals as well.
For non-locals, it should be equal to the current symbol minus the last
segment.

Ported from mozsearch/semanticdb-kotlinc@63c9b65

Reduces the debug log clutter.

Ported from https://github.com//mozsearch/semanticdb-kotlinc/commit/4b7afc21337cba75c2c7b724043bcc9204eb9eb0

This should reduce memory usage a bit and avoids lots of “given file is
not under the sourceroot” clutter when running the tests.

Ported from https://github.com//mozsearch/semanticdb-kotlinc/commit/65b1898de26c02929f1fbd445a3c9c92e8fd3bab

scip-kotlinc changes:

getContainingSymbol was moved from o.j.k.fir.analysis.checkers to
org.jetbrains.kotlin.fir.resolve.

CompilerPluginRegistrar now has a virtual pluginId which must match the
CommandLineProcessor.

Ported from mozsearch/semanticdb-kotlinc@cde86db

- Remove unnecessary cast to FirClass in SymbolsCache (FirClassSymbol.fir already returns FirClass)
- Use parameterless toClassLikeSymbol() overload in AnalyzerCheckers
- Remove unnecessary inline modifier from test snippet

Ported from mozsearch/semanticdb-kotlinc@eb68377

…age collector

Previously handleException constructed a fresh CompilerConfiguration()
to retrieve the message collector key, which always fell back to
PrintingMessageCollector(System.err) — ignoring whatever collector the
actual compilation was configured with.

Pass the real CompilerConfiguration into PostAnalysisExtension so that
exception messages are routed through the same collector as all other
diagnostics. Also change the severity from EXCEPTION to WARNING, since
EXCEPTION is treated as isError=true by the Kotlin compiler and would
cause the build to fail — contrary to the plugin's intentional
"log-and-continue" behaviour.

The exception test now captures compiler output via result.messages and
asserts that the warning is actually emitted with the expected content,
rather than only checking the exit code.

Ported from mozsearch/semanticdb-kotlinc@e6b72f5

This is a test for 88f6272784c5e56eb135019a4b0dda5dd0808017.

Ported from mozsearch/semanticdb-kotlinc@2c1f170

…turn types

- local functions: verifies that named functions declared inside a
  function body receive local symbols (local0, local1), that explicit
  return type references are emitted (kotlin/Int#), and that call-site
  references resolve to the same local symbol. Also adds a
  SemanticdbSymbolsTest entry confirming the locals counter increments
  correctly.

- user-defined class as return type: verifies that SemanticSimple-
  FunctionChecker emits a REFERENCE occurrence for a user-defined class
  appearing in the return-type position (sample/MyClass#), exercising
  the returnTypeRef path that built-in types do not reach.

Ported from mozsearch/semanticdb-kotlinc@ff23a7e

displayName() in SemanticdbTextDocumentBuilder fell through to
`firBasedSymbol.toString()` for two symbol types, producing strings
like "FirTypeAliasSymbol sample/MyAlias" and "FirTypeParameterSymbol T"
instead of the short name.

- Broaden the FirClassSymbol branch to FirClassLikeSymbol so that
  FirTypeAliasSymbol (a subtype of FirClassLikeSymbol but not of
  FirClassSymbol) also uses classId.shortClassName.
- Add an explicit FirTypeParameterSymbol branch that returns
  symbol.name.asString().

Add tests that exercise the corrected paths:

- typealias: first test for SemanticTypeAliasChecker; verifies the
  DEFINITION occurrence and SymbolInformation (including the now-correct
  displayName "MyAlias"). Notes that type-alias references in value
  declarations are not currently tracked (the property checker resolves
  aliases to their expansion).

- type parameters: first test for SemanticTypeParameterChecker; verifies
  the DEFINITION occurrence and SymbolInformation (displayName "T").
  Notes that T in type-annotation positions does not produce REFERENCE
  occurrences because toClassLikeSymbol() returns null for type
  parameters.

Ported from mozsearch/semanticdb-kotlinc@20b1d20

…rrences

- Fix SemanticSimpleFunctionChecker and SemanticPropertyChecker to emit
  REFERENCE occurrences for extension receiver types
- Add SemanticEnumEntryChecker to emit DEFINITION occurrences for enum
  entries (previously missing from AnalyzerDeclarationCheckers)
- Fix semanticdbKind() to return Kind.ENUM_MEMBER for FirEnumEntry (FirEnumEntry
  extends FirVariable, which would otherwise map to Kind.LOCAL)
- Fix SemanticClassReferenceExpressionChecker (is/as operators) to use the
  already-extracted typeRef and source locals consistently rather than
  re-accessing expression.conversionTypeRef on each line
- Extract emitTypeRef() helper to reduce duplication across checkers; use it
  for supertype references in SemanticClassLikeChecker so they get the same
  fake-source guard as other type reference emissions
- Add tests for extension receivers, enum entries, named/unnamed companion
  objects, string template references, and is/as type references

Ported from mozsearch/semanticdb-kotlinc@d823d35

…uators

- multiple supertype references: verifies that SemanticClassLikeChecker
  emits REFERENCE occurrences for every entry in superTypeRefs, and that
  SymbolInformation.overriddenSymbols is populated for all supertypes
- three-way overload disambiguator: extends the existing two-overload
  coverage to three, verifying the (+1)/(+2) suffix counting logic in
  methodDisambiguator()

Ported from mozsearch/semanticdb-kotlinc@24b6433

The "non-0-based" phrasing predates this branch but is misleading: line
helpers return 1-based values (callers subtract 1 for protobuf), while
column helpers already return 0-based values. Make the docstrings say
what they actually return.

Ported from mozsearch/semanticdb-kotlinc@3059517

Add two test cases:

- enum entry with body (Op.PLUS { override ... }): the body is modeled
  as a synthetic anonymous subclass, so the entry gets a multi-line
  enclosing_range and the overridden member surfaces as a local symbol
  rather than a sample/Op#PLUS.apply() global. Asserts both.
- multi-line generic class declaration: general coverage we lacked for
  generic classes. Asserts enclosing_range for the class, its type
  parameter, and a member.

Ported from mozsearch/semanticdb-kotlinc@979f1f3

Changes in scip-kotlinc:
- Rename FirSimpleFunction -> FirNamedFunction in SemanticSimpleFunctionChecker
- Replace removed isLocalMember with isLocalDeclaredInBlock
  (moved within org.jetbrains.kotlin.fir.analysis.checkers.declaration)

Ported from mozsearch/semanticdb-kotlinc@832bf44

Changes to scip-kotlinc:
- -Xcontext-parameters argument isn't required anymore
- FirAllModifierRenderer now takes an argument

Ported from mozsearch/semanticdb-kotlinc@89427a5