Struct Driver

Source

pub struct Driver { /* private fields */ }

Expand description

Pipeline orchestration wrapper around ProjectDb.

Implementations§

Source §

impl Driver

Source

pub fn new() -> Self

Create a new driver with an empty database.

Source

pub fn from_db(db: ProjectDb) -> Self

Create a driver from an existing database.

Source

pub fn set_analysis_options(&mut self, options: AnalysisOptions)

Set the analysis options (e.g. a registered host manifest + external check severity) used by analyze. An input write — dependent queries recompute on next read.

Source

pub fn db(&self) -> &ProjectDb

Borrow the underlying database.

Source

pub fn db_mut(&mut self) -> &mut ProjectDb

Mutably borrow the underlying database.

Salsa’s dependency tracking invalidates derived queries on input writes, so no manual cache invalidation happens here.

Source

pub fn into_db(self) -> ProjectDb

Consume the driver and return the underlying database.

Source

pub fn discover<F>( &mut self, entry: &str, read_file: F, ) -> Result<(), DiscoverError>
where F: FnMut(&str) -> Result<String, Error>,

Discover all files reachable via INCLUDEs from the entry point.

Source

pub fn discover_native( &mut self, tree: &dyn SourceTree, ) -> Result<(), DiscoverError>

Discover a native .brink project: enumerate tree (sorted, root-relative keys, scoped to tree’s own constructor-held root — issue #1371) and load every file — no INCLUDE BFS, since native has no INCLUDEs. tree must be constructed with the project’s source root (RealFs::new/GitRev::new) — see native_source_root to derive it from an entry path.

Source

pub fn analyze(&mut self) -> &AnalysisResult

Run cross-file analysis on all files (memoized by the db’s analysis query — an unchanged project returns the cached result).

Source

pub fn analyze_project(&self, file_ids: &[FileId]) -> AnalysisResult

Run analysis on a specific subset of files (one project). Not cached.

Module-aware and options-honoring (issue #1553): the pass runs with the db’s own ProjectDb::module_map and registered AnalysisOptions, so the DefinitionIds it mints key this db’s per-def queries and the declared dialect/types/lints reach it — the same contract analyze has. A bare brink_analyzer::analyze/analyze_with_options here was module-blind and dropped the options entirely, which for a native .brink project (whose module is its path, always declared) mints a different identity space than the db’s — see ProjectDb::module_map’s doc.

Stem-collision diagnostics (E085) are folded in from ProjectDb::module_map_diagnostics, scoped to file_ids, for the same reason: the analyzer is handed the finished map and cannot re-derive them.

Source

pub fn analysis_inputs_for( &self, file_ids: &[FileId], ) -> Vec<(FileId, HirFile, SymbolManifest)>

Snapshot analysis inputs for a subset of files.

Source

pub fn analysis_inputs(&self) -> Vec<(FileId, HirFile, SymbolManifest)>

Snapshot all analysis inputs.

Source

pub fn compute_projects(&self) -> Vec<(FileId, Vec<FileId>)>

Compute independent projects: ink files by INCLUDE reachability, native .brink files as one project (issue #1562). See brink_db::ProjectDb::compute_projects.

Source