Expand description
Context-aware SQL completion engine.
Public entry point is complete: given the source text, parsed
statements, resolved scopes, catalog snapshot, and cursor offset, it
returns a flat list of Items. The LSP server crate maps each
Item onto an LSP CompletionItem; using a plain struct here keeps
this crate free of LSP wire dependencies and unit-testable in
milliseconds.
Re-exports§
pub use engine::complete;pub use engine::complete_with_derived;pub use item::Item;pub use item::ItemKind;
Modules§
- context
- Detect what kinds of completion items make sense at the cursor.
- contexts
- Context-aware special completions.
- create_
index - CREATE INDEX sub-phase detector.
- create_
table - CREATE TABLE sub-phase analyzer.
- detectors
- Phase/context detector functions – one function per SQL
sub-construct (
CREATE X ... <cursor>,ALTER X ... <cursor>, slot predicates, item-pushing helpers). Extracted fromengine.rsin one large batch (was inline there) – pure code motion, no behavior change. Deliberately not yet subdivided by statement family (CREATE vs ALTER vs …); that’s a lower-risk follow-up now that this code has no cross-file coupling left to worry about, only internal reorganization within one file. - engine
- Completion engine.
- fallback
- Text-level fallback scope builder.
- item
- Crate-local representation of one completion entry.
- json_
path - JSON key completion for
col->'<cursor>'/col->>'<cursor>'chains. - merge
- MERGE statement completion.
- phase
- Phase-aware completion context.
- plpgsql_
locals - Extract PL/pgSQL function parameters and DECLARE’d locals from the buffer text so completion can surface them inside a function body.
- render
- Markdown renderers for table and column completion items.
- source_
tables - Build a synthetic catalog from CREATE TABLE statements in the current file (and a few text-scanned object families) so completion / hover / validation work even offline – before the DB knows about anything, or when there is no DB connection configured at all.
- sources
- Item producers, one per kind of source data.