Skip to content

code-editor.md

Latest commit

 

History

History
66 lines (49 loc) · 2.98 KB

code-editor.md

File metadata and controls

66 lines (49 loc) · 2.98 KB

Code Editor

Overview

Query Pilot uses CodeMirror 6 for SQL editing with dialect-specific features.

Architecture

src/components/CodeEditor/
├── core/                    # Core editor logic
├── languages/
│   ├── sql/                 # SQL language support
│   │   ├── context.ts       # SQL context analyzer (the "Brain")
│   │   ├── completion.ts    # Intelligent autocomplete
│   │   ├── linter-strategy.ts       # Two-stage dialect linter entrypoint
│   │   ├── unified-linter.ts        # Fast pass + deferred semantic pass
│   │   ├── dialect-lint-adapter.ts  # Canonical SQL + offset remapping
│   │   ├── metadataProvider.ts      # Schema metadata
│   │   └── dialect-validators/      # Dialect-specific validators
├── services/
│   └── linter-coordinator.ts # Deduped semantic IPC batching + cache
└── languages/dbml/           # DBML language for ERD

Multi-Dialect Linting

All SQL dialects now use the same two-stage CodeMirror pipeline:

  1. Fast pass: local token-aware checks in the editor for cheap feedback (SELECT *, invalid *, keyword typos, destructive updates/deletes without WHERE).
  2. Semantic pass: deferred Rust sql_validate call with schema-aware validation and nested-query traversal.

For PostgreSQL, the dialect adapter canonicalizes parser-hostile but semantics-preserving syntax before the Rust pass. Current examples include EXPLAIN ANALYSE ... normalization to EXPLAIN ANALYZE ....

Semantic diagnostics are computed against canonical SQL and then remapped back to raw editor offsets before display.

Smart SQL Features

  • Context-aware autocomplete: Table/column suggestions from active connection
  • Real-time semantic linting: Validates against actual schema
  • Fast local diagnostics: Cheap lint rules stay responsive while typing
  • Hover tooltips: Table/column information on hover
  • Symbol table tracking: CTE, EXPLAIN, and nested subquery reference resolution

Key Components

File Purpose
context.ts SQL context analyzer - determines cursor position context
completion.ts Autocomplete provider with schema awareness
linter-strategy.ts Public dialect linter entrypoint returning fast + semantic extensions
unified-linter.ts CodeMirror lint sources for fast pass and semantic pass
dialect-lint-adapter.ts Canonical SQL preparation, offset mapping, and fast diagnostics
linter-coordinator.ts Semantic-only IPC batching, caching, and schema sync
metadataProvider.ts Bridges schema metadata to editor features

DBML Support

Separate language support in languages/dbml/ for Entity Relationship Diagram editing.

Adding Dialect Support

  1. Create validator in dialect-validators/
  2. Implement linter strategy in linter-strategy.ts
  3. Register in the dialect detection logic
  4. Add any dialect-specific canonicalization to dialect-lint-adapter.ts