Implementation:TobikoData Sqlmesh SQLMeshDialect

Knowledge Sources	TobikoData_Sqlmesh
Domains	Web_UI, Code_Editor, SQL_Syntax, Autocomplete
Last Updated	2026-02-07 20:00 GMT

Overview

A CodeMirror language extension that provides SQL syntax highlighting and intelligent autocomplete for SQLMesh-specific keywords and model references.

Description

SQLMeshDialect is a sophisticated CodeMirror extension that extends standard SQL language support with SQLMesh-specific syntax and intelligent autocomplete functionality. It defines a custom SQL dialect that includes SQLMesh keywords (MODEL, kind, dialect, etc.), SQLMesh types (FULL, INCREMENTAL_BY_TIME_RANGE, etc.), and provides context-aware suggestions for model names, column names, and configuration keywords. The extension integrates with a SQLGlot worker for advanced parsing capabilities.

The autocomplete system is context-aware, providing different suggestions based on cursor position and surrounding code. When typing after a dot (.), it suggests column names for known models or model names for partial matches. Within MODEL definitions, it provides SQLMesh-specific keywords and configuration options. The extension also handles model and column name matching with support for both quoted and unquoted identifiers, making it flexible for different SQL naming conventions.

Usage

Use this extension to enhance the code editor with SQLMesh-aware syntax highlighting and autocomplete. It should be added to CodeMirror's extension array when configuring the editor for SQL files in a SQLMesh project.

Code Reference

Source Location

Repository: TobikoData_Sqlmesh
File: web/client/src/library/components/editor/extensions/SQLMeshDialect.ts

Signature

export type ExtensionSQLMeshDialect = (
  models: Map<string, Model>,
  allModelsNames: Array<{ label: string; type: string }>,
  allColumnsNames: Array<{ label: string; type: string }>,
  options?: { types: string; keywords: string },
  dialects?: string[],
) => LanguageSupport

export const SQLMeshDialect: ExtensionSQLMeshDialect

export function SQLMeshDialectCleanUp(): void

export function getSQLMeshModelKeywords(
  dialects?: string[]
): Map<string, Completion[]>

Import

import { SQLMeshDialect, SQLMeshDialectCleanUp } from '@components/editor/extensions/SQLMeshDialect'

I/O Contract

Inputs

Name	Type	Required	Description
models	Map<string, Model>	Yes	Map of model names to model objects for column suggestions
allModelsNames	Array<{label, type}>	Yes	List of all model names for autocomplete
allColumnsNames	Array<{label, type}>	Yes	List of all column names for autocomplete
options	{types, keywords}	No	Additional SQL types and keywords to include
dialects	string[]	No	List of available SQL dialects for dialect suggestions

Outputs

Name	Type	Description
LanguageSupport	LanguageSupport	CodeMirror language support object with autocomplete

SQLMesh Keywords

The extension recognizes the following SQLMesh-specific keywords:

const SQLMeshKeywords =
  'path threshold jinja_query_begin number_of_rows jinja_end not_null ' +
  'forall criteria length unique_values interval_unit unique_key ' +
  'columns grain grains references metric tags audit model name kind ' +
  'owner cron start storage_format time_column partitioned_by pre post ' +
  'batch_size audits dialect'

SQLMesh Types

Model kind types recognized by the extension:

FULL: Full refresh models
INCREMENTAL_BY_TIME_RANGE: Time-based incremental models
INCREMENTAL_BY_UNIQUE_KEY: Key-based incremental models
VIEW: View models
SEED: Seed data models
EMBEDDED: Embedded models

Autocomplete Behavior

Dot Notation Completion

When typing after a dot, suggests columns or models:

// Example: model_name.column_name
// Provides column suggestions for the model
if (isNotNil(maybeModel)) {
  options = maybeModel.columns.map(column => ({
    label: column.name,
    type: 'column',
  }))
}

MODEL Block Completion

Inside MODEL() blocks, provides SQLMesh configuration keywords:

// Detects MODEL blocks and provides context-specific suggestions
const isInModel = matchModels
  .filter(str => str.includes(word.text))
  .some(([start, end]) => ctx.pos >= start && ctx.pos <= end)

if (isInModel) {
  // Provide MODEL-specific keywords, kind options, or dialect options
  suggestions = SQLMeshModelDictionary.get('keywords')
}

Kind and Dialect Completion

Provides specialized suggestions after 'kind' and 'dialect' keywords:

After kind: Suggests FULL, INCREMENTAL_BY_TIME_RANGE, VIEW, SEED, EMBEDDED
After dialect: Suggests available SQL dialects from the dialects array

Worker Integration

The extension registers a message handler with the SQLGlot worker:

const handler = function getTokensFromSQLGlot(e: MessageEvent): void {
  if (e.data.topic === 'parse') {
    // TODO: set parsed tree and use it to improve editor autocomplete
  }
}

sqlglotWorker.addEventListener('message', handler)

Cleanup

Always call SQLMeshDialectCleanUp() when unmounting the editor to remove event listeners and prevent memory leaks.

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment