QQL — Qdrant Query Language for VS Code

Syntax highlighting, live linting, and autocompletion for QQL — a SQL-like query language for the Qdrant vector database.

QQL is to Qdrant what SQL is to Postgres.

Features

Syntax Highlighting

All 127 QQL keywords tokenized into distinct scopes — statements, query expressions, filters, DDL, formula expressions, and boolean/null constants. Strings, numbers, comments, operators, and dotted paths (field.nested, items[].name) are all highlighted correctly.

Token type	Scope
Statements (`QUERY`, `UPSERT`, `CREATE`, …)	`keyword.control`
Clauses (`FROM`, `WHERE`, `LIMIT`, …)	`keyword.other`
Query modes (`HYBRID`, `MMR`, `FUSION`, …)	`keyword.query`
DDL (`COLLECTION`, `HNSW`, `QUANTIZATION`, …)	`keyword.ddl`
Filters (`MATCH`, `NESTED`, `GEO_BBOX`, …)	`keyword.filter`
Formula (`CASE WHEN`, `ABS`, `GEO_DISTANCE`, …)	`keyword.formula`
Booleans (`true`, `false`)	`constant.language.boolean`
Null	`constant.language.null`
Strings	`string.quoted.single` / `.double`
Numbers (int, float, sci notation)	`constant.numeric`
Comments (`-- …`)	`comment.line`

Live Diagnostics

Every .qql file is parsed in real-time by the same WASM parser that powers qql-core. Parse errors appear as red squiggly underlines with the exact error code, message, and position from the Rust compiler pipeline.

QUERY missing FROM docs;   ← QQL-PARSE-UNEXPECTED: Expected FROM

Errors update within 300ms of typing
Error spans are precise (byte-level from the Rust parser, converted to VS Code positions)
Zero external dependencies — the 1.35 MB WASM binary is bundled directly

Autocompletion

127 keyword completions — every QQL keyword is available at your cursor.

19 snippet templates — common query patterns insert with placeholders:

Snippet	What it generates
`QUERY NEAREST`	Basic semantic search with `TEXT`/`USING`/`LIMIT`
`QUERY HYBRID`	Hybrid dense+sparse RRF search
`QUERY HYBRID DBSF`	Hybrid with DBSF fusion
`CTE + FUSION`	Multi-stage retrieval with `WITH`/`PREFETCH`/`FUSION RRF`
`CTE + RERANK`	Two-stage retrieval with cross-encoder reranking
`QUERY RECOMMEND`	Recommendation from positive/negative examples
`QUERY MMR`	Maximal Marginal Relevance diversity search
`QUERY FORMULA`	Score shaping with arithmetic and payload fields
`QUERY POINTS`	Direct point ID retrieval
`QUERY ORDER BY`	Ordered payload scan
`QUERY SAMPLE`	Random sampling
`UPSERT INTO`	Point upsert with auto-embedding
`CREATE COLLECTION`	Collection DDL
`CREATE COLLECTION HYBRID`	Hybrid dense+sparse collection DDL
`CREATE INDEX`	Payload index creation
`SCROLL`	Cursor-based pagination
`COUNT`	Point counting with filter
`DELETE`	Point deletion by filter

Language Configuration

Comment toggle: Ctrl+/ inserts --
Bracket matching: {}, [], () highlight pairs and auto-close
String auto-close: ' and " auto-close with the matching quote

Requirements

VS Code 1.85 or newer
No additional dependencies — the extension bundles everything

Supported File Types

.qql files — full highlighting, linting, and completions

Extension Size

Component	Size
WASM parser binary	1.35 MB
JS loader + providers	~14 KB
TextMate grammar	4.2 KB
Total VSIX	846 KB

The WASM binary is compiled from qql-core (the Rust reference implementation) and contains the complete QQL v1 parser — lexer, recursive-descent parser, Pratt formula parser, and AST builder — all running in-process.

How It Works

.qql file → VS Code onDidChangeTextDocument
                │
                ▼ (300ms debounce)
          qql-wasm analyze()
                │
                ▼
        ┌──────────────┐
        │  Rust parser  │  (1.35 MB WASM binary)
        │  • Lexer      │
        │  • Parser     │
        │  • AST build  │
        └──────┬───────┘
               │
               ▼
    { valid, error: { code, message, start, end }, tokens, … }
               │
               ▼
      VS Code DiagnosticCollection (red squiggly)

Parsing happens synchronously in the extension host (Node.js)
No network calls — the parser is fully self-contained
No file writes — everything is in-memory

Installation

From VS Code Marketplace (recommended)

Open VS Code
Go to Extensions (Ctrl+Shift+X)
Search for QQL
Click Install

Or from the command line:

code --install-extension srimon12.qql-lang

From `.vsix` (offline / local build)

# Download the .vsix from GitHub Releases
code --install-extension qql-lang-0.1.0.vsix

Build from source

git clone https://github.com/srimon12/qql-rs
cd qql-rs/editors/vscode

# Build the WASM parser
wasm-pack build ../../crates/qql-wasm --release --target nodejs --out-dir wasm

# Install dependencies
npm install

# Compile TypeScript
npm run vscode:prepublish

# Package
npx vsce package

# Install locally
code --install-extension qql-lang-0.1.0.vsix

| Project | Description | |---------|-------------| | `qql-rs` | Rust reference implementation — parser, planner, runtime, CLI, edge, bindings | | `qql-go` | Go implementation — gateway, RPC, policy engine, MCP server |

License

MIT — see LICENSE for details.

QQL — Qdrant Query Language

srimon