Add DuckDB engine support with database-backed catalog

This commit adds comprehensive DuckDB support to sqlc using a
database-backed approach with required analyzer, similar to PostgreSQL's
optional analyzer pattern but made mandatory for DuckDB.

Key features:
- Database-backed catalog using DuckDB connections
- Required analyzer for type inference and schema information
- TiDB parser for SQL parsing (shared with MySQL engine)
- DuckDB reserved keywords implementation
- Type normalization for DuckDB-specific types
- Example project demonstrating basic usage

Implementation details:
- Parser: Uses TiDB parser, supports -- and /* */ comments
- Catalog: Minimal implementation, no pre-generated types
- Analyzer: Required component, connects via go-duckdb driver
- Converter: Reuses Dolphin/MySQL AST converter
- Reserved keywords: Based on DuckDB 1.3.0 specification

Files created:
- internal/engine/duckdb/parse.go - Parser implementation
- internal/engine/duckdb/catalog.go - Minimal catalog
- internal/engine/duckdb/convert.go - AST converter
- internal/engine/duckdb/reserved.go - Reserved keywords
- internal/engine/duckdb/analyzer/analyze.go - Database analyzer
- examples/duckdb/basic/ - Example project
- DUCKDB_SUPPORT.md - Comprehensive documentation

Files modified:
- internal/config/config.go - Added EngineDuckDB constant
- internal/compiler/engine.go - Registered DuckDB with analyzer
- go.mod - Added github.com/marcboeker/go-duckdb v1.8.5

Requirements:
- Database connection is required (not optional)
- Configuration must include database.uri parameter
- Run 'go mod tidy' to download dependencies (network required)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

DUCKDB_SUPPORT.md

@@ -0,0 +1,241 @@
+# DuckDB Support for sqlc
+
+This document describes the DuckDB engine implementation for sqlc.
+
+## Overview
+
+DuckDB support has been added to sqlc using a database-backed approach, similar to PostgreSQL's analyzer pattern. Unlike MySQL and SQLite which use Go-based catalogs, DuckDB relies entirely on database connections for type inference and schema information.
+
+## Implementation Details
+
+### Core Components
+
+1. **Parser** (`/internal/engine/duckdb/parse.go`)
+   - Uses the TiDB parser (same as MySQL/Dolphin engine)
+   - Implements the `Parser` interface with `Parse()`, `CommentSyntax()`, and `IsReservedKeyword()` methods
+   - Supports `--` and `/* */` comment styles (DuckDB standard)
+
+2. **Catalog** (`/internal/engine/duckdb/catalog.go`)
+   - Minimal catalog implementation
+   - Sets "main" as the default schema and "memory" as the default catalog
+   - Does not include pre-generated types/functions (database-backed only)
+
+3. **Analyzer** (`/internal/engine/duckdb/analyzer/analyze.go`)
+   - **REQUIRED** for DuckDB engine (not optional like PostgreSQL)
+   - Connects to DuckDB database via `github.com/marcboeker/go-duckdb`
+   - Uses PREPARE and DESCRIBE to analyze queries
+   - Queries column metadata from prepared statements
+   - Normalizes DuckDB types to sqlc-compatible types
+
+4. **AST Converter** (`/internal/engine/duckdb/convert.go`)
+   - Copied from Dolphin/MySQL implementation
+   - Converts TiDB parser AST to sqlc universal AST
+
+5. **Reserved Keywords** (`/internal/engine/duckdb/reserved.go`)
+   - DuckDB reserved keywords based on official documentation
+   - Includes LAMBDA (reserved as of DuckDB 1.3.0)
+   - Can be queried from DuckDB using `SELECT * FROM duckdb_keywords()`
+
+## Configuration
+
+### Engine Registration
+
+Added `EngineDuckDB` constant to `/internal/config/config.go`:
+```go
+const (
+    EngineDuckDB     Engine = "duckdb"
+    EngineMySQL      Engine = "mysql"
+    EnginePostgreSQL Engine = "postgresql"
+    EngineSQLite     Engine = "sqlite"
+)
+```
+
+### Compiler Integration
+
+Registered in `/internal/compiler/engine.go` with required database analyzer:
+```go
+case config.EngineDuckDB:
+    c.parser = duckdb.NewParser()
+    c.catalog = duckdb.NewCatalog()
+    c.selector = newDefaultSelector()
+    // DuckDB requires database analyzer
+    if conf.Database == nil {
+        return nil, fmt.Errorf("duckdb engine requires database configuration")
+    }
+    if conf.Analyzer.Database == nil || *conf.Analyzer.Database {
+        c.analyzer = analyzer.Cached(
+            duckdbanalyze.New(c.client, *conf.Database),
+            combo.Global,
+            *conf.Database,
+        )
+    }
+```
+
+## Usage Example
+
+### sqlc.yaml Configuration
+
+```yaml
+version: "2"
+sql:
+  - name: "basic"
+    engine: "duckdb"
+    schema: "schema/"
+    queries: "query/"
+    database:
+      uri: ":memory:"  # or path to .db file
+    gen:
+      go:
+        out: "db"
+        package: "db"
+        emit_json_tags: true
+        emit_interface: true
+```
+
+### Schema Example
+
+```sql
+CREATE TABLE authors (
+  id   INTEGER PRIMARY KEY,
+  name VARCHAR NOT NULL,
+  bio  TEXT
+);
+```
+
+### Query Example
+
+```sql
+-- name: GetAuthor :one
+SELECT * FROM authors
+WHERE id = $1 LIMIT 1;
+
+-- name: ListAuthors :many
+SELECT * FROM authors
+ORDER BY name;
+
+-- name: CreateAuthor :exec
+INSERT INTO authors (name, bio)
+VALUES ($1, $2);
+```
+
+## Key Differences from Other Engines
+
+### vs PostgreSQL
+- **PostgreSQL**: Optional database analyzer, rich Go-based catalog with pg_catalog
+- **DuckDB**: Required database analyzer, minimal catalog
+
+### vs MySQL/SQLite
+- **MySQL/SQLite**: Go-based catalog with built-in functions
+- **DuckDB**: Database-backed only, no Go-based catalog
+
+## Type Mapping
+
+DuckDB types are normalized to sqlc-compatible types:
+
+| DuckDB Type | sqlc Type |
+|-------------|-----------|
+| INTEGER, INT, INT4 | integer |
+| BIGINT, INT8, LONG | bigint |
+| SMALLINT, INT2, SHORT | smallint |
+| TINYINT, INT1 | tinyint |
+| DOUBLE, FLOAT8 | double |
+| REAL, FLOAT4, FLOAT | real |
+| VARCHAR, TEXT, STRING | varchar |
+| BOOLEAN, BOOL | boolean |
+| DATE | date |
+| TIME | time |
+| TIMESTAMP | timestamp |
+| TIMESTAMPTZ | timestamptz |
+| BLOB, BYTEA, BINARY | bytea |
+| UUID | uuid |
+| JSON | json |
+| DECIMAL, NUMERIC | decimal |
+
+## Dependencies
+
+Added to `go.mod`:
+```go
+github.com/marcboeker/go-duckdb v1.8.5
+```
+
+## Setup Instructions
+
+1. **Install dependencies** (requires network access):
+   ```bash
+   go mod tidy
+   ```
+
+2. **Build sqlc**:
+   ```bash
+   go build ./cmd/sqlc
+   ```
+
+3. **Run code generation**:
+   ```bash
+   ./sqlc generate
+   ```
+
+## Testing
+
+An example project is provided in `/examples/duckdb/basic/` with:
+- Schema definitions
+- Sample queries
+- sqlc.yaml configuration
+
+To test:
+```bash
+cd examples/duckdb/basic
+sqlc generate
+```
+
+## Database Requirements
+
+DuckDB engine **requires** a database connection. You must configure:
+```yaml
+database:
+  uri: "path/to/database.db"  # or ":memory:" for in-memory
+```
+
+Without this configuration, the compiler will return an error:
+```
+duckdb engine requires database configuration
+```
+
+## Limitations
+
+1. **Network dependency**: Requires network access to download go-duckdb initially
+2. **Parameter type inference**: DuckDB doesn't provide parameter types without execution, so parameters are typed as "any" by the analyzer
+3. **Parser limitations**: Uses TiDB parser which may not support all DuckDB-specific syntax (STRUCT, LIST, UNION types may require custom handling)
+
+## Future Enhancements
+
+1. Improve parameter type inference
+2. Add support for DuckDB-specific types (STRUCT, LIST, UNION, MAP)
+3. Support DuckDB extensions
+4. Add DuckDB-specific selector for custom column handling
+5. Improve error messages with DuckDB-specific error codes
+
+## Files Modified/Created
+
+### Created:
+- `/internal/engine/duckdb/parse.go`
+- `/internal/engine/duckdb/catalog.go`
+- `/internal/engine/duckdb/convert.go`
+- `/internal/engine/duckdb/reserved.go`
+- `/internal/engine/duckdb/analyzer/analyze.go`
+- `/examples/duckdb/basic/schema/schema.sql`
+- `/examples/duckdb/basic/query/query.sql`
+- `/examples/duckdb/basic/sqlc.yaml`
+
+### Modified:
+- `/internal/config/config.go` - Added `EngineDuckDB` constant
+- `/internal/compiler/engine.go` - Registered DuckDB engine with analyzer
+- `/go.mod` - Added `github.com/marcboeker/go-duckdb v1.8.5`
+
+## Notes
+
+- DuckDB uses "main" as the default schema (different from PostgreSQL's "public")
+- DuckDB uses "memory" as the default catalog name
+- Comment syntax supports only `--` and `/* */`, not `#`
+- Reserved keyword LAMBDA was added in DuckDB 1.3.0
+- Reserved keyword GRANT was removed in DuckDB 1.3.0

examples/duckdb/basic/query/query.sql

@@ -0,0 +1,23 @@
+-- name: GetAuthor :one
+SELECT * FROM authors
+WHERE id = $1 LIMIT 1;
+
+-- name: ListAuthors :many
+SELECT * FROM authors
+ORDER BY name;
+
+-- name: CreateAuthor :exec
+INSERT INTO authors (
+  name, bio
+) VALUES (
+  $1, $2
+);
+
+-- name: UpdateAuthor :exec
+UPDATE authors
+SET name = $1, bio = $2
+WHERE id = $3;
+
+-- name: DeleteAuthor :exec
+DELETE FROM authors
+WHERE id = $1;

examples/duckdb/basic/schema/schema.sql

@@ -0,0 +1,5 @@
+CREATE TABLE authors (
+  id   INTEGER PRIMARY KEY,
+  name VARCHAR NOT NULL,
+  bio  TEXT
+);

examples/duckdb/basic/sqlc.yaml

@@ -0,0 +1,14 @@
+version: "2"
+sql:
+  - name: "basic"
+    engine: "duckdb"
+    schema: "schema/"
+    queries: "query/"
+    database:
+      uri: ":memory:"
+    gen:
+      go:
+        out: "db"
+        package: "db"
+        emit_json_tags: true
+        emit_interface: true

go.mod

@@ -16,6 +16,7 @@ require (
 	github.com/jackc/pgx/v5 v5.7.6
 	github.com/jinzhu/inflection v1.0.0
 	github.com/lib/pq v1.10.9
+	github.com/marcboeker/go-duckdb v1.8.5
 	github.com/pganalyze/pg_query_go/v6 v6.1.0
 	github.com/pingcap/tidb/pkg/parser v0.0.0-20250324122243-d51e00e5bbf0
 	github.com/riza-io/grpc-go v0.2.0

internal/compiler/engine.go

@@ -8,6 +8,8 @@ import (
 	"github.com/sqlc-dev/sqlc/internal/config"
 	"github.com/sqlc-dev/sqlc/internal/dbmanager"
 	"github.com/sqlc-dev/sqlc/internal/engine/dolphin"
+	"github.com/sqlc-dev/sqlc/internal/engine/duckdb"
+	duckdbanalyze "github.com/sqlc-dev/sqlc/internal/engine/duckdb/analyzer"
 	"github.com/sqlc-dev/sqlc/internal/engine/postgresql"
 	pganalyze "github.com/sqlc-dev/sqlc/internal/engine/postgresql/analyzer"
 	"github.com/sqlc-dev/sqlc/internal/engine/sqlite"
@@ -37,6 +39,21 @@ func NewCompiler(conf config.SQL, combo config.CombinedSettings) (*Compiler, err
 	}
 
 	switch conf.Engine {
+	case config.EngineDuckDB:
+		c.parser = duckdb.NewParser()
+		c.catalog = duckdb.NewCatalog()
+		c.selector = newDefaultSelector()
+		// DuckDB requires database analyzer
+		if conf.Database == nil {
+			return nil, fmt.Errorf("duckdb engine requires database configuration")
+		}
+		if conf.Analyzer.Database == nil || *conf.Analyzer.Database {
+			c.analyzer = analyzer.Cached(
+				duckdbanalyze.New(c.client, *conf.Database),
+				combo.Global,
+				*conf.Database,
+			)
+		}
 	case config.EngineSQLite:
 		c.parser = sqlite.NewParser()
 		c.catalog = sqlite.NewCatalog()

internal/config/config.go

@@ -51,6 +51,7 @@ func (p *Paths) UnmarshalYAML(unmarshal func(interface{}) error) error {
 }
 
 const (
+	EngineDuckDB     Engine = "duckdb"
 	EngineMySQL      Engine = "mysql"
 	EnginePostgreSQL Engine = "postgresql"
 	EngineSQLite     Engine = "sqlite"