Add comprehensive PostgreSQL catalog compatibility to enable Metabase…

.gitignore

@@ -27,3 +27,18 @@ Thumbs.db
 *.test
 *.out
 coverage.html
+
+# Local config files with credentials
+duckgres_r2_test.yaml
+duckgres_*.yaml
+!duckgres.yaml.example
+
+# Local convenience scripts
+run-duckgres.sh
+
+# Local Docker setup
+docker-compose-local.yml
+Dockerfile
+
+# Local docs (may contain project-specific notes)
+docs/

CLAUDE.md

@@ -85,10 +85,14 @@ Three-tier configuration (highest to lowest priority):
 ## Testing
 
 ```bash
-# Build
-go build -o duckgres .
+# Build and run with DuckLake config (kills existing, tails logs)
+./run-duckgres.sh
+
+# Build and run with limited tables for faster Metabase/Grafana testing
+./run-duckgres.sh 5   # Only show 5 tables (DUCKGRES_TABLE_LIMIT=5)
 
-# Run on non-standard port
+# Manual build and run
+go build -o duckgres .
 ./duckgres --port 35437
 
 # Connect with psql

duckgres.yaml.example

@@ -0,0 +1,42 @@
+# Duckgres Configuration Example
+# Copy this file to duckgres.yaml and fill in your values
+
+host: "0.0.0.0"
+port: 5432
+
+# Directory for DuckDB database files
+data_dir: "./data"
+
+# User credentials (username: password)
+users:
+  postgres: "your_password_here"
+
+# Optional: TLS certificates (auto-generated if not specified)
+# tls_cert: "./certs/server.crt"
+# tls_key: "./certs/server.key"
+
+# Optional: DuckDB extensions to load
+extensions:
+  - ducklake
+  # - spatial
+  # - httpfs
+
+# Optional: DuckLake configuration for cloud-native lakehouse
+# ducklake:
+#   # PostgreSQL metadata store connection string
+#   metadata_store: "postgres:host=your-host.com port=5432 dbname=ducklake user=your_user password=YOUR_PASSWORD sslmode=require"
+#
+#   # Object store URL (S3, R2, GCS, etc.)
+#   object_store: "s3://your-bucket/"
+#
+#   # For AWS S3:
+#   # s3_region: "us-east-1"
+#   # s3_access_key: "YOUR_ACCESS_KEY"
+#   # s3_secret_key: "YOUR_SECRET_KEY"
+#   # s3_endpoint: ""  # Leave empty for AWS, set for S3-compatible services
+#   # s3_use_ssl: true
+#
+#   # For Cloudflare R2:
+#   # r2_account_id: "YOUR_ACCOUNT_ID"
+#   # r2_access_key: "YOUR_ACCESS_KEY"
+#   # r2_secret_key: "YOUR_SECRET_KEY"

main.go

@@ -40,7 +40,7 @@ type RateLimitFileConfig struct {
 
 type DuckLakeFileConfig struct {
 	MetadataStore string `yaml:"metadata_store"` // e.g., "postgres:host=localhost user=ducklake password=secret dbname=ducklake"
-	ObjectStore   string `yaml:"object_store"`   // e.g., "s3://bucket/path/" for S3/MinIO storage
+	ObjectStore   string `yaml:"object_store"`   // e.g., "s3://bucket/path/" for S3/MinIO or "r2://bucket/path/" for R2
 
 	// S3 credential provider: "config" (explicit) or "credential_chain" (AWS SDK)
 	S3Provider string `yaml:"s3_provider"`
@@ -56,6 +56,11 @@ type DuckLakeFileConfig struct {
 	// Credential chain provider settings (AWS SDK credential chain)
 	S3Chain   string `yaml:"s3_chain"`   // e.g., "env;config" - which credential sources to check
 	S3Profile string `yaml:"s3_profile"` // AWS profile name for config chain
+
+	// R2 (Cloudflare) settings for r2:// URLs
+	R2AccountID string `yaml:"r2_account_id"` // Cloudflare account ID
+	R2AccessKey string `yaml:"r2_access_key"` // R2 access key ID
+	R2SecretKey string `yaml:"r2_secret_key"` // R2 secret access key
 }
 
 // loadConfigFile loads configuration from a YAML file
@@ -221,6 +226,17 @@ func main() {
 		if fileCfg.DuckLake.S3Profile != "" {
 			cfg.DuckLake.S3Profile = fileCfg.DuckLake.S3Profile
 		}
+
+		// Apply R2 (Cloudflare) config
+		if fileCfg.DuckLake.R2AccountID != "" {
+			cfg.DuckLake.R2AccountID = fileCfg.DuckLake.R2AccountID
+		}
+		if fileCfg.DuckLake.R2AccessKey != "" {
+			cfg.DuckLake.R2AccessKey = fileCfg.DuckLake.R2AccessKey
+		}
+		if fileCfg.DuckLake.R2SecretKey != "" {
+			cfg.DuckLake.R2SecretKey = fileCfg.DuckLake.R2SecretKey
+		}
 	}
 
 	// Apply environment variables (override config file)
@@ -274,6 +290,16 @@ func main() {
 	if v := os.Getenv("DUCKGRES_DUCKLAKE_S3_PROFILE"); v != "" {
 		cfg.DuckLake.S3Profile = v
 	}
+	// R2 (Cloudflare) environment variables
+	if v := os.Getenv("DUCKGRES_DUCKLAKE_R2_ACCOUNT_ID"); v != "" {
+		cfg.DuckLake.R2AccountID = v
+	}
+	if v := os.Getenv("DUCKGRES_DUCKLAKE_R2_ACCESS_KEY"); v != "" {
+		cfg.DuckLake.R2AccessKey = v
+	}
+	if v := os.Getenv("DUCKGRES_DUCKLAKE_R2_SECRET_KEY"); v != "" {
+		cfg.DuckLake.R2SecretKey = v
+	}
 
 	// Apply CLI flags (highest priority)
 	if *host != "" {

server/conn.go

@@ -260,7 +260,7 @@ func (c *clientConn) handleStartup() error {
 
 func (c *clientConn) sendInitialParams() {
 	params := map[string]string{
-		"server_version":          "15.0 (Duckgres)",
+		"server_version":          "15.0",
 		"server_encoding":         "UTF8",
 		"client_encoding":         "UTF8",
 		"DateStyle":               "ISO, MDY",
@@ -357,6 +357,10 @@ func (c *clientConn) handleQuery(body []byte) error {
 
 	log.Printf("[%s] Query: %s", c.username, query)
 
+	// Fix JDBC driver escaping of regex operators (e.g., \!~ -> !~)
+	// PostgreSQL JDBC sometimes sends escaped operators which pg_query can't parse
+	query = fixJDBCEscapedOperators(query)
+
 	// Transpile PostgreSQL SQL to DuckDB-compatible SQL
 	tr := c.newTranspiler(false)
 	result, err := tr.Transpile(query)
@@ -1757,3 +1761,16 @@ func isAlterTableNotTableError(err error) bool {
 	return strings.Contains(msg, "cannot use alter table") &&
 		strings.Contains(msg, "not a table")
 }
+
+// fixJDBCEscapedOperators removes backslash escaping from regex operators.
+// PostgreSQL JDBC driver sometimes sends escaped operators like \!~ which
+// pg_query cannot parse. This converts them back to standard operators.
+func fixJDBCEscapedOperators(query string) string {
+	// Fix escaped regex operators: \!~ -> !~, \!~* -> !~*
+	query = strings.ReplaceAll(query, `\!~*`, `!~*`)
+	query = strings.ReplaceAll(query, `\!~`, `!~`)
+	// Fix escaped LIKE operators: \!~~ -> !~~, \!~~* -> !~~*
+	query = strings.ReplaceAll(query, `\!~~*`, `!~~*`)
+	query = strings.ReplaceAll(query, `\!~~`, `!~~`)
+	return query
+}