Add COPY protocol, graceful shutdown, and enhanced pg_catalog support

CLAUDE.md

@@ -0,0 +1,124 @@
+# Claude Code Context for Duckgres
+
+This file provides context for Claude Code sessions working on this codebase.
+
+## Project Overview
+
+Duckgres is a PostgreSQL wire protocol server backed by DuckDB. It allows any PostgreSQL client (psql, pgAdmin, lib/pq, psycopg2, JDBC, etc.) to connect and execute queries against DuckDB databases.
+
+## Architecture
+
+```
+PostgreSQL Client → TLS → Duckgres Server → DuckDB (per-user database)
+```
+
+### Key Components
+
+- **main.go**: Entry point, configuration loading (CLI flags, env vars, YAML)
+- **server/server.go**: Server struct, connection handling, graceful shutdown
+- **server/conn.go**: Client connection handling, query execution, COPY protocol
+- **server/protocol.go**: PostgreSQL wire protocol message encoding/decoding
+- **server/catalog.go**: pg_catalog compatibility (views, functions, query rewriting)
+- **server/types.go**: Type OID mapping between DuckDB and PostgreSQL
+- **server/ratelimit.go**: Rate limiting for brute-force protection
+- **server/tls.go**: Auto-generation of self-signed TLS certificates
+
+## PostgreSQL Wire Protocol
+
+The server implements the PostgreSQL v3 protocol:
+
+### Message Types (server/protocol.go)
+- **Frontend (client→server)**: Query, Parse, Bind, Describe, Execute, Sync, Close, CopyData, CopyDone
+- **Backend (server→client)**: AuthOK, RowDescription, DataRow, CommandComplete, ReadyForQuery, CopyInResponse, CopyOutResponse
+
+### Query Flow
+1. Client sends Query message ('Q')
+2. Server parses SQL, rewrites pg_catalog references
+3. Server executes via DuckDB's database/sql driver
+4. Server sends RowDescription + DataRow messages
+5. Server sends CommandComplete + ReadyForQuery
+
+### Extended Query Protocol
+Supports prepared statements (Parse/Bind/Execute) for parameterized queries and binary result formats.
+
+## pg_catalog Compatibility (server/catalog.go)
+
+psql and other clients expect PostgreSQL system catalogs. We provide compatibility by:
+
+1. **Creating views** in main schema that mirror pg_catalog tables:
+   - `pg_database`, `pg_class_full`, `pg_collation`, `pg_policy`, `pg_roles`
+   - `pg_statistic_ext`, `pg_publication`, `pg_publication_rel`, `pg_inherits`, etc.
+
+2. **Creating macros** for PostgreSQL functions:
+   - `pg_get_userbyid`, `pg_table_is_visible`, `format_type`, `pg_get_expr`
+   - `obj_description`, `col_description`, `pg_get_indexdef`, etc.
+
+3. **Query rewriting** to replace PostgreSQL-specific syntax:
+   - `pg_catalog.pg_class` → `pg_class_full`
+   - `OPERATOR(pg_catalog.~)` → `~`
+   - `::pg_catalog.regtype` → `::VARCHAR`
+
+## COPY Protocol (server/conn.go)
+
+Supports bulk data transfer:
+- **COPY TO STDOUT**: Streams query results to client
+- **COPY FROM STDIN**: Receives data from client, inserts row by row
+- Supports CSV format with HEADER option
+
+## Configuration
+
+Three-tier configuration (highest to lowest priority):
+1. CLI flags (`--port`, `--config`, etc.)
+2. Environment variables (`DUCKGRES_PORT`, etc.)
+3. YAML config file
+4. Built-in defaults
+
+## Testing
+
+```bash
+# Build
+go build -o duckgres .
+
+# Run on non-standard port
+./duckgres --port 35437
+
+# Connect with psql
+PGPASSWORD=postgres psql "host=127.0.0.1 port=35437 user=postgres sslmode=require"
+
+# Test commands
+\dt          # List tables
+\d tablename # Describe table
+\l           # List databases
+```
+
+## Common Development Tasks
+
+### Adding a new pg_catalog view
+1. Add view creation SQL in `initPgCatalog()` in `catalog.go`
+2. Add regex pattern to rewrite `pg_catalog.viewname` to `viewname`
+3. Add the replacement in `rewritePgCatalogQuery()`
+
+### Adding a new PostgreSQL function
+1. Add `CREATE MACRO` in the `functions` slice in `initPgCatalog()`
+2. Add function name to `pgCatalogFunctions` slice for query rewriting
+
+### Adding protocol support
+1. Add message type constant in `protocol.go`
+2. Add write function (e.g., `writeCopyData()`)
+3. Handle in message loop in `conn.go`
+
+## Dependencies
+
+- `github.com/duckdb/duckdb-go/v2` - DuckDB Go driver
+- `gopkg.in/yaml.v3` - YAML config parsing
+
+## Known Limitations
+
+- Single process (all users share one process)
+- No replication
+- Some pg_catalog tables are stubs (return empty)
+- Type OID mapping is incomplete (some types show as "unknown")
+
+## TODO Reference
+
+See `TODO.md` for the full feature roadmap and known issues.

README.md

@@ -7,11 +7,13 @@ A PostgreSQL wire protocol compatible server backed by DuckDB. Connect with any
 - **PostgreSQL Wire Protocol**: Full compatibility with PostgreSQL clients
 - **TLS Encryption**: Required TLS connections with auto-generated self-signed certificates
 - **Per-User Databases**: Each authenticated user gets their own isolated DuckDB database file
-- **Password Authentication**: MD5 password authentication
+- **Password Authentication**: Cleartext password authentication over TLS
 - **Extended Query Protocol**: Support for prepared statements, binary format, and parameterized queries
+- **COPY Protocol**: Bulk data import/export with `COPY FROM STDIN` and `COPY TO STDOUT`
 - **DuckDB Extensions**: Configurable extension loading (ducklake enabled by default)
 - **DuckLake Integration**: Auto-attach DuckLake catalogs for lakehouse workflows
 - **Rate Limiting**: Built-in protection against brute-force attacks
+- **Graceful Shutdown**: Waits for in-flight queries before exiting
 - **Flexible Configuration**: YAML config files, environment variables, and CLI flags
 
 ## Quick Start
@@ -144,6 +146,46 @@ ATTACH 'ducklake:postgres:host=localhost dbname=ducklake' (DATA_PATH 's3://my-bu
 
 See [DuckLake documentation](https://ducklake.select/docs/stable/duckdb/usage/connecting) for more details.
 
+## COPY Protocol
+
+Duckgres supports PostgreSQL's COPY protocol for efficient bulk data import and export:
+
+```sql
+-- Export data to stdout (tab-separated)
+COPY tablename TO STDOUT;
+
+-- Export as CSV with headers
+COPY tablename TO STDOUT WITH CSV HEADER;
+
+-- Export query results
+COPY (SELECT * FROM tablename WHERE id > 100) TO STDOUT WITH CSV;
+
+-- Import data from stdin
+COPY tablename FROM STDIN;
+
+-- Import CSV with headers
+COPY tablename FROM STDIN WITH CSV HEADER;
+```
+
+This works with psql's `\copy` command and programmatic COPY operations from PostgreSQL drivers.
+
+## Graceful Shutdown
+
+Duckgres handles shutdown signals (SIGINT, SIGTERM) gracefully:
+
+- Stops accepting new connections immediately
+- Waits for in-flight queries to complete (default 30s timeout)
+- Logs active connection count during shutdown
+- Closes all database connections cleanly
+
+The shutdown timeout can be configured:
+
+```go
+cfg := server.Config{
+    ShutdownTimeout: 60 * time.Second,
+}
+```
+
 ## Rate Limiting
 
 Built-in rate limiting protects against brute-force authentication attacks:
@@ -219,6 +261,7 @@ GROUP BY name;
 - `DROP TABLE/INDEX/VIEW`
 - `ALTER TABLE`
 - `BEGIN/COMMIT/ROLLBACK` (DuckDB transaction support)
+- `COPY` - Bulk data loading and export (see below)
 
 ### PostgreSQL Compatibility
 - Extended query protocol (prepared statements)

TODO.md

@@ -16,26 +16,26 @@
 
 ### Protocol Compatibility
 - [ ] **Binary Format Support**: Encode results in binary format for better performance with some clients
-- [ ] **COPY Protocol**: Support `COPY FROM`/`COPY TO` for bulk data loading
+- [x] **COPY Protocol**: Support `COPY FROM`/`COPY TO` for bulk data loading
 - [ ] **Cancel Request Handling**: Properly cancel long-running queries
 
 ### Compatibility
 - [x] **System Catalog Emulation**: Basic `pg_catalog` compatibility for psql
   - [x] `\dt` (list tables) - working
   - [x] `\l` (list databases) - working
-  - [ ] `\d <table>` (describe table) - needs more pg_class columns
+  - [x] `\d <table>` (describe table) - working
 - [ ] **Information Schema**: Emulate PostgreSQL's `information_schema`
 - [ ] **Session Variables**: Support `SET` commands (timezone, search_path, etc.)
 - [ ] **Type OID Mapping**: Proper PostgreSQL OID mapping for all DuckDB types
 
 ### Features
-- [ ] **Extensions**: Load DuckDB extensions on startup
+- [x] **Extensions**: Load DuckDB extensions on startup
 
 ### Operations
 - [ ] **Hot Reload**: Reload config without restart
 - [ ] **Admin Commands**: `\duckgres status`, `\duckgres users`, etc.
 - [ ] **Docker Image**: Official container image
-- [ ] **Graceful Shutdown**: Finish in-flight queries before shutdown
+- [x] **Graceful Shutdown**: Finish in-flight queries before shutdown
 
 ## Medium Priority
 
@@ -73,7 +73,7 @@
 ## Known Issues
 
 - [ ] Some PostgreSQL drivers may fail with unsupported OIDs
-- [ ] `\d` commands in psql don't work (need system catalog)
+- [x] `\d` commands in psql don't work (need system catalog) - fixed
 - [ ] Transaction isolation may differ from PostgreSQL behavior
 - [ ] Large result sets may cause memory issues (no streaming)