apache / apache/datafusion-sqlparser-rs
DorisSQL: add Doris Dialect
- 主要语言
- Rust
- 星标
- 3.5k
- 派生
- 772
- 平均合并
- 4 天 9 小时
- 30 天内合并 PR
- 17
描述
# Doris SQL: Support Doris Dialect
## Summary
Add Doris SQL dialect, covering the core Doris syntax needed to parse common Doris DDL and load statements.
## Motivation
Apache Doris has SQL syntax that differs from the generic SQL dialect, especially around `CREATE TABLE` clauses, key models, partition/distribution definitions, table properties, and Doris-specific load statements. A dedicated `DorisDialect` would let users parse Doris SQL without weakening generic parser behavior.
## Proposed Scope
- Add `DorisDialect`
- Support Doris `CREATE TABLE` clauses:
- `ENGINE = OLAP`
- `DUPLICATE KEY`, `UNIQUE KEY`, `AGGREGATE KEY`
- table `COMMENT`
- `PARTITION BY RANGE/LIST`
- `AUTO PARTITION`
- `DISTRIBUTED BY HASH/RANDOM`
- `BUCKETS` / `BUCKETS AUTO`
- `PROPERTIES (...)`
- Support Doris column options such as `AUTO_INCREMENT`
- Support Doris `LOAD DATA INFILE`
- Support Doris `CREATE ROUTINE LOAD`
- Preserve generic dialect behavior where Doris-only syntax should remain rejected
## Parser / AST Changes
- Add Doris-specific AST structures for key model, partition, distribution, table properties, and load statements
- Extend `CreateTable` with optional Doris-specific fields
- Add `Display` support for canonical SQL rendering
- Add span support for newly introduced AST nodes where applicable
## Tests
Add Doris parser tests covering:
- Positive Doris parsing cases
- Generic/ANSI rejection cases for Doris-only syntax
- AST structure assertions
- Round-trip canonical rendering
- Escaping behavior for table comments containing single quotes
## Implementation Plan
This feature is delivered as a stacked PR series. Each PR builds on the previous one.
- PR 1: Doris dialect skeleton + AUTO_INCREMENT AST cleanup
- PR 2: CREATE TABLE column options
- PR 3: CREATE TABLE key model and distribution
- PR 4: CREATE TABLE partition clauses + ON UPDATE + double-quoted COMMENT
- PR 5: LOAD DATA INFILE
- PR 6: CREATE ROUTINE LOAD
贡献指南
这个仓库没有索引到贡献指南
调研方向
先从计划中的 PR 1 Doris 方言骨架和 AUTO_INCREMENT AST 清理开始,然后按照分层的实现计划继续完成 CREATE TABLE 子句和 Doris 加载语句。完成的标准是覆盖 Doris 专用 AST、CreateTable 字段、Display 和 span 支持、解析器的通过和拒绝测试、AST 断言、往返渲染以及注释转义用例。
由索引模型根据 Issue 内容生成。
评估
- 技术栈
- rust, sql
- 领域
- databases
- Issue 类型
- 功能
- 难度
- 5/5
- 预计耗时
- 一周以上
- 活跃度
- 冷清
- 描述清晰度
- 基本清楚
- 新手友好度
- 35/100