Refactor column storage to row storage (#49)

* refactor: clear useless code and dependencies

* feat: impl eval_column for tp

* feat: impl storage for tp

* feat: impl dml(insert/values/seq_scan/project) and ddl(create) for tp

* feat: impl dml(filter) for tp

* feat: impl dml(sort) for tp

* feat: impl dml(limit) for tp

* feat: `SeqScan` supports Projection and Limit pushdown

* style: new dql

* feat: impl dql(hash_join) for tp

* feat: impl `on filter` for dql(hash_join)

* feat: impl `repeat join` for dql(hash_join)

* style: remove ap

* perf: use Arc pointers to encapsulate `DataValue` and `ColumnCatalog` to avoid copy loss

* docs: rewrite README.md and fix `hash_join` bug

* style: version reset

Author: KKould

Cargo.toml

@@ -2,46 +2,33 @@
 
 [package]
 name = "kip-sql"
-version = "0.1.0"
+version = "0.0.1-alpha.0"
 edition = "2021"
 
 [lib]
 doctest = false
 
 [dependencies]
-log = "^0.4"
 sqlparser = "0.34.0"
 thiserror = "1"
 parking_lot = "0.12.1"
 itertools = "0.10"
 sqllogictest = "0.11.1"
-rust_decimal = "1"
-paste = "1.0.11"
 tracing = "0.1.37"
-pin-project = "1.1.0"
 chrono = "0.4.26"
-roaring = "0.10.1"
-educe = "0.4"
-num-traits = "0.2"
-num-derive = "0.3"
-anyhow = "1.0.71"
 tokio = { version = "1.28.2", features = ["full"] }
-tokio-process = "0.2.5"
 serde = { version = "1", features = ["derive", "rc"] }
 serde_json = "1"
 async-trait = "0.1.68"
 integer-encoding = "3.0.4"
-arrow = { version = "28", features = ["prettyprint", "simd"] }
 strum_macros = "0.24"
 ordered-float = "3.0"
 petgraph = "0.6.3"
 futures-async-stream = "0.2.6"
-async-channel = "1.8.0"
-async-backtrace = "0.2.6"
 futures = "0.3.25"
-futures-lite = "1.12.0"
 ahash = "0.8.3"
 lazy_static = "1.4.0"
+comfy-table = "7.0.1"
 
 [dev-dependencies]
 tokio-test = "0.4.2"

README.md

@@ -1,71 +1,84 @@
 # KipSQL 
 
-> build the SQL layer of KipDB database.
+> Lightweight SQL calculation engine, as the SQL layer of KipDB, implemented with TalentPlan's TinySQL as the reference standard
 
+### Architecture
+![architecture](./static/images/architecture.png)
 
-> 目前处于探索阶段
-轻量级SQL计算引擎，作为KipDB的SQL层，以TalentPlan的TinySQL作为参考标准进行实现
-
-## Example
-```rust
-let kipsql = Database::new_on_mem();
-
-let _ = kipsql.run("create table t1 (a int, b int)").await?;
-let _ = kipsql.run("create table t2 (c int, d int)").await?;
-let _ = kipsql.run("insert into t1 (b, a) values (1, 1), (3, 3), (5, 4)").await?;
-let _ = kipsql.run("insert into t2 (d, c) values (1, 2), (2, 3), (5, 6)").await?;
-
-println!("full t1:");
-let vec_batch_full_fields_t1 = kipsql.run("select * from t1").await?;
-print_batches(&vec_batch_full_fields_t1)?;
-
-println!("full t2:");
-let vec_batch_full_fields_t2 = kipsql.run("select * from t2").await?;
-print_batches(&vec_batch_full_fields_t2)?;
-
-println!("projection_and_filter:");
-let vec_batch_projection_a = kipsql.run("select a from t1 where a <= b").await?;
-print_batches(&vec_batch_projection_a)?;
-
-println!("projection_and_sort:");
-let vec_batch_projection_a = kipsql.run("select a from t1 order by a").await?;
-print_batches(&vec_batch_projection_a)?;
-
-println!("limit:");
-let vec_batch_limit=kipsql.run("select * from t1 limit 1 offset 1").await?;
-print_batches(&vec_batch_limit)?;
-
-println!("inner join:");
-let vec_batch_inner_join = kipsql.run("select * from t1 inner join t2 on a = c").await?;
-print_batches(&vec_batch_inner_join)?;
+### Get Started
+Install rust toolchain first.
+```
+cargo run
+```
+test command
+```mysql
+create table t1 (a int, b int);
 
-println!("left join:");
-let vec_batch_left_join = kipsql.run("select * from t1 left join t2 on a = c").await?;
-print_batches(&vec_batch_left_join)?;
+insert into t1 (a, b) values (1, 1), (5, 3), (5, 2);
 
-println!("right join:");
-let vec_batch_right_join = kipsql.run("select * from t1 right join t2 on a = c and a > 1").await?;
-print_batches(&vec_batch_right_join)?;
+select * from t1;
 
-println!("full join:");
-let vec_batch_full_join = kipsql.run("select d, b from t1 full join t2 on a = c and a > 1").await?;
-print_batches(&vec_batch_full_join)?;            
+select * from t1 order by a asc nulls first
 ```
 
+![demo](./static/images/demo.png)
+
 ### Features
-- Select
-  - Filter
-  - Limit
-  - Sort
-  - Projection
-  - TableScan
-  - Join (HashJoin)
-    - Inner
-    - Left
-    - Right
-    - Full 
-- Insert
-- CreateTable
+- DDL
+  - Create
+    - [x] CreateTable
+    - [ ] CreateIndex
+  - Drop
+- DQL
+  - [x] Select
+  - [x] Where
+  - [ ] Distinct
+  - [ ] Aggregation: count()/sum()/avg()/min()/max()
+  - [ ] Subquery
+  - [x] Join: Inner/Left/Right/Full Cross(x)
+  - [ ] Group By
+  - [ ] Having
+  - [x] Order By
+  - [x] Limit
+- DML
+  - [x] Insert
+  - [ ] Update
+  - [ ] Delete
+- DataTypes
+  - Invalid
+  - SqlNull
+  - Boolean
+  - Tinyint
+  - UTinyint
+  - Smallint
+  - USmallint
+  - Integer
+  - UInteger
+  - Bigint
+  - UBigint
+  - Float
+  - Double
+  - Varchar
+- Optimizer rules
+  - Limit Project Transpose
+  - Eliminate Limits
+  - Push Limit Through Join
+  - Push Limit Into Scan
+  - Combine Filters
+  - Column Pruning
+  - Collapse Project
+- Executors
+  - [x] CreateTable
+  - [x] SeqScan
+  - [ ] IndexScan
+  - [x] Filter
+  - [x] Project
+  - [x] Limit
+  - [x] Hash Join
+  - [x] Insert
+  - [x] Values
+  - [ ] Update
+  - [ ] Delete
 
 ### Thanks For
 - [Fedomn/sqlrs](https://github.com/Fedomn/sqlrs): 主要参考资料，Optimizer、Executor均参考自sqlrs的设计