feat: simple UPDATE implementation

README.md

@@ -19,7 +19,7 @@ minisql>
 
 I plan to implement more features of traditional relational databases in the future as part of this project simply to learn and discovery how various features I have grown acustomed to over the years are implemented under the hood. However, currently only a very small number of features are implemented:
 
-- simple SQL parser (partial support for `CREATE TABLE`, `INSERT`, `SELECT` queries)
+- simple SQL parser (partial support for `CREATE TABLE`, `INSERT`, `SELECT`, `UPDATE` queries)
 - only tables supported, no indexes (this means all selects are scanning whole tables for now)
 - only `int4`, `int8` and `varchar` columns supported
 - no primary key support (tables internally use row ID as key in B tree data structure)
@@ -30,13 +30,13 @@ I plan to implement more features of traditional relational databases in the fut
 
 ### Planned features:
 
-- support additional basic query types such as `UPDATE`, `DELETE`, `DROP TABLE`
+- support additional basic query types such as `DELETE`, `DROP TABLE`
 - support `NULL` values
 - B+ tree and support indexes (starting with unique and primary)
 - more column types starting with simpler ones such as `bool` and `timestamp`
 - support bigger column types such as `text` that can overflow to more pages via linked list data structure
 - joins such as `INNER`, `LEFT`, `RIGHT`
-- support `ORDER BY`, `GROUP BY`
+- support `ORDER BY`, `LIMIT`, `GROUP BY`
 - dedicate first 100B of root page for config similar to how sqlite does it
 - support altering tables
 - transactions
@@ -85,7 +85,6 @@ Insert more than a single page worth of data:
 ```sh
 minisql> insert into foo(id, email, age) values(1, '[email protected]', 35), (2, '[email protected]', 32), (3, '[email protected]', 27), (4, '[email protected]', 32), (5, '[email protected]', 27), (6, '[email protected]', 32), (7, '[email protected]', 27),  (8, '[email protected]', 32), (9, '[email protected]', 27),  (10, '[email protected]', 32), (11, '[email protected]', 27),  (12, '[email protected]', 32), (13, '[email protected]', 27), (14, '[email protected]', 27), (15, '[email protected]', 27)
 Rows affected: 15
-
 minisql>
 ```
 
@@ -101,3 +100,11 @@ minisql> select * from foo
 minisql>
 ```
 
+Update rows:
+
+```sh
+minisql> update foo set id = 45 where id = 75
+Rows affected: 0
+minisql>
+```
+

internal/pkg/database/database.go

@@ -324,12 +324,7 @@ func (d *Database) executeUpdate(ctx context.Context, stmt minisql.Statement) (m
 		return minisql.StatementResult{}, errTableDoesNotExist
 	}
 
-	if err := aTable.Update(ctx, stmt); err != nil {
-		return minisql.StatementResult{}, err
-	}
-
-	// TODO - calculate rows affected properly
-	return minisql.StatementResult{RowsAffected: 0}, nil
+	return aTable.Update(ctx, stmt)
 }
 
 func (d *Database) executeDelete(ctx context.Context, stmt minisql.Statement) (minisql.StatementResult, error) {
@@ -338,10 +333,5 @@ func (d *Database) executeDelete(ctx context.Context, stmt minisql.Statement) (m
 		return minisql.StatementResult{}, errTableDoesNotExist
 	}
 
-	if err := aTable.Delete(ctx, stmt); err != nil {
-		return minisql.StatementResult{}, err
-	}
-
-	// TODO - calculate rows affected properly
-	return minisql.StatementResult{RowsAffected: 0}, nil
+	return aTable.Delete(ctx, stmt)
 }

internal/pkg/minisql/cursor.go

@@ -35,7 +35,7 @@ func (c *Cursor) LeafNodeInsert(ctx context.Context, key uint64, aRow *Row) erro
 	}
 	aPage.LeafNode.Header.Cells += 1
 
-	err = saveToCell(ctx, &aPage.LeafNode.Cells[c.CellIdx], key, aRow)
+	err = saveToCell(&aPage.LeafNode.Cells[c.CellIdx], key, aRow)
 	return err
 }
 
@@ -98,7 +98,7 @@ func (c *Cursor) LeafNodeSplitInsert(ctx context.Context, key uint64, aRow *Row)
 		destCell := &destPage.LeafNode.Cells[cellIdx]
 
 		if i == c.CellIdx {
-			if err := saveToCell(ctx, destCell, key, aRow); err != nil {
+			if err := saveToCell(destCell, key, aRow); err != nil {
 				return err
 			}
 		} else if i > c.CellIdx {
@@ -134,7 +134,7 @@ func (c *Cursor) LeafNodeSplitInsert(ctx context.Context, key uint64, aRow *Row)
 	return c.Table.InternalNodeInsert(ctx, parentPageIdx, newPageIdx)
 }
 
-func saveToCell(ctx context.Context, cell *Cell, key uint64, aRow *Row) error {
+func saveToCell(cell *Cell, key uint64, aRow *Row) error {
 	rowBuf, err := aRow.Marshal()
 	if err != nil {
 		return err
@@ -144,32 +144,42 @@ func saveToCell(ctx context.Context, cell *Cell, key uint64, aRow *Row) error {
 	return nil
 }
 
-func (c *Cursor) fetchRow(ctx context.Context) (Row, error) {
+func updateCell(cell *Cell, aRow *Row) error {
+	rowBuf, err := aRow.Marshal()
+	if err != nil {
+		return err
+	}
+	copy(cell.Value[:], rowBuf)
+	return nil
+}
+
+func (c *Cursor) fetchRow(ctx context.Context) (Row, *Cell, error) {
 	aPage, err := c.Table.pager.GetPage(ctx, c.Table, c.PageIdx)
 	if err != nil {
-		return Row{}, err
+		return Row{}, nil, err
 	}
 	aRow := NewRow(c.Table.Columns)
 
 	if err := UnmarshalRow(aPage.LeafNode.Cells[c.CellIdx].Value[:], &aRow); err != nil {
-		return Row{}, err
+		return Row{}, nil, err
 	}
+	destCell := &aPage.LeafNode.Cells[c.CellIdx]
 
 	// There are still more cells in the page, move cursor to next cell and return
 	if c.CellIdx < aPage.LeafNode.Header.Cells-1 {
 		c.CellIdx += 1
-		return aRow, nil
+		return aRow, destCell, nil
 	}
 
 	// If there is no leaf page to the right, set end of table flag and return
 	if aPage.LeafNode.Header.NextLeaf == 0 {
 		c.EndOfTable = true
-		return aRow, nil
+		return aRow, destCell, nil
 	}
 
 	// Otherwise, we try to move the cursor to the next leaf page
 	c.PageIdx = aPage.LeafNode.Header.NextLeaf
 	c.CellIdx = 0
 
-	return aRow, nil
+	return aRow, destCell, nil
 }

internal/pkg/minisql/delete.go

@@ -5,7 +5,7 @@ import (
 	"fmt"
 )
 
-func (t *Table) Delete(ctx context.Context, stmt Statement) error {
+func (t *Table) Delete(ctx context.Context, stmt Statement) (StatementResult, error) {
 	fmt.Println("TODO - implement DELETE")
-	return fmt.Errorf("not implemented")
+	return StatementResult{}, fmt.Errorf("not implemented")
 }

internal/pkg/minisql/insert_test.go

@@ -26,7 +26,7 @@ func TestTable_Insert(t *testing.T) {
 	stmt := Statement{
 		Kind:      Insert,
 		TableName: "foo",
-		Fields:    []string{"id", "email", "age"},
+		Fields:    columnNames(testColumns...),
 		Inserts:   [][]any{aRow.Values},
 	}
 
@@ -58,7 +58,7 @@ func TestTable_Insert_MultiInsert(t *testing.T) {
 	stmt := Statement{
 		Kind:      Insert,
 		TableName: "foo",
-		Fields:    []string{"id", "email", "age"},
+		Fields:    columnNames(testColumns...),
 		Inserts:   [][]any{aRow.Values, aRow2.Values, aRow3.Values},
 	}
 
@@ -116,7 +116,7 @@ func TestTable_Insert_SplitRootLeaf(t *testing.T) {
 	stmt := Statement{
 		Kind:      Insert,
 		TableName: "foo",
-		Fields:    []string{"id", "email", "age"},
+		Fields:    columnNames(testColumns...),
 		Inserts:   [][]any{aRow.Values},
 	}
 
@@ -198,19 +198,20 @@ func TestTable_Insert_SplitLeaf(t *testing.T) {
 		return old
 	}, nil)
 
-	// Insert test rows
+	// Batch insert test rows
+	stmt := Statement{
+		Kind:      Insert,
+		TableName: "foo",
+		Fields:    columnNames(testBigColumns...),
+		Inserts:   [][]any{},
+	}
 	for _, aRow := range rows {
-		stmt := Statement{
-			Kind:      Insert,
-			TableName: "foo",
-			Fields:    []string{"id", "email", "name", "description"},
-			Inserts:   [][]any{aRow.Values},
-		}
-
-		err := aTable.Insert(ctx, stmt)
-		require.NoError(t, err)
+		stmt.Inserts = append(stmt.Inserts, aRow.Values)
 	}
 
+	err := aTable.Insert(ctx, stmt)
+	require.NoError(t, err)
+
 	// Assert root node
 	assert.Equal(t, 3, int(aRootPage.InternalNode.Header.KeysNum))
 	assert.True(t, aRootPage.InternalNode.Header.IsRoot)
@@ -293,7 +294,7 @@ func TestTable_Insert_SplitInternalNode_CreateNewRoot(t *testing.T) {
 	stmt := Statement{
 		Kind:      Insert,
 		TableName: "foo",
-		Fields:    []string{"id", "email", "name", "description"},
+		Fields:    columnNames(testBigColumns...),
 		Inserts:   [][]any{},
 	}
 	for _, aRow := range rows {

internal/pkg/minisql/minisql_test.go

@@ -65,6 +65,14 @@ func init() {
 	}
 }
 
+func columnNames(columns ...Column) []string {
+	names := make([]string, 0, len(columns))
+	for _, aColumn := range columns {
+		names = append(names, aColumn.Name)
+	}
+	return names
+}
+
 type dataGen struct {
 	*gofakeit.Faker
 }
@@ -77,14 +85,6 @@ func newDataGen(seed uint64) *dataGen {
 	return &g
 }
 
-func (g *dataGen) Rows(number int) []Row {
-	rows := make([]Row, 0, number)
-	for i := 0; i < number; i++ {
-		rows = append(rows, g.Row())
-	}
-	return rows
-}
-
 func (g *dataGen) Row() Row {
 	return Row{
 		Columns: testColumns,
@@ -96,10 +96,20 @@ func (g *dataGen) Row() Row {
 	}
 }
 
-func (g *dataGen) BigRows(number int) []Row {
+func (g *dataGen) Rows(number int) []Row {
+	// Make sure all rows will have unique ID, this is important in some tests
+	idMap := map[int64]struct{}{}
 	rows := make([]Row, 0, number)
 	for i := 0; i < number; i++ {
-		rows = append(rows, g.BigRow())
+		aRow := g.Row()
+		_, ok := idMap[aRow.Values[0].(int64)]
+		for ok {
+			aRow = g.Row()
+			_, ok = idMap[aRow.Values[0].(int64)]
+		}
+		rows = append(rows, aRow)
+		idMap[aRow.Values[0].(int64)] = struct{}{}
+
 	}
 	return rows
 }
@@ -116,6 +126,23 @@ func (g *dataGen) BigRow() Row {
 	}
 }
 
+func (g *dataGen) BigRows(number int) []Row {
+	// Make sure all rows will have unique ID, this is important in some tests
+	idMap := map[int64]struct{}{}
+	rows := make([]Row, 0, number)
+	for i := 0; i < number; i++ {
+		aRow := g.BigRow()
+		_, ok := idMap[aRow.Values[0].(int64)]
+		for ok {
+			aRow = g.BigRow()
+			_, ok = idMap[aRow.Values[0].(int64)]
+		}
+		rows = append(rows, aRow)
+		idMap[aRow.Values[0].(int64)] = struct{}{}
+	}
+	return rows
+}
+
 func newInternalPageWithCells(iCells []ICell, rightChildIdx uint32) *Page {
 	aRoot := NewInternalNode()
 	aRoot.Header.KeysNum = uint32(len(iCells))

internal/pkg/minisql/minisqltest/minisqltest.go

@@ -29,6 +29,29 @@ var (
 			Name: "age",
 		},
 	}
+
+	testBigColumns = []minisql.Column{
+		{
+			Kind: minisql.Int8,
+			Size: 8,
+			Name: "id",
+		},
+		{
+			Kind: minisql.Varchar,
+			Size: 255,
+			Name: "name",
+		},
+		{
+			Kind: minisql.Varchar,
+			Size: 255,
+			Name: "email",
+		},
+		{
+			Kind: minisql.Varchar,
+			Size: minisql.PageSize - 6 - 8 - 4*8 - 8 - 255 - 255,
+			Name: "description",
+		},
+	}
 )
 
 type DataGen struct {
@@ -43,25 +66,64 @@ func NewDataGen(seed uint64) *DataGen {
 	return &g
 }
 
+func (g *DataGen) Row() minisql.Row {
+	return minisql.Row{
+		Columns: testColumns,
+		Values: []any{
+			g.Int64(),
+			g.Email(),
+			int32(g.IntRange(18, 100)),
+		},
+	}
+}
+
 func (g *DataGen) Rows(number int) []minisql.Row {
+	// Make sure all rows will have unique ID, this is important in some tests
+	idMap := map[int64]struct{}{}
 	rows := make([]minisql.Row, 0, number)
 	for i := 0; i < number; i++ {
-		rows = append(rows, g.Row())
+		aRow := g.Row()
+		_, ok := idMap[aRow.Values[0].(int64)]
+		for ok {
+			aRow = g.Row()
+			_, ok = idMap[aRow.Values[0].(int64)]
+		}
+		rows = append(rows, aRow)
+		idMap[aRow.Values[0].(int64)] = struct{}{}
+
 	}
 	return rows
 }
 
-func (g *DataGen) Row() minisql.Row {
+func (g *DataGen) BigRow() minisql.Row {
 	return minisql.Row{
-		Columns: testColumns,
+		Columns: testBigColumns,
 		Values: []any{
 			g.Int64(),
 			g.Email(),
-			int32(g.IntRange(18, 100)),
+			g.Name(),
+			g.Sentence(15),
 		},
 	}
 }
 
+func (g *DataGen) BigRows(number int) []minisql.Row {
+	// Make sure all rows will have unique ID, this is important in some tests
+	idMap := map[int64]struct{}{}
+	rows := make([]minisql.Row, 0, number)
+	for i := 0; i < number; i++ {
+		aRow := g.BigRow()
+		_, ok := idMap[aRow.Values[0].(int64)]
+		for ok {
+			aRow = g.BigRow()
+			_, ok = idMap[aRow.Values[0].(int64)]
+		}
+		rows = append(rows, aRow)
+		idMap[aRow.Values[0].(int64)] = struct{}{}
+	}
+	return rows
+}
+
 func (g *DataGen) NewRootLeafPageWithCells(cells, rowSize int) *minisql.Page {
 	aRootLeaf := minisql.NewLeafNode(uint64(rowSize))
 	aRootLeaf.Header.Header.IsRoot = true