CSV type affinity (#102)

Use sqlite-createtable-parser compiled to Wasm to parse the CREATE TABLE statement.

ext/csv/arg_test.go

@@ -40,6 +40,8 @@ func Test_uintArg(t *testing.T) {
 }
 
 func Test_boolArg(t *testing.T) {
+	t.Parallel()
+
 	tests := []struct {
 		arg string
 		key string
@@ -76,6 +78,8 @@ func Test_boolArg(t *testing.T) {
 }
 
 func Test_runeArg(t *testing.T) {
+	t.Parallel()
+
 	tests := []struct {
 		arg string
 		key string

ext/csv/csv.go

@@ -12,6 +12,7 @@ import (
 	"fmt"
 	"io"
 	"io/fs"
+	"strconv"
 	"strings"
 
 	"github.com/ncruces/go-sqlite3"
@@ -93,6 +94,8 @@ func RegisterFS(db *sqlite3.Conn, fsys fs.FS) {
 				}
 			}
 			schema = getSchema(header, columns, row)
+		} else {
+			table.typs = getColumnAffinities(schema)
 		}
 
 		err = db.DeclareVTab(schema)
@@ -113,6 +116,7 @@ type table struct {
 	fsys   fs.FS
 	name   string
 	data   string
+	typs   []affinity
 	comma  rune
 	header bool
 }
@@ -226,7 +230,40 @@ func (c *cursor) RowID() (int64, error) {
 
 func (c *cursor) Column(ctx *sqlite3.Context, col int) error {
 	if col < len(c.row) {
-		ctx.ResultText(c.row[col])
+		var typ affinity
+		if col < len(c.table.typs) {
+			typ = c.table.typs[col]
+		}
+
+		txt := c.row[col]
+		if typ == blob {
+			ctx.ResultText(txt)
+			return nil
+		}
+		if txt == "" {
+			return nil
+		}
+
+		switch typ {
+		case numeric, integer:
+			if strings.TrimLeft(txt, "+-0123456789") == "" {
+				if i, err := strconv.ParseInt(txt, 10, 64); err == nil {
+					ctx.ResultInt64(i)
+					return nil
+				}
+			}
+			fallthrough
+		case real:
+			if strings.TrimLeft(txt, "+-.0123456789Ee") == "" {
+				if f, err := strconv.ParseFloat(txt, 64); err == nil {
+					ctx.ResultFloat(f)
+					return nil
+				}
+			}
+			fallthrough
+		case text:
+			ctx.ResultText(c.row[col])
+		}
 	}
 	return nil
 }

ext/csv/csv_test.go

@@ -113,6 +113,50 @@ Robert	"Griesemer"	"gri"`
 	}
 }
 
+func TestAffinity(t *testing.T) {
+	t.Parallel()
+
+	db, err := sqlite3.Open(":memory:")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer db.Close()
+
+	csv.Register(db)
+
+	const data = "01\n0.10\ne"
+	err = db.Exec(`
+		CREATE VIRTUAL TABLE temp.nums USING csv(
+			data   = ` + sqlite3.Quote(data) + `,
+			schema = 'CREATE TABLE x(a numeric)'
+		)`)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	stmt, _, err := db.Prepare(`SELECT * FROM temp.nums`)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer stmt.Close()
+
+	if stmt.Step() {
+		if got := stmt.ColumnText(0); got != "1" {
+			t.Errorf("got %q want 1", got)
+		}
+	}
+	if stmt.Step() {
+		if got := stmt.ColumnText(0); got != "0.1" {
+			t.Errorf("got %q want 0.1", got)
+		}
+	}
+	if stmt.Step() {
+		if got := stmt.ColumnText(0); got != "e" {
+			t.Errorf("got %q want e", got)
+		}
+	}
+}
+
 func TestRegister_errors(t *testing.T) {
 	t.Parallel()
 

ext/csv/types.go

@@ -0,0 +1,54 @@
+package csv
+
+import (
+	_ "embed"
+	"strings"
+
+	"github.com/ncruces/go-sqlite3/util/vtabutil"
+)
+
+type affinity byte
+
+const (
+	blob    affinity = 0
+	text    affinity = 1
+	numeric affinity = 2
+	integer affinity = 3
+	real    affinity = 4
+)
+
+func getColumnAffinities(schema string) []affinity {
+	tab, err := vtabutil.Parse(schema)
+	if err != nil {
+		return nil
+	}
+	defer tab.Close()
+
+	types := make([]affinity, tab.NumColumns())
+	for i := range types {
+		col := tab.Column(i)
+		types[i] = getAffinity(col.Type())
+	}
+	return types
+}
+
+func getAffinity(declType string) affinity {
+	// https://sqlite.org/datatype3.html#determination_of_column_affinity
+	if declType == "" {
+		return blob
+	}
+	name := strings.ToUpper(declType)
+	if strings.Contains(name, "INT") {
+		return integer
+	}
+	if strings.Contains(name, "CHAR") || strings.Contains(name, "CLOB") || strings.Contains(name, "TEXT") {
+		return text
+	}
+	if strings.Contains(name, "BLOB") {
+		return blob
+	}
+	if strings.Contains(name, "REAL") || strings.Contains(name, "FLOA") || strings.Contains(name, "DOUB") {
+		return real
+	}
+	return numeric
+}

ext/csv/types_test.go

@@ -0,0 +1,35 @@
+package csv
+
+import (
+	_ "embed"
+	"testing"
+)
+
+func Test_getAffinity(t *testing.T) {
+	tests := []struct {
+		decl string
+		want affinity
+	}{
+		{"", blob},
+		{"INTEGER", integer},
+		{"TINYINT", integer},
+		{"TEXT", text},
+		{"CHAR", text},
+		{"CLOB", text},
+		{"BLOB", blob},
+		{"REAL", real},
+		{"FLOAT", real},
+		{"DOUBLE", real},
+		{"NUMERIC", numeric},
+		{"DECIMAL", numeric},
+		{"BOOLEAN", numeric},
+		{"DATETIME", numeric},
+	}
+	for _, tt := range tests {
+		t.Run(tt.decl, func(t *testing.T) {
+			if got := getAffinity(tt.decl); got != tt.want {
+				t.Errorf("getAffinity() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}