Regular expression extension. (#114)

Author: ncruces

README.md

@@ -45,12 +45,16 @@ Go, wazero and [`x/sys`](https://pkg.go.dev/golang.org/x/sys) are the _only_ run
   reads data [line-by-line](https://github.com/asg017/sqlite-lines).
 - [`github.com/ncruces/go-sqlite3/ext/pivot`](https://pkg.go.dev/github.com/ncruces/go-sqlite3/ext/pivot)
   creates [pivot tables](https://github.com/jakethaw/pivot_vtab).
+- [`github.com/ncruces/go-sqlite3/ext/regexp`](https://pkg.go.dev/github.com/ncruces/go-sqlite3/ext/regexp)
+  provides regular expression functions.
 - [`github.com/ncruces/go-sqlite3/ext/statement`](https://pkg.go.dev/github.com/ncruces/go-sqlite3/ext/statement)
   creates [parameterized views](https://github.com/0x09/sqlite-statement-vtab).
 - [`github.com/ncruces/go-sqlite3/ext/stats`](https://pkg.go.dev/github.com/ncruces/go-sqlite3/ext/stats)
   provides [statistics](https://www.oreilly.com/library/view/sql-in-a/9780596155322/ch04s02.html) functions.
 - [`github.com/ncruces/go-sqlite3/ext/unicode`](https://pkg.go.dev/github.com/ncruces/go-sqlite3/ext/unicode)
   provides [Unicode aware](https://sqlite.org/src/dir/ext/icu) functions.
+- [`github.com/ncruces/go-sqlite3/ext/uuid`](https://pkg.go.dev/github.com/ncruces/go-sqlite3/ext/uuid)
+  generates [UUIDs](https://en.wikipedia.org/wiki/Universally_unique_identifier).
 - [`github.com/ncruces/go-sqlite3/ext/zorder`](https://pkg.go.dev/github.com/ncruces/go-sqlite3/ext/zorder)
   maps multidimensional data to one dimension.
 - [`github.com/ncruces/go-sqlite3/vfs/adiantum`](https://pkg.go.dev/github.com/ncruces/go-sqlite3/vfs/adiantum)

ext/regexp/regexp.go

@@ -0,0 +1,77 @@
+// Package regexp provides additional regular expression functions.
+//
+// It provides the following Unicode aware functions:
+//   - regexp_like(),
+//   - regexp_substr(),
+//   - regexp_replace(),
+//   - and a REGEXP operator.
+//
+// The implementation uses Go [regexp/syntax] for regular expressions.
+//
+// https://github.com/nalgeon/sqlean/blob/main/docs/regexp.md
+package regexp
+
+import (
+	"regexp"
+
+	"github.com/ncruces/go-sqlite3"
+)
+
+// Register registers Unicode aware functions for a database connection.
+func Register(db *sqlite3.Conn) {
+	flags := sqlite3.DETERMINISTIC | sqlite3.INNOCUOUS
+
+	db.CreateFunction("regexp", 2, flags, regex)
+	db.CreateFunction("regexp_like", 2, flags, regexLike)
+	db.CreateFunction("regexp_substr", 2, flags, regexSubstr)
+	db.CreateFunction("regexp_replace", 3, flags, regexReplace)
+}
+
+func load(ctx sqlite3.Context, i int, expr string) (*regexp.Regexp, error) {
+	re, ok := ctx.GetAuxData(i).(*regexp.Regexp)
+	if !ok {
+		r, err := regexp.Compile(expr)
+		if err != nil {
+			return nil, err
+		}
+		re = r
+		ctx.SetAuxData(0, r)
+	}
+	return re, nil
+}
+
+func regex(ctx sqlite3.Context, arg ...sqlite3.Value) {
+	re, err := load(ctx, 0, arg[0].Text())
+	if err != nil {
+		ctx.ResultError(err)
+	} else {
+		ctx.ResultBool(re.Match(arg[1].RawText()))
+	}
+}
+
+func regexLike(ctx sqlite3.Context, arg ...sqlite3.Value) {
+	re, err := load(ctx, 1, arg[1].Text())
+	if err != nil {
+		ctx.ResultError(err)
+	} else {
+		ctx.ResultBool(re.Match(arg[0].RawText()))
+	}
+}
+
+func regexSubstr(ctx sqlite3.Context, arg ...sqlite3.Value) {
+	re, err := load(ctx, 1, arg[1].Text())
+	if err != nil {
+		ctx.ResultError(err)
+	} else {
+		ctx.ResultRawText(re.Find(arg[0].RawText()))
+	}
+}
+
+func regexReplace(ctx sqlite3.Context, arg ...sqlite3.Value) {
+	re, err := load(ctx, 1, arg[1].Text())
+	if err != nil {
+		ctx.ResultError(err)
+	} else {
+		ctx.ResultRawText(re.ReplaceAll(arg[0].RawText(), arg[2].RawText()))
+	}
+}

ext/regexp/regexp_test.go

@@ -0,0 +1,75 @@
+package regexp
+
+import (
+	"testing"
+
+	"github.com/ncruces/go-sqlite3"
+	"github.com/ncruces/go-sqlite3/driver"
+	_ "github.com/ncruces/go-sqlite3/embed"
+	_ "github.com/ncruces/go-sqlite3/internal/testcfg"
+)
+
+func TestRegister(t *testing.T) {
+	t.Parallel()
+
+	db, err := driver.Open(":memory:", func(conn *sqlite3.Conn) error {
+		Register(conn)
+		return nil
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer db.Close()
+
+	tests := []struct {
+		test string
+		want string
+	}{
+		{`'Hello' REGEXP 'elo'`, "0"},
+		{`'Hello' REGEXP 'ell'`, "1"},
+		{`'Hello' REGEXP 'el.'`, "1"},
+		{`regexp_like('Hello', 'elo')`, "0"},
+		{`regexp_like('Hello', 'ell')`, "1"},
+		{`regexp_like('Hello', 'el.')`, "1"},
+		{`regexp_substr('Hello', 'el.')`, "ell"},
+		{`regexp_replace('Hello', 'llo', 'll')`, "Hell"},
+	}
+
+	for _, tt := range tests {
+		var got string
+		err := db.QueryRow(`SELECT ` + tt.test).Scan(&got)
+		if err != nil {
+			t.Fatal(err)
+		}
+		if got != tt.want {
+			t.Errorf("got %q, want %q", got, tt.want)
+		}
+	}
+}
+
+func TestRegister_errors(t *testing.T) {
+	t.Parallel()
+
+	db, err := driver.Open(":memory:", func(conn *sqlite3.Conn) error {
+		Register(conn)
+		return nil
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer db.Close()
+
+	tests := []string{
+		`'' REGEXP ?`,
+		`regexp_like('', ?)`,
+		`regexp_substr('', ?)`,
+		`regexp_replace('', ?, '')`,
+	}
+
+	for _, tt := range tests {
+		err := db.QueryRow(`SELECT `+tt, `\`).Scan(nil)
+		if err == nil {
+			t.Fatal("want error")
+		}
+	}
+}

ext/unicode/unicode.go

@@ -111,7 +111,7 @@ func regex(ctx sqlite3.Context, arg ...sqlite3.Value) {
 			return
 		}
 		re = r
-		ctx.SetAuxData(0, re)
+		ctx.SetAuxData(0, r)
 	}
 	ctx.ResultBool(re.Match(arg[1].RawText()))
 }

ext/uuid/uuid_test.go

@@ -10,6 +10,8 @@ import (
 )
 
 func Test_generate(t *testing.T) {
+	t.Parallel()
+
 	db, err := driver.Open(":memory:", func(conn *sqlite3.Conn) error {
 		Register(conn)
 		return nil
@@ -130,6 +132,8 @@ func Test_generate(t *testing.T) {
 }
 
 func Test_convert(t *testing.T) {
+	t.Parallel()
+
 	db, err := driver.Open(":memory:", func(conn *sqlite3.Conn) error {
 		Register(conn)
 		return nil

func_test.go

@@ -130,8 +130,8 @@ func ExampleContext_SetAuxData() {
 				ctx.ResultError(err)
 				return
 			}
-			ctx.SetAuxData(0, r)
 			re = r
+			ctx.SetAuxData(0, r)
 		}
 		ctx.ResultBool(re.Match(arg[1].RawText()))
 	})