Unicode tests, fixes.

Author: ncruces

ext/unicode/unicode.go

@@ -1,9 +1,17 @@
-// Package unicode provides a replacement for the SQLite ICU extension.
+// Package unicode provides an alternative to the SQLite ICU extension.
 //
 // Provides Unicode aware:
 //   - upper and lower functions,
-//   - LIKE and REGEX operators,
+//   - LIKE and REGEXP operators,
 //   - collation sequences.
+//
+// This package is not 100% compatible with the ICU extension:
+//   - upper and lower use [strings.ToUpper], [strings.ToLower] and [cases];
+//   - the LIKE operator follows [strings.EqualFold] rules;
+//   - the REGEXP operator uses Go [regex/syntax];
+//   - collation sequences use [collate].
+//
+// Expect subtle differences (e.g.) in the handling of Turkish case folding.
 package unicode
 
 import (
@@ -20,7 +28,7 @@ import (
 )
 
 // Register registers Unicode aware functions for a database connection.
-func Register(db sqlite3.Conn) {
+func Register(db *sqlite3.Conn) {
 	flags := sqlite3.DETERMINISTIC | sqlite3.INNOCUOUS
 
 	db.CreateFunction("like", 2, flags, like)
@@ -58,7 +66,7 @@ func upper(ctx sqlite3.Context, arg ...sqlite3.Value) {
 	}
 	cs, ok := ctx.GetAuxData(1).(cases.Caser)
 	if !ok {
-		t, err := language.Parse(arg[0].Text())
+		t, err := language.Parse(arg[1].Text())
 		if err != nil {
 			ctx.ResultError(err)
 			return
@@ -77,7 +85,7 @@ func lower(ctx sqlite3.Context, arg ...sqlite3.Value) {
 	}
 	cs, ok := ctx.GetAuxData(1).(cases.Caser)
 	if !ok {
-		t, err := language.Parse(arg[0].Text())
+		t, err := language.Parse(arg[1].Text())
 		if err != nil {
 			ctx.ResultError(err)
 			return
@@ -137,7 +145,8 @@ func like2regex(pattern string, escape rune) string {
 	var re strings.Builder
 	start := 0
 	literal := false
-	re.WriteString(`(?is)`) // case insensitive, . matches any character
+	re.Grow(len(pattern) + 10)
+	re.WriteString(`(?is)\A`) // case insensitive, . matches any character
 	for i, r := range pattern {
 		if start < 0 {
 			start = i
@@ -164,5 +173,6 @@ func like2regex(pattern string, escape rune) string {
 	if start >= 0 {
 		re.WriteString(regexp.QuoteMeta(pattern[start:]))
 	}
+	re.WriteString(`\z`)
 	return re.String()
 }

ext/unicode/unicode_test.go

@@ -1,25 +1,134 @@
 package unicode
 
-import "testing"
+import (
+	"errors"
+	"testing"
+
+	"github.com/ncruces/go-sqlite3"
+	_ "github.com/ncruces/go-sqlite3/embed"
+)
+
+func TestRegister(t *testing.T) {
+	t.Parallel()
+
+	db, err := sqlite3.Open(":memory:")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer db.Close()
+
+	exec := func(fn string) string {
+		stmt, _, err := db.Prepare(`SELECT ` + fn)
+		if err != nil {
+			t.Fatal(err)
+		}
+		defer stmt.Close()
+
+		if stmt.Step() {
+			return stmt.ColumnText(0)
+		}
+		t.Fatal(stmt.Err())
+		return ""
+	}
+
+	Register(db)
+
+	tests := []struct {
+		test string
+		want string
+	}{
+		{`upper('hello')`, "HELLO"},
+		{`lower('HELLO')`, "hello"},
+		{`upper('привет')`, "ПРИВЕТ"},
+		{`lower('ПРИВЕТ')`, "привет"},
+		{`upper('istanbul')`, "ISTANBUL"},
+		{`upper('istanbul', 'tr-TR')`, "İSTANBUL"},
+		{`lower('Dünyanın İlk Borsası', 'tr-TR')`, "dünyanın ilk borsası"},
+		{`upper('Dünyanın İlk Borsası', 'tr-TR')`, "DÜNYANIN İLK BORSASI"},
+		{`'Hello' REGEXP 'ell'`, "1"},
+		{`'Hello' REGEXP 'el.'`, "1"},
+		{`'Hello' LIKE 'hel_'`, "0"},
+		{`'Hello' LIKE 'hel%'`, "1"},
+		{`'Hello' LIKE 'h_llo'`, "1"},
+		{`'Hello' LIKE 'hello'`, "1"},
+		{`'Привет' LIKE 'ПРИВЕТ'`, "1"},
+		{`'100%' LIKE '100|%' ESCAPE '|'`, "1"},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.test, func(t *testing.T) {
+			if got := exec(tt.test); got != tt.want {
+				t.Errorf("exec(%q) = %q, want %q", tt.test, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestRegister_error(t *testing.T) {
+	t.Parallel()
+
+	db, err := sqlite3.Open(":memory:")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer db.Close()
+
+	Register(db)
+
+	err = db.Exec(`SELECT upper('hello', 'enUS')`)
+	if err == nil {
+		t.Error("want error")
+	}
+	if !errors.Is(err, sqlite3.ERROR) {
+		t.Errorf("got %v, want sqlite3.ERROR", err)
+	}
+
+	err = db.Exec(`SELECT lower('hello', 'enUS')`)
+	if err == nil {
+		t.Error("want error")
+	}
+	if !errors.Is(err, sqlite3.ERROR) {
+		t.Errorf("got %v, want sqlite3.ERROR", err)
+	}
+
+	err = db.Exec(`SELECT 'hello' REGEXP '\'`)
+	if err == nil {
+		t.Error("want error")
+	}
+	if !errors.Is(err, sqlite3.ERROR) {
+		t.Errorf("got %v, want sqlite3.ERROR", err)
+	}
+
+	err = db.Exec(`SELECT 'hello' LIKE 'HELLO' ESCAPE '\\' `)
+	if err == nil {
+		t.Error("want error")
+	}
+	if !errors.Is(err, sqlite3.ERROR) {
+		t.Errorf("got %v, want sqlite3.ERROR", err)
+	}
+}
 
 func Test_like2regex(t *testing.T) {
+	const prefix = `(?is)\A`
+	const sufix = `\z`
 	tests := []struct {
 		pattern string
 		escape  rune
 		want    string
 	}{
-		{`a`, -1, `(?is)a`},
-		{`a.`, -1, `(?is)a\.`},
-		{`a%`, -1, `(?is)a.*`},
-		{`a\`, -1, `(?is)a\\`},
-		{`a_b`, -1, `(?is)a.b`},
-		{`a|b`, '|', `(?is)ab`},
-		{`a|_`, '|', `(?is)a_`},
+		{`a`, -1, `a`},
+		{`a.`, -1, `a\.`},
+		{`a%`, -1, `a.*`},
+		{`a\`, -1, `a\\`},
+		{`a_b`, -1, `a.b`},
+		{`a|b`, '|', `ab`},
+		{`a|_`, '|', `a_`},
 	}
 	for _, tt := range tests {
 		t.Run(tt.pattern, func(t *testing.T) {
-			if got := like2regex(tt.pattern, tt.escape); got != tt.want {
-				t.Errorf("like2regex() = %v, want %v", got, tt.want)
+			want := prefix + tt.want + sufix
+			if got := like2regex(tt.pattern, tt.escape); got != want {
+				t.Errorf("like2regex() = %q, want %q", got, want)
 			}
 		})
 	}