import into: enable [] glob matching

pkg/executor/importer/import.go

@@ -1050,7 +1050,7 @@ func (e *LoadDataController) InitDataFiles(ctx context.Context) error {
 		fileNameKey = strings.Trim(u.Path, "/")
 	}
 	// try to find pattern error in advance
-	_, err2 = filepath.Match(stringutil.EscapeGlobExceptAsterisk(fileNameKey), "")
+	_, err2 = filepath.Match(stringutil.EscapeGlobQuestionMark(fileNameKey), "")
 	if err2 != nil {
 		return exeerrors.ErrLoadDataInvalidURI.GenWithStackByArgs(plannercore.ImportIntoDataSource,
 			"Glob pattern error: "+err2.Error())
@@ -1063,7 +1063,8 @@ func (e *LoadDataController) InitDataFiles(ctx context.Context) error {
 	s := e.dataStore
 	var totalSize int64
 	dataFiles := []*mydump.SourceFileMeta{}
-	idx := strings.IndexByte(fileNameKey, '*')
+	// check glob pattern is present in filename.
+	idx := strings.IndexAny(fileNameKey, "*[")
 	// simple path when the path represent one file
 	sourceType := e.getSourceType()
 	if idx == -1 {
@@ -1098,7 +1099,7 @@ func (e *LoadDataController) InitDataFiles(ctx context.Context) error {
 		// when import from server disk, all entries in parent directory should have READ
 		// access, else walkDir will fail
 		// we only support '*', in order to reuse glob library manually escape the path
-		escapedPath := stringutil.EscapeGlobExceptAsterisk(fileNameKey)
+		escapedPath := stringutil.EscapeGlobQuestionMark(fileNameKey)
 		err := s.WalkDir(ctx, &storage.WalkOption{ObjPrefix: commonPrefix, SkipSubDir: true},
 			func(remotePath string, size int64) error {
 				// we have checked in LoadDataExec.Next

pkg/executor/importer/import_test.go

@@ -414,6 +414,28 @@ func TestSupportedSuffixForServerDisk(t *testing.T) {
 	require.NoError(t, os.Chmod(path.Join(tempDir, "no-perm"), 0o400))
 	c.Path = path.Join(tempDir, "server-*.csv")
 	require.NoError(t, c.InitDataFiles(ctx))
+	// test glob matching pattern [12]
+	err = os.WriteFile(path.Join(tempDir, "glob-1.csv"), []byte("1,1"), 0o644)
+	require.NoError(t, err)
+	err = os.WriteFile(path.Join(tempDir, "glob-2.csv"), []byte("2,2"), 0o644)
+	require.NoError(t, err)
+	err = os.WriteFile(path.Join(tempDir, "glob-3.csv"), []byte("3,3"), 0o644)
+	require.NoError(t, err)
+	c.Path = path.Join(tempDir, "glob-[12].csv")
+	require.NoError(t, c.InitDataFiles(ctx))
+	gotPath := make([]string, 0, len(c.dataFiles))
+	for _, f := range c.dataFiles {
+		gotPath = append(gotPath, f.Path)
+	}
+	require.ElementsMatch(t, []string{"glob-1.csv", "glob-2.csv"}, gotPath)
+	// test glob matching pattern [2-3]
+	c.Path = path.Join(tempDir, "glob-[2-3].csv")
+	require.NoError(t, c.InitDataFiles(ctx))
+	gotPath = make([]string, 0, len(c.dataFiles))
+	for _, f := range c.dataFiles {
+		gotPath = append(gotPath, f.Path)
+	}
+	require.ElementsMatch(t, []string{"glob-2.csv", "glob-3.csv"}, gotPath)
 }
 
 func TestGetDataSourceType(t *testing.T) {

pkg/executor/test/loadremotetest/multi_file_test.go

@@ -54,14 +54,6 @@ func (s *mockGCSSuite) TestFilenameAsterisk() {
 		Content: []byte("5\ttest5\n" +
 			"6\ttest6"),
 	})
-	s.server.CreateObject(fakestorage.Object{
-		ObjectAttrs: fakestorage.ObjectAttrs{
-			BucketName: "test-multi-load",
-			Name:       "not.me.[1-9].tsv",
-		},
-		Content: []byte("7\ttest7\n" +
-			"8\ttest8"),
-	})
 	s.server.CreateObject(fakestorage.Object{
 		ObjectAttrs: fakestorage.ObjectAttrs{
 			BucketName: "not-me",
@@ -88,14 +80,14 @@ func (s *mockGCSSuite) TestFilenameAsterisk() {
 		"2 test2", "4 test4", "6 test6",
 	))
 
-	// only '*' is supported in pattern matching
+	// only `*` and `[]` is supported in pattern matching
 	s.tk.MustExec("TRUNCATE TABLE multi_load.t;")
-	sql = fmt.Sprintf(`LOAD DATA INFILE 'gs://test-multi-load/not.me.[1-9].tsv?endpoint=%s'
+	sql = fmt.Sprintf(`LOAD DATA INFILE 'gs://test-multi-load/db.tbl.00[13].tsv?endpoint=%s'
 		INTO TABLE multi_load.t with thread=1;`, gcsEndpoint)
 	s.tk.MustExec(sql)
 	s.Equal(uint64(0), s.tk.Session().GetSessionVars().StmtCtx.LastInsertID)
 	s.tk.MustQuery("SELECT * FROM multi_load.t;").Check(testkit.Rows(
-		"7 test7", "8 test8",
+		"1 test1", "2 test2", "5 test5", "6 test6",
 	))
 }
 

pkg/util/stringutil/string_util.go

@@ -546,13 +546,12 @@ func LowerOneStringExcludeEscapeChar(str []byte, escapeChar byte) byte {
 	return actualEscapeChar
 }
 
-// EscapeGlobExceptAsterisk escapes '?', '[', ']' for a glob path pattern.
-func EscapeGlobExceptAsterisk(s string) string {
+// EscapeGlobQuestionMark escapes '?' for a glob path pattern.
+func EscapeGlobQuestionMark(s string) string {
 	var buf strings.Builder
 	buf.Grow(len(s))
 	for _, c := range s {
-		switch c {
-		case '?', '[', ']':
+		if c == '?' {
 			buf.WriteByte('\\')
 		}
 		buf.WriteRune(c)

pkg/util/stringutil/string_util_test.go

@@ -194,15 +194,15 @@ func TestBuildStringFromLabels(t *testing.T) {
 	}
 }
 
-func TestEscapeGlobExceptAsterisk(t *testing.T) {
+func TestEscapeGlobQuestionMark(t *testing.T) {
 	cases := [][2]string{
 		{"123", "123"},
 		{"12*3", "12*3"},
 		{"12?", `12\?`},
-		{`[1-2]`, `\[1-2\]`},
+		{`[1-2]`, `[1-2]`},
 	}
 	for _, pair := range cases {
-		require.Equal(t, pair[1], EscapeGlobExceptAsterisk(pair[0]))
+		require.Equal(t, pair[1], EscapeGlobQuestionMark(pair[0]))
 	}
 }