support setting whether the data table in csv / xls / xlsx files contains a header row

2026-05-18 00:34:28 +08:00 · 2025-08-08 20:11:31 +08:00
parent 351cebe169
commit ecf6fbd187
13 changed files with 688 additions and 105 deletions
@@ -13,7 +13,8 @@ import (

 // CsvFileBasicDataTable defines the structure of csv data table
 type CsvFileBasicDataTable struct {
-	allLines [][]string
+	allLines     [][]string
+	hasTitleLine bool
 }

 // CsvFileBasicDataTableRow defines the structure of csv data table row
@@ -34,7 +35,11 @@ func (t *CsvFileBasicDataTable) DataRowCount() int {
 		return 0
 	}

-	return len(t.allLines) - 1
+	if t.hasTitleLine {
+		return len(t.allLines) - 1
+	} else {
+		return len(t.allLines)
+	}
 }

 // HeaderColumnNames returns the header column name list
@@ -43,14 +48,24 @@ func (t *CsvFileBasicDataTable) HeaderColumnNames() []string {
 		return nil
 	}

-	return t.allLines[0]
+	if t.hasTitleLine {
+		return t.allLines[0]
+	} else {
+		return nil
+	}
 }

 // DataRowIterator returns the iterator of data row
 func (t *CsvFileBasicDataTable) DataRowIterator() datatable.BasicDataTableRowIterator {
+	startIndex := -1
+
+	if t.hasTitleLine {
+		startIndex = 0
+	}
+
 	return &CsvFileBasicDataTableRowIterator{
 		dataTable:    t,
-		currentIndex: 0,
+		currentIndex: startIndex,
 	}
 }

@@ -95,18 +110,19 @@ func (t *CsvFileBasicDataTableRowIterator) Next() datatable.BasicDataTableRow {
 }

 // CreateNewCsvBasicDataTable returns comma separated values data table by io readers
-func CreateNewCsvBasicDataTable(ctx core.Context, reader io.Reader) (datatable.BasicDataTable, error) {
-	return createNewCsvFileBasicDataTable(ctx, reader, ',')
+func CreateNewCsvBasicDataTable(ctx core.Context, reader io.Reader, hasTitleLine bool) (datatable.BasicDataTable, error) {
+	return createNewCsvFileBasicDataTable(ctx, reader, ',', hasTitleLine)
 }

 // CreateNewCustomCsvBasicDataTable returns character separated values data table by io readers
-func CreateNewCustomCsvBasicDataTable(allLines [][]string) datatable.BasicDataTable {
+func CreateNewCustomCsvBasicDataTable(allLines [][]string, hasTitleLine bool) datatable.BasicDataTable {
 	return &CsvFileBasicDataTable{
-		allLines: allLines,
+		allLines:     allLines,
+		hasTitleLine: hasTitleLine,
 	}
 }

-func createNewCsvFileBasicDataTable(ctx core.Context, reader io.Reader, separator rune) (*CsvFileBasicDataTable, error) {
+func createNewCsvFileBasicDataTable(ctx core.Context, reader io.Reader, separator rune, hasTitleLine bool) (*CsvFileBasicDataTable, error) {
 	csvReader := csv.NewReader(reader)
 	csvReader.Comma = separator
 	csvReader.FieldsPerRecord = -1
@@ -133,6 +149,7 @@ func createNewCsvFileBasicDataTable(ctx core.Context, reader io.Reader, separato
 	}

 	return &CsvFileBasicDataTable{
-		allLines: allLines,
+		allLines:     allLines,
+		hasTitleLine: hasTitleLine,
 	}, nil
 }
@@ -14,7 +14,17 @@ func TestCsvFileBasicDataTableDataRowCount(t *testing.T) {
 		{"A1", "B1", "C1"},
 		{"A2", "B2", "C2"},
 		{"A3", "B3", "C3"},
-	})
+	}, false)
+
+	assert.Equal(t, 3, datatable.DataRowCount())
+}
+
+func TestCsvFileBasicDataTableDataRowCount_HasTitleLine(t *testing.T) {
+	datatable := CreateNewCustomCsvBasicDataTable([][]string{
+		{"A1", "B1", "C1"},
+		{"A2", "B2", "C2"},
+		{"A3", "B3", "C3"},
+	}, true)

 	assert.Equal(t, 2, datatable.DataRowCount())
 }
@@ -22,14 +32,16 @@ func TestCsvFileBasicDataTableDataRowCount(t *testing.T) {
 func TestCsvFileBasicDataTableDataRowCount_OnlyHeaderLine(t *testing.T) {
 	datatable := CreateNewCustomCsvBasicDataTable([][]string{
 		{"A1", "B1", "C1"},
-	})
+	}, true)

 	assert.Equal(t, 0, datatable.DataRowCount())
 }

 func TestCsvFileBasicDataTableDataRowCount_EmptyContent(t *testing.T) {
-	datatable := CreateNewCustomCsvBasicDataTable([][]string{})
+	datatable := CreateNewCustomCsvBasicDataTable([][]string{}, false)
+	assert.Equal(t, 0, datatable.DataRowCount())

+	datatable = CreateNewCustomCsvBasicDataTable([][]string{}, true)
 	assert.Equal(t, 0, datatable.DataRowCount())
 }

@@ -38,14 +50,16 @@ func TestCsvFileBasicDataTableHeaderColumnNames(t *testing.T) {
 		{"A1", "B1", "C1"},
 		{"A2", "B2", "C2"},
 		{"A3", "B3", "C3"},
-	})
+	}, true)

 	assert.EqualValues(t, []string{"A1", "B1", "C1"}, datatable.HeaderColumnNames())
 }

 func TestCsvFileBasicDataTableHeaderColumnNames_EmptyContent(t *testing.T) {
-	datatable := CreateNewCustomCsvBasicDataTable([][]string{})
+	datatable := CreateNewCustomCsvBasicDataTable([][]string{}, false)
+	assert.Nil(t, datatable.HeaderColumnNames())

+	datatable = CreateNewCustomCsvBasicDataTable([][]string{}, true)
 	assert.Nil(t, datatable.HeaderColumnNames())
 }

@@ -54,7 +68,34 @@ func TestCsvFileBasicDataTableRowIterator(t *testing.T) {
 		{"A1", "B1", "C1"},
 		{"A2", "B2", "C2"},
 		{"A3", "B3", "C3"},
-	})
+	}, false)
+
+	iterator := datatable.DataRowIterator()
+	assert.True(t, iterator.HasNext())
+
+	// data row 1
+	assert.NotNil(t, iterator.Next())
+	assert.True(t, iterator.HasNext())
+
+	// data row 2
+	assert.NotNil(t, iterator.Next())
+	assert.True(t, iterator.HasNext())
+
+	// data row 3
+	assert.NotNil(t, iterator.Next())
+	assert.False(t, iterator.HasNext())
+
+	// not existed data row 4
+	assert.Nil(t, iterator.Next())
+	assert.False(t, iterator.HasNext())
+}
+
+func TestCsvFileBasicDataTableRowIterator_HasTitleLine(t *testing.T) {
+	datatable := CreateNewCustomCsvBasicDataTable([][]string{
+		{"A1", "B1", "C1"},
+		{"A2", "B2", "C2"},
+		{"A3", "B3", "C3"},
+	}, true)

 	iterator := datatable.DataRowIterator()
 	assert.True(t, iterator.HasNext())
@@ -81,7 +122,7 @@ func TestCsvFileBasicDataTableRowColumnCount(t *testing.T) {
 		{"A1", "B1", "C1"},
 		{"A2", "B2", "C2"},
 		{"A3", "B3", "C3"},
-	})
+	}, true)

 	iterator := datatable.DataRowIterator()

@@ -97,7 +138,32 @@ func TestCsvFileBasicDataTableRowGetData(t *testing.T) {
 		{"A1", "B1", "C1"},
 		{"A2", "B2", "C2"},
 		{"A3", "B3", "C3"},
-	})
+	}, false)
+
+	iterator := datatable.DataRowIterator()
+
+	row1 := iterator.Next()
+	assert.Equal(t, "A1", row1.GetData(0))
+	assert.Equal(t, "B1", row1.GetData(1))
+	assert.Equal(t, "C1", row1.GetData(2))
+
+	row2 := iterator.Next()
+	assert.Equal(t, "A2", row2.GetData(0))
+	assert.Equal(t, "B2", row2.GetData(1))
+	assert.Equal(t, "C2", row2.GetData(2))
+
+	row3 := iterator.Next()
+	assert.Equal(t, "A3", row3.GetData(0))
+	assert.Equal(t, "B3", row3.GetData(1))
+	assert.Equal(t, "C3", row3.GetData(2))
+}
+
+func TestCsvFileBasicDataTableRowGetData_HasTitleLine(t *testing.T) {
+	datatable := CreateNewCustomCsvBasicDataTable([][]string{
+		{"A1", "B1", "C1"},
+		{"A2", "B2", "C2"},
+		{"A3", "B3", "C3"},
+	}, true)

 	iterator := datatable.DataRowIterator()

@@ -117,7 +183,7 @@ func TestCsvFileBasicDataTableRowGetData_GetNotExistedColumnData(t *testing.T) {
 		{"A1", "B1", "C1"},
 		{"A2", "B2", "C2"},
 		{"A3", "B3", "C3"},
-	})
+	}, true)

 	iterator := datatable.DataRowIterator()

@@ -130,7 +196,7 @@ func TestCreateNewCsvBasicDataTable(t *testing.T) {
 	reader := bytes.NewReader([]byte("A1,B1,C1\n" +
 		"A2,B2,C2\n" +
 		"A3,B3,C3\n"))
-	datatable, err := CreateNewCsvBasicDataTable(context, reader)
+	datatable, err := CreateNewCsvBasicDataTable(context, reader, true)
 	assert.Nil(t, err)

 	assert.Equal(t, 2, datatable.DataRowCount())
@@ -160,7 +226,7 @@ func TestCreateNewCsvBasicDataTable_SkipBlankLine(t *testing.T) {
 		"A2,B2,C2\n" +
 		"\n" +
 		"A3,B3,C3\n"))
-	datatable, err := CreateNewCsvBasicDataTable(context, reader)
+	datatable, err := CreateNewCsvBasicDataTable(context, reader, true)
 	assert.Nil(t, err)

 	assert.Equal(t, 2, datatable.DataRowCount())