ref #1096, reduce memory usage by about 50% for large data spreadsheet

cell.go

@@ -375,8 +375,18 @@ func (f *File) sharedStringsLoader() (err error) {
 	if path, ok := f.tempFiles.Load(defaultXMLPathSharedStrings); ok {
 		f.Pkg.Store(defaultXMLPathSharedStrings, f.readBytes(defaultXMLPathSharedStrings))
 		f.tempFiles.Delete(defaultXMLPathSharedStrings)
-		err = os.Remove(path.(string))
-		f.SharedStrings, f.sharedStringItemMap = nil, nil
+		if err = os.Remove(path.(string)); err != nil {
+			return
+		}
+		f.SharedStrings = nil
+	}
+	if f.sharedStringTemp != nil {
+		if err := f.sharedStringTemp.Close(); err != nil {
+			return err
+		}
+		f.tempFiles.Delete(defaultTempFileSST)
+		f.sharedStringItem, err = nil, os.Remove(f.sharedStringTemp.Name())
+		f.sharedStringTemp = nil
 	}
 	return
 }

cell_test.go

@@ -2,6 +2,7 @@ package excelize
 
 import (
 	"fmt"
+	"os"
 	"path/filepath"
 	"reflect"
 	"strconv"
@@ -653,9 +654,12 @@ func TestFormattedValue2(t *testing.T) {
 func TestSharedStringsError(t *testing.T) {
 	f, err := OpenFile(filepath.Join("test", "Book1.xlsx"), Options{UnzipXMLSizeLimit: 128})
 	assert.NoError(t, err)
+	tempFile, ok := f.tempFiles.Load(defaultXMLPathSharedStrings)
+	assert.True(t, ok)
 	f.tempFiles.Store(defaultXMLPathSharedStrings, "")
-	assert.Equal(t, "1", f.getFromStringItemMap(1))
-
+	assert.Equal(t, "1", f.getFromStringItem(1))
+	// Cleanup undelete temporary files
+	assert.NoError(t, os.Remove(tempFile.(string)))
 	// Test reload the file error on set cell cell and rich text. The error message was different between macOS and Windows.
 	err = f.SetCellValue("Sheet1", "A19", "A19")
 	assert.Error(t, err)
@@ -663,6 +667,50 @@ func TestSharedStringsError(t *testing.T) {
 	f.tempFiles.Store(defaultXMLPathSharedStrings, "")
 	err = f.SetCellRichText("Sheet1", "A19", []RichTextRun{})
 	assert.Error(t, err)
-
 	assert.NoError(t, f.Close())
+
+	f, err = OpenFile(filepath.Join("test", "Book1.xlsx"), Options{UnzipXMLSizeLimit: 128})
+	assert.NoError(t, err)
+	rows, err := f.Rows("Sheet1")
+	assert.NoError(t, err)
+	const maxUint16 = 1<<16 - 1
+	for rows.Next() {
+		if rows.CurrentRow() == 19 {
+			_, err := rows.Columns()
+			assert.NoError(t, err)
+			// Test get cell value from string item with invalid offset
+			f.sharedStringItem[1] = []uint{maxUint16 - 1, maxUint16}
+			assert.Equal(t, "1", f.getFromStringItem(1))
+			break
+		}
+	}
+	assert.NoError(t, rows.Close())
+	// Test shared string item temporary files has been closed before close the workbook
+	assert.NoError(t, f.sharedStringTemp.Close())
+	assert.Error(t, f.Close())
+	// Cleanup undelete temporary files
+	f.tempFiles.Range(func(k, v interface{}) bool {
+		return assert.NoError(t, os.Remove(v.(string)))
+	})
+
+	f, err = OpenFile(filepath.Join("test", "Book1.xlsx"), Options{UnzipXMLSizeLimit: 128})
+	assert.NoError(t, err)
+	rows, err = f.Rows("Sheet1")
+	assert.NoError(t, err)
+	for rows.Next() {
+		if rows.CurrentRow() == 19 {
+			_, err := rows.Columns()
+			assert.NoError(t, err)
+			break
+		}
+	}
+	assert.NoError(t, rows.Close())
+	assert.NoError(t, f.sharedStringTemp.Close())
+	// Test shared string item temporary files has been closed before set the cell value
+	assert.Error(t, f.SetCellValue("Sheet1", "A1", "A1"))
+	assert.Error(t, f.Close())
+	// Cleanup undelete temporary files
+	f.tempFiles.Range(func(k, v interface{}) bool {
+		return assert.NoError(t, os.Remove(v.(string)))
+	})
 }

excelize.go

@@ -32,30 +32,31 @@ import (
 // File define a populated spreadsheet file struct.
 type File struct {
 	sync.Mutex
-	options             *Options
-	xmlAttr             map[string][]xml.Attr
-	checked             map[string]bool
-	sheetMap            map[string]string
-	streams             map[string]*StreamWriter
-	tempFiles           sync.Map
-	CalcChain           *xlsxCalcChain
-	Comments            map[string]*xlsxComments
-	ContentTypes        *xlsxTypes
-	Drawings            sync.Map
-	Path                string
-	SharedStrings       *xlsxSST
-	sharedStringsMap    map[string]int
-	sharedStringItemMap *sync.Map
-	Sheet               sync.Map
-	SheetCount          int
-	Styles              *xlsxStyleSheet
-	Theme               *xlsxTheme
-	DecodeVMLDrawing    map[string]*decodeVmlDrawing
-	VMLDrawing          map[string]*vmlDrawing
-	WorkBook            *xlsxWorkbook
-	Relationships       sync.Map
-	Pkg                 sync.Map
-	CharsetReader       charsetTranscoderFn
+	options          *Options
+	xmlAttr          map[string][]xml.Attr
+	checked          map[string]bool
+	sheetMap         map[string]string
+	streams          map[string]*StreamWriter
+	tempFiles        sync.Map
+	CalcChain        *xlsxCalcChain
+	Comments         map[string]*xlsxComments
+	ContentTypes     *xlsxTypes
+	Drawings         sync.Map
+	Path             string
+	SharedStrings    *xlsxSST
+	sharedStringsMap map[string]int
+	sharedStringItem [][]uint
+	sharedStringTemp *os.File
+	Sheet            sync.Map
+	SheetCount       int
+	Styles           *xlsxStyleSheet
+	Theme            *xlsxTheme
+	DecodeVMLDrawing map[string]*decodeVmlDrawing
+	VMLDrawing       map[string]*vmlDrawing
+	WorkBook         *xlsxWorkbook
+	Relationships    sync.Map
+	Pkg              sync.Map
+	CharsetReader    charsetTranscoderFn
 }
 
 type charsetTranscoderFn func(charset string, input io.Reader) (rdr io.Reader, err error)

file.go

@@ -85,6 +85,11 @@ func (f *File) SaveAs(name string, opt ...Options) error {
 // Close closes and cleanup the open temporary file for the spreadsheet.
 func (f *File) Close() error {
 	var err error
+	if f.sharedStringTemp != nil {
+		if err := f.sharedStringTemp.Close(); err != nil {
+			return err
+		}
+	}
 	f.tempFiles.Range(func(k, v interface{}) bool {
 		if err = os.Remove(v.(string)); err != nil {
 			return false

rows.go

@@ -16,12 +16,12 @@ import (
 	"encoding/xml"
 	"fmt"
 	"io"
+	"io/ioutil"
 	"log"
 	"math"
 	"math/big"
 	"os"
 	"strconv"
-	"sync"
 
 	"github.com/mohae/deepcopy"
 )
@@ -280,23 +280,30 @@ func (f *File) Rows(sheet string) (*Rows, error) {
 	return &rows, nil
 }
 
-// getFromStringItemMap build shared string item map from system temporary
+// getFromStringItem build shared string item offset list from system temporary
 // file at one time, and return value by given to string index.
-func (f *File) getFromStringItemMap(index int) string {
-	if f.sharedStringItemMap != nil {
-		if value, ok := f.sharedStringItemMap.Load(index); ok {
-			return value.(string)
+func (f *File) getFromStringItem(index int) string {
+	if f.sharedStringTemp != nil {
+		if len(f.sharedStringItem) <= index {
+			return strconv.Itoa(index)
 		}
-		return strconv.Itoa(index)
+		offsetRange := f.sharedStringItem[index]
+		buf := make([]byte, offsetRange[1]-offsetRange[0])
+		if _, err := f.sharedStringTemp.ReadAt(buf, int64(offsetRange[0])); err != nil {
+			return strconv.Itoa(index)
+		}
+		return string(buf)
 	}
-	f.sharedStringItemMap = &sync.Map{}
 	needClose, decoder, tempFile, err := f.xmlDecoder(defaultXMLPathSharedStrings)
 	if needClose && err == nil {
 		defer tempFile.Close()
 	}
+	f.sharedStringItem = [][]uint{}
+	f.sharedStringTemp, _ = ioutil.TempFile(os.TempDir(), "excelize-")
+	f.tempFiles.Store(defaultTempFileSST, f.sharedStringTemp.Name())
 	var (
 		inElement string
-		i         int
+		i, offset uint
 	)
 	for {
 		token, _ := decoder.Token()
@@ -309,12 +316,16 @@ func (f *File) getFromStringItemMap(index int) string {
 			if inElement == "si" {
 				si := xlsxSI{}
 				_ = decoder.DecodeElement(&si, &xmlElement)
-				f.sharedStringItemMap.Store(i, si.String())
+
+				startIdx := offset
+				n, _ := f.sharedStringTemp.WriteString(si.String())
+				offset += uint(n)
+				f.sharedStringItem = append(f.sharedStringItem, []uint{startIdx, offset})
 				i++
 			}
 		}
 	}
-	return f.getFromStringItemMap(index)
+	return f.getFromStringItem(index)
 }
 
 // xmlDecoder creates XML decoder by given path in the zip from memory data
@@ -454,7 +465,7 @@ func (c *xlsxC) getValueFrom(f *File, d *xlsxSST, raw bool) (string, error) {
 			xlsxSI := 0
 			xlsxSI, _ = strconv.Atoi(c.V)
 			if _, ok := f.tempFiles.Load(defaultXMLPathSharedStrings); ok {
-				return f.formattedValue(c.S, f.getFromStringItemMap(xlsxSI), raw), nil
+				return f.formattedValue(c.S, f.getFromStringItem(xlsxSI), raw), nil
 			}
 			if len(d.SI) > xlsxSI {
 				return f.formattedValue(c.S, d.SI[xlsxSI].String(), raw), nil

templates.go

@@ -30,6 +30,7 @@ const (
 	defaultXMLPathSharedStrings = "xl/sharedStrings.xml"
 	defaultXMLPathStyles        = "xl/styles.xml"
 	defaultXMLPathWorkbook      = "xl/workbook.xml"
+	defaultTempFileSST          = "sharedStrings"
 )
 
 const templateDocpropsApp = `<Properties xmlns="http://schemas.openxmlformats.org/officeDocument/2006/extended-properties" xmlns:vt="http://schemas.openxmlformats.org/officeDocument/2006/docPropsVTypes"><TotalTime>0</TotalTime><Application>Go Excelize</Application></Properties>`