From 2cfcf9eb5ff2f332dad0c6adead53ef0500001db Mon Sep 17 00:00:00 2001
From: xuri <xuri.me@gmail.com>
Date: Wed, 16 Jun 2021 15:03:50 +0000
Subject: encode the escaped string literal which not permitted in an XML 1.0
 document

---
 cell.go     |  3 ++-
 lib.go      | 46 ++++++++++++++++++++++++++++++++++++++++++----
 lib_test.go | 13 +++++++++++++
 3 files changed, 57 insertions(+), 5 deletions(-)

diff --git a/cell.go b/cell.go
index f94b81e..4dec093 100644
--- a/cell.go
+++ b/cell.go
@@ -288,6 +288,7 @@ func (f *File) setSharedString(val string) int {
 	}
 	sst.Count++
 	sst.UniqueCount++
+	val = bstrMarshal(val)
 	t := xlsxT{Val: val}
 	// Leading and ending space(s) character detection.
 	if len(val) > 0 && (val[0] == 32 || val[len(val)-1] == 32) {
@@ -315,7 +316,7 @@ func setCellStr(value string) (t string, v string, ns xml.Attr) {
 		}
 	}
 	t = "str"
-	v = value
+	v = bstrMarshal(value)
 	return
 }
 
diff --git a/lib.go b/lib.go
index baa62bf..5c9bbf6 100644
--- a/lib.go
+++ b/lib.go
@@ -456,6 +456,11 @@ func isNumeric(s string) (bool, int) {
 	return true, p
 }
 
+var (
+	bstrExp       = regexp.MustCompile(`_x[a-zA-Z\d]{4}_`)
+	bstrEscapeExp = regexp.MustCompile(`x[a-zA-Z\d]{4}_`)
+)
+
 // bstrUnmarshal parses the binary basic string, this will trim escaped string
 // literal which not permitted in an XML 1.0 document. The basic string
 // variant type can store any valid Unicode character. Unicode characters
@@ -468,15 +473,13 @@ func isNumeric(s string) (bool, int) {
 // initial underscore shall itself be escaped (i.e. stored as _x005F_). For
 // example: The string literal _x0008_ would be stored as _x005F_x0008_.
 func bstrUnmarshal(s string) (result string) {
-	bstrExp := regexp.MustCompile(`_x[a-zA-Z0-9]{4}_`)
-	escapeExp := regexp.MustCompile(`x[a-zA-Z0-9]{4}_`)
 	matches, l, cursor := bstrExp.FindAllStringSubmatchIndex(s, -1), len(s), 0
 	for _, match := range matches {
 		result += s[cursor:match[0]]
 		subStr := s[match[0]:match[1]]
 		if subStr == "_x005F_" {
 			cursor = match[1]
-			if l > match[1]+6 && !escapeExp.MatchString(s[match[1]:match[1]+6]) {
+			if l > match[1]+6 && !bstrEscapeExp.MatchString(s[match[1]:match[1]+6]) {
 				result += subStr
 				continue
 			}
@@ -487,7 +490,7 @@ func bstrUnmarshal(s string) (result string) {
 			cursor = match[1]
 			v, err := strconv.Unquote(`"\u` + s[match[0]+2:match[1]-1] + `"`)
 			if err != nil {
-				if l > match[1]+6 && escapeExp.MatchString(s[match[1]:match[1]+6]) {
+				if l > match[1]+6 && bstrEscapeExp.MatchString(s[match[1]:match[1]+6]) {
 					result += subStr[:6]
 					cursor = match[1] + 6
 					continue
@@ -512,6 +515,41 @@ func bstrUnmarshal(s string) (result string) {
 	return result
 }
 
+// bstrMarshal encode the escaped string literal which not permitted in an XML
+// 1.0 document.
+func bstrMarshal(s string) (result string) {
+	matches, l, cursor := bstrExp.FindAllStringSubmatchIndex(s, -1), len(s), 0
+	for _, match := range matches {
+		result += s[cursor:match[0]]
+		subStr := s[match[0]:match[1]]
+		if subStr == "_x005F_" {
+			cursor = match[1]
+			if match[1]+6 <= l && bstrEscapeExp.MatchString(s[match[1]:match[1]+6]) {
+				_, err := strconv.Unquote(`"\u` + s[match[1]+1:match[1]+5] + `"`)
+				if err == nil {
+					result += subStr + "x005F" + subStr
+					continue
+				}
+			}
+			result += subStr + "x005F_"
+			continue
+		}
+		if bstrExp.MatchString(subStr) {
+			cursor = match[1]
+			_, err := strconv.Unquote(`"\u` + s[match[0]+2:match[1]-1] + `"`)
+			if err == nil {
+				result += "_x005F" + subStr
+				continue
+			}
+			result += subStr
+		}
+	}
+	if cursor < l {
+		result += s[cursor:]
+	}
+	return result
+}
+
 // Stack defined an abstract data type that serves as a collection of elements.
 type Stack struct {
 	list *list.List
diff --git a/lib_test.go b/lib_test.go
index ad20946..315688f 100644
--- a/lib_test.go
+++ b/lib_test.go
@@ -258,3 +258,16 @@ func TestBstrUnmarshal(t *testing.T) {
 		assert.Equal(t, expected, bstrUnmarshal(bstr))
 	}
 }
+
+func TestBstrMarshal(t *testing.T) {
+	bstrs := map[string]string{
+		"*_xG05F_*":       "*_xG05F_*",
+		"*_x0008_*":       "*_x005F_x0008_*",
+		"*_x005F_*":       "*_x005F_x005F_*",
+		"*_x005F_xG006_*": "*_x005F_x005F_xG006_*",
+		"*_x005F_x0006_*": "*_x005F_x005F_x005F_x0006_*",
+	}
+	for bstr, expected := range bstrs {
+		assert.Equal(t, expected, bstrMarshal(bstr))
+	}
+}
-- 
cgit v1.2.1