dump: support unescape string (go-mysql-org#104)

Author: siddontang

cmd/go-canal/main.go

@@ -29,6 +29,8 @@ var tableDB = flag.String("table_db", "test", "database for dump tables")
 var ignoreTables = flag.String("ignore_tables", "", "ignore tables, must be database.table format, separated by comma")
 
 func main() {
+flag.Parse()
+
 cfg := canal.NewDefaultConfig()
 cfg.Addr = fmt.Sprintf("%s:%d", *host, *port)
 cfg.User = *user

dump/dump_test.go

@@ -139,7 +139,7 @@ func (s *schemaTestSuite) TestParseValue(c *C) {
 str := `'abc\\',''`
 values, err := parseValues(str)
 c.Assert(err, IsNil)
-c.Assert(values, DeepEquals, []string{`'abc\\'`, `''`})
+c.Assert(values, DeepEquals, []string{`'abc\'`, `''`})
 
 str = `123,'\Z#÷QÎx£. Æ‘ÇoPâÅ_\r—\\','','qn'`
 values, err = parseValues(str)

dump/parser.go

@@ -109,10 +109,12 @@ func parseValues(str string) ([]string, error) {
 // read string until another single quote
 j := i + 1
 
+escaped := false
 for j < len(str) {
 if str[j] == '\\' {
 // skip escaped character
 j += 2
+escaped = true
 continue
 } else if str[j] == '\'' {
 break
@@ -125,7 +127,11 @@ func parseValues(str string) ([]string, error) {
 return nil, fmt.Errorf("parse quote values error")
 }
 
-values = append(values, str[i:j+1])
+value := str[i : j+1]
+if escaped {
+value = unescapeString(value)
+}
+values = append(values, value)
 // skip ' and ,
 i = j + 2
 }
@@ -135,3 +141,48 @@ func parseValues(str string) ([]string, error) {
 
 return values, nil
 }
+
+// unescapeString un-escapes the string.
+// mysqldump will escape the string when dumps,
+// Refer http://dev.mysql.com/doc/refman/5.7/en/string-literals.html
+func unescapeString(s string) string {
+i := 0
+
+value := make([]byte, 0, len(s))
+for i < len(s) {
+if s[i] == '\\' {
+j := i + 1
+if j == len(s) {
+// The last char is \, remove
+break
+}
+
+value = append(value, unescapeChar(s[j]))
+i += 2
+} else {
+value = append(value, s[i])
+i++
+}
+}
+
+return string(value)
+}
+
+func unescapeChar(ch byte) byte {
+// \" \' \\ \n \0 \b \Z \r \t ==> escape to one char
+switch ch {
+case 'n':
+ch = '\n'
+case '0':
+ch = 0
+case 'b':
+ch = 8
+case 'Z':
+ch = 26
+case 'r':
+ch = '\r'
+case 't':
+ch = '\t'
+}
+return ch
+}