Optimizing regexp calls to improve performance (#1532)

calc.go

@@ -193,6 +193,24 @@ var (
 			return fmt.Sprintf("R[%d]C[%d]", row, col), nil
 		},
 	}
+	formularFormats = []*regexp.Regexp{
+		regexp.MustCompile(`^(\d+)$`),
+		regexp.MustCompile(`^=(.*)$`),
+		regexp.MustCompile(`^<>(.*)$`),
+		regexp.MustCompile(`^<=(.*)$`),
+		regexp.MustCompile(`^>=(.*)$`),
+		regexp.MustCompile(`^<(.*)$`),
+		regexp.MustCompile(`^>(.*)$`),
+	}
+	formularCriterias = []byte{
+		criteriaEq,
+		criteriaEq,
+		criteriaNe,
+		criteriaLe,
+		criteriaGe,
+		criteriaL,
+		criteriaG,
+	}
 )
 
 // calcContext defines the formula execution context.
@@ -1654,33 +1672,11 @@ func formulaCriteriaParser(exp string) (fc *formulaCriteria) {
 	if exp == "" {
 		return
 	}
-	if match := regexp.MustCompile(`^(\d+)$`).FindStringSubmatch(exp); len(match) > 1 {
+	for i, re := range formularFormats {
-		fc.Type, fc.Condition = criteriaEq, match[1]
+		if match := re.FindStringSubmatch(exp); len(match) > 1 {
-		return
+			fc.Type, fc.Condition = formularCriterias[i], match[1]
-	}
+			return
-	if match := regexp.MustCompile(`^=(.*)$`).FindStringSubmatch(exp); len(match) > 1 {
+		}
-		fc.Type, fc.Condition = criteriaEq, match[1]
-		return
-	}
-	if match := regexp.MustCompile(`^<>(.*)$`).FindStringSubmatch(exp); len(match) > 1 {
-		fc.Type, fc.Condition = criteriaNe, match[1]
-		return
-	}
-	if match := regexp.MustCompile(`^<=(.*)$`).FindStringSubmatch(exp); len(match) > 1 {
-		fc.Type, fc.Condition = criteriaLe, match[1]
-		return
-	}
-	if match := regexp.MustCompile(`^>=(.*)$`).FindStringSubmatch(exp); len(match) > 1 {
-		fc.Type, fc.Condition = criteriaGe, match[1]
-		return
-	}
-	if match := regexp.MustCompile(`^<(.*)$`).FindStringSubmatch(exp); len(match) > 1 {
-		fc.Type, fc.Condition = criteriaL, match[1]
-		return
-	}
-	if match := regexp.MustCompile(`^>(.*)$`).FindStringSubmatch(exp); len(match) > 1 {
-		fc.Type, fc.Condition = criteriaG, match[1]
-		return
 	}
 	if strings.Contains(exp, "?") {
 		exp = strings.ReplaceAll(exp, "?", ".")

sheet.go

@@ -977,6 +977,7 @@ func (f *File) searchSheet(name, value string, regSearch bool) (result []string,
 	if sst, err = f.sharedStringsReader(); err != nil {
 		return
 	}
+	regex := regexp.MustCompile(value)
 	decoder := f.xmlNewDecoder(bytes.NewReader(f.readBytes(name)))
 	for {
 		var token xml.Token
@@ -1001,7 +1002,6 @@ func (f *File) searchSheet(name, value string, regSearch bool) (result []string,
 				_ = decoder.DecodeElement(&colCell, &xmlElement)
 				val, _ := colCell.getValueFrom(f, sst, false)
 				if regSearch {
-					regex := regexp.MustCompile(value)
 					if !regex.MatchString(val) {
 						continue
 					}

table.go

@@ -23,6 +23,13 @@ import (
 	"unicode/utf8"
 )
 
+var (
+	expressionFormat = regexp.MustCompile(`"(?:[^"]|"")*"|\S+`)
+	conditionFormat  = regexp.MustCompile(`(or|\|\|)`)
+	blankFormat      = regexp.MustCompile("blanks|nonblanks")
+	matchFormat      = regexp.MustCompile("[*?]")
+)
+
 // parseTableOptions provides a function to parse the format settings of the
 // table with default value.
 func parseTableOptions(opts *Table) (*Table, error) {
@@ -400,8 +407,7 @@ func (f *File) autoFilter(sheet, ref string, columns, col int, opts []AutoFilter
 			return fmt.Errorf("incorrect index of column '%s'", opt.Column)
 		}
 		fc := &xlsxFilterColumn{ColID: offset}
-		re := regexp.MustCompile(`"(?:[^"]|"")*"|\S+`)
+		token := expressionFormat.FindAllString(opt.Expression, -1)
-		token := re.FindAllString(opt.Expression, -1)
 		if len(token) != 3 && len(token) != 7 {
 			return fmt.Errorf("incorrect number of tokens in criteria '%s'", opt.Expression)
 		}
@@ -484,8 +490,7 @@ func (f *File) parseFilterExpression(expression string, tokens []string) ([]int,
 		// expressions).
 		conditional := 0
 		c := tokens[3]
-		re, _ := regexp.Match(`(or|\|\|)`, []byte(c))
+		if conditionFormat.Match([]byte(c)) {
-		if re {
 			conditional = 1
 		}
 		expression1, token1, err := f.parseFilterTokens(expression, tokens[:3])
@@ -533,7 +538,7 @@ func (f *File) parseFilterTokens(expression string, tokens []string) ([]int, str
 	}
 	token := tokens[2]
 	// Special handling for Blanks/NonBlanks.
-	re, _ := regexp.Match("blanks|nonblanks", []byte(strings.ToLower(token)))
+	re := blankFormat.Match([]byte(strings.ToLower(token)))
 	if re {
 		// Only allow Equals or NotEqual in this context.
 		if operator != 2 && operator != 5 {
@@ -558,7 +563,7 @@ func (f *File) parseFilterTokens(expression string, tokens []string) ([]int, str
 	}
 	// If the string token contains an Excel match character then change the
 	// operator type to indicate a non "simple" equality.
-	re, _ = regexp.Match("[*?]", []byte(token))
+	re = matchFormat.Match([]byte(token))
 	if operator == 2 && re {
 		operator = 22
 	}