Created

extractx/number.go

@@ -0,0 +1,110 @@
+package extractx
+
+import (
+	"regexp"
+	"strconv"
+	"strings"
+)
+
+var rxNumber = regexp.MustCompile(`\-?\d+[\d.,]*\d*`)
+
+// 提取的内容默认为 1,234.56 格式的数字，未实现根据国家标准实现提取
+// https://zhuanlan.zhihu.com/p/157980325
+func clean(s string) string {
+	s = strings.TrimSpace(s)
+	if s == "" {
+		return s
+	}
+	s = strings.ReplaceAll(s, ",", "")
+	n := len(s)
+	if s[n-1:] == "." {
+		s = s[n-2 : n-1]
+	}
+	return s
+}
+
+func Number(s string) string {
+	if s == "" {
+		return ""
+	}
+	return clean(rxNumber.FindString(s))
+}
+
+func Numbers(s string) []string {
+	if s == "" {
+		return []string{}
+	}
+
+	matches := rxNumber.FindAllString(s, -1)
+	if matches == nil {
+		return []string{}
+	}
+
+	for i, v := range matches {
+		matches[i] = clean(v)
+	}
+	return matches
+}
+
+func Float64(s string) float64 {
+	if s = Number(s); s != "" {
+		if v, err := strconv.ParseFloat(s, 64); err == nil {
+			return v
+		}
+	}
+	return 0
+}
+
+func Float32(s string) float32 {
+	if s = Number(s); s != "" {
+		if v, err := strconv.ParseFloat(s, 64); err == nil {
+			return float32(v)
+		}
+	}
+	return 0
+}
+
+func Int64(s string) int64 {
+	if s = Number(s); s != "" {
+		if v, err := strconv.ParseInt(s, 10, 64); err == nil {
+			return v
+		}
+	}
+	return 0
+}
+
+func Int32(s string) int32 {
+	if s = Number(s); s != "" {
+		if v, err := strconv.ParseInt(s, 10, 32); err == nil {
+			return int32(v)
+		}
+	}
+	return 0
+}
+
+func Int16(s string) int16 {
+	if s = Number(s); s != "" {
+		if v, err := strconv.ParseInt(s, 10, 16); err == nil {
+			return int16(v)
+		}
+	}
+	return 0
+}
+
+func Int8(s string) int8 {
+	if s = Number(s); s != "" {
+		if v, err := strconv.ParseInt(s, 10, 16); err == nil {
+			return int8(v)
+		}
+	}
+	return 0
+}
+
+func Int(s string) int {
+	if s = Number(s); s != "" {
+		if v, err := strconv.ParseInt(s, 10, 16); err == nil {
+			return int(v)
+		}
+	}
+	return 0
+}

extractx/number_test.go

@@ -0,0 +1,59 @@
+package extractx
+
+import (
+	"github.com/stretchr/testify/assert"
+	"testing"
+)
+
+func TestNumber(t *testing.T) {
+	testCases := []struct {
+		tag      string
+		string   string
+		expected string
+	}{
+		{"t1", "123", "123"},
+		{"t2", "12.3", "12.3"},
+		{"t3", "1,234.3", "1234.3"},
+		{"t4", "          ab 1 123", "1"},
+		{"t5", ".", ""},
+		{"t6", ",", ""},
+		{"t7", ".,", ""},
+		{"t8", "100 23.", "100"},
+		{"t8.1", "$100 $23.", "100"},
+		{"t9", "-1", "-1"},
+		{"t10", "-1-1", "-1"}, // todo maybe is empty
+		{"t11", "+1", "1"},
+		{"t12", "+1+1", "1"}, // todo maybe is empty
+		{"t13", "1.0 out of 5 stars", "1.0"},
+	}
+	for _, testCase := range testCases {
+		n := Number(testCase.string)
+		assert.Equal(t, testCase.expected, n, testCase.tag)
+	}
+}
+
+func TestNumbers(t *testing.T) {
+	testCases := []struct {
+		tag      string
+		string   string
+		expected []string
+	}{
+		{"t1", "123", []string{"123"}},
+		{"t2", "12.3", []string{"12.3"}},
+		{"t3", "1,234.3", []string{"1234.3"}},
+		{"t4", "          ab 1 123", []string{"1", "123"}},
+		{"t5", "          ab .1 123", []string{"1", "123"}},
+		{"t5", "          ab ,1 123", []string{"1", "123"}},
+		{"t6", "          ab 1. 123", []string{"1", "123"}},
+		{"t7", "$100,$200", []string{"100", "200"}},
+		{"t8", "1,2.3,4", []string{"12.34"}},
+		{"t9", "1, 2.3, 4", []string{"1", "2.3", "4"}},
+		{"t10", "-123,4", []string{"-1234"}},
+		{"t11", "1-1", []string{"1", "-1"}},  // todo May be return empty string
+		{"t12", "N1-1", []string{"1", "-1"}}, // todo May be return empty string
+	}
+	for _, testCase := range testCases {
+		n := Numbers(testCase.string)
+		assert.Equal(t, testCase.expected, n, testCase.tag)
+	}
+}