From 15c1537bf0bc96f53edf29474b945e3f9dccff38 Mon Sep 17 00:00:00 2001 From: dudaodong Date: Thu, 23 Feb 2023 10:23:38 +0800 Subject: [PATCH] doc: add doc for SplitWords and WordCount --- docs/strutil.md | 418 +++++++++++++++++++++--------------- docs/strutil_zh-CN.md | 424 ++++++++++++++++++++++--------------- strutil/string.go | 71 +++++++ strutil/string_internal.go | 33 +++ strutil/string_test.go | 34 +++ 5 files changed, 647 insertions(+), 333 deletions(-) diff --git a/docs/strutil.md b/docs/strutil.md index 0c29fa2..cc42660 100644 --- a/docs/strutil.md +++ b/docs/strutil.md @@ -42,6 +42,8 @@ import ( - [Wrap](#Wrap) - [Unwrap](#Unwrap) - [SplitEx](#SplitEx) +- [SplitWords](#SplitWords) +- [WordCount](#WordCount)
@@ -61,19 +63,19 @@ func After(s, char string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.After("lancet", "") - fmt.Println(s1) //lancet + s1 := strutil.After("lancet", "") + fmt.Println(s1) //lancet - s2 := strutil.After("github.com/test/lancet", "/") - fmt.Println(s2) //test/lancet + s2 := strutil.After("github.com/test/lancet", "/") + fmt.Println(s2) //test/lancet - s3 := strutil.After("github.com/test/lancet", "test") - fmt.Println(s3) // /lancet + s3 := strutil.After("github.com/test/lancet", "test") + fmt.Println(s3) // /lancet } ``` @@ -91,19 +93,19 @@ func AfterLast(s, char string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.AfterLast("lancet", "") - fmt.Println(s1) //lancet + s1 := strutil.AfterLast("lancet", "") + fmt.Println(s1) //lancet - s2 := strutil.AfterLast("github.com/test/lancet", "/") - fmt.Println(s2) //lancet + s2 := strutil.AfterLast("github.com/test/lancet", "/") + fmt.Println(s2) //lancet - s3 := strutil.AfterLast("github.com/test/test/lancet", "test") - fmt.Println(s3) // /test/lancet + s3 := strutil.AfterLast("github.com/test/test/lancet", "test") + fmt.Println(s3) // /test/lancet } ``` @@ -121,19 +123,19 @@ func Before(s, char string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.Before("lancet", "") - fmt.Println(s1) //lancet + s1 := strutil.Before("lancet", "") + fmt.Println(s1) //lancet - s2 := strutil.Before("github.com/test/lancet", "/") - fmt.Println(s2) //github.com + s2 := strutil.Before("github.com/test/lancet", "/") + fmt.Println(s2) //github.com - s3 := strutil.Before("github.com/test/lancet", "test") - fmt.Println(s3) // github.com/ + s3 := strutil.Before("github.com/test/lancet", "test") + fmt.Println(s3) // github.com/ } ``` @@ -151,19 +153,19 @@ func BeforeLast(s, char string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.BeforeLast("lancet", "") - fmt.Println(s1) //lancet + s1 := strutil.BeforeLast("lancet", "") + fmt.Println(s1) //lancet - s2 := strutil.BeforeLast("github.com/test/lancet", "/") - fmt.Println(s2) //github.com/test + s2 := strutil.BeforeLast("github.com/test/lancet", "/") + fmt.Println(s2) //github.com/test - s3 := strutil.BeforeLast("github.com/test/test/lancet", "test") - fmt.Println(s3) //github.com/test/ + s3 := strutil.BeforeLast("github.com/test/test/lancet", "test") + fmt.Println(s3) //github.com/test/ } ``` @@ -181,25 +183,25 @@ func CamelCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.CamelCase("foo_bar") - fmt.Println(s1) //fooBar + s1 := strutil.CamelCase("foo_bar") + fmt.Println(s1) //fooBar - s2 := strutil.CamelCase("Foo-Bar") - fmt.Println(s2) //fooBar + s2 := strutil.CamelCase("Foo-Bar") + fmt.Println(s2) //fooBar - s3 := strutil.CamelCase("Foo&bar") - fmt.Println(s3) //fooBar + s3 := strutil.CamelCase("Foo&bar") + fmt.Println(s3) //fooBar - s4 := strutil.CamelCase("foo bar") - fmt.Println(s4) //fooBar + s4 := strutil.CamelCase("foo bar") + fmt.Println(s4) //fooBar - s4 := strutil.CamelCase("Foo-#1😄$_%^&*(1bar") - fmt.Println(s4) //foo11Bar + s4 := strutil.CamelCase("Foo-#1😄$_%^&*(1bar") + fmt.Println(s4) //foo11Bar ``` ### Capitalize @@ -216,19 +218,19 @@ func Capitalize(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.Capitalize("foo") - fmt.Println(s1) //foo + s1 := strutil.Capitalize("foo") + fmt.Println(s1) //foo - s2 := strutil.Capitalize("Foo") - fmt.Println(s2) //foo + s2 := strutil.Capitalize("Foo") + fmt.Println(s2) //foo - s3 := strutil.Capitalize("FOo" - fmt.Println(s3) //fOo + s3 := strutil.Capitalize("FOo" + fmt.Println(s3) //fOo } ``` @@ -246,17 +248,17 @@ func IsString(v interface{}) bool ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - fmt.Println(strutil.IsString("lancet")) //true - fmt.Println(strutil.IsString("")) //true + fmt.Println(strutil.IsString("lancet")) //true + fmt.Println(strutil.IsString("")) //true - fmt.Println(strutil.IsString(1)) //false - fmt.Println(strutil.IsString("")) //false - fmt.Println(strutil.IsString([]string{})) //false + fmt.Println(strutil.IsString(1)) //false + fmt.Println(strutil.IsString("")) //false + fmt.Println(strutil.IsString([]string{})) //false } ``` @@ -274,22 +276,22 @@ func KebabCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.KebabCase("Foo Bar-") - fmt.Println(s1) //foo-bar + s1 := strutil.KebabCase("Foo Bar-") + fmt.Println(s1) //foo-bar - s2 := strutil.KebabCase("foo_Bar") - fmt.Println(s2) //foo-bar + s2 := strutil.KebabCase("foo_Bar") + fmt.Println(s2) //foo-bar - s3 := strutil.KebabCase("fooBar") - fmt.Println(s3) //foo-bar + s3 := strutil.KebabCase("fooBar") + fmt.Println(s3) //foo-bar - s4 := strutil.KebabCase("__FOO_BAR__") - fmt.Println(s4) //foo-bar + s4 := strutil.KebabCase("__FOO_BAR__") + fmt.Println(s4) //foo-bar } ``` @@ -307,22 +309,22 @@ func KebabCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.UpperKebabCase("Foo Bar-") - fmt.Println(s1) //FOO-BAR + s1 := strutil.UpperKebabCase("Foo Bar-") + fmt.Println(s1) //FOO-BAR - s2 := strutil.UpperKebabCase("foo_Bar") - fmt.Println(s2) //FOO-BAR + s2 := strutil.UpperKebabCase("foo_Bar") + fmt.Println(s2) //FOO-BAR - s3 := strutil.UpperKebabCase("fooBar") - fmt.Println(s3) //FOO-BAR + s3 := strutil.UpperKebabCase("fooBar") + fmt.Println(s3) //FOO-BAR - s4 := strutil.UpperKebabCase("__FOO_BAR__") - fmt.Println(s4) //FOO-BAR + s4 := strutil.UpperKebabCase("__FOO_BAR__") + fmt.Println(s4) //FOO-BAR } ``` @@ -340,22 +342,22 @@ func LowerFirst(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.LowerFirst("foo") - fmt.Println(s1) //foo + s1 := strutil.LowerFirst("foo") + fmt.Println(s1) //foo - s2 := strutil.LowerFirst("BAR") - fmt.Println(s2) //bAR + s2 := strutil.LowerFirst("BAR") + fmt.Println(s2) //bAR - s3 := strutil.LowerFirst("FOo") - fmt.Println(s3) //fOo + s3 := strutil.LowerFirst("FOo") + fmt.Println(s3) //fOo - s4 := strutil.LowerFirst("fOo大") - fmt.Println(s4) //fOo大 + s4 := strutil.LowerFirst("fOo大") + fmt.Println(s4) //fOo大 } ``` @@ -373,22 +375,22 @@ func UpperFirst(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.UpperFirst("foo") - fmt.Println(s1) //Foo + s1 := strutil.UpperFirst("foo") + fmt.Println(s1) //Foo - s2 := strutil.UpperFirst("bAR") - fmt.Println(s2) //BAR + s2 := strutil.UpperFirst("bAR") + fmt.Println(s2) //BAR - s3 := strutil.UpperFirst("FOo") - fmt.Println(s3) //FOo + s3 := strutil.UpperFirst("FOo") + fmt.Println(s3) //FOo - s4 := strutil.UpperFirst("fOo大") - fmt.Println(s4) //FOo大 + s4 := strutil.UpperFirst("fOo大") + fmt.Println(s4) //FOo大 } ``` @@ -543,16 +545,16 @@ func ReverseStr(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.ReverseStr("abc") - fmt.Println(s1) //cba + s1 := strutil.ReverseStr("abc") + fmt.Println(s1) //cba - s2 := strutil.ReverseStr("12345") - fmt.Println(s2) //54321 + s2 := strutil.ReverseStr("12345") + fmt.Println(s2) //54321 } ``` @@ -570,25 +572,25 @@ func SnakeCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.SnakeCase("Foo Bar-") - fmt.Println(s1) //foo_bar + s1 := strutil.SnakeCase("Foo Bar-") + fmt.Println(s1) //foo_bar - s2 := strutil.SnakeCase("foo_Bar") - fmt.Println(s2) //foo_bar + s2 := strutil.SnakeCase("foo_Bar") + fmt.Println(s2) //foo_bar - s3 := strutil.SnakeCase("fooBar") - fmt.Println(s3) //foo_bar + s3 := strutil.SnakeCase("fooBar") + fmt.Println(s3) //foo_bar - s4 := strutil.SnakeCase("__FOO_BAR__") - fmt.Println(s4) //foo_bar + s4 := strutil.SnakeCase("__FOO_BAR__") + fmt.Println(s4) //foo_bar - s5 := strutil.SnakeCase("Foo-#1😄$_%^&*(1bar") - fmt.Println(s5) //foo_1_1_bar + s5 := strutil.SnakeCase("Foo-#1😄$_%^&*(1bar") + fmt.Println(s5) //foo_1_1_bar } ``` @@ -606,25 +608,25 @@ func SnakeCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.UpperSnakeCase("Foo Bar-") - fmt.Println(s1) //FOO_BAR + s1 := strutil.UpperSnakeCase("Foo Bar-") + fmt.Println(s1) //FOO_BAR - s2 := strutil.UpperSnakeCase("foo_Bar") - fmt.Println(s2) //FOO_BAR + s2 := strutil.UpperSnakeCase("foo_Bar") + fmt.Println(s2) //FOO_BAR - s3 := strutil.UpperSnakeCase("fooBar") - fmt.Println(s3) //FOO_BAR + s3 := strutil.UpperSnakeCase("fooBar") + fmt.Println(s3) //FOO_BAR - s4 := strutil.UpperSnakeCase("__FOO_BAR__") - fmt.Println(s4) //FOO_BAR + s4 := strutil.UpperSnakeCase("__FOO_BAR__") + fmt.Println(s4) //FOO_BAR - s5 := strutil.UpperSnakeCase("Foo-#1😄$_%^&*(1bar") - fmt.Println(s5) //FOO_1_1_BAR + s5 := strutil.UpperSnakeCase("Foo-#1😄$_%^&*(1bar") + fmt.Println(s5) //FOO_1_1_BAR } ``` @@ -642,25 +644,25 @@ func Wrap(str string, wrapWith string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.Wrap("ab", "") - fmt.Println(s1) //ab + s1 := strutil.Wrap("ab", "") + fmt.Println(s1) //ab - s2 := strutil.Wrap("", "*") - fmt.Println(s2) //"" + s2 := strutil.Wrap("", "*") + fmt.Println(s2) //"" - s3 := strutil.Wrap("ab", "*") - fmt.Println(s3) //*ab* + s3 := strutil.Wrap("ab", "*") + fmt.Println(s3) //*ab* - s4 := strutil.Wrap("ab", "\"") - fmt.Println(s4) //\"ab\" + s4 := strutil.Wrap("ab", "\"") + fmt.Println(s4) //\"ab\" - s5 := strutil.Wrap("ab", "'") - fmt.Println(s5) //'ab' + s5 := strutil.Wrap("ab", "'") + fmt.Println(s5) //'ab' } ``` @@ -678,25 +680,25 @@ func Unwrap(str string, wrapToken string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.Unwrap("ab", "") - fmt.Println(s1) //ab + s1 := strutil.Unwrap("ab", "") + fmt.Println(s1) //ab - s2 := strutil.Unwrap("ab", "*") - fmt.Println(s2) //ab + s2 := strutil.Unwrap("ab", "*") + fmt.Println(s2) //ab - s3 := strutil.Unwrap("**ab**", "*") - fmt.Println(s3) //*ab* + s3 := strutil.Unwrap("**ab**", "*") + fmt.Println(s3) //*ab* - s4 := strutil.Unwrap("*ab", "*") - fmt.Println(s4) //*ab + s4 := strutil.Unwrap("*ab", "*") + fmt.Println(s4) //*ab - s5 := strutil.Unwrap("***", "**") - fmt.Println(s5) //*** + s5 := strutil.Unwrap("***", "**") + fmt.Println(s5) //*** } ``` @@ -714,24 +716,110 @@ func SplitEx(s, sep string, removeEmptyString bool) []string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - arr1 := strutil.SplitEx(" a b c ", "", true) - fmt.Println(arr1) //[]string{} + arr1 := strutil.SplitEx(" a b c ", "", true) + fmt.Println(arr1) //[]string{} - arr2 := strutil.SplitEx(" a b c ", " ", false) - fmt.Println(arr2) //[]string{"", "a", "b", "c", ""} + arr2 := strutil.SplitEx(" a b c ", " ", false) + fmt.Println(arr2) //[]string{"", "a", "b", "c", ""} - arr3 := strutil.SplitEx(" a b c ", " ", true) - fmt.Println(arr3) //[]string{"a", "b", "c"} + arr3 := strutil.SplitEx(" a b c ", " ", true) + fmt.Println(arr3) //[]string{"a", "b", "c"} - arr4 := strutil.SplitEx(" a = b = c = ", " = ", false) - fmt.Println(arr4) //[]string{" a", "b", "c", ""} + arr4 := strutil.SplitEx(" a = b = c = ", " = ", false) + fmt.Println(arr4) //[]string{" a", "b", "c", ""} - arr5 := strutil.SplitEx(" a = b = c = ", " = ", true) - fmt.Println(arr5) //[]string{" a", "b", "c"} + arr5 := strutil.SplitEx(" a = b = c = ", " = ", true) + fmt.Println(arr5) //[]string{" a", "b", "c"} +} +``` + +### SplitWords + +

Splits a string into words, word only contains alphabetic characters.

+ +Signature: + +```go +func SplitWords(s string) []string +``` + +Example: + +```go +import ( + "fmt" + "github.com/duke-git/lancet/strutil" +) + +func main() { + result1 := strutil.SplitWords("a word") + result2 := strutil.SplitWords("I'am a programmer") + result3 := strutil.SplitWords("Bonjour, je suis programmeur") + result4 := strutil.SplitWords("a -b-c' 'd'e") + result5 := strutil.SplitWords("你好，我是一名码农") + result6 := strutil.SplitWords("こんにちは，私はプログラマーです") + + fmt.Println(result1) + fmt.Println(result2) + fmt.Println(result3) + fmt.Println(result4) + fmt.Println(result5) + fmt.Println(result6) + + // Output: + // [a word] + // [I'am a programmer] + // [Bonjour je suis programmeur] + // [a b-c' d'e] + // [] + // [] +} +``` + +### WordCount + +

Return the number of meaningful word, word only contains alphabetic characters.

+ +Signature: + +```go +func WordCount(s string) int +``` + +Example: + +```go +import ( + "fmt" + "github.com/duke-git/lancet/strutil" +) + +func main() { + result1 := strutil.WordCount("a word") + result2 := strutil.WordCount("I'am a programmer") + result3 := strutil.WordCount("Bonjour, je suis programmeur") + result4 := strutil.WordCount("a -b-c' 'd'e") + result5 := strutil.WordCount("你好，我是一名码农") + result6 := strutil.WordCount("こんにちは，私はプログラマーです") + + fmt.Println(result1) + fmt.Println(result2) + fmt.Println(result3) + fmt.Println(result4) + fmt.Println(result5) + fmt.Println(result6) + + // Output: + // 2 + // 3 + // 4 + // 3 + // 0 + // 0 } ``` diff --git a/docs/strutil_zh-CN.md b/docs/strutil_zh-CN.md index 423da4e..595e64d 100644 --- a/docs/strutil_zh-CN.md +++ b/docs/strutil_zh-CN.md @@ -42,6 +42,8 @@ import ( - [Wrap](#Wrap) - [Unwrap](#Unwrap) - [SplitEx](#SplitEx) +- [SplitWords](#SplitWords) +- [WordCount](#WordCount)
@@ -61,19 +63,19 @@ func After(s, char string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.After("lancet", "") - fmt.Println(s1) //lancet + s1 := strutil.After("lancet", "") + fmt.Println(s1) //lancet - s2 := strutil.After("github.com/test/lancet", "/") - fmt.Println(s2) //test/lancet + s2 := strutil.After("github.com/test/lancet", "/") + fmt.Println(s2) //test/lancet - s3 := strutil.After("github.com/test/lancet", "test") - fmt.Println(s3) // /lancet + s3 := strutil.After("github.com/test/lancet", "test") + fmt.Println(s3) // /lancet } ``` @@ -91,19 +93,19 @@ func AfterLast(s, char string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.AfterLast("lancet", "") - fmt.Println(s1) //lancet + s1 := strutil.AfterLast("lancet", "") + fmt.Println(s1) //lancet - s2 := strutil.AfterLast("github.com/test/lancet", "/") - fmt.Println(s2) //lancet + s2 := strutil.AfterLast("github.com/test/lancet", "/") + fmt.Println(s2) //lancet - s3 := strutil.AfterLast("github.com/test/test/lancet", "test") - fmt.Println(s3) // /lancet + s3 := strutil.AfterLast("github.com/test/test/lancet", "test") + fmt.Println(s3) // /lancet } ``` @@ -121,19 +123,19 @@ func Before(s, char string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.Before("lancet", "") - fmt.Println(s1) //lancet + s1 := strutil.Before("lancet", "") + fmt.Println(s1) //lancet - s2 := strutil.Before("github.com/test/lancet", "/") - fmt.Println(s2) //github.com + s2 := strutil.Before("github.com/test/lancet", "/") + fmt.Println(s2) //github.com - s3 := strutil.Before("github.com/test/lancet", "test") - fmt.Println(s3) // github.com/ + s3 := strutil.Before("github.com/test/lancet", "test") + fmt.Println(s3) // github.com/ } ``` @@ -151,19 +153,19 @@ func BeforeLast(s, char string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.BeforeLast("lancet", "") - fmt.Println(s1) //lancet + s1 := strutil.BeforeLast("lancet", "") + fmt.Println(s1) //lancet - s2 := strutil.BeforeLast("github.com/test/lancet", "/") - fmt.Println(s2) //github.com/test + s2 := strutil.BeforeLast("github.com/test/lancet", "/") + fmt.Println(s2) //github.com/test - s3 := strutil.BeforeLast("github.com/test/test/lancet", "test") - fmt.Println(s3) //github.com/test/ + s3 := strutil.BeforeLast("github.com/test/test/lancet", "test") + fmt.Println(s3) //github.com/test/ } ``` @@ -181,25 +183,25 @@ func CamelCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.CamelCase("foo_bar") - fmt.Println(s1) //fooBar + s1 := strutil.CamelCase("foo_bar") + fmt.Println(s1) //fooBar - s2 := strutil.CamelCase("Foo-Bar") - fmt.Println(s2) //fooBar + s2 := strutil.CamelCase("Foo-Bar") + fmt.Println(s2) //fooBar - s3 := strutil.CamelCase("Foo&bar") - fmt.Println(s3) //fooBar + s3 := strutil.CamelCase("Foo&bar") + fmt.Println(s3) //fooBar - s4 := strutil.CamelCase("foo bar") - fmt.Println(s4) //fooBar + s4 := strutil.CamelCase("foo bar") + fmt.Println(s4) //fooBar - s4 := strutil.CamelCase("Foo-#1😄$_%^&*(1bar") - fmt.Println(s4) //foo11Bar + s4 := strutil.CamelCase("Foo-#1😄$_%^&*(1bar") + fmt.Println(s4) //foo11Bar } ``` @@ -217,19 +219,19 @@ func Capitalize(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.Capitalize("foo") - fmt.Println(s1) //foo + s1 := strutil.Capitalize("foo") + fmt.Println(s1) //foo - s2 := strutil.Capitalize("Foo") - fmt.Println(s2) //foo + s2 := strutil.Capitalize("Foo") + fmt.Println(s2) //foo - s3 := strutil.Capitalize("FOo" - fmt.Println(s3) //fOo + s3 := strutil.Capitalize("FOo" + fmt.Println(s3) //fOo } ``` @@ -247,17 +249,17 @@ func IsString(v interface{}) bool ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - fmt.Println(strutil.IsString("lancet")) //true - fmt.Println(strutil.IsString("")) //true + fmt.Println(strutil.IsString("lancet")) //true + fmt.Println(strutil.IsString("")) //true - fmt.Println(strutil.IsString(1)) //false - fmt.Println(strutil.IsString("")) //false - fmt.Println(strutil.IsString([]string{})) //false + fmt.Println(strutil.IsString(1)) //false + fmt.Println(strutil.IsString("")) //false + fmt.Println(strutil.IsString([]string{})) //false } ``` @@ -275,22 +277,22 @@ func KebabCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.KebabCase("Foo Bar-") - fmt.Println(s1) //foo-bar + s1 := strutil.KebabCase("Foo Bar-") + fmt.Println(s1) //foo-bar - s2 := strutil.KebabCase("foo_Bar") - fmt.Println(s2) //foo-bar + s2 := strutil.KebabCase("foo_Bar") + fmt.Println(s2) //foo-bar - s3 := strutil.KebabCase("fooBar") - fmt.Println(s3) //foo-bar + s3 := strutil.KebabCase("fooBar") + fmt.Println(s3) //foo-bar - s4 := strutil.KebabCase("__FOO_BAR__") - fmt.Println(s4) //foo-bar + s4 := strutil.KebabCase("__FOO_BAR__") + fmt.Println(s4) //foo-bar } ``` @@ -308,22 +310,22 @@ func KebabCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.UpperKebabCase("Foo Bar-") - fmt.Println(s1) //FOO-BAR + s1 := strutil.UpperKebabCase("Foo Bar-") + fmt.Println(s1) //FOO-BAR - s2 := strutil.UpperKebabCase("foo_Bar") - fmt.Println(s2) //FOO-BAR + s2 := strutil.UpperKebabCase("foo_Bar") + fmt.Println(s2) //FOO-BAR - s3 := strutil.UpperKebabCase("fooBar") - fmt.Println(s3) //FOO-BAR + s3 := strutil.UpperKebabCase("fooBar") + fmt.Println(s3) //FOO-BAR - s4 := strutil.UpperKebabCase("__FOO_BAR__") - fmt.Println(s4) //FOO-BAR + s4 := strutil.UpperKebabCase("__FOO_BAR__") + fmt.Println(s4) //FOO-BAR } ``` @@ -341,22 +343,22 @@ func LowerFirst(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.LowerFirst("foo") - fmt.Println(s1) //foo + s1 := strutil.LowerFirst("foo") + fmt.Println(s1) //foo - s2 := strutil.LowerFirst("BAR") - fmt.Println(s2) //bAR + s2 := strutil.LowerFirst("BAR") + fmt.Println(s2) //bAR - s3 := strutil.LowerFirst("FOo") - fmt.Println(s3) //fOo + s3 := strutil.LowerFirst("FOo") + fmt.Println(s3) //fOo - s4 := strutil.LowerFirst("fOo大") - fmt.Println(s4) //fOo大 + s4 := strutil.LowerFirst("fOo大") + fmt.Println(s4) //fOo大 } ``` @@ -374,22 +376,22 @@ func UpperFirst(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.UpperFirst("foo") - fmt.Println(s1) //Foo + s1 := strutil.UpperFirst("foo") + fmt.Println(s1) //Foo - s2 := strutil.UpperFirst("bAR") - fmt.Println(s2) //BAR + s2 := strutil.UpperFirst("bAR") + fmt.Println(s2) //BAR - s3 := strutil.UpperFirst("FOo") - fmt.Println(s3) //FOo + s3 := strutil.UpperFirst("FOo") + fmt.Println(s3) //FOo - s4 := strutil.UpperFirst("fOo大") - fmt.Println(s4) //FOo大 + s4 := strutil.UpperFirst("fOo大") + fmt.Println(s4) //FOo大 } ``` @@ -408,7 +410,7 @@ func Pad(source string, size int, padStr string) string ```go import ( "fmt" - "github.com/duke-git/lancet/v2/strutil" + "github.com/duke-git/lancet/strutil" ) func main() { @@ -453,7 +455,7 @@ func PadEnd(source string, size int, padStr string) string ```go import ( "fmt" - "github.com/duke-git/lancet/v2/strutil" + "github.com/duke-git/lancet/strutil" ) func main() { @@ -499,7 +501,7 @@ func PadStart(source string, size int, padStr string) string ```go import ( "fmt" - "github.com/duke-git/lancet/v2/strutil" + "github.com/duke-git/lancet/strutil" ) func main() { @@ -544,16 +546,16 @@ func ReverseStr(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.ReverseStr("abc") - fmt.Println(s1) //cba + s1 := strutil.ReverseStr("abc") + fmt.Println(s1) //cba - s2 := strutil.ReverseStr("12345") - fmt.Println(s2) //54321 + s2 := strutil.ReverseStr("12345") + fmt.Println(s2) //54321 } ``` @@ -571,25 +573,25 @@ func SnakeCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.SnakeCase("Foo Bar-") - fmt.Println(s1) //foo_bar + s1 := strutil.SnakeCase("Foo Bar-") + fmt.Println(s1) //foo_bar - s2 := strutil.SnakeCase("foo_Bar") - fmt.Println(s2) //foo_bar + s2 := strutil.SnakeCase("foo_Bar") + fmt.Println(s2) //foo_bar - s3 := strutil.SnakeCase("fooBar") - fmt.Println(s3) //foo_bar + s3 := strutil.SnakeCase("fooBar") + fmt.Println(s3) //foo_bar - s4 := strutil.SnakeCase("__FOO_BAR__") - fmt.Println(s4) //foo_bar + s4 := strutil.SnakeCase("__FOO_BAR__") + fmt.Println(s4) //foo_bar - s5 := strutil.SnakeCase("Foo-#1😄$_%^&*(1bar") - fmt.Println(s5) //foo_1_1_bar + s5 := strutil.SnakeCase("Foo-#1😄$_%^&*(1bar") + fmt.Println(s5) //foo_1_1_bar } ``` @@ -607,25 +609,25 @@ func SnakeCase(s string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.UpperSnakeCase("Foo Bar-") - fmt.Println(s1) //FOO_BAR + s1 := strutil.UpperSnakeCase("Foo Bar-") + fmt.Println(s1) //FOO_BAR - s2 := strutil.UpperSnakeCase("foo_Bar") - fmt.Println(s2) //FOO_BAR + s2 := strutil.UpperSnakeCase("foo_Bar") + fmt.Println(s2) //FOO_BAR - s3 := strutil.UpperSnakeCase("fooBar") - fmt.Println(s3) //FOO_BAR + s3 := strutil.UpperSnakeCase("fooBar") + fmt.Println(s3) //FOO_BAR - s4 := strutil.UpperSnakeCase("__FOO_BAR__") - fmt.Println(s4) //FOO_BAR + s4 := strutil.UpperSnakeCase("__FOO_BAR__") + fmt.Println(s4) //FOO_BAR - s5 := strutil.UpperSnakeCase("Foo-#1😄$_%^&*(1bar") - fmt.Println(s5) //FOO_1_1_BAR + s5 := strutil.UpperSnakeCase("Foo-#1😄$_%^&*(1bar") + fmt.Println(s5) //FOO_1_1_BAR } ``` @@ -643,25 +645,25 @@ func Wrap(str string, wrapWith string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.Wrap("ab", "") - fmt.Println(s1) //ab + s1 := strutil.Wrap("ab", "") + fmt.Println(s1) //ab - s2 := strutil.Wrap("", "*") - fmt.Println(s2) //"" + s2 := strutil.Wrap("", "*") + fmt.Println(s2) //"" - s3 := strutil.Wrap("ab", "*") - fmt.Println(s3) //*ab* + s3 := strutil.Wrap("ab", "*") + fmt.Println(s3) //*ab* - s4 := strutil.Wrap("ab", "\"") - fmt.Println(s4) //\"ab\" + s4 := strutil.Wrap("ab", "\"") + fmt.Println(s4) //\"ab\" - s5 := strutil.Wrap("ab", "'") - fmt.Println(s5) //'ab' + s5 := strutil.Wrap("ab", "'") + fmt.Println(s5) //'ab' } ``` @@ -679,25 +681,25 @@ func Unwrap(str string, wrapToken string) string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - s1 := strutil.Unwrap("ab", "") - fmt.Println(s1) //ab + s1 := strutil.Unwrap("ab", "") + fmt.Println(s1) //ab - s2 := strutil.Unwrap("ab", "*") - fmt.Println(s2) //ab + s2 := strutil.Unwrap("ab", "*") + fmt.Println(s2) //ab - s3 := strutil.Unwrap("**ab**", "*") - fmt.Println(s3) //*ab* + s3 := strutil.Unwrap("**ab**", "*") + fmt.Println(s3) //*ab* - s4 := strutil.Unwrap("*ab", "*") - fmt.Println(s4) //*ab + s4 := strutil.Unwrap("*ab", "*") + fmt.Println(s4) //*ab - s5 := strutil.Unwrap("***", "**") - fmt.Println(s5) //*** + s5 := strutil.Unwrap("***", "**") + fmt.Println(s5) //*** } ``` @@ -715,24 +717,110 @@ func SplitEx(s, sep string, removeEmptyString bool) []string ```go import ( - "fmt" - "github.com/duke-git/lancet/strutil" + "fmt" + "github.com/duke-git/lancet/strutil" ) func main() { - arr1 := strutil.SplitEx(" a b c ", "", true) - fmt.Println(arr1) //[]string{} + arr1 := strutil.SplitEx(" a b c ", "", true) + fmt.Println(arr1) //[]string{} - arr2 := strutil.SplitEx(" a b c ", " ", false) - fmt.Println(arr2) //[]string{"", "a", "b", "c", ""} + arr2 := strutil.SplitEx(" a b c ", " ", false) + fmt.Println(arr2) //[]string{"", "a", "b", "c", ""} - arr3 := strutil.SplitEx(" a b c ", " ", true) - fmt.Println(arr3) //[]string{"a", "b", "c"} + arr3 := strutil.SplitEx(" a b c ", " ", true) + fmt.Println(arr3) //[]string{"a", "b", "c"} - arr4 := strutil.SplitEx(" a = b = c = ", " = ", false) - fmt.Println(arr4) //[]string{" a", "b", "c", ""} + arr4 := strutil.SplitEx(" a = b = c = ", " = ", false) + fmt.Println(arr4) //[]string{" a", "b", "c", ""} - arr5 := strutil.SplitEx(" a = b = c = ", " = ", true) - fmt.Println(arr5) //[]string{" a", "b", "c"} + arr5 := strutil.SplitEx(" a = b = c = ", " = ", true) + fmt.Println(arr5) //[]string{" a", "b", "c"} +} +``` + +### SplitWords + +

将字符串拆分为单词，只支持字母字符单词。

+ +函数签名: + +```go +func SplitWords(s string) []string +``` + +示例: + +```go +import ( + "fmt" + "github.com/duke-git/lancet/strutil" +) + +func main() { + result1 := strutil.SplitWords("a word") + result2 := strutil.SplitWords("I'am a programmer") + result3 := strutil.SplitWords("Bonjour, je suis programmeur") + result4 := strutil.SplitWords("a -b-c' 'd'e") + result5 := strutil.SplitWords("你好，我是一名码农") + result6 := strutil.SplitWords("こんにちは，私はプログラマーです") + + fmt.Println(result1) + fmt.Println(result2) + fmt.Println(result3) + fmt.Println(result4) + fmt.Println(result5) + fmt.Println(result6) + + // Output: + // [a word] + // [I'am a programmer] + // [Bonjour je suis programmeur] + // [a b-c' d'e] + // [] + // [] +} +``` + +### WordCount + +

返回有意义单词的数量，只支持字母字符单词。

+ +函数签名: + +```go +func WordCount(s string) int +``` + +示例: + +```go +import ( + "fmt" + "github.com/duke-git/lancet/strutil" +) + +func main() { + result1 := strutil.WordCount("a word") + result2 := strutil.WordCount("I'am a programmer") + result3 := strutil.WordCount("Bonjour, je suis programmeur") + result4 := strutil.WordCount("a -b-c' 'd'e") + result5 := strutil.WordCount("你好，我是一名码农") + result6 := strutil.WordCount("こんにちは，私はプログラマーです") + + fmt.Println(result1) + fmt.Println(result2) + fmt.Println(result3) + fmt.Println(result4) + fmt.Println(result5) + fmt.Println(result6) + + // Output: + // 2 + // 3 + // 4 + // 3 + // 0 + // 0 } ``` diff --git a/strutil/string.go b/strutil/string.go index 567a064..bf80be4 100644 --- a/strutil/string.go +++ b/strutil/string.go @@ -253,3 +253,74 @@ func SplitEx(s, sep string, removeEmptyString bool) []string { return ret } + +// SplitWords splits a string into words, word only contains alphabetic characters. +func SplitWords(s string) []string { + var word string + var words []string + var r rune + var size, pos int + + isWord := false + + for len(s) > 0 { + r, size = utf8.DecodeRuneInString(s) + + switch { + case isLetter(r): + if !isWord { + isWord = true + word = s + pos = 0 + } + + case isWord && (r == '\'' || r == '-'): + // is word + + default: + if isWord { + isWord = false + words = append(words, word[:pos]) + } + } + + pos += size + s = s[size:] + } + + if isWord { + words = append(words, word[:pos]) + } + + return words +} + +// WordCount return the number of meaningful word, word only contains alphabetic characters. +func WordCount(s string) int { + var r rune + var size, count int + + isWord := false + + for len(s) > 0 { + r, size = utf8.DecodeRuneInString(s) + + switch { + case isLetter(r): + if !isWord { + isWord = true + count++ + } + + case isWord && (r == '\'' || r == '-'): + // is word + + default: + isWord = false + } + + s = s[size:] + } + + return count +} diff --git a/strutil/string_internal.go b/strutil/string_internal.go index 45ea245..4e3fe13 100644 --- a/strutil/string_internal.go +++ b/strutil/string_internal.go @@ -132,3 +132,36 @@ func padAtPosition(str string, length int, padStr string, position int) string { return leftPad + str + rightPad } + +// isLetter checks r is a letter but not CJK character. +func isLetter(r rune) bool { + if !unicode.IsLetter(r) { + return false + } + + switch { + // cjk char: /[\u3040-\u30ff\u3400-\u4dbf\u4e00-\u9fff\uf900-\ufaff\uff66-\uff9f]/ + + // hiragana and katakana (Japanese only) + case r >= '\u3034' && r < '\u30ff': + return false + + // CJK unified ideographs extension A (Chinese, Japanese, and Korean) + case r >= '\u3400' && r < '\u4dbf': + return false + + // CJK unified ideographs (Chinese, Japanese, and Korean) + case r >= '\u4e00' && r < '\u9fff': + return false + + // CJK compatibility ideographs (Chinese, Japanese, and Korean) + case r >= '\uf900' && r < '\ufaff': + return false + + // half-width katakana (Japanese only) + case r >= '\uff66' && r < '\uff9f': + return false + } + + return true +} diff --git a/strutil/string_test.go b/strutil/string_test.go index f44c040..1799eef 100644 --- a/strutil/string_test.go +++ b/strutil/string_test.go @@ -295,3 +295,37 @@ func TestSplitEx(t *testing.T) { assert.Equal([]string{" a", "b", "c", ""}, SplitEx(" a = b = c = ", " = ", false)) assert.Equal([]string{" a", "b", "c"}, SplitEx(" a = b = c = ", " = ", true)) } + +func TestSplitWords(t *testing.T) { + assert := internal.NewAssert(t, "TestSplitWords") + + cases := map[string][]string{ + "a word": {"a", "word"}, + "I'am a programmer": {"I'am", "a", "programmer"}, + "Bonjour, je suis programmeur": {"Bonjour", "je", "suis", "programmeur"}, + "a -b-c' 'd'e": {"a", "b-c'", "d'e"}, + "你好，我是一名码农": nil, + "こんにちは，私はプログラマーです": nil, + } + + for k, v := range cases { + assert.Equal(v, SplitWords(k)) + } +} + +func TestWordCount(t *testing.T) { + assert := internal.NewAssert(t, "TestSplitWords") + + cases := map[string]int{ + "a word": 2, // {"a", "word"}, + "I'am a programmer": 3, // {"I'am", "a", "programmer"}, + "Bonjour, je suis programmeur": 4, // {"Bonjour", "je", "suis", "programmeur"}, + "a -b-c' 'd'e": 3, // {"a", "b-c'", "d'e"}, + "你好，我是一名码农": 0, // nil, + "こんにちは，私はプログラマーです": 0, // nil, + } + + for k, v := range cases { + assert.Equal(v, WordCount(k)) + } +}