Package uniseg

Constants

These constants define whether a given text may be broken into the next line. If the break is optional (LineCanBreak), you may choose to break or not based on your own criteria, for example, if the text has reached the available width.

const (
    LineDontBreak = iota // You may not break the line here.
    LineCanBreak         // You may or may not break the line here.
    LineMustBreak        // You must break the line here.
)

The bit masks used to extract boundary information returned by Step.

const (
    MaskLine     = 3
    MaskWord     = 4
    MaskSentence = 8
)

The number of bits to shift the boundary information returned by Step to obtain the monospace width of the grapheme cluster.

const ShiftWidth = 4

Variables

EastAsianAmbiguousWidth specifies the monospace width for East Asian characters classified as Ambiguous. The default is 1 but some rare fonts render them with a width of 2.

var EastAsianAmbiguousWidth = 1

func FirstGraphemeCluster ¶

func FirstGraphemeCluster(b []byte, state int) (cluster, rest []byte, width, newState int)

FirstGraphemeCluster returns the first grapheme cluster found in the given byte slice according to the rules of Unicode Standard Annex #29, Grapheme Cluster Boundaries. This function can be called continuously to extract all grapheme clusters from a byte slice, as illustrated in the example below.

If you don't know the current state, for example when calling the function for the first time, you must pass -1. For consecutive calls, pass the state and rest slice returned by the previous call.

The "rest" slice is the sub-slice of the original byte slice "b" starting after the last byte of the identified grapheme cluster. If the length of the "rest" slice is 0, the entire byte slice "b" has been processed. The "cluster" byte slice is the sub-slice of the input slice containing the identified grapheme cluster.

The returned width is the width of the grapheme cluster for most monospace fonts where a value of 1 represents one character cell.

Given an empty byte slice "b", the function returns nil values.

While slightly less convenient than using the Graphemes class, this function has much better performance and makes no allocations. It lends itself well to large byte slices.

b := []byte("🇩🇪🏳️‍🌈!")
state := -1
var c []byte
for len(b) > 0 {
    var width int
    c, b, width, state = uniseg.FirstGraphemeCluster(b, state)
    fmt.Println(string(c), width)
}

🇩🇪 2
🏳️‍🌈 2
! 1

func FirstGraphemeClusterInString ¶

func FirstGraphemeClusterInString(str string, state int) (cluster, rest string, width, newState int)

FirstGraphemeClusterInString is like FirstGraphemeCluster but its input and outputs are strings.

str := "🇩🇪🏳️‍🌈!"
state := -1
var c string
for len(str) > 0 {
    var width int
    c, str, width, state = uniseg.FirstGraphemeClusterInString(str, state)
    fmt.Println(c, width)
}

🇩🇪 2
🏳️‍🌈 2
! 1

func FirstLineSegment ¶

func FirstLineSegment(b []byte, state int) (segment, rest []byte, mustBreak bool, newState int)

FirstLineSegment returns the prefix of the given byte slice after which a decision to break the string over to the next line can or must be made, according to the rules of Unicode Standard Annex #14. This is used to implement line breaking.

Line breaking, also known as word wrapping, is the process of breaking a section of text into lines such that it will fit in the available width of a page, window or other display area.

The returned "segment" may not be broken into smaller parts, unless no other breaking opportunities present themselves, in which case you may break by grapheme clusters (using the FirstGraphemeCluster function to determine the grapheme clusters).

The "mustBreak" flag indicates whether you MUST break the line after the given segment (true), for example after newline characters, or you MAY break the line after the given segment (false).

This function can be called continuously to extract all non-breaking sub-sets from a byte slice, as illustrated in the example below.

If you don't know the current state, for example when calling the function for the first time, you must pass -1. For consecutive calls, pass the state and rest slice returned by the previous call.

The "rest" slice is the sub-slice of the original byte slice "b" starting after the last byte of the identified line segment. If the length of the "rest" slice is 0, the entire byte slice "b" has been processed. The "segment" byte slice is the sub-slice of the input slice containing the identified line segment.

Given an empty byte slice "b", the function returns nil values.

Note that in accordance with UAX #14 LB3, the final segment will end with "mustBreak" set to true. You can choose to ignore this by checking if the length of the "rest" slice is 0 and calling HasTrailingLineBreak or HasTrailingLineBreakInString on the last rune.

Note also that this algorithm may break within grapheme clusters. This is addressed in Section 8.2 Example 6 of UAX #14. To avoid this, you can use the Step function instead.

b := []byte("First line.\nSecond line.")
state := -1
var (
    c         []byte
    mustBreak bool
)
for len(b) > 0 {
    c, b, mustBreak, state = uniseg.FirstLineSegment(b, state)
    fmt.Printf("(%s)", string(c))
    if mustBreak {
        fmt.Print("!")
    }
}

(First )(line.
)!(Second )(line.)!

func FirstLineSegmentInString ¶

func FirstLineSegmentInString(str string, state int) (segment, rest string, mustBreak bool, newState int)

FirstLineSegmentInString is like FirstLineSegment but its input and outputs are strings.

str := "First line.\nSecond line."
state := -1
var (
    c         string
    mustBreak bool
)
for len(str) > 0 {
    c, str, mustBreak, state = uniseg.FirstLineSegmentInString(str, state)
    fmt.Printf("(%s)", c)
    if mustBreak {
        fmt.Println(" < must break")
    } else {
        fmt.Println(" < may break")
    }
}

(First ) < may break
(line.
) < must break
(Second ) < may break
(line.) < must break

func FirstSentence ¶

func FirstSentence(b []byte, state int) (sentence, rest []byte, newState int)

FirstSentence returns the first sentence found in the given byte slice according to the rules of Unicode Standard Annex #29, Sentence Boundaries. This function can be called continuously to extract all sentences from a byte slice, as illustrated in the example below.

If you don't know the current state, for example when calling the function for the first time, you must pass -1. For consecutive calls, pass the state and rest slice returned by the previous call.

The "rest" slice is the sub-slice of the original byte slice "b" starting after the last byte of the identified sentence. If the length of the "rest" slice is 0, the entire byte slice "b" has been processed. The "sentence" byte slice is the sub-slice of the input slice containing the identified sentence.

Given an empty byte slice "b", the function returns nil values.

b := []byte("This is sentence 1.0. And this is sentence two.")
state := -1
var c []byte
for len(b) > 0 {
    c, b, state = uniseg.FirstSentence(b, state)
    fmt.Printf("(%s)\n", string(c))
}

(This is sentence 1.0. )
(And this is sentence two.)

func FirstSentenceInString ¶

func FirstSentenceInString(str string, state int) (sentence, rest string, newState int)

FirstSentenceInString is like FirstSentence but its input and outputs are strings.

str := "This is sentence 1.0. And this is sentence two."
state := -1
var c string
for len(str) > 0 {
    c, str, state = uniseg.FirstSentenceInString(str, state)
    fmt.Printf("(%s)\n", c)
}

(This is sentence 1.0. )
(And this is sentence two.)

func FirstWord ¶

func FirstWord(b []byte, state int) (word, rest []byte, newState int)

FirstWord returns the first word found in the given byte slice according to the rules of Unicode Standard Annex #29, Word Boundaries. This function can be called continuously to extract all words from a byte slice, as illustrated in the example below.

If you don't know the current state, for example when calling the function for the first time, you must pass -1. For consecutive calls, pass the state and rest slice returned by the previous call.

The "rest" slice is the sub-slice of the original byte slice "b" starting after the last byte of the identified word. If the length of the "rest" slice is 0, the entire byte slice "b" has been processed. The "word" byte slice is the sub-slice of the input slice containing the identified word.

Given an empty byte slice "b", the function returns nil values.

b := []byte("Hello, world!")
state := -1
var c []byte
for len(b) > 0 {
    c, b, state = uniseg.FirstWord(b, state)
    fmt.Printf("(%s)\n", string(c))
}

(Hello)
(,)
( )
(world)
(!)

func FirstWordInString ¶

func FirstWordInString(str string, state int) (word, rest string, newState int)

FirstWordInString is like FirstWord but its input and outputs are strings.

str := "Hello, world!"
state := -1
var c string
for len(str) > 0 {
    c, str, state = uniseg.FirstWordInString(str, state)
    fmt.Printf("(%s)\n", c)
}

(Hello)
(,)
( )
(world)
(!)

func GraphemeClusterCount ¶

func GraphemeClusterCount(s string) (n int)

GraphemeClusterCount returns the number of user-perceived characters (grapheme clusters) for the given string.

n := uniseg.GraphemeClusterCount("🇩🇪🏳️‍🌈")
fmt.Println(n)

2

func HasTrailingLineBreak ¶

func HasTrailingLineBreak(b []byte) bool

HasTrailingLineBreak returns true if the last rune in the given byte slice is one of the hard line break code points defined in LB4 and LB5 of UAX #14.

func HasTrailingLineBreakInString ¶

func HasTrailingLineBreakInString(str string) bool

HasTrailingLineBreakInString is like HasTrailingLineBreak but for a string.

func ReverseString ¶

func ReverseString(s string) string

ReverseString reverses the given string while observing grapheme cluster boundaries.

func Step ¶

func Step(b []byte, state int) (cluster, rest []byte, boundaries int, newState int)

Step returns the first grapheme cluster (user-perceived character) found in the given byte slice. It also returns information about the boundary between that grapheme cluster and the one following it as well as the monospace width of the grapheme cluster. There are three types of boundary information: word boundaries, sentence boundaries, and line breaks. This function is therefore a combination of FirstGraphemeCluster, FirstWord, FirstSentence, and FirstLineSegment.

The "boundaries" return value can be evaluated as follows:

• boundaries&MaskWord != 0: The boundary is a word boundary.
• boundaries&MaskWord == 0: The boundary is not a word boundary.
• boundaries&MaskSentence != 0: The boundary is a sentence boundary.
• boundaries&MaskSentence == 0: The boundary is not a sentence boundary.
• boundaries&MaskLine == LineDontBreak: You must not break the line at the boundary.
• boundaries&MaskLine == LineMustBreak: You must break the line at the boundary.
• boundaries&MaskLine == LineCanBreak: You may or may not break the line at the boundary.
• boundaries >> ShiftWidth: The width of the grapheme cluster for most monospace fonts where a value of 1 represents one character cell.

This function can be called continuously to extract all grapheme clusters from a byte slice, as illustrated in the examples below.

If you don't know which state to pass, for example when calling the function for the first time, you must pass -1. For consecutive calls, pass the state and rest slice returned by the previous call.

The "rest" slice is the sub-slice of the original byte slice "b" starting after the last byte of the identified grapheme cluster. If the length of the "rest" slice is 0, the entire byte slice "b" has been processed. The "cluster" byte slice is the sub-slice of the input slice containing the first identified grapheme cluster.

Given an empty byte slice "b", the function returns nil values.

While slightly less convenient than using the Graphemes class, this function has much better performance and makes no allocations. It lends itself well to large byte slices.

Note that in accordance with UAX #14 LB3, the final segment will end with a mandatory line break (boundaries&MaskLine == LineMustBreak). You can choose to ignore this by checking if the length of the "rest" slice is 0 and calling HasTrailingLineBreak or HasTrailingLineBreakInString on the last rune.

b := []byte("🇩🇪🏳️‍🌈!")
state := -1
var c []byte
for len(b) > 0 {
    var boundaries int
    c, b, boundaries, state = uniseg.Step(b, state)
    fmt.Println(string(c), boundaries>>uniseg.ShiftWidth)
}

🇩🇪 2
🏳️‍🌈 2
! 1

b := []byte("First line.\nSecond line.")
state := -1
var (
    c          []byte
    boundaries int
)
for len(b) > 0 {
    c, b, boundaries, state = uniseg.Step(b, state)
    fmt.Print(string(c))
    if boundaries&uniseg.MaskLine == uniseg.LineCanBreak {
        fmt.Print("|")
    } else if boundaries&uniseg.MaskLine == uniseg.LineMustBreak {
        fmt.Print("‖")
    }
}

First |line.
‖Second |line.‖

b := []byte("This is sentence 1.0. And this is sentence two.")
state := -1
var (
    c          []byte
    boundaries int
)
for len(b) > 0 {
    c, b, boundaries, state = uniseg.Step(b, state)
    fmt.Print(string(c))
    if boundaries&uniseg.MaskSentence != 0 {
        fmt.Print("|")
    }
}

This is sentence 1.0. |And this is sentence two.|

b := []byte("Hello, world!")
state := -1
var (
    c          []byte
    boundaries int
)
for len(b) > 0 {
    c, b, boundaries, state = uniseg.Step(b, state)
    fmt.Print(string(c))
    if boundaries&uniseg.MaskWord != 0 {
        fmt.Print("|")
    }
}

Hello|,| |world|!|

func StepString ¶

func StepString(str string, state int) (cluster, rest string, boundaries int, newState int)

StepString is like Step but its input and outputs are strings.

str := "🇩🇪🏳️‍🌈!"
state := -1
var c string
for len(str) > 0 {
    var boundaries int
    c, str, boundaries, state = uniseg.StepString(str, state)
    fmt.Println(c, boundaries>>uniseg.ShiftWidth)
}

🇩🇪 2
🏳️‍🌈 2
! 1

str := "First line.\nSecond line."
state := -1
var (
    c          string
    boundaries int
)
for len(str) > 0 {
    c, str, boundaries, state = uniseg.StepString(str, state)
    fmt.Print(c)
    if boundaries&uniseg.MaskLine == uniseg.LineCanBreak {
        fmt.Print("|")
    } else if boundaries&uniseg.MaskLine == uniseg.LineMustBreak {
        fmt.Print("‖")
    }
}

First |line.
‖Second |line.‖

str := "This is sentence 1.0. And this is sentence two."
state := -1
var (
    c          string
    boundaries int
)
for len(str) > 0 {
    c, str, boundaries, state = uniseg.StepString(str, state)
    fmt.Print(c)
    if boundaries&uniseg.MaskSentence != 0 {
        fmt.Print("|")
    }
}

This is sentence 1.0. |And this is sentence two.|

str := "Hello, world!"
state := -1
var (
    c          string
    boundaries int
)
for len(str) > 0 {
    c, str, boundaries, state = uniseg.StepString(str, state)
    fmt.Print(c)
    if boundaries&uniseg.MaskWord != 0 {
        fmt.Print("|")
    }
}

Hello|,| |world|!|

func StringWidth ¶

func StringWidth(s string) (width int)

StringWidth returns the monospace width for the given string, that is, the number of same-size cells to be occupied by the string.

fmt.Println(uniseg.StringWidth("Hello, 世界"))

11

type Graphemes ¶

Graphemes implements an iterator over Unicode grapheme clusters, or user-perceived characters. While iterating, it also provides information about word boundaries, sentence boundaries, line breaks, and monospace character widths.

After constructing the class via NewGraphemes for a given string "str", Graphemes.Next is called for every grapheme cluster in a loop until it returns false. Inside the loop, information about the grapheme cluster as well as boundary information and character width is available via the various methods (see examples below).

This class basically wraps the StepString parser and provides a convenient interface to it. If you are only interested in some parts of this package's functionality, using the specialized functions starting with "First" is almost always faster.

type Graphemes struct {
    // contains filtered or unexported fields
}

g := uniseg.NewGraphemes("🇩🇪🏳️‍🌈")
for g.Next() {
    fmt.Println(g.Str())
}

🇩🇪
🏳️‍🌈

g := uniseg.NewGraphemes("First line.\nSecond line.")
for g.Next() {
    fmt.Print(g.Str())
    if g.LineBreak() == uniseg.LineCanBreak {
        fmt.Print("|")
    } else if g.LineBreak() == uniseg.LineMustBreak {
        fmt.Print("‖")
    }
}
if g.LineBreak() == uniseg.LineMustBreak {
    fmt.Print("\nNo clusters left. LineMustBreak")
}
g.Reset()
if g.LineBreak() == uniseg.LineDontBreak {
    fmt.Print("\nIterator has been reset. LineDontBreak")
}

First |line.
‖Second |line.‖
No clusters left. LineMustBreak
Iterator has been reset. LineDontBreak

g := uniseg.NewGraphemes("This is sentence 1.0. And this is sentence two.")
for g.Next() {
    fmt.Print(g.Str())
    if g.IsSentenceBoundary() {
        fmt.Print("|")
    }
}

This is sentence 1.0. |And this is sentence two.|

g := uniseg.NewGraphemes("Hello, world!")
for g.Next() {
    fmt.Print(g.Str())
    if g.IsWordBoundary() {
        fmt.Print("|")
    }
}

Hello|,| |world|!|

func NewGraphemes ¶

func NewGraphemes(str string) *Graphemes

NewGraphemes returns a new grapheme cluster iterator.

func (*Graphemes) Bytes ¶

func (g *Graphemes) Bytes() []byte

Bytes returns a byte slice which corresponds to the current grapheme cluster. If the iterator is already past the end or Graphemes.Next has not yet been called, nil is returned.

func (*Graphemes) IsSentenceBoundary ¶

func (g *Graphemes) IsSentenceBoundary() bool

IsSentenceBoundary returns true if a sentence ends after the current grapheme cluster.

func (*Graphemes) IsWordBoundary ¶

func (g *Graphemes) IsWordBoundary() bool

IsWordBoundary returns true if a word ends after the current grapheme cluster.

func (*Graphemes) LineBreak ¶

func (g *Graphemes) LineBreak() int

LineBreak returns whether the line can be broken after the current grapheme cluster. A value of LineDontBreak means the line may not be broken, a value of LineMustBreak means the line must be broken, and a value of LineCanBreak means the line may or may not be broken.

func (*Graphemes) Next ¶

func (g *Graphemes) Next() bool

Next advances the iterator by one grapheme cluster and returns false if no clusters are left. This function must be called before the first cluster is accessed.

func (*Graphemes) Positions ¶

func (g *Graphemes) Positions() (int, int)

Positions returns the interval of the current grapheme cluster as byte positions into the original string. The first returned value "from" indexes the first byte and the second returned value "to" indexes the first byte that is not included anymore, i.e. str[from:to] is the current grapheme cluster of the original string "str". If Graphemes.Next has not yet been called, both values are 0. If the iterator is already past the end, both values are 1.

func (*Graphemes) Reset ¶

func (g *Graphemes) Reset()

Reset puts the iterator into its initial state such that the next call to Graphemes.Next sets it to the first grapheme cluster again.

func (*Graphemes) Runes ¶

func (g *Graphemes) Runes() []rune

Runes returns a slice of runes (code points) which corresponds to the current grapheme cluster. If the iterator is already past the end or Graphemes.Next has not yet been called, nil is returned.

func (*Graphemes) Str ¶

func (g *Graphemes) Str() string

Str returns a substring of the original string which corresponds to the current grapheme cluster. If the iterator is already past the end or Graphemes.Next has not yet been called, an empty string is returned.

func (*Graphemes) Width ¶

func (g *Graphemes) Width() int

Width returns the monospace width of the current grapheme cluster.