1
0
mirror of https://github.com/fumiama/jieba.git synced 2026-06-05 00:32:51 +08:00

added AddWord/DeleteWord/SuggestFrequency functions, this is correpsonding to jieba commit #59aa8b69b1399569ea6b417280c993da703baba8

This commit is contained in:
Wang Bin
2015-05-08 11:57:46 +08:00
parent 3d91f615cf
commit c48eb5b4a7
2 changed files with 67 additions and 19 deletions

View File

@@ -60,28 +60,28 @@ func (jt *JiebaTokenizer) Tokenize(input []byte) analysis.TokenStream {
pos := 1
var width int
var gram string
dict := jt.seg.Dictionary()
for word := range jt.seg.Cut(string(input), jt.hmm) {
if jt.searchMode {
runes := []rune(word)
width = len(runes)
for _, step := range [2]int{2, 3} {
if width > step {
for i := 0; i < width-step+1; i++ {
gram = string(runes[i : i+step])
gramLen := len(gram)
if frequency, ok := dict.Frequency(gram); ok && frequency > 0 {
gramStart := start + len(string(runes[:i]))
token := analysis.Token{
Term: []byte(gram),
Start: gramStart,
End: gramStart + gramLen,
Position: pos,
Type: detectTokenType(gram),
}
rv = append(rv, &token)
pos++
if width <= step {
continue
}
for i := 0; i < width-step+1; i++ {
gram = string(runes[i : i+step])
gramLen := len(gram)
if frequency, ok := jt.seg.Frequency(gram); ok && frequency > 0 {
gramStart := start + len(string(runes[:i]))
token := analysis.Token{
Term: []byte(gram),
Start: gramStart,
End: gramStart + gramLen,
Position: pos,
Type: detectTokenType(gram),
}
rv = append(rv, &token)
pos++
}
}
}