func cutDetail(sentence string) []WordTag {
result := make([]WordTag, 0)
re_han := regexp.MustCompile(`\p{Han}+`)
re_skip := regexp.MustCompile(`[[\.[:digit:]]+|[:alnum:]]+`)
re_eng := regexp.MustCompile(`[[:alnum:]]`)
re_num := regexp.MustCompile(`[\.[:digit:]]+`)
blocks := jiebago.RegexpSplit(re_han, sentence)
for _, blk := range blocks {
if re_han.MatchString(blk) {
for _, wordTag := range __cut(blk) {
result = append(result, wordTag)
}
} else {
for _, x := range jiebago.RegexpSplit(re_skip, blk) {
if len(x) == 0 {
continue
}
switch {
case re_num.MatchString(x):
result = append(result, WordTag{x, "m"})
case re_eng.MatchString(x):
result = append(result, WordTag{x, "eng"})
default:
result = append(result, WordTag{x, "x"})
}
}
}
}
return result
}
func cut(sentence string, HMM bool) []WordTag {
result := make([]WordTag, 0)
re_han := regexp.MustCompile(`([\p{Han}+[:alnum:]+#&\._]+)`)
re_skip := regexp.MustCompile(`(\r\n|\s)`)
re_eng := regexp.MustCompile(`[[:alnum:]]`)
re_num := regexp.MustCompile(`[\.[:digit:]]+`)
blocks := jiebago.RegexpSplit(re_han, sentence)
var cut_block cutAction
if HMM {
cut_block = cut_DAG
} else {
cut_block = cut_DAG_NO_HMM
}
for _, blk := range blocks {
if re_han.MatchString(blk) {
for _, wordTag := range cut_block(blk) {
result = append(result, wordTag)
}
} else {
for _, x := range jiebago.RegexpSplit(re_skip, blk) {
if re_skip.MatchString(x) {
result = append(result, WordTag{x, "x"})
} else {
for _, xx := range x {
s := string(xx)
switch {
case re_num.MatchString(s):
result = append(result, WordTag{s, "m"})
case re_eng.MatchString(x):
result = append(result, WordTag{x, "eng"})
break
default:
result = append(result, WordTag{s, "x"})
}
}
}
}
}
}
return result
}