1
This commit is contained in:
Alex Yatskov 2016-12-21 22:41:27 -08:00
parent 269ad27cfa
commit f222e002c6

View File

@ -68,9 +68,11 @@ func (e *daijirinExtractor) extractTerms(entry epwingEntry) []dbTerm {
} }
var tags []string var tags []string
if matches := e.annotExp.FindStringSubmatch(strings.Split(entry.Text, "\n")[0]); matches != nil { for _, split := range strings.Split(entry.Text, "\n") {
for _, tag := range strings.Split(matches[1], "・") { if matches := e.annotExp.FindStringSubmatch(split); matches != nil {
tags = append(tags, tag) for _, tag := range strings.Split(matches[1], "・") {
tags = append(tags, tag)
}
} }
} }
@ -83,6 +85,8 @@ func (e *daijirinExtractor) extractTerms(entry epwingEntry) []dbTerm {
} }
e.exportTags(&term, tags) e.exportTags(&term, tags)
e.exportRules(&term, tags)
terms = append(terms, term) terms = append(terms, term)
} }
@ -96,6 +100,8 @@ func (e *daijirinExtractor) extractTerms(entry epwingEntry) []dbTerm {
} }
e.exportTags(&term, tags) e.exportTags(&term, tags)
e.exportRules(&term, tags)
terms = append(terms, term) terms = append(terms, term)
} }
} }
@ -108,6 +114,46 @@ func (*daijirinExtractor) extractKanji(entry epwingEntry) []dbKanji {
return nil return nil
} }
func (e *daijirinExtractor) exportRules(term *dbTerm, tags []string) {
v5 := []string{
"動ワ五",
"動カ下二", "動カ五",
"動ガ下二", "動ガ五",
"動サ五",
"動タ五",
"動ナ五",
"動バ五",
"動マ五",
"動ラ五",
}
v1 := []string{
"動バ下一",
}
tagLoop:
for _, tag := range tags {
if tag == "形" {
term.addRules("adj-i")
continue tagLoop
}
for _, v := range v5 {
if strings.HasPrefix(tag, v) {
term.addRules("v5")
continue tagLoop
}
}
for _, v := range v1 {
if strings.HasPrefix(tag, v) {
term.addRules("v1")
continue tagLoop
}
}
}
}
func (e *daijirinExtractor) exportTags(term *dbTerm, tags []string) { func (e *daijirinExtractor) exportTags(term *dbTerm, tags []string) {
parsed := []string{ parsed := []string{
"並立助", "並立助",