alex/yomichan-import
1
Fork 0
Code Releases Activity

wip

Browse Source
This commit is contained in:
Alex Yatskov 2016-12-27 17:52:50 -08:00
parent fe530cd2d3
commit 4835f58758
1 changed files with 8 additions and 123 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

131
daijirin.go
View File

@ -34,16 +34,18 @@ type daijirinExtractor struct {
metaExp *regexp.Regexp
v5Exp *regexp.Regexp
v1Exp *regexp.Regexp
tagExp *regexp.Regexp
}
func makeDaijirinExtractor() epwingExtractor {
return &daijirinExtractor{
partsExp: regexp.MustCompile(`([^（【〖]+)(?:【(.*)】)?(?:〖(.*)〗)?(?:（(.*)）)?`),
readGroupExp: regexp.MustCompile(`[-・]+`),
expVarExp: regexp.MustCompile(`\((.*)\)`),
metaExp: regexp.MustCompile(`（(.*)）`),
v5Exp: regexp.MustCompile(`(動.五)|(動..二)`),
v1Exp: regexp.MustCompile(`動..一`),
expVarExp: regexp.MustCompile(`\(([^\)]*)\)`),
metaExp: regexp.MustCompile(`（([^）]*)）`),
v5Exp: regexp.MustCompile(`(動.[四五](\[[^\]]+\])?)|(動..二)`),
v1Exp: regexp.MustCompile(`(動..一)`),
tagExp: regexp.MustCompile(`(動.[四五](\[[^\]]+\])?)|(動..二)|(動..一)|(名)|(形動)|(副)|(連語)|(形)|(枕詞)|(代)|(感)|(接尾)|(助動)|(接続)|(接頭)|(連体)|(終助)|(接助)|(副助)|(係助)|(格助)|(間投助)`),
}
}
@ -138,126 +140,9 @@ func (*daijirinExtractor) getRevision() string {
}
func (e *daijirinExtractor) exportTags(term *dbTerm, tags []string) {
parsed := []string{
"並立助",
"代",
"係助",
"副",
"副助",
"助動",
"動",
"動ア上一",
"動ア下一",
"動ア下二",
"動ア五［ハ四］",
"動カ上一",
"動カ上二",
"動カ下一",
"動カ下二",
"動カ五",
"動カ五［四］",
"動カ四",
"動カ変",
"動ガ上一",
"動ガ上二",
"動ガ下一",
"動ガ下二",
"動ガ五",
"動ガ五［四］",
"動ガ四",
"動サ上一",
"動サ下一",
"動サ下二",
"動サ五",
"動サ五［四］",
"動サ四",
"動サ変",
"動サ特活",
"動ザ上一",
"動ザ上二",
"動ザ下一",
"動ザ下二",
"動タ上一",
"動タ上二",
"動タ下一",
"動タ下二",
"動タ五［四］",
"動タ四",
"動ダ上二",
"動ダ下一",
"動ダ下二",
"動ナ上一",
"動ナ下一",
"動ナ下二",
"動ナ五",
"動ナ五［四］",
"動ハ上一",
"動ハ上二",
"動ハ下一",
"動ハ下二",
"動ハ四",
"動ハ特活",
"動バ上一",
"動バ上二",
"動バ下一",
"動バ下二",
"動バ五［四］",
"動バ四",
"動マ上一",
"動マ上二",
"動マ下一",
"動マ下二",
"動マ五",
"動マ五［四］",
"動マ四",
"動マ特活",
"動マ特活",
"動ヤ上一",
"動ヤ上二",
"動ヤ下二",
"動ラ上一",
"動ラ上二",
"動ラ下一",
"動ラ下二",
"動ラ五",
"動ラ五［四］",
"動ラ四",
"動ラ変",
"動ラ特活",
"動ワ上一",
"動ワ上二",
"動ワ下一",
"動ワ下二",
"動ワ五",
"動ワ五［ハ四］",
"動五［四］",
"動特活",
"動詞五［四］段型活用",
"名",
"形",
"形ク",
"形シク",
"形動",
"形動タリ",
"形動ナリ",
"感",
"接助",
"接尾",
"接続",
"接頭",
"枕詞",
"格助",
"終助",
"連体",
"連語",
"間投助",
}
for _, tag := range tags {
for _, p := range parsed {
if tag == p {
term.addTags(tag)
}
if e.tagExp.MatchString(tag) {
term.addTags(tag)
}
}
}