Add japanese document language

This commit is contained in:
eikek
2021-07-28 20:05:38 +02:00
parent 0104cdc825
commit f994d4b248
7 changed files with 66 additions and 2 deletions

View File

@ -22,7 +22,7 @@ object DateFind {
def findDates(text: String, lang: Language): Stream[Pure, NerDateLabel] =
TextSplitter
.splitToken(text, " \t.,\n\r/".toSet)
.splitToken(text, " \t.,\n\r/年月日".toSet)
.filter(w => lang != Language.Latvian || w.value != "gada")
.sliding(3)
.filter(_.size == 3)
@ -89,6 +89,7 @@ object DateFind {
case Language.Swedish => ymd.or(dmy).or(mdy)
case Language.Dutch => dmy.or(ymd).or(mdy)
case Language.Latvian => dmy.or(lavLong).or(ymd)
case Language.Japanese => ymd
}
p.read(parts) match {
case Result.Success(sds, _) =>

View File

@ -50,6 +50,8 @@ object MonthName {
russian
case Language.Latvian =>
latvian
case Language.Japanese =>
japanese
}
private val numbers = List(
@ -290,4 +292,19 @@ object MonthName {
List("novembris", "nov."),
List("decembris", "dec.")
)
private val japanese = List(
List("1", "一"),
List("2", "二"),
List("3", "三"),
List("4", "四"),
List("5", "五"),
List("6", "六"),
List("7", "七"),
List("8", "八"),
List("9", "九"),
List("10", "十"),
List("11", "十一"),
List("12", "十二")
)
}

View File

@ -143,4 +143,29 @@ class DateFindSpec extends FunSuite {
)
}
test("find japanese dates") {
assertEquals(
DateFind
.findDates("some text in japanese 2021.7.21 and more", Language.Japanese)
.toVector,
Vector(
NerDateLabel(
LocalDate.of(2021, 7, 21),
NerLabel("2021.7.21", NerTag.Date, 22, 31)
)
)
)
assertEquals(
DateFind
.findDates("some text in japanese 2021年7月21日 and more", Language.Japanese)
.toVector,
Vector(
NerDateLabel(
LocalDate.of(2021, 7, 21),
NerLabel("2021年7月21", NerTag.Date, 22, 31)
)
)
)
}
}