2021-01-01 22:31:58 +00:00
|
|
|
package yomichan
|
2016-08-07 01:17:02 +00:00
|
|
|
|
|
|
|
import (
|
2016-12-19 01:46:40 +00:00
|
|
|
"os"
|
2016-08-07 01:17:02 +00:00
|
|
|
|
2022-07-04 03:59:33 +00:00
|
|
|
"foosoft.net/projects/jmdict"
|
2016-08-07 01:17:02 +00:00
|
|
|
)
|
|
|
|
|
2017-09-10 20:25:11 +00:00
|
|
|
const jmnedictRevision = "jmnedict1"
|
2016-12-24 05:52:49 +00:00
|
|
|
|
2017-09-10 20:25:11 +00:00
|
|
|
func jmnedictBuildTagMeta(entities map[string]string) dbTagList {
|
|
|
|
var tags dbTagList
|
2016-12-17 23:48:13 +00:00
|
|
|
|
|
|
|
for name, value := range entities {
|
2017-09-10 20:25:11 +00:00
|
|
|
tag := dbTag{Name: name, Notes: value}
|
2016-12-17 23:48:13 +00:00
|
|
|
|
|
|
|
switch name {
|
2016-12-18 01:08:38 +00:00
|
|
|
case "company", "fem", "given", "masc", "organization", "person", "place", "product", "station", "surname", "unclass", "work":
|
2016-12-18 01:39:24 +00:00
|
|
|
tag.Category = "name"
|
2016-12-17 23:48:13 +00:00
|
|
|
tag.Order = 4
|
|
|
|
}
|
|
|
|
|
2017-09-10 20:25:11 +00:00
|
|
|
tags = append(tags, tag)
|
2016-12-17 23:48:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return tags
|
|
|
|
}
|
|
|
|
|
2016-12-18 19:46:47 +00:00
|
|
|
func jmnedictExtractTerms(enamdictEntry jmdict.JmnedictEntry) []dbTerm {
|
2016-11-05 20:13:13 +00:00
|
|
|
var terms []dbTerm
|
2016-08-07 01:17:02 +00:00
|
|
|
|
|
|
|
convert := func(reading jmdict.JmnedictReading, kanji *jmdict.JmnedictKanji) {
|
|
|
|
if kanji != nil && hasString(kanji.Expression, reading.Restrictions) {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
2016-11-05 20:13:13 +00:00
|
|
|
var term dbTerm
|
2017-10-12 23:48:58 +00:00
|
|
|
term.Sequence = enamdictEntry.Sequence
|
|
|
|
term.addTermTags(reading.Information...)
|
2016-11-05 20:13:13 +00:00
|
|
|
|
2016-08-07 01:17:02 +00:00
|
|
|
if kanji == nil {
|
2016-11-05 20:13:13 +00:00
|
|
|
term.Expression = reading.Reading
|
2016-08-07 01:17:02 +00:00
|
|
|
} else {
|
2016-11-05 20:13:13 +00:00
|
|
|
term.Expression = kanji.Expression
|
|
|
|
term.Reading = reading.Reading
|
2017-10-12 23:48:58 +00:00
|
|
|
term.addTermTags(kanji.Information...)
|
2016-08-07 01:17:02 +00:00
|
|
|
|
2016-11-05 20:13:13 +00:00
|
|
|
for _, priority := range kanji.Priorities {
|
|
|
|
if hasString(priority, reading.Priorities) {
|
2017-10-12 23:48:58 +00:00
|
|
|
term.addTermTags(priority)
|
2016-11-05 20:13:13 +00:00
|
|
|
}
|
|
|
|
}
|
2016-08-07 01:17:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
for _, trans := range enamdictEntry.Translations {
|
2023-01-22 20:14:33 +00:00
|
|
|
for _, translation := range trans.Translations {
|
|
|
|
term.Glossary = append(term.Glossary, translation)
|
|
|
|
}
|
2017-10-12 23:48:58 +00:00
|
|
|
term.addDefinitionTags(trans.NameTypes...)
|
2016-08-07 01:17:02 +00:00
|
|
|
}
|
|
|
|
|
2016-11-05 20:13:13 +00:00
|
|
|
terms = append(terms, term)
|
2016-08-07 01:17:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if len(enamdictEntry.Kanji) > 0 {
|
|
|
|
for _, kanji := range enamdictEntry.Kanji {
|
|
|
|
for _, reading := range enamdictEntry.Readings {
|
|
|
|
convert(reading, &kanji)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
for _, reading := range enamdictEntry.Readings {
|
|
|
|
convert(reading, nil)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-11-05 20:13:13 +00:00
|
|
|
return terms
|
2016-08-07 01:17:02 +00:00
|
|
|
}
|
|
|
|
|
2017-06-26 00:22:17 +00:00
|
|
|
func jmnedictExportDb(inputPath, outputPath, language, title string, stride int, pretty bool) error {
|
2016-12-19 01:46:40 +00:00
|
|
|
reader, err := os.Open(inputPath)
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
defer reader.Close()
|
|
|
|
|
2016-08-07 01:17:02 +00:00
|
|
|
dict, entities, err := jmdict.LoadJmnedictNoTransform(reader)
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
|
2016-11-05 20:13:13 +00:00
|
|
|
var terms dbTermList
|
2017-06-10 23:53:31 +00:00
|
|
|
for _, entry := range dict.Entries {
|
|
|
|
terms = append(terms, jmnedictExtractTerms(entry)...)
|
2016-08-07 01:17:02 +00:00
|
|
|
}
|
|
|
|
|
2016-12-29 01:45:33 +00:00
|
|
|
if title == "" {
|
|
|
|
title = "JMnedict"
|
|
|
|
}
|
|
|
|
|
2017-09-10 20:25:11 +00:00
|
|
|
recordData := map[string]dbRecordList{
|
2017-09-10 20:45:06 +00:00
|
|
|
"term": terms.crush(),
|
|
|
|
"tag": jmnedictBuildTagMeta(entities).crush(),
|
2017-09-10 20:25:11 +00:00
|
|
|
}
|
|
|
|
|
2016-11-05 20:13:13 +00:00
|
|
|
return writeDb(
|
2017-06-26 00:22:17 +00:00
|
|
|
outputPath,
|
2016-11-06 00:09:23 +00:00
|
|
|
title,
|
2017-09-10 20:25:11 +00:00
|
|
|
jmnedictRevision,
|
2017-10-29 18:42:36 +00:00
|
|
|
true,
|
2017-09-10 20:25:11 +00:00
|
|
|
recordData,
|
2016-12-29 01:45:33 +00:00
|
|
|
stride,
|
2016-12-19 01:31:27 +00:00
|
|
|
pretty,
|
2016-11-05 20:13:13 +00:00
|
|
|
)
|
2016-08-07 01:17:02 +00:00
|
|
|
}
|