1
This commit is contained in:
Alex Yatskov 2016-08-24 09:02:26 -07:00
parent d81d94fbcd
commit 1253ab2912
4 changed files with 101 additions and 75 deletions

View File

@ -31,11 +31,13 @@ import (
)
const (
REF_STEP_COUNT = 50000
BANK_STRIDE = 50000
DB_VERSION = 0
)
type termJson struct {
Refs int `json:"refs"`
type termIndex struct {
Version int `json:"version"`
Banks int `json:"banks"`
Entities [][]string `json:"ents"`
defs [][]string
}
@ -68,12 +70,10 @@ func (s *termSource) addTagsPri(tags ...string) {
}
}
func buildTermJson(entries []termSource, entities map[string]string) termJson {
var dict termJson
for name, value := range entities {
ent := []string{name, value}
dict.Entities = append(dict.Entities, ent)
func buildTermIndex(entries []termSource, entities map[string]string) termIndex {
dict := termIndex{
Version: DB_VERSION,
Banks: bankCount(len(entries)),
}
for _, e := range entries {
@ -82,20 +82,15 @@ func buildTermJson(entries []termSource, entities map[string]string) termJson {
dict.defs = append(dict.defs, def)
}
dict.Refs = len(dict.defs) / REF_STEP_COUNT
for name, value := range entities {
ent := []string{name, value}
dict.Entities = append(dict.Entities, ent)
}
return dict
}
func marshalJson(obj interface{}, pretty bool) ([]byte, error) {
if pretty {
return json.MarshalIndent(obj, "", " ")
}
return json.Marshal(obj)
}
func outputTermJson(outputDir string, entries []termSource, entities map[string]string, pretty bool) error {
func outputTermIndex(outputDir string, entries []termSource, entities map[string]string, pretty bool) error {
if err := os.MkdirAll(outputDir, 0755); err != nil {
return err
}
@ -106,8 +101,7 @@ func outputTermJson(outputDir string, entries []termSource, entities map[string]
}
defer outputIndex.Close()
dict := buildTermJson(entries, entities)
dict := buildTermIndex(entries, entities)
indexBytes, err := marshalJson(dict, pretty)
if err != nil {
return err
@ -118,16 +112,15 @@ func outputTermJson(outputDir string, entries []termSource, entities map[string]
}
defCnt := len(dict.defs)
for i := 0; i < defCnt; i += REF_STEP_COUNT {
outputRef, err := os.Create(path.Join(outputDir, fmt.Sprintf("ref_%d.json", i/REF_STEP_COUNT)))
for i := 0; i < defCnt; i += BANK_STRIDE {
outputRef, err := os.Create(path.Join(outputDir, fmt.Sprintf("bank_%d.json", i/BANK_STRIDE+1)))
if err != nil {
return err
}
defer outputRef.Close()
indexSrc := i
indexDst := i + REF_STEP_COUNT
indexDst := i + BANK_STRIDE
if indexDst > defCnt {
indexDst = defCnt
}
@ -145,6 +138,23 @@ func outputTermJson(outputDir string, entries []termSource, entities map[string]
return nil
}
func marshalJson(obj interface{}, pretty bool) ([]byte, error) {
if pretty {
return json.MarshalIndent(obj, "", " ")
}
return json.Marshal(obj)
}
func bankCount(defCount int) int {
count := defCount / BANK_STRIDE
if defCount%BANK_STRIDE > 0 {
count += 1
}
return count
}
func hasString(needle string, haystack []string) bool {
for _, value := range haystack {
if needle == value {

View File

@ -98,5 +98,5 @@ func outputEdictJson(outputDir string, reader io.Reader, flags int) error {
entries = append(entries, convertEdictEntry(e)...)
}
return outputTermJson(outputDir, entries, entities, flags&flagPrettyJson == flagPrettyJson)
return outputTermIndex(outputDir, entries, entities, flags&flagPrettyJson == flagPrettyJson)
}

View File

@ -84,5 +84,5 @@ func outputJmnedictJson(outputDir string, reader io.Reader, flags int) error {
entries = append(entries, convertJmnedictEntry(e)...)
}
return outputTermJson(outputDir, entries, entities, flags&flagPrettyJson == flagPrettyJson)
return outputTermIndex(outputDir, entries, entities, flags&flagPrettyJson == flagPrettyJson)
}

View File

@ -23,25 +23,20 @@
package main
import (
"encoding/json"
"fmt"
"io"
"os"
"path"
"strconv"
"strings"
"github.com/FooSoft/jmdict"
)
type kanjiDefJson struct {
Character string `json:"c"`
Onyomi string `json:"o"`
Kunyomi string `json:"k"`
Tags string `json:"t"`
Meanings []string `json:"m"`
}
type kanjiJson struct {
Defs []kanjiDefJson `json:"d"`
type kanjiIndex struct {
Version int `json:"version"`
Banks int `json:"banks"`
defs [][]string
}
type kanjiSource struct {
@ -60,46 +55,69 @@ func (s *kanjiSource) addTags(tags ...string) {
}
}
func buildKanjiJson(kanji []kanjiSource) kanjiJson {
var dict kanjiJson
for _, k := range kanji {
def := kanjiDefJson{
Character: k.Character,
Onyomi: strings.Join(k.Onyomi, " "),
Kunyomi: strings.Join(k.Kunyomi, " "),
Tags: strings.Join(k.Tags, " "),
Meanings: k.Meanings,
func buildKanjiIndex(entries []kanjiSource) kanjiIndex {
dict := kanjiIndex{
Version: DB_VERSION,
Banks: bankCount(len(entries)),
}
dict.Defs = append(dict.Defs, def)
for _, e := range entries {
def := []string{e.Character, strings.Join(e.Onyomi, " "), strings.Join(e.Kunyomi, " "), strings.Join(e.Tags, " ")}
def = append(def, e.Meanings...)
dict.defs = append(dict.defs, def)
}
return dict
}
func outputKanjiJson(writer io.Writer, kanji []kanjiSource, pretty bool) error {
dict := buildKanjiJson(kanji)
var (
bytes []byte
err error
)
if pretty {
bytes, err = json.MarshalIndent(dict, "", " ")
} else {
bytes, err = json.Marshal(dict)
func outputKanjiIndex(outputDir string, entries []kanjiSource, pretty bool) error {
if err := os.MkdirAll(outputDir, 0755); err != nil {
return err
}
outputIndex, err := os.Create(path.Join(outputDir, "index.json"))
if err != nil {
return err
}
defer outputIndex.Close()
dict := buildKanjiIndex(entries)
indexBytes, err := marshalJson(dict, pretty)
if err != nil {
return err
}
_, err = writer.Write(bytes)
if _, err = outputIndex.Write(indexBytes); err != nil {
return err
}
defCnt := len(dict.defs)
for i := 0; i < defCnt; i += BANK_STRIDE {
outputRef, err := os.Create(path.Join(outputDir, fmt.Sprintf("bank_%d.json", i/BANK_STRIDE+1)))
if err != nil {
return err
}
defer outputRef.Close()
indexSrc := i
indexDst := i + BANK_STRIDE
if indexDst > defCnt {
indexDst = defCnt
}
refBytes, err := marshalJson(dict.defs[indexSrc:indexDst], pretty)
if err != nil {
return err
}
if _, err = outputRef.Write(refBytes); err != nil {
return err
}
}
return nil
}
func convertKanjidicCharacter(kanjidicCharacter jmdict.KanjidicCharacter) kanjiSource {
character := kanjiSource{Character: kanjidicCharacter.Literal}
@ -151,17 +169,15 @@ func convertKanjidicCharacter(kanjidicCharacter jmdict.KanjidicCharacter) kanjiS
}
func outputKanjidicJson(outputDir string, reader io.Reader, flags int) error {
// dict, err := jmdict.LoadKanjidic(reader)
// if err != nil {
// return err
// }
// var kanji []kanjiSource
// for _, kanjidicCharacter := range dict.Characters {
// kanji = append(kanji, convertKanjidicCharacter(kanjidicCharacter))
// }
// return outputKanjiJson(writer, kanji, flags&flagPrettyJson == flagPrettyJson)
return nil
dict, err := jmdict.LoadKanjidic(reader)
if err != nil {
return err
}
var kanji []kanjiSource
for _, kanjidicCharacter := range dict.Characters {
kanji = append(kanji, convertKanjidicCharacter(kanjidicCharacter))
}
return outputKanjiIndex(outputDir, kanji, flags&flagPrettyJson == flagPrettyJson)
}