alex/jmdict
1
Fork 0
Code Releases Activity

WIP

Browse Source
This commit is contained in:
Alex Yatskov 2016-07-31 14:01:52 -07:00
parent 8efc580f80
commit 74b76375a8
3 changed files with 14 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
common.go
View File

@ -28,7 +28,9 @@ import (
"regexp"
)
func parseEntries(reader io.Reader, callback func(decoder *xml.Decoder, element *xml.StartElement) error) (map[string]string, error) {
type Parser func(decoder *xml.Decoder, element *xml.StartElement) error
func parseEntries(reader io.Reader, transform bool, callback Parser) (map[string]string, error) {
decoder := xml.NewDecoder(reader)
for {
@ -40,7 +42,7 @@ func parseEntries(reader io.Reader, callback func(decoder *xml.Decoder, element
switch startElement := token.(type) {
case xml.Directive:
directive := token.(xml.Directive)
decoder.Entity = parseEntities(&directive)
decoder.Entity = parseEntities(&directive, transform)
case xml.StartElement:
if err := callback(decoder, &startElement); err != nil {
return nil, err
@ -51,13 +53,17 @@ func parseEntries(reader io.Reader, callback func(decoder *xml.Decoder, element
return decoder.Entity, nil
}
func parseEntities(d *xml.Directive) map[string]string {
func parseEntities(d *xml.Directive, transform bool) map[string]string {
re := regexp.MustCompile("<!ENTITY\\s([0-9\\-A-z]+)\\s\"(.+)\">")
matches := re.FindAllStringSubmatch(string(*d), -1)
entities := make(map[string]string)
for _, match := range matches {
entities[match[1]] = match[2]
if transform {
entities[match[1]] = match[2]
} else {
entities[match[1]] = match[1]
}
}
return entities

4
edict.go
View File

@ -230,10 +230,10 @@ type edictSense struct {
Glossary []edictGlossary `xml:"gloss"`
}
func LoadEdict(reader io.Reader) ([]edictEntry, map[string]string, error) {
func LoadEdict(reader io.Reader, transform bool) ([]edictEntry, map[string]string, error) {
var entries []edictEntry
entities, err := parseEntries(reader, func(decoder *xml.Decoder, element *xml.StartElement) error {
entities, err := parseEntries(reader, transform, func(decoder *xml.Decoder, element *xml.StartElement) error {
if element.Name.Local != "entry" {
return nil
}

4
enamdict.go
View File

@ -129,10 +129,10 @@ type enamTranslation struct {
Language string `xml:"lang,attr"`
}
func LoadEnamdict(reader io.Reader) ([]enamdictEntry, map[string]string, error) {
func LoadEnamdict(reader io.Reader, transform bool) ([]enamdictEntry, map[string]string, error) {
var entries []enamdictEntry
entities, err := parseEntries(reader, func(decoder *xml.Decoder, element *xml.StartElement) error {
entities, err := parseEntries(reader, transform, func(decoder *xml.Decoder, element *xml.StartElement) error {
if element.Name.Local != "entry" {
return nil
}