Move all the code to a lib folder
This commit is contained in:
parent
e963d00014
commit
9d1f1ad5c0
31 changed files with 123 additions and 98 deletions
66
lib/parser/language.go
Normal file
66
lib/parser/language.go
Normal file
|
@ -0,0 +1,66 @@
|
|||
package parser
|
||||
|
||||
import (
|
||||
"io/ioutil"
|
||||
"strings"
|
||||
|
||||
"github.com/meskio/epubgo"
|
||||
"github.com/rainycape/cld2"
|
||||
)
|
||||
|
||||
func GuessLang(epub *epubgo.Epub, orig_langs []string) []string {
|
||||
spine, err := epub.Spine()
|
||||
if err != nil {
|
||||
return orig_langs
|
||||
}
|
||||
|
||||
var err_spine error
|
||||
err_spine = nil
|
||||
langs := []string{}
|
||||
for err_spine == nil {
|
||||
html, err := spine.Open()
|
||||
err_spine = spine.Next()
|
||||
if err != nil {
|
||||
continue
|
||||
}
|
||||
defer html.Close()
|
||||
|
||||
buff, err := ioutil.ReadAll(html)
|
||||
if err != nil {
|
||||
continue
|
||||
}
|
||||
langs = append(langs, cld2.Detect(string(buff)))
|
||||
}
|
||||
|
||||
lang := commonLang(langs)
|
||||
if lang != "un" && differentLang(lang, orig_langs) {
|
||||
return []string{lang}
|
||||
}
|
||||
return orig_langs
|
||||
}
|
||||
|
||||
func commonLang(langs []string) string {
|
||||
count := map[string]int{}
|
||||
for _, l := range langs {
|
||||
count[l]++
|
||||
}
|
||||
|
||||
lang := "un"
|
||||
maxcount := 0
|
||||
for l, c := range count {
|
||||
if c > maxcount && l != "un" {
|
||||
lang = l
|
||||
maxcount = c
|
||||
}
|
||||
}
|
||||
return lang
|
||||
}
|
||||
|
||||
func differentLang(lang string, orig_langs []string) bool {
|
||||
orig_lang := "un"
|
||||
if len(orig_langs) > 0 && len(orig_langs) >= 2 {
|
||||
orig_lang = strings.ToLower(orig_langs[0][0:2])
|
||||
}
|
||||
|
||||
return orig_lang != lang
|
||||
}
|
Reference in a new issue