1
0
mirror of https://github.com/fumiama/jieba.git synced 2026-06-05 00:32:51 +08:00

refactor, removed duplicated codes

This commit is contained in:
Wang Bin
2015-02-26 17:59:34 +08:00
parent ed0bc44921
commit dd7e4ec802
2 changed files with 13 additions and 29 deletions

View File

@@ -1,12 +1,8 @@
package analyse
import (
"bufio"
"os"
"path/filepath"
"github.com/wangbin/jiebago"
"sort"
"strconv"
"strings"
)
var (
@@ -33,39 +29,27 @@ func (loader *IDFLoader) newPath(idfFilePath string) error {
if loader.Path == idfFilePath {
return nil
}
idfFile, err := os.Open(idfFilePath)
wtfs, err := jiebago.ParseDictFile(idfFilePath)
if err != nil {
return err
}
scanner := bufio.NewScanner(idfFile)
freqs := make([]float64, 0)
for scanner.Scan() {
line := scanner.Text()
words := strings.Split(line, " ")
word, freqStr := words[0], words[1]
freq, err := strconv.ParseFloat(freqStr, 64)
if err != nil {
continue
}
loader.Freq[word] = freq
freqs = append(freqs, freq)
}
if err := scanner.Err(); err != nil {
return err
for _, wtf := range wtfs {
loader.Freq[wtf.Word] = wtf.Freq
freqs = append(freqs, wtf.Freq)
}
sort.Float64s(freqs)
loader.Median = freqs[len(freqs)/2]
return nil
}
func SetIdf(idfFilePath string) error {
if !filepath.IsAbs(idfFilePath) {
pwd, err := os.Getwd()
if err != nil {
return err
}
idfFilePath = filepath.Clean(filepath.Join(pwd, idfFilePath))
func SetIdf(idfFileName string) error {
idfFilePath, err := jiebago.DictPath(idfFileName)
if err != nil {
return err
}
return idfLoader.newPath(idfFilePath)
}

View File

@@ -32,7 +32,7 @@ func ParseDictFile(dictFilePath string) (wtfs []*WordTagFreq, err error) {
if err != nil {
return
}
defer dictFile.Close()
scanner := bufio.NewScanner(dictFile)
for scanner.Scan() {
line := scanner.Text()