func main() {
tokenizer, err := goseg.NewTokenizerFromFile(dictDataFilePath)
if err != nil {
log.Fatal(err)
}
for _, item := range allData {
fis, err := ioutil.ReadDir(item.folder)
termFreq := make(map[string]int)
if err != nil {
log.Fatal(err)
}
for _, fi := range fis {
if fi.IsDir() {
continue
}
filePath := filepath.Join(item.folder, fi.Name())
mail := mailfile.NewPOP3Mail(filePath)
if err = mail.Parse(); err != nil {
log.Fatal(err)
}
post, err := mailpost.Parse(mail)
mail.Close()
if err != nil {
log.Fatalf("Err: %v, Mail:%s", err, mail.Path())
}
words := tokenizer.Cut([]rune(post.Subject + " " + post.Content))
for _, word := range words {
key := strings.Trim(word, cutset)
if len(key) > 1 {
termFreq[key] = termFreq[key] + 1
}
}
}
pairList := sortMapByValue(termFreq)
output, err := os.OpenFile(item.output, os.O_WRONLY|os.O_TRUNC|os.O_CREATE, 0644)
writer := bufio.NewWriter(output)
offset := int(float64(len(pairList)) * topTermRatio)
for _, pair := range pairList[offset:] {
if len([]rune(pair.Key)) > termMinLength {
writer.WriteString(pair.Key + " " + strconv.Itoa(pair.Value) + "\n")
}
}
writer.Flush()
output.Close()
}
}
func main() {
classifier := bayesian.NewClassifier(Good, Bad)
tokenizer, err := goseg.NewTokenizerFromFile(dictDataFilePath)
if err != nil {
log.Fatal(err)
}
for _, item := range trainingData {
log.Printf("Traning %s", item.folder)
totalNum := 0
var totalSize int64
startTime := time.Now()
fis, err := ioutil.ReadDir(item.folder)
if err != nil {
log.Fatal(err)
}
for _, fi := range fis {
if fi.IsDir() {
continue
}
totalSize += fi.Size()
filePath := filepath.Join(item.folder, fi.Name())
mail := mailfile.NewPOP3Mail(filePath)
if err = mail.Parse(); err != nil {
log.Fatal(err)
}
post, err := mailpost.Parse(mail)
mail.Close()
if err != nil {
log.Fatalf("Err: %v, Mail:%s", err, mail.Path())
}
words := analyzer.Normalize(tokenizer.Cut([]rune(post.Subject+" "+post.Content)), cutset)
classifier.Learn(words, item.class)
totalNum += 1
}
elapsed := time.Now().Sub(startTime)
fmt.Printf("TotalNum: %d\n", totalNum)
fmt.Printf("Elapsed: %s, TPS(Mail): %f, TPS(FileSize): %s\n",
time.Now().Sub(startTime),
float64(totalNum)/(float64(elapsed)/float64(time.Second)),
fileutil.Humanize(uint64(float64(totalSize)/(float64(elapsed)/float64(time.Second)))))
}
classifier.WriteToFile(output)
}
func (cif *ContentInspectionFilter) Filter(mail mailfile.Mail) Result {
log.Printf("Run %s, Mail:%s\n", cif, mail.Name())
cif.total.Inc(1)
post, err := mailpost.Parse(mail)
mail.Close()
if err != nil {
cif.malformed.Inc(1)
log.Printf("ContentInspectionFilter: Err:%v, Mail:%s\n", err, mail.Path())
return cif.next.Filter(mail)
}
class := cif.anlz.Test(post.Subject + " " + post.Content)
cif.counters[class].Inc(1)
if cif.allPass || analyzer.Good == class {
return cif.next.Filter(mail)
}
return Quarantine
}
func (cih *ContentInspectionUpdater) Update(mail mailfile.Mail) {
if leaner, ok := cih.anlz.(analyzer.Learner); ok {
log.Printf("Run %s, Mail:%s\n", cih, mail.Name())
cih.total.Inc(1)
post, err := mailpost.Parse(mail)
mail.Close()
if err != nil {
cih.malformed.Inc(1)
log.Printf("ContentInspectionUpdater: Err:%v, Mail:%s\n", err, mail.Path())
return
}
leaner.Learn(post.Subject, cih.class)
leaner.Learn(post.Content, cih.class)
err = os.Remove(mail.Path())
if err != nil {
log.Println(err)
}
}
}
func main() {
anlz, err := analyzer.NewBayesianAnalyzer(traningDataFilePath, dictDataFilePath)
if err != nil {
log.Fatal(err)
}
for _, item := range testData {
log.Printf("Testing %s", item.folder)
totalNum, totalError, totalNeutral := 0, 0, 0
var totalSize int64
fis, err := ioutil.ReadDir(item.folder)
if err != nil {
log.Fatal(err)
}
startTime := time.Now()
for _, fi := range fis {
if fi.IsDir() {
continue
}
totalSize += fi.Size()
totalNum += 1
mailFilePath := filepath.Join(item.folder, fi.Name())
mail := mailfile.NewPOP3Mail(mailFilePath)
if err = mail.Parse(); err != nil {
log.Fatal(err)
}
post, err := mailpost.Parse(mail)
mail.Close()
if err != nil {
log.Fatalf("Err: %v, Mail:%s", err, mail.Path())
}
class := anlz.Test(post.Subject + " " + post.Content)
switch {
case analyzer.Neutral == class:
totalNeutral += 1
fmt.Println(ansi.Color(mailFilePath, "cyan+b"))
case item.class != class:
totalError += 1
fmt.Println(ansi.Color(mailFilePath, "red+b"))
}
}
elapsed := time.Now().Sub(startTime)
fmt.Printf("TotalNum: %d, TotalError: %d, ErrRate: %f, TotalNeutral:%d, Confident:%f\n",
totalNum, totalError, float64(totalError)/float64(totalNum), totalNeutral, float64(totalNum-totalNeutral)/float64(totalNum))
fmt.Printf("Elapsed: %s, TPS(Mail): %f, TPS(FileSize): %s\n",
time.Now().Sub(startTime),
float64(totalNum)/(float64(elapsed)/float64(time.Second)),
fileutil.Humanize(uint64(float64(totalSize)/(float64(elapsed)/float64(time.Second)))))
}
}
