start loading mails and classifying them

glide.lock

@@ -1,6 +1,8 @@
-hash: 0fbdb475136d5d29feb99b15dafd533580a8bbbf4fcfd801ddeeeb24e9a840c0
-updated: 2017-02-20T22:35:31.503897743Z
+hash: d18ff656d998425dece8d9d7fbdf5e7c0dd9a2a41b0fc791d55b242d1f8915e7
+updated: 2017-02-22T20:04:09.761903217Z
 imports:
+- name: github.com/jbrukh/bayesian
+  version: bf3f261f9a9c61145c60d47665b0518cc32c774f
 - name: github.com/luksen/maildir
   version: 5297d9c3091c7d4891c9d4f6fa743d500c038d6f
 testImports: []

glide.yaml

@@ -1,3 +1,4 @@
 package: github.com/carlostrub/sisyphus
 import:
 - package: github.com/luksen/maildir
+- package: github.com/jbrukh/bayesian

main.go

@@ -1,42 +1,95 @@
 package main
 
 import (
-	"fmt"
+	"bufio"
 	"log"
+	"strings"
 
+	"github.com/jbrukh/bayesian"
 	"github.com/luksen/maildir"
 )
 
+const (
+	// good is the class of good mails that are not supposed to be Spam
+	good bayesian.Class = "Good"
+	// junk is the class of Spam mails
+	junk bayesian.Class = "Junk"
+)
+
 var (
-	// Maildirs holds a set of mail directories to handle.
-	Maildirs []string
+	// Maildir holds a set of mail directories to handle.
+	Maildir = "/usr/home/cs/Maildir.TEST"
+
+	// processed is a map of e-mail IDs and true if processed already.
+	processed map[string]bool
 )
 
-func main() {
-	Maildirs = []string{"/usr/home/cs/Maildir.TEST"}
+// Mails contains the keys of all mails in the Junk.cur and cur directories.
+type Mails struct {
+	Junk, Good []string
+}
+
+// Classifiers contains the classifiers for mail subjects and bodies
+type Classifiers struct {
+	Subject, Body *bayesian.Classifier
+}
+
+// LoadMails loads all mail keys from the Maildir directory for processing.
+func LoadMails() (m Mails, err error) {
+
+	m.Junk, err = maildir.Dir(Maildir + "/.Junk").Keys()
+	if err != nil {
+		return m, err
+	}
+
+	m.Good, err = maildir.Dir(Maildir).Keys()
+	if err != nil {
+		return m, err
+	}
 
-	var err error
-	var Bad, Good []string
+	return m, nil
+}
+
+// Learn initially classifies all mails and returns the respective classifiers.
+func (m Mails) Learn() (c Classifiers, err error) {
+	return
+}
+
+func cleanText(t string) (c string, err error) {
+	return
+}
+
+// getContent reads mails' subjects and bodies and returns the respective
+// slices of strings
+func getContent(keys []string) (s, b []string, err error) {
+	for _, k := range keys {
 
-	for _, dir := range Maildirs {
-		var keysBad, keysGood []string
-		keysBad, err = maildir.Dir(dir + "/.Junk").Keys()
+		message, err := maildir.Dir(Maildir).Message(k)
 		if err != nil {
-			log.Fatal(err)
+			return s, b, err
 		}
 
-		Bad = append(Bad, keysBad...)
+		// get Subject
+		subject := message.Header.Get("Subject")
+		s = append(s, strings.Split(subject, " ")...)
 
-		keysGood, err = maildir.Dir(dir).Keys()
-		if err != nil {
-			log.Fatal(err)
+		// get Body
+		bScanner := bufio.NewScanner(message.Body)
+		for bScanner.Scan() {
+			b = append(b, strings.Split(bScanner.Text(), " ")...)
 		}
+	}
+
+	return s, b, nil
+}
+
+func main() {
 
-		Good = append(Good, keysGood...)
+	_, err := LoadMails()
+	if err != nil {
+		log.Fatal(err)
 	}
 
-	fmt.Println("Bad guys:")
-	fmt.Println(Bad)
-	fmt.Println("Good guys:")
-	fmt.Println(Good)
+	// Create a classifier
+	//classifier := bayesian.NewClassifier(Good, Junk)
 }