v0.1 working draft

Author: thc

README.md

@@ -0,0 +1,5 @@
+module projects/subayes
+
+go 1.20
+
+require github.com/navossoc/bayesian v0.0.0-20230423142728-ab66f8feaf97

go.mod

@@ -0,0 +1,2 @@
+github.com/navossoc/bayesian v0.0.0-20230423142728-ab66f8feaf97 h1:6CBjPos6l0GnoMCMiQPbhPRhHXFaa7RFAGwpRwveVlI=
+github.com/navossoc/bayesian v0.0.0-20230423142728-ab66f8feaf97/go.mod h1:P1c1lcW3JeYIRbVw98K6qNHJq/3hX4ru5SCQc84ZbZo=

go.sum

@@ -0,0 +1,115 @@
+package main
+
+import (
+	"bufio"
+	"errors"
+	"flag"
+	"fmt"
+	"os"
+	"regexp"
+
+	"github.com/navossoc/bayesian"
+)
+
+var (
+	db, data            string
+	learnSpam, learnHam bool
+	Spam                bayesian.Class = "Spam"
+	Ham                 bayesian.Class = "Ham"
+	rxp                                = regexp.MustCompile(" |'|,|\t|\n")
+)
+
+func main() {
+
+	// db is path for storing data classes
+	flag.StringVar(&db, "db", "db", " db path")
+	// data is the file to be read
+	flag.StringVar(&data, "d", "data", "data filename")
+	// choosing between learning Spam or Han (write db/classes files)
+	flag.BoolVar(&learnSpam, "learnSpam", false, "Learn Spam subjects")
+	flag.BoolVar(&learnHam, "learnHam", false, "Learn Ham subjects")
+	// Default is to read stdin line per line for classification
+	flag.Parse()
+
+	K := bayesian.NewClassifier(Ham, Spam)
+
+	switch {
+
+	case learnHam && learnSpam:
+		errcheck(errors.New("Please choose learn Ham or Spam, not Both !"))
+
+	case learnHam:
+		errcheck(learn(K, db, data, Ham))
+		showClassesCount(K)
+
+	case learnSpam:
+		errcheck(learn(K, db, data, Spam))
+		showClassesCount(K)
+
+	case !learnHam && !learnSpam:
+		errcheck(K.ReadClassFromFile(Spam, db))
+		errcheck(K.ReadClassFromFile(Ham, db))
+		showClassesCount(K)
+
+		scanner := bufio.NewScanner(os.Stdin)
+
+		for scanner.Scan() { // read line per line
+			text := scanner.Text()
+
+			if len(text) > 3 { // Minimum 'Re:'
+
+				spl := rxp.Split(text, -1)
+
+				fmt.Printf("%v: %s\n",
+					classify(K, spl),
+					text)
+			}
+		}
+		if err := scanner.Err(); err != nil {
+			errcheck(err)
+		}
+	}
+}
+
+func learn(c *bayesian.Classifier, xdb string, input string, class bayesian.Class) (err error) {
+
+	err = c.ReadClassFromFile(class, xdb)
+	// if db/class don't exist, we will create it, so any err is acceptable
+	// errcheck(err)
+
+	in, err := os.ReadFile(input) // in  type is []byte
+	errcheck(err)
+
+	ins := string(in) // ins type is string
+
+	indata := rxp.Split(ins, -1) // indata is []string
+	c.Learn(indata, class)
+
+	err = c.WriteClassToFile(class, xdb)
+	errcheck(err)
+
+	return nil
+}
+func classify(c *bayesian.Classifier, pattern []string) bayesian.Class {
+	//  ProbScores return scores ([]float64), indexofclass, strict(?)
+	_, likelyb, _ := c.ProbScores(pattern)
+	return c.Classes[likelyb]
+}
+
+func showClassesCount(c *bayesian.Classifier) {
+	fmt.Printf("INFO classifier corpus : ")
+	for i := 0; i < len(c.Classes); i++ {
+		fmt.Printf(" [ %v -> %d items ]",
+			c.Classes[i],
+			c.WordCount()[i])
+
+	}
+	fmt.Println()
+}
+
+func errcheck(e error) {
+	if e != nil {
+		fmt.Println(e)
+		os.Exit(-1)
+	}
+}

main.go

@@ -0,0 +1,5 @@
+module projects/subayes/test
+
+go 1.20
+
+require github.com/navossoc/bayesian v0.0.0-20230423142728-ab66f8feaf97

test/go.mod

@@ -0,0 +1,2 @@
+github.com/navossoc/bayesian v0.0.0-20230423142728-ab66f8feaf97 h1:6CBjPos6l0GnoMCMiQPbhPRhHXFaa7RFAGwpRwveVlI=
+github.com/navossoc/bayesian v0.0.0-20230423142728-ab66f8feaf97/go.mod h1:P1c1lcW3JeYIRbVw98K6qNHJq/3hX4ru5SCQc84ZbZo=

test/go.sum

@@ -0,0 +1,67 @@
+package main
+
+import (
+	"fmt"
+
+	"github.com/navossoc/bayesian"
+)
+
+const (
+	Good bayesian.Class = "Good"
+	Bad  bayesian.Class = "Bad"
+	// Grey bayesian.Class = "Grey"
+)
+
+func main() {
+
+	classifier := bayesian.NewClassifier(Good, Bad)
+
+	goodStuff := []string{"tall", "rich", "handsome", "tennis", "chicken", "theatre", "reading", "books", "money", "suits", "golf", "eating", "wine"}
+	badStuff := []string{"poor", "smelly", "ugly", "playing", "games", "burgers", "tv", "magazines", "debts", "jogging", "foot", "drinking", "beer", "Zdob<C4><85>d<C5><BA>pobli\xc5\xbcu"}
+	greyStuff := []string{"man", "girl", "young", "citizen"}
+
+	classifier.Learn(goodStuff, Good)
+	classifier.Learn(badStuff, Bad)
+	classifier.Learn(greyStuff, Bad)
+
+	// classifier.Learn(greyStuff, Grey)
+
+	//classifier.WriteToFile("data")
+
+	fmt.Printf("classifier learned :\n")
+	for i := 0; i < len(classifier.Classes); i++ {
+		fmt.Printf("  %d words for classes %v \n",
+			classifier.WordCount()[i],
+			classifier.Classes[i])
+	}
+	fmt.Println()
+
+	tests := [][]string{
+		{"tall", "girl", "watching", "tv", "eating", "chicken", "wearing", "jogging"},
+		{"small", "handsome", "man", "playing", "games", "before", "tv"},
+		{"unknows", "words"},
+	}
+
+	for _, pattern := range tests {
+
+		fmt.Printf("test : %v\n", pattern)
+
+		probs, likelyb, _ := classifier.ProbScores(pattern)
+		fmt.Printf("probs ")
+		for i := 0; i < len(classifier.Classes); i++ {
+			fmt.Printf("%s(%.2f) ",
+				classifier.Classes[i], probs[i])
+		}
+		fmt.Printf("=> Class : %s\n\n",
+			classifier.Classes[likelyb])
+	}
+}
+
+// func round(num float64) int {
+// 	return int(num + math.Copysign(0.5, num))
+// }
+
+// func toFixed(num float64, precision int) float64 {
+// 	output := math.Pow(10, float64(precision))
+// 	return float64(round(num*output)) / output
+// }