Rework api, various tweaks.

README.md

@@ -32,15 +32,12 @@ func main() {
 
 func printResult(phrase string) {
 	description := "is appropriate"
-	if isInappropriate(phrase) {
+	if moderation.IsInappropriate(phrase) {
 		description = "is NOT appropriate"
 	}
 	fmt.Printf("\"%s\" %s.\n", phrase, description)
 }
 
-func isInappropriate(phrase string) bool {
-	return moderation.Analyze(phrase).IsInappropriate()
-}
 ```
 
 ```console
@@ -55,7 +52,7 @@ $ go run hello_world.go
 Accuracy was evaluated based on a [dataset of moderated comments](https://raw.githubusercontent.com/vzhou842/profanity-check/master/profanity_check/data/clean_data.csv).
 **Package**|**Accuracy**|**Comment**
 :-----:|:-----:|:-----:
-https://github.com/finnbear/moderation|90.50%|Current version is not stable
+https://github.com/finnbear/moderation|90.49%|Current version is not stable
 https://github.com/TwinProduction/go-away|74.83%|Many false positives from combined words like "push it"
 
 

comparison/comparison_test.go

@@ -6,20 +6,16 @@ import (
 	"os"
 	"testing"
 
-    "github.com/finnbear/moderation"
-    "github.com/TwinProduction/go-away"
+	"github.com/TwinProduction/go-away"
+	"github.com/finnbear/moderation"
 )
 
-func moderationIsInappropriate(phrase string) bool {
-	return moderation.Analyze(phrase).IsInappropriate()
-}
-
 func TestModerationWikipedia(t *testing.T) {
-    testWikipedia(t, moderationIsInappropriate)
+	testWikipedia(t, moderation.IsInappropriate)
 }
 
 func TestGoAwayWikipedia(t *testing.T) {
-    testWikipedia(t, goaway.IsProfane)
+	testWikipedia(t, goaway.IsProfane)
 }
 
 func testWikipedia(t *testing.T, isInappropriate func(string) bool) {

comparison/go.mod

@@ -2,6 +2,8 @@ module comparison
 
 go 1.15
 
+replace github.com/finnbear/moderation => ../
+
 require (
 	github.com/TwinProduction/go-away v1.0.1
 	github.com/finnbear/moderation v0.5.0

examples/.gitignore

@@ -0,0 +1 @@
+moderation.prof

examples/hello_world.go

@@ -22,12 +22,8 @@ func main() {
 
 func printResult(phrase string) {
 	description := "is appropriate"
-	if isInappropriate(phrase) {
+	if moderation.IsInappropriate(phrase) {
 		description = "is NOT appropriate"
 	}
 	fmt.Printf("\"%s\" %s.\n", phrase, description)
 }
-
-func isInappropriate(phrase string) bool {
-	return moderation.Analyze(phrase).IsInappropriate()
-}

examples/pprof.go

@@ -19,8 +19,8 @@ func main() {
 	defer pprof.StopCPUProfile()
 
 	for i := 0; i < 100000; i++ {
-		moderation.Analyze("hello")
-		moderation.Analyze("sh1t")
-		moderation.Analyze("Hello John Doe, I hope you're feeling well, as I come today bearing shitty news regarding your favorite chocolate chip cookie brand")
+		moderation.IsInappropriate("hello")
+		moderation.IsInappropriate("sh1t")
+		moderation.IsInappropriate("Hello John Doe, I hope you're feeling well, as I come today bearing shitty news regarding your favorite chocolate chip cookie brand")
 	}
 }

moderation.go

@@ -9,7 +9,12 @@ import (
 	"unicode"
 )
 
-var tree *radix.Tree
+var (
+	// The threshold where a phrase is considered inappropriate
+	InappropriateThreshold int = 1
+
+	tree *radix.Tree
+)
 
 func init() {
 	tree = radix.New()
@@ -41,14 +46,15 @@ var replacements = [...]string{
 var removeAccentsTransform = transform.Chain(norm.NFD, runes.Remove(runes.In(unicode.Mn)), norm.NFC)
 
 // Analyze analyzes a given phrase for moderation purposes
-func Analyze(text string) (analysis Analysis) {
+func IsInappropriate(text string) bool {
 	buf := make([]byte, 0, len(text))
 	_, n, _ := transform.Append(removeAccentsTransform, buf, []byte(text))
 	text = string(buf[:n])
 	lastSepMin := 0
 	lastSepMax := 0
 
 	var matches radix.Queue
+	inappropriateLevel := 0
 
 	var lastMatchable byte
 	for _, textRune := range text {
@@ -120,7 +126,7 @@ func Analyze(text string) (analysis Analysis) {
 							if next != nil {
 								if next.Word() {
 									if next.Depth() > 4 || (next.Depth() > 3 && next.Start() != 's') || (next.Depth() >= lastSepMin && next.Depth() <= lastSepMax) {
-										analysis.InappropriateLevel += int(next.Data())
+										inappropriateLevel += int(next.Data())
 									}
 								}
 
@@ -136,5 +142,5 @@ func Analyze(text string) (analysis Analysis) {
 			}
 		}
 	}
-	return
+	return inappropriateLevel >= InappropriateThreshold
 }

moderation_bench_test.go

@@ -4,9 +4,7 @@ import (
 	"testing"
 )
 
-func IsProfane(phrase string) bool {
-	return Analyze(phrase).IsInappropriate()
-}
+var IsProfane = IsInappropriate
 
 // from https://github.com/TwinProduction/go-away/blob/master/goaway_bench_test.go
 func BenchmarkIsProfaneWhenShortStringHasNoProfanity(b *testing.B) {

moderation_test.go

@@ -62,9 +62,9 @@ func TestAnalyze(t *testing.T) {
 		{"We were seated at breakfast one morning, my wife and I, when the maid brought in a telegram. It was from Sherlock Holmes and ran in this way", false},
 	}
 	for _, testCase := range testCases {
-		analysis := Analyze(testCase.phrase)
-		if analysis.IsInappropriate() != testCase.offensive {
-			t.Errorf("phrase=\"%s\" analysis offensive=%v actual offensive=%v", testCase.phrase, analysis.IsInappropriate(), testCase.offensive)
+		inappropriate := IsInappropriate(testCase.phrase)
+		if inappropriate != testCase.offensive {
+			t.Errorf("phrase=\"%s\" analysis offensive=%v actual offensive=%v", testCase.phrase, inappropriate, testCase.offensive)
 		}
 	}
 }
@@ -89,8 +89,7 @@ func TestAnalyzeWikipedia(t *testing.T) {
 		}
 		phrase := fields[1]
 		offensive := fields[0] == "1"
-		analysis := Analyze(phrase)
-		if analysis.IsInappropriate() == offensive {
+		if IsInappropriate(phrase) == offensive {
 			correct++
 		} else {
 			//fmt.Printf("phrase=\"%s\" analysis offensive=%v actual offensive=%v", phrase, analysis.IsInappropriate(), offensive)