Move beidermorse/common pkg to internal/bmpm

BREAKING: end users will have to rename accuracy constants
f1monkey · Feb 22, 2023 · 54fc71b · 54fc71b
1 parent 0fde288
commit 54fc71b
Show file tree

Hide file tree

Showing 17 changed files with 7,479 additions and 7,439 deletions.
diff --git a/README.md b/README.md
@@ -104,7 +104,6 @@ import (
 	"github.com/f1monkey/phonetic/beidermorse"
 	"github.com/f1monkey/phonetic/beidermorse/beidermorseash"
 	"github.com/f1monkey/phonetic/beidermorse/beidermorsesep"
-	"github.com/f1monkey/phonetic/beidermorse/common"
 )
 
 func main() {
@@ -119,7 +118,7 @@ func main() {
 	// prints: [orangi oragi orongi orogi orYngi Yrangi Yrongi YrYngi oranxi oronxi orani oroni oranii oronii oranzi oronzi urangi urongi]
 
 	// USE ENCODER WITH "GENERIC" RULESET WITH "EXACT" ACCURACY
-	genEncoder, err = beidermorse.NewEncoder(beidermorse.WithAccuracy(common.Exact))
+	genEncoder, err = beidermorse.NewEncoder(beidermorse.WithAccuracy(beidermorse.Exact))
 	if err != nil {
 		panic(err)
 	}
@@ -130,7 +129,7 @@ func main() {
 	// USE ENCODER WITH "GENERIC" RULESET WITH "EXACT" ACCURACY
 	// AND "ENGLISH" LANGUAGE
 	genEncoder, err = beidermorse.NewEncoder(
-		beidermorse.WithAccuracy(common.Exact),
+		beidermorse.WithAccuracy(beidermorse.Exact),
 		beidermorse.WithLang(beidermorse.English),
 	)
 	if err != nil {
@@ -140,7 +139,6 @@ func main() {
 	fmt.Println(result)
 	// prints: [orenk orenge orendS orendZe oronk oronge orondS orondZe orank orange orandS orandZe arenk arenge arendS arendZe aronk aronge arondS arondZe arank arange arandS arandZe]
 
-
 	// USE ENCODER WITH "ASHKENAZI" RULESET
 	ashEncoder, err := beidermorseash.NewEncoder()
 	if err != nil {

diff --git a/beidermorse/beidermorseash/encoder.go b/beidermorse/beidermorseash/encoder.go
@@ -6,22 +6,34 @@ import (
 	"fmt"
 	"math/bits"
 
-	"github.com/f1monkey/phonetic/beidermorse/common"
+	"github.com/f1monkey/phonetic/internal/bmpm"
 	"github.com/f1monkey/phonetic/internal/exrunes"
 )
 
 var ErrInvalidMode = fmt.Errorf("invalid name mode")
 var ErrInvalidAccuracy = fmt.Errorf("invalid accuracy value")
 
+// Accuracy exact or approximate matching
+type Accuracy bmpm.Accuracy
+
+const (
+	Exact  Accuracy = "exact"  // exact matching rules
+	Approx Accuracy = "approx" // approx matching (results in more tokens)
+)
+
+func (a Accuracy) Valid() bool {
+	return a == Exact || a == Approx
+}
+
 type Encoder struct {
-	accuracy common.Accuracy
+	accuracy Accuracy
 	lang     Lang
 }
 
 // NewEncoder create new encoder instance
 func NewEncoder(opts ...EncoderOption) (*Encoder, error) {
 	result := &Encoder{
-		accuracy: common.Approx,
+		accuracy: Approx,
 	}
 
 	for _, opt := range opts {
@@ -45,7 +57,7 @@ func MustNewEncoder(opts ...EncoderOption) *Encoder {
 // Encode transform a passed string to a slice of phonetic tokens
 func (e *Encoder) Encode(input string) []string {
 	langDetector := detectLangFunc()
-	lang := common.Lang(e.lang)
+	lang := bmpm.Lang(e.lang)
 	if lang == 0 {
 		lang = langDetector(input)
 	}
@@ -54,10 +66,10 @@ func (e *Encoder) Encode(input string) []string {
 
 	buf := exrunes.NewBuffer(200)
 
-	tokens := common.MakeTokens(
-		input, common.Ashkenazi,
-		e.accuracy,
-		common.Ruleset{Main: main, Final1: final1, Final2: final2, Discards: Discards, DetectLang: langDetector},
+	tokens := bmpm.MakeTokens(
+		input, bmpm.Ashkenazi,
+		bmpm.Accuracy(e.accuracy),
+		bmpm.Ruleset{Main: main, Final1: final1, Final2: final2, Discards: Discards, DetectLang: langDetector},
 		lang,
 		false,
 		buf,
@@ -85,7 +97,7 @@ func (e *Encoder) SetOption(opt EncoderOption) error {
 type EncoderOption func(e *Encoder) error
 
 // WithAccuracy Set encoder accuracy
-func WithAccuracy(a common.Accuracy) EncoderOption {
+func WithAccuracy(a Accuracy) EncoderOption {
 	return func(e *Encoder) error {
 		if !a.Valid() {
 			return fmt.Errorf("%w: %q", ErrInvalidAccuracy, a)
@@ -104,35 +116,35 @@ func WithLang(l Lang) EncoderOption {
 }
 
 func getRules(
-	accuracy common.Accuracy,
-	lang common.Lang,
-) (common.Rules, common.Rules, common.Rules) {
-	var main, final1, final2 common.Rules
+	accuracy Accuracy,
+	lang bmpm.Lang,
+) (bmpm.Rules, bmpm.Rules, bmpm.Rules) {
+	var main, final1, final2 bmpm.Rules
 
 	langCount := bits.OnesCount64(uint64(lang))
 	if langCount > 1 {
-		lang = common.Lang(Any)
+		lang = bmpm.Lang(Any)
 	}
 	main = Rules[lang]
 
-	if accuracy == common.Approx {
+	if accuracy == Approx {
 		final1 = FinalRules.Approx.First
 		final2 = FinalRules.Approx.Second[lang]
-	} else if accuracy == common.Exact {
+	} else if accuracy == Exact {
 		final1 = FinalRules.Exact.First
 		final2 = FinalRules.Exact.Second[lang]
 	}
 
 	return main, final1, final2
 }
 
-func detectLangFunc() common.DetectLangFunc {
-	return func(input string) common.Lang {
+func detectLangFunc() bmpm.DetectLangFunc {
+	return func(input string) bmpm.Lang {
 		all := All
 		rules := LangRules
 
 		runes := []rune(input)
-		remaining := common.Lang(all)
+		remaining := bmpm.Lang(all)
 		for _, rule := range rules {
 			if rule.Matcher == nil {
 				continue

diff --git a/beidermorse/beidermorseash/encoder_test.go b/beidermorse/beidermorseash/encoder_test.go
@@ -4,46 +4,45 @@ import (
 	"fmt"
 	"testing"
 
-	"github.com/f1monkey/phonetic/beidermorse/common"
 	"github.com/stretchr/testify/require"
 )
 
 func Benchmark_Encoder_Encode_Ru_Approx(b *testing.B) {
-	e := MustNewEncoder(WithAccuracy(common.Approx))
+	e := MustNewEncoder(WithAccuracy(Approx))
 	for i := 0; i < b.N; i++ {
 		e.Encode("апельсин")
 	}
 }
 
 func Benchmark_Encoder_Encode_Ru_Exact(b *testing.B) {
-	e := MustNewEncoder(WithAccuracy(common.Exact))
+	e := MustNewEncoder(WithAccuracy(Exact))
 	for i := 0; i < b.N; i++ {
 		e.Encode("апельсин")
 	}
 }
 
 func Benchmark_Encoder_Encode_En_Approx(b *testing.B) {
-	e := MustNewEncoder(WithAccuracy(common.Approx))
+	e := MustNewEncoder(WithAccuracy(Approx))
 	for i := 0; i < b.N; i++ {
 		e.Encode("orange")
 	}
 }
 
 func Benchmark_Encoder_Encode_En_Exact(b *testing.B) {
-	e := MustNewEncoder(WithAccuracy(common.Exact))
+	e := MustNewEncoder(WithAccuracy(Exact))
 	for i := 0; i < b.N; i++ {
 		e.Encode("orange")
 	}
 }
 
 func Test_Encoder_Encode(t *testing.T) {
 	cases := []struct {
-		accuracy common.Accuracy
+		accuracy Accuracy
 		input    string
 		expected []string
 	}{
 		{
-			accuracy: common.Approx,
+			accuracy: Approx,
 			input:    "orange",
 			expected: []string{
 				"orangi",
@@ -63,7 +62,7 @@ func Test_Encoder_Encode(t *testing.T) {
 			},
 		},
 		{
-			accuracy: common.Exact,
+			accuracy: Exact,
 			input:    "orange",
 			expected: []string{
 				"orange",
@@ -73,7 +72,7 @@ func Test_Encoder_Encode(t *testing.T) {
 			},
 		},
 		{
-			accuracy: common.Exact,
+			accuracy: Exact,
 			input:    "van der orange",
 			expected: []string{
 				"vander",
@@ -90,7 +89,7 @@ func Test_Encoder_Encode(t *testing.T) {
 			},
 		},
 		{
-			accuracy: common.Approx,
+			accuracy: Approx,
 			input:    "test",
 			expected: []string{
 				"tist",
@@ -100,20 +99,20 @@ func Test_Encoder_Encode(t *testing.T) {
 			},
 		},
 		{
-			accuracy: common.Exact,
+			accuracy: Exact,
 			input:    "test",
 			expected: []string{
 				"teSt",
 				"test",
 			},
 		},
 		{
-			accuracy: common.Exact,
+			accuracy: Exact,
 			input:    "апельсин",
 			expected: []string{"apelsin"},
 		},
 		{
-			accuracy: common.Approx,
+			accuracy: Approx,
 			input:    "апельсин",
 			expected: []string{
 				"apYlzn",