Fix capital letters for Armenian transliteration

Add Chinese transliteration via Pinyin conversion library
Update variable names and comments for interface-based transliteration
2021-10-05 09:09:19 -07:00 · 2021-10-04 22:26:16 -07:00 · 2021-10-04 20:23:54 -07:00 · 2021-10-04 20:06:08 -07:00
9 changed files with 232 additions and 141 deletions
--- a/README.md
+++ b/README.md
@@ -58,6 +58,7 @@ Since the PineTime does not have enough space to store all unicode glyphs, it on
 - French
 - Armenian
 - Korean
+- Chinese
 - Emoji

 Place the desired map names in an array as `notifs.translit.use`. They will be evaluated in order. You can also put custom transliterations in `notifs.translit.custom`. These take priority over any other maps. The `notifs.translit` config section should look like this:
--- a/go.mod
+++ b/go.mod
@@ -15,6 +15,7 @@ require (
 	github.com/mattn/go-isatty v0.0.13 // indirect
 	github.com/mattn/go-runewidth v0.0.13 // indirect
 	github.com/mitchellh/mapstructure v1.4.1
+	github.com/mozillazg/go-pinyin v0.18.0
 	github.com/rs/zerolog v1.23.0
 	github.com/sirupsen/logrus v1.8.1 // indirect
 	github.com/spf13/cast v1.4.1 // indirect
--- a/go.sum
+++ b/go.sum
@@ -242,6 +242,8 @@ github.com/mitchellh/mapstructure v1.4.1/go.mod h1:bFUtVrKA4DC2yAKiSyO/QUcy7e+RR
 github.com/modern-go/concurrent v0.0.0-20180228061459-e0a39a4cb421/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
 github.com/modern-go/reflect2 v0.0.0-20180701023420-4b7aa43c6742/go.mod h1:bx2lNnkwVCuqBIxFjflWJWanXIb3RllmbCylyMrvgv0=
 github.com/modern-go/reflect2 v1.0.1/go.mod h1:bx2lNnkwVCuqBIxFjflWJWanXIb3RllmbCylyMrvgv0=
+github.com/mozillazg/go-pinyin v0.18.0 h1:hQompXO23/0ohH8YNjvfsAITnCQImCiR/Fny8EhIeW0=
+github.com/mozillazg/go-pinyin v0.18.0/go.mod h1:iR4EnMMRXkfpFVV5FMi4FNB6wGq9NV6uDWbUuPhP4Yc=
 github.com/muka/go-bluetooth v0.0.0-20210812063148-b6c83362e27d h1:EG/xyWjHT19rkUpwsWSkyiCCmyqNwFovr9m10rhyOxU=
 github.com/muka/go-bluetooth v0.0.0-20210812063148-b6c83362e27d/go.mod h1:dMCjicU6vRBk34dqOmIZm0aod6gUwZXOXzBROqGous0=
 github.com/nfnt/resize v0.0.0-20180221191011-83c6a9932646/go.mod h1:jpp1/29i3P1S/RLdc7JQKbRpFeM1dOBd8T9ki5s+AY8=
--- a/notifs.go
+++ b/notifs.go
@@ -73,7 +73,7 @@ func initNotifRelay(dev *infinitime.Device) error {
 			}

 			maps := viper.GetStringSlice("notifs.translit.use")
-			translit.Maps["custom"] = translit.Map(viper.GetStringSlice("notifs.translit.custom"))
+			translit.Transliterators["custom"] = translit.Map(viper.GetStringSlice("notifs.translit.custom"))
 			sender = translit.Transliterate(sender, maps...)
 			summary = translit.Transliterate(summary, maps...)
 			body = translit.Transliterate(body, maps...)
--- a/socket.go
+++ b/socket.go
@@ -171,7 +171,7 @@ func handleConnection(conn net.Conn, dev *infinitime.Device) {
 				break
 			}
 			maps := viper.GetStringSlice("notifs.translit.use")
-			translit.Maps["custom"] = translit.Map(viper.GetStringSlice("notifs.translit.custom"))
+			translit.Transliterators["custom"] = translit.Map(viper.GetStringSlice("notifs.translit.custom"))
 			title := translit.Transliterate(reqData.Title, maps...)
 			body := translit.Transliterate(reqData.Body, maps...)
 			// Send notification to watch
--- a/translit/armenian.go
+++ b/translit/armenian.go
@@ -0,0 +1,138 @@
+package translit
+
+import (
+	"strings"
+)
+
+type ArmenianTranslit struct{}
+
+var armenianMap = []string{
+	"աու", "au",
+	"բու", "bu",
+	"գու", "gu",
+	"դու", "du",
+	"եու", "eu",
+	"զու", "zu",
+	"էու", "eu",
+	"ըու", "yu",
+	"թու", "tu",
+	"ժու", "ju",
+	"իու", "iu",
+	"լու", "lu",
+	"խու", "xu",
+	"ծու", "cu",
+	"կու", "ku",
+	"հու", "hu",
+	"ձու", "dzu",
+	"ղու", "xu",
+	"ճու", "cu",
+	"մու", "mu",
+	"յու", "yu",
+	"նու", "nu",
+	"շու", "shu",
+	"չու", "chu",
+	"պու", "pu",
+	"ջու", "ju",
+	"ռու", "ru",
+	"սու", "su",
+	"վու", "vu",
+	"տու", "tu",
+	"րու", "ru",
+	"ցու", "cu",
+	"փու", "pu",
+	"քու", "qu",
+	"օու", "ou",
+	"ևու", "eu",
+	"ֆու", "fu",
+	"ոու", "vou",
+	"ու", "u",
+	"բո", "bo",
+	"գո", "go",
+	"դո", "do",
+	"զո", "zo",
+	"թո", "to",
+	"ժո", "jo",
+	"լո", "lo",
+	"խո", "xo",
+	"ծո", "co",
+	"կո", "ko",
+	"հո", "ho",
+	"ձո", "dzo",
+	"ղո", "xo",
+	"ճո", "co",
+	"մո", "mo",
+	"յո", "yo",
+	"նո", "no",
+	"շո", "so",
+	"չո", "co",
+	"պո", "po",
+	"ջո", "jo",
+	"ռո", "ro",
+	"սո", "so",
+	"վո", "vo",
+	"տո", "to",
+	"րո", "ro",
+	"ցո", "co",
+	"փո", "po",
+	"քո", "qo",
+	"ևո", "eo",
+	"ֆո", "fo",
+	"ո", "vo",
+	"եւ", "ev",
+	"եվ", "ev",
+	"ա", "a",
+	"բ", "b",
+	"գ", "g",
+	"դ", "d",
+	"ե", "e",
+	"զ", "z",
+	"է", "e",
+	"ը", "y",
+	"թ", "t",
+	"ժ", "j",
+	"ի", "i",
+	"լ", "l",
+	"խ", "x",
+	"ծ", "c",
+	"կ", "k",
+	"հ", "h",
+	"ձ", "dz",
+	"ղ", "x",
+	"ճ", "c",
+	"մ", "m",
+	"յ", "y",
+	"ն", "n",
+	"շ", "sh",
+	"չ", "ch",
+	"պ", "p",
+	"ջ", "j",
+	"ռ", "r",
+	"ս", "s",
+	"վ", "v",
+	"տ", "t",
+	"ր", "r",
+	"ց", "c",
+	"փ", "p",
+	"ք", "q",
+	"օ", "o",
+	"և", "ev",
+	"ֆ", "f",
+	"ւ", "",
+}
+
+func init() {
+	lower := armenianMap
+	for i, val := range lower {
+		if i%2 == 1 {
+			continue
+		}
+		capital := strings.Title(val)
+		if capital != val {
+			armenianMap = append(armenianMap, capital, strings.Title(armenianMap[i+1]))
+		}
+	}
+}
+
+func (at *ArmenianTranslit) Transliterate(s string) string {
+	return strings.NewReplacer(armenianMap...).Replace(s)
+}
--- a/translit/chinese.go
+++ b/translit/chinese.go
@@ -0,0 +1,42 @@
+package translit
+
+import (
+	"bytes"
+	"strings"
+	"unicode"
+
+	"github.com/mozillazg/go-pinyin"
+)
+
+// ChineseTranslit implements Transliterator using a pinyin
+// conversion library.
+type ChineseTranslit struct{}
+
+func (ct *ChineseTranslit) Transliterate(s string) string {
+	// Create buffer for final output
+	outBuf := &bytes.Buffer{}
+	// Create buffer to temporarily store chinese characters
+	tmpBuf := &bytes.Buffer{}
+	// For every character in string
+	for _, char := range s {
+		// If character in Han range
+		if unicode.Is(unicode.Han, char) {
+			// Write character to temporary buffer
+			tmpBuf.WriteRune(char)
+		} else {
+			// If buffer contains characters
+			if tmpBuf.Len() > 0 {
+				// Convert to pinyin (without tones)
+				out := pinyin.LazyConvert(tmpBuf.String(), nil)
+				// Write space-separated string to output
+				outBuf.WriteString(strings.Join(out, " "))
+				// Reset temporary buffer
+				tmpBuf.Reset()
+			}
+			// Write character to output
+			outBuf.WriteRune(char)
+		}
+	}
+	// Return output string
+	return outBuf.String()
+}
--- a/translit/korean.go
+++ b/translit/korean.go
@@ -294,10 +294,14 @@ func translitSpecialProvisions(previousEnding rune, nextInitial rune) (string, b
 		switch previousEnding {
 		case 0x11A8:
 			return "ngn", true // ㄱ
-		case 0x11AE: // ㄷ
-		case 0x11BA: // ㅅ
-		case 0x11BD: // ㅈ
-		case 0x11BE: // ㅊ
+		case 0x11AE:
+			fallthrough // ㄷ
+		case 0x11BA:
+			fallthrough // ㅅ
+		case 0x11BD:
+			fallthrough // ㅈ
+		case 0x11BE:
+			fallthrough // ㅊ
 		case 0x11C0: // ㅌ
 			return "nn", true
 		case 0x11AF:
@@ -309,17 +313,24 @@ func translitSpecialProvisions(previousEnding rune, nextInitial rune) (string, b
 		}
 	case 0x1105: // ㄹ
 		switch previousEnding {
-		case 0x11A8: // ㄱ
-		case 0x11AB: // ㄴ
+		case 0x11A8:
+			fallthrough // ㄱ
+		case 0x11AB:
+			fallthrough // ㄴ
 		case 0x11AF: // ㄹ
 			return "ll", true
-		case 0x11AE: // ㄷ
-		case 0x11BA: // ㅅ
-		case 0x11BD: // ㅈ
-		case 0x11BE: // ㅊ
+		case 0x11AE:
+			fallthrough // ㄷ
+		case 0x11BA:
+			fallthrough // ㅅ
+		case 0x11BD:
+			fallthrough // ㅈ
+		case 0x11BE:
+			fallthrough // ㅊ
 		case 0x11C0: // ㅌ
 			return "nn", true
-		case 0x11B7: // ㅁ
+		case 0x11B7:
+			fallthrough // ㅁ
 		case 0x11B8: // ㅂ
 			return "mn", true
 		case 0x11BC:
@@ -331,10 +342,14 @@ func translitSpecialProvisions(previousEnding rune, nextInitial rune) (string, b
 		switch previousEnding {
 		case 0x11A8:
 			return "ngm", true // ㄱ
-		case 0x11AE: // ㄷ
-		case 0x11BA: // ㅅ
-		case 0x11BD: // ㅈ
-		case 0x11BE: // ㅊ
+		case 0x11AE:
+			fallthrough // ㄷ
+		case 0x11BA:
+			fallthrough // ㅅ
+		case 0x11BD:
+			fallthrough // ㅈ
+		case 0x11BE:
+			fallthrough // ㅊ
 		case 0x11C0: // ㅌ
 			return "nm", true
 		case 0x11B8:
@@ -372,10 +387,14 @@ func translitSpecialProvisions(previousEnding rune, nextInitial rune) (string, b
 		}
 	case 0x1110: // ㅌ
 		switch previousEnding {
-		case 0x11AE: // ㄷ
-		case 0x11BA: // ㅅ
-		case 0x11BD: // ㅈ
-		case 0x11BE: // ㅊ
+		case 0x11AE:
+			fallthrough // ㄷ
+		case 0x11BA:
+			fallthrough // ㅅ
+		case 0x11BD:
+			fallthrough // ㅈ
+		case 0x11BE:
+			fallthrough // ㅊ
 		case 0x11C0: // ㅌ
 			return "t-t", true
 		default:
@@ -391,7 +410,6 @@ func translitSpecialProvisions(previousEnding rune, nextInitial rune) (string, b
 	default:
 		return "", false
 	}
-	return "", false
 }

 // Decompose a syllable into several jamo. Does nothing if that isn't possible.
--- a/translit/translit.go
+++ b/translit/translit.go
@@ -9,7 +9,7 @@ func Transliterate(s string, useMaps ...string) string {
 	// Create variable to store modified string
 	out := s
 	// If custom map exists
-	if customMap, ok := Maps["custom"]; ok {
+	if customMap, ok := Transliterators["custom"]; ok {
 		// Perform transliteration with it
 		out = customMap.Transliterate(out)
 	}
@@ -20,7 +20,7 @@ func Transliterate(s string, useMaps ...string) string {
 			continue
 		}
 		// Get requested map
-		translitMap, ok := Maps[useMap]
+		translitMap, ok := Transliterators[useMap]
 		if !ok {
 			continue
 		}
@@ -47,9 +47,9 @@ func (mt Map) Transliterate(s string) string {
 	return strings.NewReplacer(mt...).Replace(s)
 }

-// Maps stores transliteration maps as slices to preserve order.
-// Some of these maps were sourced from https://codeberg.org/Freeyourgadget/Gadgetbridge
-var Maps = map[string]Transliterator{
+// Transliterators stores transliterator implementations for each supported language.
+// Some of these were sourced from https://codeberg.org/Freeyourgadget/Gadgetbridge
+var Transliterators = map[string]Transliterator{
 	"eASCII": Map{
 		"œ", "oe",
 		"ª", "a",
@@ -323,119 +323,6 @@ var Maps = map[string]Transliterator{
 		"ÿ", "y",
 		"ç", "c",
 	},
-	"Armenian": Map{
-		"աու", "au",
-		"բու", "bu",
-		"գու", "gu",
-		"դու", "du",
-		"եու", "eu",
-		"զու", "zu",
-		"էու", "eu",
-		"ըու", "yu",
-		"թու", "tu",
-		"ժու", "ju",
-		"իու", "iu",
-		"լու", "lu",
-		"խու", "xu",
-		"ծու", "cu",
-		"կու", "ku",
-		"հու", "hu",
-		"ձու", "dzu",
-		"ղու", "xu",
-		"ճու", "cu",
-		"մու", "mu",
-		"յու", "yu",
-		"նու", "nu",
-		"շու", "shu",
-		"չու", "chu",
-		"պու", "pu",
-		"ջու", "ju",
-		"ռու", "ru",
-		"սու", "su",
-		"վու", "vu",
-		"տու", "tu",
-		"րու", "ru",
-		"ցու", "cu",
-		"փու", "pu",
-		"քու", "qu",
-		"օու", "ou",
-		"ևու", "eu",
-		"ֆու", "fu",
-		"ոու", "vou",
-		"ու", "u",
-		"բո", "bo",
-		"գո", "go",
-		"դո", "do",
-		"զո", "zo",
-		"թո", "to",
-		"ժո", "jo",
-		"լո", "lo",
-		"խո", "xo",
-		"ծո", "co",
-		"կո", "ko",
-		"հո", "ho",
-		"ձո", "dzo",
-		"ղո", "xo",
-		"ճո", "co",
-		"մո", "mo",
-		"յո", "yo",
-		"նո", "no",
-		"շո", "so",
-		"չո", "co",
-		"պո", "po",
-		"ջո", "jo",
-		"ռո", "ro",
-		"սո", "so",
-		"վո", "vo",
-		"տո", "to",
-		"րո", "ro",
-		"ցո", "co",
-		"փո", "po",
-		"քո", "qo",
-		"ևո", "eo",
-		"ֆո", "fo",
-		"ո", "vo",
-		"եւ", "ev",
-		"եվ", "ev",
-		"ա", "a",
-		"բ", "b",
-		"գ", "g",
-		"դ", "d",
-		"ե", "e",
-		"զ", "z",
-		"է", "e",
-		"ը", "y",
-		"թ", "t",
-		"ժ", "j",
-		"ի", "i",
-		"լ", "l",
-		"խ", "x",
-		"ծ", "c",
-		"կ", "k",
-		"հ", "h",
-		"ձ", "dz",
-		"ղ", "x",
-		"ճ", "c",
-		"մ", "m",
-		"յ", "y",
-		"ն", "n",
-		"շ", "sh",
-		"չ", "ch",
-		"պ", "p",
-		"ջ", "j",
-		"ռ", "r",
-		"ս", "s",
-		"վ", "v",
-		"տ", "t",
-		"ր", "r",
-		"ց", "c",
-		"փ", "p",
-		"ք", "q",
-		"օ", "o",
-		"և", "ev",
-		"ֆ", "f",
-		"ւ", "",
-	},
 	"Emoji": Map{
 		"😂", ":')",
 		"😊", ":)",
@@ -465,4 +352,6 @@ var Maps = map[string]Transliterator{
 		"💤", ":zzz:",
 	},
 	"Korean":   &KoreanTranslit{},
+	"Chinese":  &ChineseTranslit{},
+	"Armenian": &ArmenianTranslit{},
 }
Author	SHA1	Message	Date
Arsen Musayelyan	01bf493c77	Fix capital letters for Armenian transliteration	2021-10-05 09:09:19 -07:00
Arsen Musayelyan	b6e9ad6160	Add Chinese transliteration via Pinyin conversion library	2021-10-04 22:26:16 -07:00
Arsen Musayelyan	c56c0ae198	Update variable names and comments for interface-based transliteration	2021-10-04 20:23:54 -07:00
Arsen Musayelyan	6b94030b83	Fix Korean transliteration	2021-10-04 20:06:08 -07:00