Add support to show suggestions for inputs that already have native language characters

govarnam/constants.go

@@ -98,7 +98,7 @@ func getVSTLookupDirs() []string {
 	}
 }
 
-//FindVSTDir Get the VST storing directory
+// FindVSTDir Get the VST storing directory
 func FindVSTDir() (string, error) {
 	for _, loc := range getVSTLookupDirs() {
 		if dirExists(loc) {

govarnam/govarnam.go

@@ -14,6 +14,7 @@ import (
 	"sort"
 	"strings"
 	"time"
+	"unicode"
 	"unicode/utf8"
 
 	// sqlite3
@@ -25,6 +26,7 @@ type LangRules struct {
 	Virama               string
 	IndicDigits          bool
 	PatternLongestLength int // Longest length of pattern in VST
+	UnicodeBlock         unicode.RangeTable
 }
 
 // SchemeDetails of VST
@@ -247,8 +249,8 @@ func (varnam *Varnam) setDefaultConfig() {
 	varnam.DictionaryMatchExact = false
 
 	varnam.LangRules.IndicDigits = false
-
 	varnam.LangRules.Virama, _ = varnam.getVirama()
+	varnam.LangRules.UnicodeBlock = varnam.getUnicodeBlock()
 
 	if varnam.SchemeDetails.LangCode == "ml" {
 		varnam.RegisterPatternWordPartializer(varnam.mlPatternWordPartializer)

govarnam/govarnam_ml_test.go

@@ -478,6 +478,7 @@ func TestMLRecentlyLearnedWords(t *testing.T) {
 	}
 
 	result, err = varnam.GetRecentlyLearntWords(context.Background(), 4, len(words))
+	checkError(err)
 	assertEqual(t, result[0].Word, "ആലപ്പുഴ")
 }
 
@@ -495,3 +496,15 @@ func TestMLGetSuggestions(t *testing.T) {
 
 	assertEqual(t, result[0].Word, "ആലപ്പുഴ")
 }
+
+func TestMLNativePartialWordsInInput(t *testing.T) {
+	varnam := getVarnamInstance("ml")
+
+	words := []string{"ആലപ്പുഴ", "പുസ്തകം"}
+	for _, word := range words {
+		varnam.Learn(word, 0)
+	}
+
+	assertEqual(t, varnam.TransliterateAdvanced("ആലppu").DictionarySuggestions[0].Word, "ആലപ്പുഴ")
+	assertEqual(t, varnam.TransliterateAdvanced("puസ്ത").DictionarySuggestions[0].Word, "പുസ്തകം")
+}

govarnam/govarnam_ml.go → govarnam/lang_specific_rules.go

@@ -1,5 +1,7 @@
 package govarnam
 
+import "unicode"
+
 /**
  * govarnam - An Indian language transliteration library
  * Copyright Subin Siby <mail at subinsb (.) com>, 2021
@@ -23,3 +25,15 @@ func (varnam *Varnam) mlPatternWordPartializer(sug *Suggestion) {
 		sug.Word = sug.Word[0:len(sug.Word)-size] + "മ"
 	}
 }
+
+func (varnam *Varnam) getUnicodeBlock() unicode.RangeTable {
+	switch varnam.SchemeDetails.LangCode {
+	case "kn":
+		return unicode.RangeTable{R16: []unicode.Range16{{0x0C80, 0x0CFF, 1}}}
+	case "ml":
+		return unicode.RangeTable{R16: []unicode.Range16{{0x0D00, 0x0D7F, 1}}}
+	default:
+		return unicode.RangeTable{}
+	}
+	// TODO add for all languages
+}

govarnam/symbol.go

@@ -12,6 +12,7 @@ import (
 	"fmt"
 	"log"
 	"strings"
+	"unicode"
 
 	"github.com/mattn/go-sqlite3"
 )
@@ -274,10 +275,17 @@ func (varnam *Varnam) tokenizeWord(ctx context.Context, word string, matchType i
 			matches := varnam.findLongestPatternMatchSymbols(ctx, sequence, matchType, acceptCondition)
 
 			if len(matches) == 0 {
-				// No matches, add a character token
-				// Note that we just add 1 character, and move on
-				token := Token{VARNAM_TOKEN_CHAR, matches, i, string(sequence[:1])}
-				results = append(results, token)
+				if unicode.In(sequence[0], &varnam.LangRules.UnicodeBlock) {
+					// This helps to get suggestions in inputs like "ആലppu"
+					character := string(sequence[0])
+					token := Token{VARNAM_TOKEN_SYMBOL, []Symbol{{Value1: character}}, i, character}
+					results = append(results, token)
+				} else {
+					// No matches, add a character token
+					// Note that we just add 1 character, and move on
+					token := Token{VARNAM_TOKEN_CHAR, matches, i, string(sequence[:1])}
+					results = append(results, token)
+				}
 
 				i++
 			} else {

govarnamgo/govarnamgo_test.go

@@ -67,7 +67,7 @@ func tearDown() {
 
 func TestMain(m *testing.M) {
 	var err error
-	testTempDir, err = os.TempDir("", "govarnam_test")
+	testTempDir, err = os.MkdirTemp("", "govarnamgo_test")
 	checkError(err)
 
 	setUp("ml")