Disambiguation

transliteration_testdata.ts

@@ -2112,7 +2112,7 @@ export const hindi: [string, string][] = [
 ["औपचारिकता", "aupacārikatā"],
 ["और", "aura"],
 ["औऱ", "auṟa"],
-["औरंगउटान", "auraṅgauṭāna"],
+["औरंगउटान", "auraṅga:uṭāna"],
 ["औरंगज़ेब", "auraṅgazēba"],
 ["औरंगाबाद", "auraṅgābāda"],
 ["औरत", "aurata"],
@@ -2152,7 +2152,7 @@ export const hindi: [string, string][] = [
 ["कंवलजीत", "kãvalajīta"],
 ["कंसर्ट", "kãsarṭa"],
 ["कंसल्टेंसी", "kãsalṭē̃sī"],
-["कइयों", "kaiyō̃"],
+["कइयों", "ka:iyō̃"],
 ["कई", "kaī"],
 ["कक्कड़", "kakkaṛa"],
 ["कक्ष", "kakṣa"],
@@ -3577,7 +3577,7 @@ export const hindi: [string, string][] = [
 ["गंधक", "gandhaka"],
 ["गंभीर", "gambhīra"],
 ["गंभीरता", "gambhīratā"],
-["गइ", "gai"],
+["गइ", "ga:i"],
 ["गई", "gaī"],
 ["गईं", "gaī̃"],
 ["गए", "gaē"],
@@ -9321,7 +9321,7 @@ export const hindi: [string, string][] = [
 ["बंधुओं", "bandhuō̃"],
 ["बंधुत्व", "bandhutva"],
 ["बंधे", "bandhē"],
-["बंबइया", "bambaiyā"],
+["बंबइया", "bamba:iyā"],
 ["बंबई", "bambaī"],
 ["बंसोडे", "bãsōḍē"],
 ["बंसोदे", "bãsōdē"],
@@ -9562,7 +9562,7 @@ export const hindi: [string, string][] = [
 ["बबली", "babalī"],
 ["बबलू", "babalū"],
 ["बम", "bama"],
-["बम्बइया", "bambaiyā"],
+["बम्बइया", "bamba:iyā"],
 ["बम्बई", "bambaī"],
 ["बमवर्षक", "bamavarṣaka"],
 ["बमों", "bamō̃"],
@@ -11291,7 +11291,7 @@ export const hindi: [string, string][] = [
 ["मुंगेर", "muṅgēra"],
 ["मुंजाल", "muñjāla"],
 ["मुंडा", "muṇḍā"],
-["मुंबइया", "mumbaiyā"],
+["मुंबइया", "mumba:iyā"],
 ["मुंबई", "mumbaī"],
 ["मुंशी", "mũśī"],
 ["मुंह", "mũha"],
@@ -13463,8 +13463,8 @@ export const hindi: [string, string][] = [
 ["वेश्यालय", "vēśyālaya"],
 ["वेषभूषा", "vēṣabhūṣā"],
 ["वेस्ट", "vēsṭa"],
-["वेस्टइंडीज", "vēsṭaiṇḍīja"],
-["वेस्टइंडीज़", "vēsṭaiṇḍīza"],
+["वेस्टइंडीज", "vēsṭa:iṇḍīja"],
+["वेस्टइंडीज़", "vēsṭa:iṇḍīza"],
 ["वैंगांकर", "vaiṅgāṅkara"],
 ["वैंडेनवर्ग", "vaiṇḍēnavarga"],
 ["वैकम", "vaikama"],
@@ -14068,7 +14068,7 @@ export const hindi: [string, string][] = [
 ["संहार", "sãhāra"],
 ["संहिता", "sãhitā"],
 ["सईद", "saīda"],
-["सउदी", "saudī"],
+["सउदी", "sa:udī"],
 ["सऊदी", "saūdī"],
 ["स्कडमोर", "skaḍamōra"],
 ["सकता", "sakatā"],
@@ -14322,7 +14322,7 @@ export const hindi: [string, string][] = [
 ["स्पाइबॉट", "spāibôṭa"],
 ["स्पाइवेयर", "spāivēyara"],
 ["सपाट", "sapāṭa"],
-["स्पायवेयरइन्फ़ो", "spāyavēyarainfō"],
+["स्पायवेयरइन्फ़ो", "spāyavēyara:infō"],
 ["स्पिटज़र", "spiṭazara"],
 ["स्पिति", "spiti"],
 ["स्पिन", "spina"],

transliterator.test.ts

@@ -14,7 +14,6 @@ it("should be good", () => {
 describe("unicode tests", () => {
     const tests = [
         ["ऱ्", "ṟ"],
-        ["र्\u200d", "r‍"], // should ZWJ be removed?
     ];
 
     it.each(tests)("check %s", (input, output) => {
@@ -59,7 +58,20 @@ describe("Malayalam special forms", () => {
     it.each(tests)("check %s", (input, output) => {
         expect(translit(input)).toEqual(output);
     });
+});
 
+describe("Devanagari special cases", () => {
+    const tests = [
+        ["बइ", "ba:i"],
+        ["बै", "bai"],
+        ["र्य", "rya"],
+        //["र्य", "r:ya"],
+        ["र्‍", "r̆"], // "eyelash R" as used in Nepali/Marathi
+    ];
+
+    it.each(tests)("check %s", (input, output) => {
+        expect(translit(input)).toEqual(output);
+    });
 });
 
 describe("npm transliteration", () => {
@@ -91,10 +103,8 @@ describe("conjuncts", () => {
     });
 });
 
-/*
 describe('hindi tests', () => {
     it.each(hindi)("check %s", (input, output) => {
         expect(translit(input)).toEqual(output);
     });
 });
-*/

transliterator.ts

@@ -46,20 +46,31 @@ const letters = expand(new Map<string | string[], Letter>([
   [["ऽ", "ഽ", "ಽ"], "’"],
   [["अ", "അ", "ಅ"], { value: "a", vowelType: "full" }],
   [["आ", "ആ", "ಆ"], { value: "ā", vowelType: "full" }],
-  [["इ", /*,*/ "ಇ"], { value: "i", vowelType: "full" }],
-  ["ഇ", {
+  [["इ", "ഇ", "ಇ"], {
     value: "i",
     vowelType: "full",
     onPrev: [
       // if previous is letter with no vowel mark
       // (so it will be -a), this is 'a:i' to distinguish from 'ai'
-      // really this should test that previous is also a Malayalam character
+      // really this should test that previous is also a script character
       // but can't do that without /v which is unsupported in Node currently.
-      [/(?<=\p{Letter})/uy, ":i"]
+      // this would be simpler if we checked upon the transliterated output...
+      [/(?<=\p{Letter})(?<![आആಆइഇಇईഈಈउഉಉ])/uy, ":i"]
     ]
   }],
   [["ई", "ഈ", "ಈ"], { value: "ī", vowelType: "full" }],
-  [["उ", "ഉ", "ಉ"], { value: "u", vowelType: "full" }],
+  [["उ", "ഉ", "ಉ"], {
+    value: "u",
+    vowelType: "full",
+    onPrev: [
+      // if previous is letter with no vowel mark
+      // (so it will be -a), this is 'a:u' to distinguish from 'au'
+      // really this should test that previous is also a script character
+      // but can't do that without /v which is unsupported in Node currently.
+      // this would be simpler if we checked upon the transliterated output...
+      [/(?<=\p{Letter})(?<![आആಆइഇಇईഈಈउഉಉ])/uy, ":u"]
+    ]
+  }],
   [["ऊ", "ഊ", "ಊ"], { value: "ū", vowelType: "full" }],
   [["ऋ", "ഋ", "ಋ"], { value: "r̥", vowelType: "semi" }],
   [["ॠ", "ൠ", "ಌೠ"], { value: "r̥̄", vowelType: "semi" }],
@@ -107,14 +118,21 @@ const letters = expand(new Map<string | string[], Letter>([
       [/(?<=്)\p{Letter}/uy, "y:"] // special medial form for suppressed vowel in Malayalam
     ]
   }],
-  [["र", /*,*/ "ರ"], { value: "r", implicitVowel: "a" }],
+  ["र", {
+    value: "r",
+    implicitVowel: "a",
+    onNext: [
+      [/(?<=्\u200d)/uy, "r̆"] // special case for "eyelash-R" in Nepali/Marathi
+    ]
+  }],
   ["ര", {
     value: "r",
     implicitVowel: "a",
     onNext: [
       [wordFinal, "ṟ"] // special final form in Malayalam
     ]
   }],
+  [["ರ"], { value: "r", implicitVowel: "a" }],
   [["ऱ", "റ", "ಱ"], { value: "ṟ", implicitVowel: "a" }], // TODO: Malayalam special final form
   [["ल", "ല", "ಲ"], { value: "l", implicitVowel: "a" }],
   [["ळ", "ള", "ಳ"], { value: "ḷ", implicitVowel: "a" }],
@@ -262,7 +280,8 @@ function expand<K>(input: Map<string | string[], K>): Map<string, K> {
 }
 
 // include Nuktas alongside character, for matching
-const regex = /((?!\p{Script=Latin})\p{L}[಼\u093c]?)(\p{M}*)/gu;
+// allow suffix ZWJ, for now
+const regex = /((?!\p{Script=Latin})\p{L}[಼\u093c]?)(\p{M}*)\u200d?/gu;
 
 export function translit(value: string): string {
   return value.replaceAll(regex, replacement);