From 4a2408f197b7f6b72225b5dcb5498d50b28a1930 Mon Sep 17 00:00:00 2001
From: lingdocs <71590811+lingdocs@users.noreply.github.com>
Date: Mon, 4 Oct 2021 21:15:13 -0400
Subject: [PATCH] get english word function

---
 package.json                     |  2 +-
 src/lib/get-english-word.test.ts | 56 ++++++++++++++++++++++++++++++
 src/lib/get-english-word.ts      | 58 ++++++++++++++++++++++++++++++++
 src/library.ts                   |  4 +++
 4 files changed, 119 insertions(+), 1 deletion(-)
 create mode 100644 src/lib/get-english-word.test.ts
 create mode 100644 src/lib/get-english-word.ts

diff --git a/package.json b/package.json
index 6fb7154..33ad4dd 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@lingdocs/pashto-inflector",
-  "version": "1.0.7",
+  "version": "1.0.8",
   "author": "lingdocs.com",
   "description": "A Pashto inflection and verb conjugation engine, inculding React components for displaying Pashto text, inflections, and conjugations",
   "homepage": "https://verbs.lingdocs.com",
diff --git a/src/lib/get-english-word.test.ts b/src/lib/get-english-word.test.ts
new file mode 100644
index 0000000..86011ca
--- /dev/null
+++ b/src/lib/get-english-word.test.ts
@@ -0,0 +1,56 @@
+import { getEnglishWord } from "./get-english-word";
+
+test("getEnglishWord", () => {
+    const tests = [
+        // irreg
+        {
+            in: {"ts":1527815251,"i":7737,"p":"سړی","f":"saRéy","g":"saRey","e":"man","c":"n. m.","ec":"man","ep":"men"},
+            out: { singular: "man", plural: "men" },
+        },
+        // ch, sh, x, s, z - es ending
+        {
+            in: {"ts":1527823620,"i":1589,"p":"بزه","f":"bza","g":"bza","e":"patch (in a garment)","c":"n. f."},
+            out: { singular: "patch", plural: "patches" },
+        },
+        {
+            in: {"ts":1527823172,"i":2066,"p":"بېله","f":"belá","g":"bela","e":"marsh, swamp","c":"n. f."},
+            out: { singular: "marsh", plural: "marshes" },
+        },
+        {
+            in: {"ts":1589885085444,"i":11290,"p":"لمبړ","f":"lUmbáR","g":"lUmbaR","e":"fox","c":"n. m. unisex"},
+            out: { singular: "fox", plural: "foxes" },
+        },
+        {
+            in: {"ts":1592642858843,"i":13644,"p":"نېک","f":"nek","g":"nek","e":"boss, master","c":"n. m. anim. unisex"},
+            out: { singular: "boss", plural: "bosses" },
+        },
+        // (consonant) w/ y - ies 
+        {
+            in: {"ts":1527812677,"i":13994,"p":"هېواد","f":"hewaad","g":"hewaad","e":"country, homeland","c":"n. m."},
+            out: { singular: "country", plural: "countries" },
+        },
+        // (vowel) w/ y - ys
+        {
+            in: {"ts":1527815417,"i":14123,"p":"ورځ","f":"wradz","g":"wradz","e":"day","c":"n. f."},
+            out: { singular: "day", plural: "days" },
+        },
+        // ends in o
+        {
+            in: {"ts":1527820648,"i":788,"p":"الو","f":"aloo","g":"aloo","e":"potato","c":"n. m.","ppp":"الوګان","ppf":"aloogáan"},
+            out: { singular: "potato", plural: "potatoes" },
+        },
+        // ends in is
+        {
+            in: {"ts":1527814761,"i":3124,"p":"تحلیل","f":"tahleel","g":"tahleel","e":"analysis","c":"n. m."},
+            out: { singular: "analysis", plural: "analyses" },
+        },
+        // only plural
+        {
+            in: {"ts":1527815008,"i":8433,"p":"شودې","f":"shoodé","g":"shoode","e":"milk","c":"n. f. pl."},
+            out: { plural: "milk" },
+        },
+    ]
+    tests.forEach((t) => {
+        expect(getEnglishWord(t.in)).toEqual(t.out);
+    });
+})
\ No newline at end of file
diff --git a/src/lib/get-english-word.ts b/src/lib/get-english-word.ts
new file mode 100644
index 0000000..a071af7
--- /dev/null
+++ b/src/lib/get-english-word.ts
@@ -0,0 +1,58 @@
+import * as T from "../types";
+
+/**
+ * returns the singular and plural english word for a Pashto entry if possible
+ * NOTE: only works with nouns and adjectives
+ * 
+ * @param entry 
+ * @returns 
+ */
+export function getEnglishWord(entry: T.DictionaryEntry): {
+    singular?: string,
+    plural: string,
+} | undefined {
+    if (!entry.c || !entry.c.includes("n.") || entry.c.includes("adj.")) {
+        return undefined;
+    }
+    if (entry.ec && entry.ep) {
+        return {
+            singular: entry.ec,
+            plural: entry.ep,
+        };
+    }
+    const base = entry.e.split(",")[0].split(";")[0].split("(")[0].trim();
+    if (entry.c.includes("pl.")) {
+        return {
+            plural: base,
+        };
+    }
+    if (base.slice(-3) === "sis") {
+        return {
+            singular: base,
+            plural: `${base.slice(0, -2)}es`,
+        };
+    }
+    if (
+        ["sh", "ch"].includes(base.slice(-2)) ||
+        ["s", "x", "z", "o"].includes(base.slice(-1))
+    ) {
+        return {
+            singular: base,
+            plural: `${base}es`,
+        };
+    }
+    if (base.slice(-1) === "y" && !isVowel(base.slice(-2, -1))) {
+        return {
+            singular: base,
+            plural: `${base.slice(0, -1)}ies`,
+        };
+    }
+    return {
+        singular: base,
+        plural: `${base}s`,
+    };
+}
+
+function isVowel(l: string): boolean {
+    return ["a", "e", "i", "o", "u"].includes(l);
+}
\ No newline at end of file
diff --git a/src/library.ts b/src/library.ts
index e9371d5..2e8ef0e 100644
--- a/src/library.ts
+++ b/src/library.ts
@@ -36,6 +36,9 @@ import {
     isUnisexSet,
     isInflectionSet,
 } from "./lib/p-text-helpers";
+import {
+    getEnglishWord,
+} from "./lib/get-english-word";
 import {
     standardizePashto,
     standardizePhonetics,
@@ -118,6 +121,7 @@ export {
     personFromVerbBlockPos,
     removeAccents,
     hasAccents,
+    getEnglishWord,
     // protobuf helpers
     readDictionary,
     writeDictionary,