From 40641faf7aec2b03793726602098393f0b9dbd2e Mon Sep 17 00:00:00 2001
From: adueck <adueck@users.noreply.github.com>
Date: Thu, 19 Jan 2023 17:24:11 +0500
Subject: [PATCH] rough try of text conversion thing

---
 website/src/App.tsx                       |   6 +-
 website/src/lib/scriptToPhonetics.ts      | 115 ++++++++++++++++++++++
 website/src/screens/ScriptToPhonetics.tsx |  62 ++++++++++++
 3 files changed, 182 insertions(+), 1 deletion(-)
 create mode 100644 website/src/lib/scriptToPhonetics.ts
 create mode 100644 website/src/screens/ScriptToPhonetics.tsx

diff --git a/website/src/App.tsx b/website/src/App.tsx
index 2a0bc3f..e5df5e9 100644
--- a/website/src/App.tsx
+++ b/website/src/App.tsx
@@ -81,6 +81,7 @@ import { searchAllInflections } from "./lib/search-all-inflections";
 import {
     addToWordlist,
 } from "./lib/wordlist-database";
+import ScriptToPhonetics from "./screens/ScriptToPhonetics";
 
 // to allow Moustrap key combos even when input fields are in focus
 Mousetrap.prototype.stopCallback = function () {
@@ -97,7 +98,7 @@ if (prod) {
 
 const possibleLandingPages = [
     "/", "/about", "/settings", "/word", "/account", "/new-entries", "/share-target", "/phrase-builder",
-    "/privacy",
+    "/privacy", "/script-to-phonetics"
 ];
 const editorOnlyPages = [
     "/edit", "/review-tasks",
@@ -656,6 +657,9 @@ class App extends Component<RouteComponentProps, State> {
                                 loadUser={this.handleLoadUser}
                             />
                         </Route>
+                        <Route path="/script-to-phonetics">
+                            <ScriptToPhonetics />
+                        </Route>
                         {this.state.user?.level === "editor" && <Route path="/edit">
                             <EntryEditor
                                 isolatedEntry={this.state.isolatedEntry}
diff --git a/website/src/lib/scriptToPhonetics.ts b/website/src/lib/scriptToPhonetics.ts
new file mode 100644
index 0000000..2ed855d
--- /dev/null
+++ b/website/src/lib/scriptToPhonetics.ts
@@ -0,0 +1,115 @@
+import { conjugateVerb, getVerbInfo, inflectWord, isNounAdjOrVerb, isPashtoScript, removeFVarients, standardizePashto } from "@lingdocs/ps-react";
+import { dictionary, allEntries } from "./dictionary";
+import {
+    Types as T,
+} from "@lingdocs/ps-react";
+import {
+    InflectionSearchResult,
+} from "../types/dictionary-types";
+import { searchPile } from "./search-pile";
+
+/**
+ * Converts some Pashto texts to phonetics by looking up each word in the dictionary and finding 
+ * the phonetic equivalent
+ * 
+ * @param p 
+ * @returns 
+ */
+export function scriptToPhonetics(p: string): string {
+    const words = splitWords(p);
+    const entries = allEntries();
+    const f = (w: string) => wordToPhonetics(w, entries);
+    return words.map(f).join(" ");
+}
+
+
+function wordToPhonetics(p: string, entries: T.DictionaryEntry[]): string {
+    if (!isPashtoScript(p)) {
+        return p;
+    }
+    const results = dictionary.exactPashtoSearch(p);
+    const entryFs = results.map(entry => removeFVarients(entry.f));
+    const inflectionsR = searchAllInflectionsCore(entries, p);
+    // TODO: also add directional prefix stuff
+    const inflections = inflectionsR.map(result => result.forms)
+            .flatMap(form => form.flatMap(x => x.matches.map(x => x.ps.f)));
+    const possibilities = [...new Set([...entryFs, ...inflections])];
+    if (possibilities.length === 0) {
+        return p;
+    }
+    return possibilities.join("/");
+}
+
+export function searchAllInflectionsCore(allDocs: T.DictionaryEntry[], searchValue: string): InflectionSearchResult[] {
+    const preSearchFun = (ps: T.PsString) => ps.p.slice(0, 2) === searchValue.slice(0, 2);
+    const searchFun = (ps: T.PsString) => ps.p === searchValue;
+    // console.time(timerLabel);
+    return allDocs.reduce((all: InflectionSearchResult[], entry) => {
+        const type = isNounAdjOrVerb(entry);
+        if (entry.c && type === "verb") {
+            try {
+                const complement = (entry.l && entry.c.includes("comp.")) ? dictionary.findOneByTs(entry.l) : undefined;
+                const verbInfo = getVerbInfo(entry, complement);
+                const initialResults = searchPile(verbInfo as any, preSearchFun);
+                if (!initialResults.length) return all;
+                const conjugation = conjugateVerb(
+                    entry,
+                    complement,
+                );
+                const forms = searchPile(
+                    conjugation as any,
+                    searchFun,
+                );
+                if (forms.length) {
+                    return [...all, { entry, forms }];
+                }
+                return all;
+            } catch (e) {
+                console.error(e);
+                console.error("error inflecting", entry.p);
+                return all;
+            }
+        }
+        if (entry.c && type === "nounAdj") {
+            const inflections = inflectWord(entry);
+            if (!inflections) return all;
+            const forms = searchPile(inflections as any, searchFun);
+            if (forms.length) {
+                return [...all, { entry, forms }];
+            }
+        }
+        return all;
+    }, []);
+}
+
+function splitWords(p: string): string[] {
+    function isP(c: string): boolean {
+        return !!c.match(/[\u0621-\u065f\u0670-\u06d3\u06d5]/);
+    }
+    const words: string[] = [];
+    let current = "";
+    let onP: boolean = true;
+    const chars = p.split("");
+    for (let char of chars) {
+        const p = isP(char);
+        if (p) {
+            if (onP) {
+                current += char;
+            } else {
+                words.push(current);
+                current = char;
+                onP = true;
+            }
+        } else {
+            if (onP) {
+                words.push(current);
+                current = char;
+                onP = false;
+            } else {
+                current += char;
+            }
+        }
+    }
+    words.push(current);
+    return words.map(standardizePashto);
+}
\ No newline at end of file
diff --git a/website/src/screens/ScriptToPhonetics.tsx b/website/src/screens/ScriptToPhonetics.tsx
new file mode 100644
index 0000000..edd8032
--- /dev/null
+++ b/website/src/screens/ScriptToPhonetics.tsx
@@ -0,0 +1,62 @@
+/**
+ * Copyright (c) 2021 lingdocs.com
+ *
+ * This source code is licensed under the GPL3 license found in the
+ * LICENSE file in the root directory of this source tree.
+ *
+ */
+
+import { FormEvent, useState } from "react";
+import { Helmet } from "react-helmet";
+import { scriptToPhonetics } from "../lib/scriptToPhonetics";
+
+const preStyle: React.CSSProperties = {
+    overflowX: "auto",
+    whiteSpace: "pre-wrap",
+    wordWrap: "break-word",
+    lineHeight: "1.5",
+};
+
+const ScriptToPhonetics = () => {
+    const [text, setText] = useState<string>("");
+    const [result, setResult] = useState<string>("");
+    function handleConversion(e: FormEvent<HTMLFormElement>) {
+        e.preventDefault();
+        setResult("Converting... Please wait...");
+        setTimeout(() => {
+            setResult(scriptToPhonetics(text));
+        }, 50);
+    }
+
+    return <div className="width-limiter">
+        <Helmet>
+            <link rel="canonical" href="https://dictionary.lingdocs.com/script-to-phonetics" />
+            <meta name="description" content="Convert Pashto Script to Phonetics" />
+            <title>Script to Phonetics - LingDocs Pashto Dictionary</title>
+        </Helmet>
+        <h2>Script to Phonetics</h2>
+        <form onSubmit={handleConversion}>
+            <div className="form-group">
+                <label htmlFor="pashto-text">Pashto Script</label>
+                <textarea
+                    className="form-control"
+                    id="pashto-text"
+                    rows={4}
+                    value={text}
+                    onChange={e => setText(e.target.value)}
+                />
+            </div>
+            <div>
+                <button type="submit" className="btn btn-primary">
+                    <i className="fas fa-exchange-alt mr-2"/> Convert
+                </button>
+            </div>
+            {result && <div className="mt-3">
+                <label>Phonetics</label>
+                <pre style={preStyle}>{result}</pre>
+            </div>}
+        </form>
+    </div>
+};
+
+export default ScriptToPhonetics;
\ No newline at end of file