Prepare for captions translation

This commit is contained in:
vlOd2 2025-12-25 22:52:50 +02:00
parent 25139cc4cc
commit 5aea772477
4 changed files with 278 additions and 9 deletions

View file

@ -5,7 +5,13 @@
"[json]": {
"editor.defaultFormatter": "esbenp.prettier-vscode"
},
"[typescriptreact]": {
"[jsonc]": {
"editor.defaultFormatter": "esbenp.prettier-vscode"
},
"[type.scriptreact]": {
"editor.defaultFormatter": "dbaeumer.vscode-eslint"
},
"[javascript]": {
"editor.defaultFormatter": "esbenp.prettier-vscode"
}
}

View file

@ -44,7 +44,7 @@ importers:
version: 1.8.0
'@p-stream/providers':
specifier: github:p-stream/providers#production
version: https://codeload.github.com/p-stream/providers/tar.gz/5fa33694229da0506e7a265ff041acdb43e25ff0
version: https://codeload.github.com/p-stream/providers/tar.gz/6a94f978c64ec025171246b7b27e5867bdf21ed1
'@plasmohq/messaging':
specifier: ^0.6.2
version: 0.6.2(react@18.3.1)
@ -1207,8 +1207,8 @@ packages:
resolution: {integrity: sha512-nn5ozdjYQpUCZlWGuxcJY/KpxkWQs4DcbMCmKojjyrYDEAGy4Ce19NN4v5MduafTwJlbKc99UA8YhSVqq9yPZA==}
engines: {node: '>=12.4.0'}
'@p-stream/providers@https://codeload.github.com/p-stream/providers/tar.gz/5fa33694229da0506e7a265ff041acdb43e25ff0':
resolution: {tarball: https://codeload.github.com/p-stream/providers/tar.gz/5fa33694229da0506e7a265ff041acdb43e25ff0}
'@p-stream/providers@https://codeload.github.com/p-stream/providers/tar.gz/6a94f978c64ec025171246b7b27e5867bdf21ed1':
resolution: {tarball: https://codeload.github.com/p-stream/providers/tar.gz/6a94f978c64ec025171246b7b27e5867bdf21ed1}
version: 3.2.0
'@pkgjs/parseargs@0.11.0':
@ -5523,7 +5523,7 @@ snapshots:
'@nolyfill/is-core-module@1.0.39': {}
'@p-stream/providers@https://codeload.github.com/p-stream/providers/tar.gz/5fa33694229da0506e7a265ff041acdb43e25ff0':
'@p-stream/providers@https://codeload.github.com/p-stream/providers/tar.gz/6a94f978c64ec025171246b7b27e5867bdf21ed1':
dependencies:
abort-controller: 3.0.0
cheerio: 1.0.0-rc.12

View file

@ -0,0 +1,201 @@
import subsrt from "subsrt-ts";
import { Caption, ContentCaption } from "subsrt-ts/dist/types/handler";
const API_URL =
"https://translate.googleapis.com/translate_a/single?client=gtx&dt=t&dj=1&ie=UTF-8&oe=UTF-8&sl=auto&tl={TARGET_LANG}&q=";
const RETRY_COUNT = 3;
const FETCH_RATE = 100;
const SUBTITLES_CACHE: Map<string, ArrayBuffer> = new Map<
string,
ArrayBuffer
>();
async function translateText(text: string): Promise<string | undefined> {
if (!text) {
return "";
}
const response = await (
await fetch(
`${API_URL.replace("TARGET_LANG", "ro")}${encodeURIComponent(text)}`,
{
method: "GET",
headers: {
Accept: "application/json",
},
},
)
).json();
if (!response) {
throw new Error("Empty response");
}
return (response.sentences as any[])
.map((s: any) => s.trans as string)
.join("");
}
async function translateCaption(caption: ContentCaption): Promise<boolean> {
(caption as any).oldText = caption.text;
let text: string | undefined;
for (let i = 0; i < RETRY_COUNT; i += 1) {
try {
text = await translateText(caption.text.replace("\n", "<br>"));
if (text) {
text = text.replace("<br>", "\n");
break;
}
} catch (error) {
console.warn("[CTR] Re-trying caption translation", caption, error);
}
}
if (!text) {
console.error("[CTR] Failed to translate caption");
caption.text = `(CAPTION COULD NOT BE TRANSLATED)\n${caption.text}}`;
return false;
}
caption.text = text.trim();
return true;
}
async function translateCaptions(captions: ContentCaption[]): Promise<boolean> {
console.log("[CTR] Translating", captions.length, "captions");
try {
const results: boolean[] = await Promise.all(
captions.map((c) => translateCaption(c)),
);
const successCount = results.filter((v) => v).length;
const failedCount = results.length - successCount;
const successPercentange = (successCount / results.length) * 100;
const failedPercentange = (failedCount / results.length) * 100;
console.log(
"[CTR] Done translating captions",
results.length,
successCount,
failedCount,
successPercentange,
failedPercentange,
);
if (failedPercentange > successPercentange) {
throw new Error("Success percentage is not acceptable");
}
} catch (error) {
console.error(
"[CTR] Could not translate",
captions.length,
"captions",
error,
);
return false;
}
return true;
}
function tryUseCached(
caption: ContentCaption,
cache: Map<string, string>,
): boolean {
const text: string | undefined = cache.get(caption.text);
if (text) {
caption.text = text;
return true;
}
return false;
}
async function translateSRTData(data: string): Promise<string | undefined> {
let captions: Caption[];
try {
captions = subsrt.parse(data);
} catch (error) {
console.error("[CTR] Failed to parse subtitle data", error);
return undefined;
}
let translatedCaptions: Caption[] | undefined = [];
const contentCaptions: ContentCaption[] = [];
const translatedCache: Map<string, string> = new Map<string, string>();
for (const caption of captions) {
translatedCaptions.push(caption);
if (caption.type !== "caption") {
continue;
}
caption.text = caption.text
.trim()
.replace("\r\n", "\n")
.replace("\r", "\n");
contentCaptions.push(caption);
}
for (let i = 0; i < contentCaptions.length; i += 1) {
if (tryUseCached(contentCaptions[i], translatedCache)) {
continue;
}
const batch: ContentCaption[] = [contentCaptions[i]];
let j;
for (j = 1; j < FETCH_RATE; j += 1) {
if (i + j >= contentCaptions.length) {
break;
}
if (tryUseCached(contentCaptions[i + j], translatedCache)) {
continue;
}
batch.push(contentCaptions[i + j]);
}
i += j;
if (!(await translateCaptions(batch))) {
translatedCaptions = undefined;
break;
}
batch.forEach((c) => translatedCache.set((c as any).oldText!, c.text));
}
return translatedCaptions
? subsrt.build(translatedCaptions, { format: "srt" })
: undefined;
}
async function compressStr(string: string): Promise<ArrayBuffer> {
const byteArray = new TextEncoder().encode(string);
const cs = new CompressionStream("deflate");
const writer = cs.writable.getWriter();
writer.write(byteArray);
writer.close();
return new Response(cs.readable).arrayBuffer();
}
async function decompressStr(byteArray: ArrayBuffer): Promise<string> {
const cs = new DecompressionStream("deflate");
const writer = cs.writable.getWriter();
writer.write(byteArray);
writer.close();
return new Response(cs.readable).arrayBuffer().then(function (arrayBuffer) {
return new TextDecoder().decode(arrayBuffer);
});
}
export async function translateSubtitles(
id: string,
srtData: string,
): Promise<string | undefined> {
const cachedData: ArrayBuffer | undefined = SUBTITLES_CACHE.get(id);
if (cachedData) {
console.log("[CTR] Using cached translation for", id);
return decompressStr(cachedData);
}
console.log("[CTR] Translating", id);
const translatedData: string | undefined = await translateSRTData(srtData);
if (!translatedData) {
return undefined;
}
console.log("[CTR] Caching translation for", id);
SUBTITLES_CACHE.set(id, await compressStr(translatedData));
return translatedData;
}

View file

@ -1,12 +1,74 @@
import { useState } from "react";
import { useCallback, useState } from "react";
import { Button } from "@/components/buttons/Button";
import { usePlayer } from "@/components/player/hooks/usePlayer";
import { PlaybackErrorPart } from "@/pages/parts/player/PlaybackErrorPart";
import { PlayerPart } from "@/pages/parts/player/PlayerPart";
import {
CaptionListItem,
PlayerMeta,
playerStatus,
} from "@/stores/player/slices/source";
import { SourceSliceSource } from "@/stores/player/utils/qualities";
const subtitlesTestMeta: PlayerMeta = {
type: "movie",
title: "Subtitles Test",
releaseYear: 2024,
tmdbId: "0",
};
const subtitlesTestSource: SourceSliceSource = {
type: "hls",
url: "http://localhost:8000/media/master.m3u8",
};
const subtitlesTestSubs: CaptionListItem[] = [
{
id: "English",
language: "en",
url: "http://localhost:8000/subs/en.srt",
needsProxy: false,
},
{
id: "Romanian",
language: "ro",
url: "http://localhost:8000/subs/ro.srt",
needsProxy: false,
},
];
// mostly empty view, add whatever you need
export default function TestView() {
const [val, setVal] = useState(false);
const player = usePlayer();
const [showPlayer, setShowPlayer] = useState(false);
const [shouldCrash, setShouldCrash] = useState(false);
if (val) throw new Error("I crashed");
if (shouldCrash) {
throw new Error("I crashed");
}
return <Button onClick={() => setVal(true)}>Crash me!</Button>;
const subtitlesTest = useCallback(async () => {
setShowPlayer(true);
player.reset();
await new Promise((r) => {
setTimeout(r, 100);
});
player.setShouldStartFromBeginning(true);
player.setMeta(subtitlesTestMeta);
player.playMedia(subtitlesTestSource, subtitlesTestSubs, null);
}, [player]);
return showPlayer ? (
<PlayerPart backUrl="/dev/">
{player && (player as any).status === playerStatus.PLAYBACK_ERROR ? (
<PlaybackErrorPart />
) : null}
</PlayerPart>
) : (
<>
<Button onClick={() => setShouldCrash(true)}>Crash me!</Button>
<Button onClick={() => subtitlesTest()}>Subtitles test</Button>
</>
);
}