Client-side language detection

Signed-off-by: marcin mikołajczak <git@mkljczk.pl>
This commit is contained in:
marcin mikołajczak
2024-05-18 15:47:36 +02:00
parent 0d2b7f026f
commit 15ad2ea91f
9 changed files with 89 additions and 16 deletions

View File

@ -27,6 +27,7 @@ const languages = Object.entries(languagesObject) as Array<[Language, string]>;
const messages = defineMessages({
languagePrompt: { id: 'compose.language_dropdown.prompt', defaultMessage: 'Select language' },
languageSuggestion: { id: 'compose.language_dropdown.suggestion', defaultMessage: '{language} (detected)' },
search: { id: 'compose.language_dropdown.search', defaultMessage: 'Search language…' },
});
@ -61,7 +62,7 @@ const LanguageDropdown: React.FC<ILanguageDropdown> = ({ composeId }) => {
],
});
const language = useCompose(composeId).language;
const { language, suggested_language: suggestedLanguage } = useCompose(composeId);
const handleClick: React.EventHandler<
React.MouseEvent<HTMLButtonElement> | React.KeyboardEvent<HTMLButtonElement>
@ -253,12 +254,18 @@ const LanguageDropdown: React.FC<ILanguageDropdown> = ({ composeId }) => {
const isSearching = searchValue !== '';
const results = search();
let buttonLabel = intl.formatMessage(messages.languagePrompt);
if (language) buttonLabel = languagesObject[language];
else if (suggestedLanguage) buttonLabel = intl.formatMessage(messages.languageSuggestion, {
language: languagesObject[suggestedLanguage as Language] || suggestedLanguage,
});
return (
<>
<Button
theme='muted'
size='xs'
text={language ? languagesObject[language] : intl.formatMessage(messages.languagePrompt)}
text={buttonLabel}
icon={require('@tabler/icons/outline/language.svg')}
secondaryIcon={require('@tabler/icons/outline/chevron-down.svg')}
title={intl.formatMessage(messages.languagePrompt)}

View File

@ -1,14 +1,17 @@
import { useLexicalComposerContext } from '@lexical/react/LexicalComposerContext';
import { $createRemarkExport } from '@mkljczk/lexical-remark';
import { type LanguageIdentificationModel } from 'fasttext.wasm.js/dist/models/language-identification/common.js';
import { $getRoot } from 'lexical';
import debounce from 'lodash/debounce';
import { useCallback, useEffect } from 'react';
import { addSuggestedQuote, setEditorState } from 'soapbox/actions/compose';
import { addSuggestedLanguage, addSuggestedQuote, setEditorState } from 'soapbox/actions/compose';
import { fetchStatus } from 'soapbox/actions/statuses';
import { useAppDispatch, useFeatures } from 'soapbox/hooks';
import { getStatusIdsFromLinksInContent } from 'soapbox/utils/status';
let lidModel: LanguageIdentificationModel;
interface IStatePlugin {
composeId: string;
isWysiwyg?: boolean;
@ -50,9 +53,35 @@ const StatePlugin: React.FC<IStatePlugin> = ({ composeId, isWysiwyg }) => {
});
}, 2000), []);
const detectLanguage = useCallback(debounce(async (text: string) => {
dispatch(async (dispatch, getState) => {
const state = getState();
const compose = state.compose.get(composeId);
if (!features.postLanguages || features.languageDetection || compose?.language) return;
const wordsLength = text.split(/\s+/).length;
if (wordsLength < 4) return;
if (!lidModel) {
// eslint-disable-next-line import/extensions
const { getLIDModel } = await import('fasttext.wasm.js/common');
lidModel = await getLIDModel();
}
if (!lidModel.model) await lidModel.load();
const { alpha2, possibility } = await lidModel.identify(text.replace(/\s+/i, ' '));
if (alpha2 && possibility > 0.5) {
dispatch(addSuggestedLanguage(composeId, alpha2));
}
});
}, 750), []);
useEffect(() => {
editor.registerUpdateListener(({ editorState }) => {
let text;
const plainText = editorState.read(() => $getRoot().getTextContent());
let text = plainText;
if (isWysiwyg) {
text = editorState.read($createRemarkExport({
handlers: {
@ -60,13 +89,12 @@ const StatePlugin: React.FC<IStatePlugin> = ({ composeId, isWysiwyg }) => {
mention: (node) => ({ type: 'text', value: node.getTextContent() }),
},
}));
} else {
text = editorState.read(() => $getRoot().getTextContent());
}
const isEmpty = text === '';
const data = isEmpty ? null : JSON.stringify(editorState.toJSON());
dispatch(setEditorState(composeId, data, text));
getQuoteSuggestions(text);
getQuoteSuggestions(plainText);
detectLanguage(plainText);
});
}, [editor]);