Import source: app/src/pages/NlpNormalizerPage
Import the existing local project snapshot. This commit organizes existing files; it does not represent newly developed functionality.
This commit is contained in:
arsam
@@ -0,0 +1,130 @@
|
||||
import { useState } from 'react';
|
||||
import { useNav, useToast } from '../../teknav-ui.jsx';
|
||||
import { iconNode } from '../../src/lib/icon.jsx';
|
||||
import clipboardIcon from '../../src/assets/icons/clipboard-copy.svg';
|
||||
import { LegalPage, LegalSection } from './shared.jsx';
|
||||
|
||||
export function NlpNormalizerPage() {
|
||||
const { navigate } = useNav();
|
||||
const toast = useToast();
|
||||
const [text, setText] = useState('');
|
||||
const [normalized, setNormalized] = useState('');
|
||||
const [options, setOptions] = useState({
|
||||
stripHtml: true,
|
||||
normalizeLetters: true,
|
||||
normalizeZwnj: true,
|
||||
cleanWhitespace: true,
|
||||
});
|
||||
|
||||
const toggle = (k) => setOptions((o) => ({ ...o, [k]: !o[k] }));
|
||||
|
||||
const handleNormalize = () => {
|
||||
let t = text || '';
|
||||
if (options.stripHtml) {
|
||||
t = t.replace(/<[^>]*>/g, ' ');
|
||||
}
|
||||
if (options.normalizeLetters) {
|
||||
t = t.replace(/ي/g, 'ی').replace(/ك/g, 'ک').replace(/ة/g, 'ه');
|
||||
}
|
||||
if (options.normalizeZwnj) {
|
||||
t = t.replace(/\s+(می|نمی)\s+/g, ' $1');
|
||||
t = t.replace(/\s+(تر|ترین|ها|های|ام|ات|اش|مان|تان|شان)\b/g, '$1');
|
||||
t = t.replace(/\b(می|نمی)\b\s+/g, '$1');
|
||||
}
|
||||
if (options.cleanWhitespace) {
|
||||
t = t.replace(/\s+/g, ' ');
|
||||
t = t.replace(/\n\s*\n+/g, '\n\n');
|
||||
}
|
||||
setNormalized(t.trim());
|
||||
toast('متن با موفقیت نرمال‌سازی شد', 'success');
|
||||
};
|
||||
|
||||
const copyToClipboard = (val) => {
|
||||
if (!val) return;
|
||||
navigator.clipboard.writeText(val);
|
||||
toast('کپی شد!', 'success');
|
||||
};
|
||||
|
||||
const textareaStyle = {
|
||||
width: '100%',
|
||||
minHeight: 180,
|
||||
background: '#FAF7F0',
|
||||
border: '1px solid #D8D0C4',
|
||||
borderRadius: 8,
|
||||
padding: '12px 16px',
|
||||
fontSize: 14,
|
||||
fontFamily: 'Vazirmatn, sans-serif',
|
||||
color: '#263238',
|
||||
outline: 'none',
|
||||
lineHeight: 1.7,
|
||||
resize: 'vertical',
|
||||
boxSizing: 'border-box',
|
||||
};
|
||||
|
||||
return (
|
||||
<LegalPage title="نرمال‌ساز متن فارسی (NLP)" subtitle="اصلاح نویسه‌ها، تنظیم فاصله‌ها و بهینه‌سازی متون فارسی برای ابزارهای پردازش زبان طبیعی و مدل‌های زبانی.">
|
||||
<div style={{ display: 'flex', flexDirection: 'column', gap: 20 }}>
|
||||
<div>
|
||||
<label style={{ fontSize: 13, fontWeight: 700, color: '#0F2A2E', display: 'block', marginBottom: 8 }}>متن ورودی:</label>
|
||||
<textarea
|
||||
value={text}
|
||||
onChange={(e) => setText(e.target.value)}
|
||||
placeholder="متن فارسی خود را برای نرمال‌سازی اینجا وارد کنید..."
|
||||
style={textareaStyle}
|
||||
/>
|
||||
</div>
|
||||
|
||||
<div style={{ background: '#FAF7F0', border: '1px solid #E4DDD2', borderRadius: 10, padding: 18 }}>
|
||||
<h3 style={{ fontSize: 14, fontWeight: 800, color: '#0F2A2E', marginBottom: 12 }}>تنظیمات نرمال‌سازی</h3>
|
||||
<div style={{ display: 'grid', gridTemplateColumns: 'repeat(auto-fit, minmax(200px, 1fr))', gap: 12 }}>
|
||||
{[
|
||||
['stripHtml', 'حذف تگ‌های HTML'],
|
||||
['normalizeLetters', 'تبدیل ی و ک عربی به فارسی'],
|
||||
['normalizeZwnj', 'تنظیم نیم‌فاصله‌های پیشوند/پسوند'],
|
||||
['cleanWhitespace', 'اصلاح فاصله‌ها و تب‌های مکرر'],
|
||||
].map(([k, label]) => (
|
||||
<label key={k} style={{ display: 'flex', alignItems: 'center', gap: 10, cursor: 'pointer', fontSize: 13, color: '#374549' }}>
|
||||
<input type="checkbox" checked={options[k]} onChange={() => toggle(k)} style={{ accentColor: '#0F6B73' }} />
|
||||
<span>{label}</span>
|
||||
</label>
|
||||
))}
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div style={{ display: 'flex', gap: 12 }}>
|
||||
<button onClick={handleNormalize} style={{ background: '#0F6B73', color: '#fff', border: 'none', borderRadius: 8, padding: '12px 28px', fontSize: 14, fontWeight: 600, cursor: 'pointer', fontFamily: 'Vazirmatn,sans-serif' }}>
|
||||
نرمال‌سازی متن
|
||||
</button>
|
||||
<button onClick={() => navigate('/tools')} style={{ background: 'transparent', color: '#5F6B6D', border: '1px solid #D8D0C4', borderRadius: 8, padding: '12px 20px', fontSize: 14, fontWeight: 600, cursor: 'pointer', fontFamily: 'Vazirmatn,sans-serif' }}>
|
||||
بازگشت
|
||||
</button>
|
||||
</div>
|
||||
|
||||
{normalized && (
|
||||
<div style={{ marginTop: 10 }}>
|
||||
<div style={{ display: 'flex', justifyContent: 'space-between', alignItems: 'center', marginBottom: 8 }}>
|
||||
<label style={{ fontSize: 13, fontWeight: 700, color: '#0F2A2E' }}>متن خروجی:</label>
|
||||
<button onClick={() => copyToClipboard(normalized)} style={{ background: '#FAF7F0', border: '1px solid #D8D0C4', borderRadius: 6, padding: '6px 12px', fontSize: 12, cursor: 'pointer', fontFamily: 'Vazirmatn,sans-serif', display: 'flex', alignItems: 'center', gap: 6 }}>
|
||||
<span className="tek-icon" style={{ ['--i']: `url("${clipboardIcon}")` }} aria-hidden="true" /> کپی خروجی
|
||||
</button>
|
||||
</div>
|
||||
<textarea
|
||||
readOnly
|
||||
value={normalized}
|
||||
style={{ ...textareaStyle, background: '#fff' }}
|
||||
/>
|
||||
</div>
|
||||
)}
|
||||
|
||||
<div style={{ borderTop: '1px solid #E4DDD2', paddingTop: 20, marginTop: 10 }}>
|
||||
<h3 style={{ fontSize: 14, fontWeight: 800, color: '#0F2A2E', marginBottom: 8 }}>ارجاع علمی و اشتراک‌گذاری</h3>
|
||||
<p style={{ color: '#5F6B6D', fontSize: 12, lineHeight: 1.7, marginBottom: 12 }}>برای استفاده از خروجی‌های این ابزار در تحقیقات یا کارهای متن‌باز خود، می‌توانید از فرمت ارجاع زیر استفاده کنید:</p>
|
||||
<div style={{ background: '#FAF7F0', border: '1px solid #E4DDD2', borderRadius: 8, padding: 12, fontSize: 11, fontFamily: 'monospace', direction: 'ltr', textAlign: 'left', overflowX: 'auto', display: 'flex', justifyContent: 'space-between', alignItems: 'center' }}>
|
||||
<code style={{ color: '#0F2A2E' }}>[Teknav Persian NLP Tool] (https://www.teknav.ir/tools/nlp-normalizer)</code>
|
||||
<button onClick={() => copyToClipboard('[Teknav Persian NLP Tool] (https://www.teknav.ir/tools/nlp-normalizer)')} style={{ background: '#fff', border: '1px solid #D8D0C4', borderRadius: 4, padding: '2px 8px', fontSize: 10, cursor: 'pointer', fontFamily: 'sans-serif' }}>Copy</button>
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
</LegalPage>
|
||||
);
|
||||
}
|
||||
Reference in New Issue
Block a user