feat(wip): render url
This commit is contained in:
@@ -1,58 +1,94 @@
|
|||||||
import { useState } from 'react';
|
import React, { useState } from 'react';
|
||||||
import axios from 'axios';
|
import axios from 'axios';
|
||||||
import VisualSelector from '../VisualSelector';
|
|
||||||
|
|
||||||
const Scraper = () => {
|
function Scraper() {
|
||||||
const [url, setUrl] = useState('');
|
const [url, setUrl] = useState('');
|
||||||
const [htmlContent, setHtmlContent] = useState('');
|
const [selections, setSelections] = useState([]);
|
||||||
const [selectedElements, setSelectedElements] = useState([]);
|
const [data, setData] = useState(null);
|
||||||
const [scrapedData, setScrapedData] = useState([]);
|
|
||||||
|
|
||||||
const handleUrlChange = (e) => {
|
const handleUrlChange = (e) => {
|
||||||
setUrl(e.target.value);
|
setUrl(e.target.value);
|
||||||
};
|
};
|
||||||
|
|
||||||
const handleFetchWebsite = async () => {
|
const handleSelectElement = (e) => {
|
||||||
try {
|
const selection = {
|
||||||
const response = await axios.post('http://localhost:8000/fetch-website', { url });
|
name: `selection_${selections.length + 1}`,
|
||||||
setHtmlContent(response.data.html);
|
selector: generateSelector(e.target),
|
||||||
} catch (error) {
|
};
|
||||||
console.error('Error:', error);
|
setSelections((prevSelections) => [...prevSelections, selection]);
|
||||||
}
|
|
||||||
};
|
};
|
||||||
|
|
||||||
const handleElementSelection = (selectedElements) => {
|
const generateSelector = (element) => {
|
||||||
setSelectedElements(selectedElements);
|
let selector = element.tagName.toLowerCase();
|
||||||
|
let parent = element.parentNode;
|
||||||
|
|
||||||
|
while (parent && parent.tagName !== 'HTML') {
|
||||||
|
const id = parent.id;
|
||||||
|
const classes = Array.from(parent.classList).join('.');
|
||||||
|
|
||||||
|
if (id) {
|
||||||
|
selector = `#${id} > ${selector}`;
|
||||||
|
break;
|
||||||
|
} else if (classes) {
|
||||||
|
selector = `.${classes} > ${selector}`;
|
||||||
|
break;
|
||||||
|
} else {
|
||||||
|
selector = `${parent.tagName.toLowerCase()} > ${selector}`;
|
||||||
|
}
|
||||||
|
|
||||||
|
parent = parent.parentNode;
|
||||||
|
}
|
||||||
|
|
||||||
|
return selector;
|
||||||
};
|
};
|
||||||
|
|
||||||
const handleScrape = async () => {
|
const handleScrape = async () => {
|
||||||
try {
|
try {
|
||||||
const response = await axios.post('http://localhost:8000/scrape', {
|
const response = await axios.post('http://localhost:3000/scrape', {
|
||||||
url,
|
url,
|
||||||
selectedElements,
|
selections,
|
||||||
});
|
});
|
||||||
setScrapedData(response.data.data);
|
setData(response.data);
|
||||||
|
console.log('Scraped Data', response.data);
|
||||||
} catch (error) {
|
} catch (error) {
|
||||||
console.error('Error:', error);
|
console.error('Error scraping:', error);
|
||||||
}
|
}
|
||||||
};
|
};
|
||||||
|
|
||||||
return (
|
return (
|
||||||
<div>
|
<div>
|
||||||
<input type="text" value={url} onChange={handleUrlChange} placeholder="Enter URL" />
|
<input
|
||||||
<button onClick={handleFetchWebsite}>Fetch Website</button>
|
type="text"
|
||||||
{htmlContent && (
|
placeholder="Enter URL to scrape"
|
||||||
<VisualSelector
|
value={url}
|
||||||
htmlContent={htmlContent}
|
onChange={handleUrlChange}
|
||||||
onElementSelection={handleElementSelection}
|
/>
|
||||||
/>
|
<button onClick={handleScrape}>Scrape</button>
|
||||||
|
<p>Click on the elements you want to scrape:</p>
|
||||||
|
<iframe
|
||||||
|
src={url}
|
||||||
|
onLoad={(e) => {
|
||||||
|
const iframeDocument = e.target.contentDocument;
|
||||||
|
iframeDocument.body.onclick = handleSelectElement;
|
||||||
|
}}
|
||||||
|
style={{ width: '100%', height: '500px' }}
|
||||||
|
/>
|
||||||
|
<p>Selected Elements:</p>
|
||||||
|
<ul>
|
||||||
|
{selections.map((selection, index) => (
|
||||||
|
<li key={index}>
|
||||||
|
{selection.name}: {selection.selector}
|
||||||
|
</li>
|
||||||
|
))}
|
||||||
|
</ul>
|
||||||
|
{data && (
|
||||||
|
<div>
|
||||||
|
<h2>Scraped Data:</h2>
|
||||||
|
<pre>{JSON.stringify(data, null, 2)}</pre>
|
||||||
|
</div>
|
||||||
)}
|
)}
|
||||||
<button onClick={handleScrape} disabled={selectedElements.length === 0}>
|
|
||||||
Scrape
|
|
||||||
</button>
|
|
||||||
<pre>{JSON.stringify(scrapedData, null, 2)}</pre>
|
|
||||||
</div>
|
</div>
|
||||||
);
|
);
|
||||||
};
|
}
|
||||||
|
|
||||||
export default Scraper;
|
export default Scraper;
|
||||||
Reference in New Issue
Block a user