Update index.js
Browse files
index.js
CHANGED
@@ -338,6 +338,74 @@ app.post('/process', upload.single('htmlFile'), (req, res) => {
|
|
338 |
}
|
339 |
});
|
340 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
341 |
const PORT = process.env.PORT || 3000;
|
342 |
app.listen(PORT, () => {
|
343 |
console.log(`Server running on http://localhost:${PORT}`);
|
|
|
338 |
}
|
339 |
});
|
340 |
|
341 |
+
// Add this function to handle script execution
|
342 |
+
function executeCheerioScript(html, scriptContent) {
|
343 |
+
try {
|
344 |
+
const $ = cheerio.load(html, {
|
345 |
+
decodeEntities: false,
|
346 |
+
xmlMode: false,
|
347 |
+
lowerCaseTags: true
|
348 |
+
});
|
349 |
+
|
350 |
+
// Create a safe function from the script content
|
351 |
+
const extractorFunction = new Function('$', `
|
352 |
+
${scriptContent}
|
353 |
+
`);
|
354 |
+
|
355 |
+
// Execute the script with Cheerio instance
|
356 |
+
const result = extractorFunction($);
|
357 |
+
return {
|
358 |
+
success: true,
|
359 |
+
data: result,
|
360 |
+
error: null
|
361 |
+
};
|
362 |
+
} catch (err) {
|
363 |
+
return {
|
364 |
+
success: false,
|
365 |
+
data: null,
|
366 |
+
error: err.message
|
367 |
+
};
|
368 |
+
}
|
369 |
+
}
|
370 |
+
|
371 |
+
// cherio extract endpoint
|
372 |
+
app.post('/extract', upload.single('htmlFile'), (req, res) => {
|
373 |
+
try {
|
374 |
+
const startTime = Date.now();
|
375 |
+
let htmlContent = req.file
|
376 |
+
? req.file.buffer.toString('utf8')
|
377 |
+
: req.body.html || '';
|
378 |
+
|
379 |
+
const extractorScript = req.body.script;
|
380 |
+
|
381 |
+
if (!htmlContent.trim()) {
|
382 |
+
return res.status(400).json({ error: 'No HTML content provided.' });
|
383 |
+
}
|
384 |
+
|
385 |
+
if (!extractorScript) {
|
386 |
+
return res.status(400).json({ error: 'No extractor script provided.' });
|
387 |
+
}
|
388 |
+
|
389 |
+
const result = executeCheerioScript(htmlContent, extractorScript);
|
390 |
+
|
391 |
+
return res.json({
|
392 |
+
success: result.success,
|
393 |
+
data: result.data,
|
394 |
+
error: result.error,
|
395 |
+
processingTime: `${Date.now() - startTime}ms`
|
396 |
+
});
|
397 |
+
|
398 |
+
} catch (err) {
|
399 |
+
console.error('Extraction failed:', err);
|
400 |
+
return res.status(500).json({
|
401 |
+
success: false,
|
402 |
+
error: 'Internal server error.',
|
403 |
+
details: err.message.substring(0, 100)
|
404 |
+
});
|
405 |
+
}
|
406 |
+
});
|
407 |
+
|
408 |
+
|
409 |
const PORT = process.env.PORT || 3000;
|
410 |
app.listen(PORT, () => {
|
411 |
console.log(`Server running on http://localhost:${PORT}`);
|