Spaces:
Running
Running
Update pdf-server.js
Browse files- pdf-server.js +155 -188
pdf-server.js
CHANGED
|
@@ -9,13 +9,35 @@ app.use(express.json({ limit: '50mb' }));
|
|
| 9 |
const PORT = 3000;
|
| 10 |
const CHROMIUM_PATH = '/usr/bin/chromium-browser';
|
| 11 |
|
| 12 |
-
//
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
const DEVICES = {
|
| 14 |
-
'desktop': { width: 1920, height: 1080
|
| 15 |
-
'laptop': { width: 1366, height: 768
|
| 16 |
-
'tablet': { width: 768, height: 1024
|
| 17 |
-
'mobile': { width: 375, height: 812
|
| 18 |
-
'mobile-android': { width: 360, height: 740, userAgent: 'Mozilla/5.0 (Linux; Android 11; Pixel 5) AppleWebKit/537.36' },
|
| 19 |
};
|
| 20 |
|
| 21 |
// Formats PDF
|
|
@@ -28,14 +50,27 @@ const FORMATS = {
|
|
| 28 |
'A5': { width: '148mm', height: '210mm' },
|
| 29 |
};
|
| 30 |
|
| 31 |
-
// ==================== HEALTH
|
| 32 |
-
app.get('/health', (req, res) => {
|
| 33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
});
|
| 35 |
|
| 36 |
-
// ====================
|
| 37 |
app.post('/generate-pdf', async (req, res) => {
|
| 38 |
let browser = null;
|
|
|
|
|
|
|
| 39 |
try {
|
| 40 |
const {
|
| 41 |
html,
|
|
@@ -45,62 +80,68 @@ app.post('/generate-pdf', async (req, res) => {
|
|
| 45 |
printBackground = true,
|
| 46 |
} = req.body;
|
| 47 |
|
|
|
|
|
|
|
| 48 |
if (!html && !url) {
|
| 49 |
return res.status(400).json({ error: 'html ou url requis' });
|
| 50 |
}
|
| 51 |
|
| 52 |
-
console.log(
|
| 53 |
-
|
| 54 |
browser = await chromium.launch({
|
| 55 |
executablePath: CHROMIUM_PATH,
|
| 56 |
headless: true,
|
| 57 |
-
args:
|
| 58 |
});
|
|
|
|
| 59 |
|
| 60 |
const page = await browser.newPage();
|
|
|
|
| 61 |
|
| 62 |
if (url) {
|
| 63 |
-
|
|
|
|
| 64 |
} else {
|
| 65 |
-
|
|
|
|
| 66 |
}
|
|
|
|
| 67 |
|
| 68 |
-
// Attendre
|
| 69 |
-
await page.evaluate(() => {
|
| 70 |
-
return Promise.all(
|
| 71 |
-
Array.from(document.images)
|
| 72 |
-
.filter(img => !img.complete)
|
| 73 |
-
.map(img => new Promise(resolve => {
|
| 74 |
-
img.onload = img.onerror = resolve;
|
| 75 |
-
setTimeout(resolve, 5000);
|
| 76 |
-
}))
|
| 77 |
-
);
|
| 78 |
-
});
|
| 79 |
-
|
| 80 |
await page.waitForTimeout(1000);
|
| 81 |
|
| 82 |
const formatConfig = FORMATS[format] || FORMATS['6x9'];
|
|
|
|
| 83 |
|
| 84 |
const pdfBuffer = await page.pdf({
|
| 85 |
width: formatConfig.width,
|
| 86 |
height: formatConfig.height,
|
| 87 |
margin: margins,
|
| 88 |
printBackground,
|
| 89 |
-
preferCSSPageSize: true,
|
| 90 |
});
|
|
|
|
| 91 |
|
| 92 |
await browser.close();
|
|
|
|
| 93 |
|
| 94 |
res.json({
|
| 95 |
success: true,
|
| 96 |
pdf: pdfBuffer.toString('base64'),
|
|
|
|
| 97 |
size_mb: (pdfBuffer.length / 1024 / 1024).toFixed(2),
|
|
|
|
| 98 |
});
|
| 99 |
|
| 100 |
} catch (error) {
|
| 101 |
-
console.error(
|
| 102 |
-
|
| 103 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
}
|
| 105 |
});
|
| 106 |
|
|
@@ -116,7 +157,6 @@ app.post('/screenshot', async (req, res) => {
|
|
| 116 |
width,
|
| 117 |
height,
|
| 118 |
selector,
|
| 119 |
-
waitFor,
|
| 120 |
delay = 0,
|
| 121 |
} = req.body;
|
| 122 |
|
|
@@ -131,12 +171,11 @@ app.post('/screenshot', async (req, res) => {
|
|
| 131 |
browser = await chromium.launch({
|
| 132 |
executablePath: CHROMIUM_PATH,
|
| 133 |
headless: true,
|
| 134 |
-
args:
|
| 135 |
});
|
| 136 |
|
| 137 |
const context = await browser.newContext({
|
| 138 |
viewport: { width: viewportWidth, height: viewportHeight },
|
| 139 |
-
userAgent: deviceConfig.userAgent,
|
| 140 |
isMobile: device.includes('mobile') || device === 'tablet',
|
| 141 |
hasTouch: device.includes('mobile') || device === 'tablet',
|
| 142 |
});
|
|
@@ -144,37 +183,27 @@ app.post('/screenshot', async (req, res) => {
|
|
| 144 |
const page = await context.newPage();
|
| 145 |
|
| 146 |
if (url) {
|
| 147 |
-
await page.goto(url, { waitUntil: '
|
| 148 |
} else {
|
| 149 |
-
await page.setContent(html, { waitUntil: '
|
| 150 |
-
}
|
| 151 |
-
|
| 152 |
-
if (waitFor) {
|
| 153 |
-
await page.waitForSelector(waitFor, { timeout: 10000 });
|
| 154 |
}
|
| 155 |
|
| 156 |
if (delay > 0) {
|
| 157 |
await page.waitForTimeout(delay);
|
| 158 |
}
|
| 159 |
|
| 160 |
-
let
|
| 161 |
-
|
| 162 |
if (selector) {
|
| 163 |
const element = await page.$(selector);
|
| 164 |
if (element) {
|
| 165 |
-
|
| 166 |
-
|
| 167 |
-
await
|
| 168 |
-
return res.json({
|
| 169 |
-
success: true,
|
| 170 |
-
screenshot: screenshot.toString('base64'),
|
| 171 |
-
device,
|
| 172 |
-
viewport: { width: viewportWidth, height: viewportHeight },
|
| 173 |
-
});
|
| 174 |
}
|
|
|
|
|
|
|
| 175 |
}
|
| 176 |
|
| 177 |
-
const screenshot = await page.screenshot(screenshotOptions);
|
| 178 |
await browser.close();
|
| 179 |
|
| 180 |
res.json({
|
|
@@ -185,130 +214,67 @@ app.post('/screenshot', async (req, res) => {
|
|
| 185 |
});
|
| 186 |
|
| 187 |
} catch (error) {
|
| 188 |
-
console.error(
|
| 189 |
-
if (browser)
|
|
|
|
|
|
|
| 190 |
res.status(500).json({ error: error.message });
|
| 191 |
}
|
| 192 |
});
|
| 193 |
|
| 194 |
-
// ====================
|
| 195 |
-
app.post('/
|
| 196 |
let browser = null;
|
| 197 |
-
const videoDir = '/tmp/videos';
|
| 198 |
-
|
| 199 |
try {
|
| 200 |
const {
|
| 201 |
url,
|
| 202 |
-
|
| 203 |
-
|
| 204 |
-
scrollSpeed = 100,
|
| 205 |
-
scrollToBottom = true,
|
| 206 |
-
actions = [],
|
| 207 |
} = req.body;
|
| 208 |
|
| 209 |
if (!url) {
|
| 210 |
return res.status(400).json({ error: 'url requis' });
|
| 211 |
}
|
| 212 |
|
| 213 |
-
// Créer le dossier vidéo
|
| 214 |
-
if (!fs.existsSync(videoDir)) {
|
| 215 |
-
fs.mkdirSync(videoDir, { recursive: true });
|
| 216 |
-
}
|
| 217 |
-
|
| 218 |
-
const deviceConfig = DEVICES[device] || DEVICES['desktop'];
|
| 219 |
-
|
| 220 |
browser = await chromium.launch({
|
| 221 |
executablePath: CHROMIUM_PATH,
|
| 222 |
headless: true,
|
| 223 |
-
args:
|
| 224 |
-
});
|
| 225 |
-
|
| 226 |
-
const context = await browser.newContext({
|
| 227 |
-
viewport: { width: deviceConfig.width, height: deviceConfig.height },
|
| 228 |
-
userAgent: deviceConfig.userAgent,
|
| 229 |
-
isMobile: device.includes('mobile') || device === 'tablet',
|
| 230 |
-
recordVideo: {
|
| 231 |
-
dir: videoDir,
|
| 232 |
-
size: { width: deviceConfig.width, height: deviceConfig.height },
|
| 233 |
-
},
|
| 234 |
});
|
| 235 |
|
| 236 |
-
const
|
| 237 |
-
await page.goto(url, { waitUntil: 'networkidle', timeout: 30000 });
|
| 238 |
-
|
| 239 |
-
// Exécuter les actions personnalisées
|
| 240 |
-
for (const action of actions) {
|
| 241 |
-
switch (action.type) {
|
| 242 |
-
case 'wait':
|
| 243 |
-
await page.waitForTimeout(action.duration || 1000);
|
| 244 |
-
break;
|
| 245 |
-
case 'click':
|
| 246 |
-
await page.click(action.selector);
|
| 247 |
-
break;
|
| 248 |
-
case 'type':
|
| 249 |
-
await page.fill(action.selector, action.text);
|
| 250 |
-
break;
|
| 251 |
-
case 'scroll':
|
| 252 |
-
await page.evaluate((y) => window.scrollBy(0, y), action.y || 500);
|
| 253 |
-
break;
|
| 254 |
-
case 'screenshot':
|
| 255 |
-
// Pause visuelle
|
| 256 |
-
await page.waitForTimeout(500);
|
| 257 |
-
break;
|
| 258 |
-
}
|
| 259 |
-
}
|
| 260 |
-
|
| 261 |
-
// Scroll automatique vers le bas
|
| 262 |
-
if (scrollToBottom) {
|
| 263 |
-
const scrollHeight = await page.evaluate(() => document.body.scrollHeight);
|
| 264 |
-
const viewportHeight = deviceConfig.height;
|
| 265 |
-
let currentScroll = 0;
|
| 266 |
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
await page.waitForTimeout(scrollSpeed);
|
| 270 |
-
currentScroll += viewportHeight / 2;
|
| 271 |
-
}
|
| 272 |
|
| 273 |
-
|
| 274 |
-
|
| 275 |
-
|
| 276 |
-
|
| 277 |
-
}
|
| 278 |
|
| 279 |
-
|
| 280 |
-
|
| 281 |
-
await browser.close();
|
| 282 |
|
| 283 |
-
|
| 284 |
-
|
| 285 |
-
const videoFile = files.find(f => f.endsWith('.webm'));
|
| 286 |
|
| 287 |
-
|
| 288 |
-
return res.status(500).json({ error: 'Video file not created' });
|
| 289 |
}
|
| 290 |
|
| 291 |
-
|
| 292 |
-
const videoBuffer = fs.readFileSync(videoPath);
|
| 293 |
-
|
| 294 |
-
// Nettoyer
|
| 295 |
-
fs.unlinkSync(videoPath);
|
| 296 |
|
| 297 |
-
res.json({
|
| 298 |
-
success: true,
|
| 299 |
-
video: videoBuffer.toString('base64'),
|
| 300 |
-
format: 'webm',
|
| 301 |
-
size_mb: (videoBuffer.length / 1024 / 1024).toFixed(2),
|
| 302 |
-
});
|
| 303 |
|
| 304 |
} catch (error) {
|
| 305 |
-
console.error(
|
| 306 |
-
if (browser)
|
|
|
|
|
|
|
| 307 |
res.status(500).json({ error: error.message });
|
| 308 |
}
|
| 309 |
});
|
| 310 |
|
| 311 |
-
// ====================
|
| 312 |
app.post('/scrape', async (req, res) => {
|
| 313 |
let browser = null;
|
| 314 |
try {
|
|
@@ -329,19 +295,15 @@ app.post('/scrape', async (req, res) => {
|
|
| 329 |
browser = await chromium.launch({
|
| 330 |
executablePath: CHROMIUM_PATH,
|
| 331 |
headless: true,
|
| 332 |
-
args:
|
| 333 |
-
});
|
| 334 |
-
|
| 335 |
-
const context = await browser.newContext({
|
| 336 |
-
viewport: { width: deviceConfig.width, height: deviceConfig.height },
|
| 337 |
-
userAgent: deviceConfig.userAgent,
|
| 338 |
});
|
| 339 |
|
| 340 |
-
const page = await
|
| 341 |
-
await page.
|
|
|
|
| 342 |
|
| 343 |
if (waitFor) {
|
| 344 |
-
await page.waitForSelector(waitFor, { timeout: 10000 });
|
| 345 |
}
|
| 346 |
|
| 347 |
const result = {
|
|
@@ -349,12 +311,11 @@ app.post('/scrape', async (req, res) => {
|
|
| 349 |
title: await page.title(),
|
| 350 |
};
|
| 351 |
|
| 352 |
-
// Extraire avec les sélecteurs
|
| 353 |
for (const [key, selector] of Object.entries(selectors)) {
|
| 354 |
try {
|
| 355 |
result[key] = await page.$$eval(selector, els => els.map(el => ({
|
| 356 |
text: el.textContent?.trim(),
|
| 357 |
-
html: el.innerHTML,
|
| 358 |
href: el.href || null,
|
| 359 |
src: el.src || null,
|
| 360 |
})));
|
|
@@ -363,59 +324,63 @@ app.post('/scrape', async (req, res) => {
|
|
| 363 |
}
|
| 364 |
}
|
| 365 |
|
| 366 |
-
//
|
| 367 |
if (javascript) {
|
| 368 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 369 |
}
|
| 370 |
|
| 371 |
await browser.close();
|
| 372 |
res.json({ success: true, ...result });
|
| 373 |
|
| 374 |
} catch (error) {
|
| 375 |
-
console.error(
|
| 376 |
-
if (browser)
|
|
|
|
|
|
|
| 377 |
res.status(500).json({ error: error.message });
|
| 378 |
}
|
| 379 |
});
|
| 380 |
|
| 381 |
-
// ====================
|
| 382 |
-
app.post('/
|
| 383 |
let browser = null;
|
| 384 |
try {
|
| 385 |
const {
|
| 386 |
url,
|
| 387 |
-
|
| 388 |
-
|
|
|
|
| 389 |
} = req.body;
|
| 390 |
|
| 391 |
if (!url) {
|
| 392 |
return res.status(400).json({ error: 'url requis' });
|
| 393 |
}
|
| 394 |
|
|
|
|
|
|
|
| 395 |
browser = await chromium.launch({
|
| 396 |
executablePath: CHROMIUM_PATH,
|
| 397 |
headless: true,
|
| 398 |
-
args:
|
| 399 |
});
|
| 400 |
|
| 401 |
-
const
|
| 402 |
-
|
| 403 |
-
|
| 404 |
-
|
| 405 |
-
|
| 406 |
-
|
| 407 |
-
|
| 408 |
-
|
| 409 |
-
|
| 410 |
-
|
| 411 |
-
|
| 412 |
-
const
|
| 413 |
-
|
| 414 |
-
|
| 415 |
-
const screenshot = await page.screenshot({ type: 'png', fullPage });
|
| 416 |
-
screenshots[device] = screenshot.toString('base64');
|
| 417 |
-
|
| 418 |
-
await context.close();
|
| 419 |
}
|
| 420 |
|
| 421 |
await browser.close();
|
|
@@ -423,19 +388,21 @@ app.post('/multi-screenshot', async (req, res) => {
|
|
| 423 |
res.json({
|
| 424 |
success: true,
|
| 425 |
screenshots,
|
| 426 |
-
|
|
|
|
| 427 |
});
|
| 428 |
|
| 429 |
} catch (error) {
|
| 430 |
-
console.error(
|
| 431 |
-
if (browser)
|
|
|
|
|
|
|
| 432 |
res.status(500).json({ error: error.message });
|
| 433 |
}
|
| 434 |
});
|
| 435 |
|
| 436 |
-
// ====================
|
| 437 |
app.listen(PORT, '0.0.0.0', () => {
|
| 438 |
-
console.log(`🚀 PDF Server running on port ${PORT}`);
|
| 439 |
-
console.log(
|
| 440 |
-
console.log(`📄 PDF Formats: ${Object.keys(FORMATS).join(', ')}`);
|
| 441 |
});
|
|
|
|
| 9 |
const PORT = 3000;
|
| 10 |
const CHROMIUM_PATH = '/usr/bin/chromium-browser';
|
| 11 |
|
| 12 |
+
// Arguments Chromium pour environnement restreint (HuggingFace)
|
| 13 |
+
const CHROMIUM_ARGS = [
|
| 14 |
+
'--no-sandbox',
|
| 15 |
+
'--disable-setuid-sandbox',
|
| 16 |
+
'--disable-dev-shm-usage',
|
| 17 |
+
'--disable-gpu',
|
| 18 |
+
'--disable-software-rasterizer',
|
| 19 |
+
'--disable-extensions',
|
| 20 |
+
'--disable-background-networking',
|
| 21 |
+
'--disable-sync',
|
| 22 |
+
'--disable-translate',
|
| 23 |
+
'--disable-crash-reporter',
|
| 24 |
+
'--disable-breakpad',
|
| 25 |
+
'--disable-features=TranslateUI',
|
| 26 |
+
'--disable-ipc-flooding-protection',
|
| 27 |
+
'--no-first-run',
|
| 28 |
+
'--no-zygote',
|
| 29 |
+
'--single-process',
|
| 30 |
+
'--deterministic-fetch',
|
| 31 |
+
'--disable-features=IsolateOrigins',
|
| 32 |
+
'--disable-site-isolation-trials',
|
| 33 |
+
];
|
| 34 |
+
|
| 35 |
+
// Devices
|
| 36 |
const DEVICES = {
|
| 37 |
+
'desktop': { width: 1920, height: 1080 },
|
| 38 |
+
'laptop': { width: 1366, height: 768 },
|
| 39 |
+
'tablet': { width: 768, height: 1024 },
|
| 40 |
+
'mobile': { width: 375, height: 812 },
|
|
|
|
| 41 |
};
|
| 42 |
|
| 43 |
// Formats PDF
|
|
|
|
| 50 |
'A5': { width: '148mm', height: '210mm' },
|
| 51 |
};
|
| 52 |
|
| 53 |
+
// ==================== HEALTH ====================
|
| 54 |
+
app.get('/health', async (req, res) => {
|
| 55 |
+
let browser = null;
|
| 56 |
+
try {
|
| 57 |
+
browser = await chromium.launch({
|
| 58 |
+
executablePath: CHROMIUM_PATH,
|
| 59 |
+
headless: true,
|
| 60 |
+
args: CHROMIUM_ARGS,
|
| 61 |
+
});
|
| 62 |
+
await browser.close();
|
| 63 |
+
res.json({ status: 'ok', chromium: 'working' });
|
| 64 |
+
} catch (error) {
|
| 65 |
+
res.json({ status: 'error', chromium: error.message });
|
| 66 |
+
}
|
| 67 |
});
|
| 68 |
|
| 69 |
+
// ==================== PDF ====================
|
| 70 |
app.post('/generate-pdf', async (req, res) => {
|
| 71 |
let browser = null;
|
| 72 |
+
const startTime = Date.now();
|
| 73 |
+
|
| 74 |
try {
|
| 75 |
const {
|
| 76 |
html,
|
|
|
|
| 80 |
printBackground = true,
|
| 81 |
} = req.body;
|
| 82 |
|
| 83 |
+
console.log(`[PDF] Starting... format=${format}`);
|
| 84 |
+
|
| 85 |
if (!html && !url) {
|
| 86 |
return res.status(400).json({ error: 'html ou url requis' });
|
| 87 |
}
|
| 88 |
|
| 89 |
+
console.log('[PDF] Launching browser...');
|
|
|
|
| 90 |
browser = await chromium.launch({
|
| 91 |
executablePath: CHROMIUM_PATH,
|
| 92 |
headless: true,
|
| 93 |
+
args: CHROMIUM_ARGS,
|
| 94 |
});
|
| 95 |
+
console.log(`[PDF] Browser launched (${Date.now() - startTime}ms)`);
|
| 96 |
|
| 97 |
const page = await browser.newPage();
|
| 98 |
+
console.log(`[PDF] Page created (${Date.now() - startTime}ms)`);
|
| 99 |
|
| 100 |
if (url) {
|
| 101 |
+
console.log(`[PDF] Loading URL: ${url}`);
|
| 102 |
+
await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
|
| 103 |
} else {
|
| 104 |
+
console.log(`[PDF] Setting HTML content (${html.length} chars)`);
|
| 105 |
+
await page.setContent(html, { waitUntil: 'domcontentloaded', timeout: 30000 });
|
| 106 |
}
|
| 107 |
+
console.log(`[PDF] Content loaded (${Date.now() - startTime}ms)`);
|
| 108 |
|
| 109 |
+
// Attendre un peu pour le rendu
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 110 |
await page.waitForTimeout(1000);
|
| 111 |
|
| 112 |
const formatConfig = FORMATS[format] || FORMATS['6x9'];
|
| 113 |
+
console.log(`[PDF] Generating with format: ${JSON.stringify(formatConfig)}`);
|
| 114 |
|
| 115 |
const pdfBuffer = await page.pdf({
|
| 116 |
width: formatConfig.width,
|
| 117 |
height: formatConfig.height,
|
| 118 |
margin: margins,
|
| 119 |
printBackground,
|
|
|
|
| 120 |
});
|
| 121 |
+
console.log(`[PDF] Generated! Size: ${pdfBuffer.length} bytes (${Date.now() - startTime}ms)`);
|
| 122 |
|
| 123 |
await browser.close();
|
| 124 |
+
browser = null;
|
| 125 |
|
| 126 |
res.json({
|
| 127 |
success: true,
|
| 128 |
pdf: pdfBuffer.toString('base64'),
|
| 129 |
+
size_bytes: pdfBuffer.length,
|
| 130 |
size_mb: (pdfBuffer.length / 1024 / 1024).toFixed(2),
|
| 131 |
+
duration_ms: Date.now() - startTime,
|
| 132 |
});
|
| 133 |
|
| 134 |
} catch (error) {
|
| 135 |
+
console.error(`[PDF] ERROR: ${error.message}`);
|
| 136 |
+
console.error(error.stack);
|
| 137 |
+
if (browser) {
|
| 138 |
+
try { await browser.close(); } catch (e) {}
|
| 139 |
+
}
|
| 140 |
+
res.status(500).json({
|
| 141 |
+
error: error.message,
|
| 142 |
+
stack: error.stack,
|
| 143 |
+
duration_ms: Date.now() - startTime,
|
| 144 |
+
});
|
| 145 |
}
|
| 146 |
});
|
| 147 |
|
|
|
|
| 157 |
width,
|
| 158 |
height,
|
| 159 |
selector,
|
|
|
|
| 160 |
delay = 0,
|
| 161 |
} = req.body;
|
| 162 |
|
|
|
|
| 171 |
browser = await chromium.launch({
|
| 172 |
executablePath: CHROMIUM_PATH,
|
| 173 |
headless: true,
|
| 174 |
+
args: CHROMIUM_ARGS,
|
| 175 |
});
|
| 176 |
|
| 177 |
const context = await browser.newContext({
|
| 178 |
viewport: { width: viewportWidth, height: viewportHeight },
|
|
|
|
| 179 |
isMobile: device.includes('mobile') || device === 'tablet',
|
| 180 |
hasTouch: device.includes('mobile') || device === 'tablet',
|
| 181 |
});
|
|
|
|
| 183 |
const page = await context.newPage();
|
| 184 |
|
| 185 |
if (url) {
|
| 186 |
+
await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
|
| 187 |
} else {
|
| 188 |
+
await page.setContent(html, { waitUntil: 'domcontentloaded' });
|
|
|
|
|
|
|
|
|
|
|
|
|
| 189 |
}
|
| 190 |
|
| 191 |
if (delay > 0) {
|
| 192 |
await page.waitForTimeout(delay);
|
| 193 |
}
|
| 194 |
|
| 195 |
+
let screenshot;
|
|
|
|
| 196 |
if (selector) {
|
| 197 |
const element = await page.$(selector);
|
| 198 |
if (element) {
|
| 199 |
+
screenshot = await element.screenshot({ type: 'png' });
|
| 200 |
+
} else {
|
| 201 |
+
screenshot = await page.screenshot({ type: 'png', fullPage });
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 202 |
}
|
| 203 |
+
} else {
|
| 204 |
+
screenshot = await page.screenshot({ type: 'png', fullPage });
|
| 205 |
}
|
| 206 |
|
|
|
|
| 207 |
await browser.close();
|
| 208 |
|
| 209 |
res.json({
|
|
|
|
| 214 |
});
|
| 215 |
|
| 216 |
} catch (error) {
|
| 217 |
+
console.error(`[SCREENSHOT] ERROR: ${error.message}`);
|
| 218 |
+
if (browser) {
|
| 219 |
+
try { await browser.close(); } catch (e) {}
|
| 220 |
+
}
|
| 221 |
res.status(500).json({ error: error.message });
|
| 222 |
}
|
| 223 |
});
|
| 224 |
|
| 225 |
+
// ==================== MULTI-SCREENSHOT ====================
|
| 226 |
+
app.post('/multi-screenshot', async (req, res) => {
|
| 227 |
let browser = null;
|
|
|
|
|
|
|
| 228 |
try {
|
| 229 |
const {
|
| 230 |
url,
|
| 231 |
+
devices = ['desktop', 'tablet', 'mobile'],
|
| 232 |
+
fullPage = true,
|
|
|
|
|
|
|
|
|
|
| 233 |
} = req.body;
|
| 234 |
|
| 235 |
if (!url) {
|
| 236 |
return res.status(400).json({ error: 'url requis' });
|
| 237 |
}
|
| 238 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 239 |
browser = await chromium.launch({
|
| 240 |
executablePath: CHROMIUM_PATH,
|
| 241 |
headless: true,
|
| 242 |
+
args: CHROMIUM_ARGS,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 243 |
});
|
| 244 |
|
| 245 |
+
const screenshots = {};
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 246 |
|
| 247 |
+
for (const device of devices) {
|
| 248 |
+
const deviceConfig = DEVICES[device] || DEVICES['desktop'];
|
|
|
|
|
|
|
|
|
|
| 249 |
|
| 250 |
+
const context = await browser.newContext({
|
| 251 |
+
viewport: { width: deviceConfig.width, height: deviceConfig.height },
|
| 252 |
+
isMobile: device.includes('mobile') || device === 'tablet',
|
| 253 |
+
});
|
|
|
|
| 254 |
|
| 255 |
+
const page = await context.newPage();
|
| 256 |
+
await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
|
|
|
|
| 257 |
|
| 258 |
+
const screenshot = await page.screenshot({ type: 'png', fullPage });
|
| 259 |
+
screenshots[device] = screenshot.toString('base64');
|
|
|
|
| 260 |
|
| 261 |
+
await context.close();
|
|
|
|
| 262 |
}
|
| 263 |
|
| 264 |
+
await browser.close();
|
|
|
|
|
|
|
|
|
|
|
|
|
| 265 |
|
| 266 |
+
res.json({ success: true, screenshots, devices });
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 267 |
|
| 268 |
} catch (error) {
|
| 269 |
+
console.error(`[MULTI-SCREENSHOT] ERROR: ${error.message}`);
|
| 270 |
+
if (browser) {
|
| 271 |
+
try { await browser.close(); } catch (e) {}
|
| 272 |
+
}
|
| 273 |
res.status(500).json({ error: error.message });
|
| 274 |
}
|
| 275 |
});
|
| 276 |
|
| 277 |
+
// ==================== SCRAPE ====================
|
| 278 |
app.post('/scrape', async (req, res) => {
|
| 279 |
let browser = null;
|
| 280 |
try {
|
|
|
|
| 295 |
browser = await chromium.launch({
|
| 296 |
executablePath: CHROMIUM_PATH,
|
| 297 |
headless: true,
|
| 298 |
+
args: CHROMIUM_ARGS,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 299 |
});
|
| 300 |
|
| 301 |
+
const page = await browser.newPage();
|
| 302 |
+
await page.setViewportSize({ width: deviceConfig.width, height: deviceConfig.height });
|
| 303 |
+
await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
|
| 304 |
|
| 305 |
if (waitFor) {
|
| 306 |
+
await page.waitForSelector(waitFor, { timeout: 10000 }).catch(() => {});
|
| 307 |
}
|
| 308 |
|
| 309 |
const result = {
|
|
|
|
| 311 |
title: await page.title(),
|
| 312 |
};
|
| 313 |
|
| 314 |
+
// Extraire avec les sélecteurs
|
| 315 |
for (const [key, selector] of Object.entries(selectors)) {
|
| 316 |
try {
|
| 317 |
result[key] = await page.$$eval(selector, els => els.map(el => ({
|
| 318 |
text: el.textContent?.trim(),
|
|
|
|
| 319 |
href: el.href || null,
|
| 320 |
src: el.src || null,
|
| 321 |
})));
|
|
|
|
| 324 |
}
|
| 325 |
}
|
| 326 |
|
| 327 |
+
// JavaScript custom
|
| 328 |
if (javascript) {
|
| 329 |
+
try {
|
| 330 |
+
result.custom = await page.evaluate(javascript);
|
| 331 |
+
} catch (e) {
|
| 332 |
+
result.custom = { error: e.message };
|
| 333 |
+
}
|
| 334 |
}
|
| 335 |
|
| 336 |
await browser.close();
|
| 337 |
res.json({ success: true, ...result });
|
| 338 |
|
| 339 |
} catch (error) {
|
| 340 |
+
console.error(`[SCRAPE] ERROR: ${error.message}`);
|
| 341 |
+
if (browser) {
|
| 342 |
+
try { await browser.close(); } catch (e) {}
|
| 343 |
+
}
|
| 344 |
res.status(500).json({ error: error.message });
|
| 345 |
}
|
| 346 |
});
|
| 347 |
|
| 348 |
+
// ==================== SCROLL & SCREENSHOT (Alternative à vidéo) ====================
|
| 349 |
+
app.post('/scroll-capture', async (req, res) => {
|
| 350 |
let browser = null;
|
| 351 |
try {
|
| 352 |
const {
|
| 353 |
url,
|
| 354 |
+
device = 'desktop',
|
| 355 |
+
steps = 5,
|
| 356 |
+
delayBetweenSteps = 500,
|
| 357 |
} = req.body;
|
| 358 |
|
| 359 |
if (!url) {
|
| 360 |
return res.status(400).json({ error: 'url requis' });
|
| 361 |
}
|
| 362 |
|
| 363 |
+
const deviceConfig = DEVICES[device] || DEVICES['desktop'];
|
| 364 |
+
|
| 365 |
browser = await chromium.launch({
|
| 366 |
executablePath: CHROMIUM_PATH,
|
| 367 |
headless: true,
|
| 368 |
+
args: CHROMIUM_ARGS,
|
| 369 |
});
|
| 370 |
|
| 371 |
+
const page = await browser.newPage();
|
| 372 |
+
await page.setViewportSize({ width: deviceConfig.width, height: deviceConfig.height });
|
| 373 |
+
await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
|
| 374 |
+
|
| 375 |
+
const screenshots = [];
|
| 376 |
+
const scrollHeight = await page.evaluate(() => document.body.scrollHeight);
|
| 377 |
+
const stepHeight = scrollHeight / steps;
|
| 378 |
+
|
| 379 |
+
for (let i = 0; i <= steps; i++) {
|
| 380 |
+
await page.evaluate((y) => window.scrollTo(0, y), i * stepHeight);
|
| 381 |
+
await page.waitForTimeout(delayBetweenSteps);
|
| 382 |
+
const screenshot = await page.screenshot({ type: 'png' });
|
| 383 |
+
screenshots.push(screenshot.toString('base64'));
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 384 |
}
|
| 385 |
|
| 386 |
await browser.close();
|
|
|
|
| 388 |
res.json({
|
| 389 |
success: true,
|
| 390 |
screenshots,
|
| 391 |
+
count: screenshots.length,
|
| 392 |
+
device,
|
| 393 |
});
|
| 394 |
|
| 395 |
} catch (error) {
|
| 396 |
+
console.error(`[SCROLL-CAPTURE] ERROR: ${error.message}`);
|
| 397 |
+
if (browser) {
|
| 398 |
+
try { await browser.close(); } catch (e) {}
|
| 399 |
+
}
|
| 400 |
res.status(500).json({ error: error.message });
|
| 401 |
}
|
| 402 |
});
|
| 403 |
|
| 404 |
+
// ==================== START ====================
|
| 405 |
app.listen(PORT, '0.0.0.0', () => {
|
| 406 |
+
console.log(`🚀 PDF Server v2 running on port ${PORT}`);
|
| 407 |
+
console.log(`📄 Endpoints: /health, /generate-pdf, /screenshot, /multi-screenshot, /scrape, /scroll-capture`);
|
|
|
|
| 408 |
});
|