mirror of
https://github.com/R0m1k3/Priceflow.git
synced 2026-10-11 17:29:14 +02:00
364 lines
14 KiB
Python
364 lines
14 KiB
Python
import asyncio
|
|
import logging
|
|
import os
|
|
from dataclasses import dataclass
|
|
from datetime import datetime
|
|
|
|
from playwright.async_api import Browser, BrowserContext, Page, async_playwright
|
|
from playwright.async_api import TimeoutError as PlaywrightTimeoutError
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
BROWSERLESS_URL = os.getenv("BROWSERLESS_URL", "ws://browserless:3000")
|
|
|
|
POPUP_SELECTORS = [
|
|
"button[aria-label='Close']",
|
|
"button[aria-label='close']",
|
|
".close-button",
|
|
".modal-close",
|
|
"svg[data-name='Close']",
|
|
"[class*='popup'] button",
|
|
"[class*='modal'] button",
|
|
"button:has-text('No, thanks')",
|
|
"button:has-text('No thanks')",
|
|
"a:has-text('No, thanks')",
|
|
"div[role='dialog'] button[aria-label='Close']",
|
|
# Amazon Interstitials (Added for robustness)
|
|
"button:has-text('Continuer les achats')",
|
|
"span:has-text('Continuer les achats')",
|
|
"a:has-text('Continuer les achats')",
|
|
"input[value='Continuer les achats']",
|
|
"input[value='Continue shopping']",
|
|
"span.a-button-inner > input.a-button-input[type='submit']",
|
|
"form:has-text('Continuer les achats') input[type='submit']",
|
|
"[aria-labelledby='continue-shopping-label']",
|
|
# Didomi / Gifi
|
|
"#didomi-notice-agree-button",
|
|
"button[id='didomi-notice-agree-button']",
|
|
"span:has-text('Accepter & Fermer')",
|
|
"button:has-text('Accepter & Fermer')",
|
|
# Common banners
|
|
"#sp-cc-accept",
|
|
"#onetrust-accept-btn-handler",
|
|
".cookie-consent-accept",
|
|
"[data-action='accept-cookies']",
|
|
"button[id*='accept']",
|
|
"button[class*='accept']",
|
|
]
|
|
|
|
|
|
@dataclass
|
|
class ScrapeConfig:
|
|
"""Configuration for scraping parameters."""
|
|
|
|
smart_scroll: bool = False
|
|
scroll_pixels: int = 350
|
|
text_length: int = 0
|
|
timeout: int = 90000
|
|
|
|
|
|
class ScraperService:
|
|
_playwright = None
|
|
_browser: Browser | None = None
|
|
_lock = asyncio.Lock()
|
|
|
|
@classmethod
|
|
async def initialize(cls):
|
|
"""Initialize the shared browser instance (Public, Thread-Safe)."""
|
|
async with cls._lock:
|
|
await cls._initialize()
|
|
|
|
@classmethod
|
|
async def _initialize(cls):
|
|
"""Internal initialization logic (Assumes lock is held)."""
|
|
if cls._browser is None:
|
|
logger.info("Initializing ScraperService shared browser...")
|
|
cls._playwright = await async_playwright().start()
|
|
cls._browser = await cls._connect_browser(cls._playwright)
|
|
logger.info("ScraperService initialized.")
|
|
|
|
@classmethod
|
|
async def shutdown(cls):
|
|
"""Shutdown the shared browser instance."""
|
|
async with cls._lock:
|
|
if cls._browser:
|
|
logger.info("Shutting down ScraperService shared browser...")
|
|
await cls._browser.close()
|
|
cls._browser = None
|
|
if cls._playwright:
|
|
await cls._playwright.stop()
|
|
cls._playwright = None
|
|
logger.info("ScraperService shutdown complete.")
|
|
|
|
@classmethod
|
|
async def _ensure_browser_connected(cls) -> bool:
|
|
"""Ensure browser is connected, reconnect if needed."""
|
|
async with cls._lock:
|
|
try:
|
|
if cls._browser is None:
|
|
logger.warning("Browser not initialized, initializing...")
|
|
await cls._initialize()
|
|
return cls._browser is not None
|
|
|
|
# Test if browser is still alive by trying to create a context
|
|
try:
|
|
test_context = await cls._browser.new_context()
|
|
await test_context.close()
|
|
return True
|
|
except Exception as e:
|
|
logger.error(f"Browser connection test failed: {e}")
|
|
logger.info("Attempting to reconnect browser...")
|
|
# Clear the old browser
|
|
cls._browser = None
|
|
if cls._playwright:
|
|
try:
|
|
await cls._playwright.stop()
|
|
except Exception:
|
|
pass
|
|
cls._playwright = None
|
|
# Reconnect
|
|
await cls._initialize()
|
|
return cls._browser is not None
|
|
except Exception as e:
|
|
logger.error(f"Failed to ensure browser connection: {e}")
|
|
return False
|
|
|
|
@staticmethod
|
|
async def scrape_item(
|
|
url: str,
|
|
selector: str | None = None,
|
|
item_id: int | None = None,
|
|
config: ScrapeConfig | None = None,
|
|
return_html: bool = False,
|
|
) -> tuple[str | None, str]:
|
|
"""
|
|
Scrapes the given URL using Browserless and Playwright.
|
|
Returns a tuple: (screenshot_path, page_text_or_html)
|
|
"""
|
|
if config is None:
|
|
config = ScrapeConfig()
|
|
|
|
# Input validation & defaults
|
|
scroll_pixels = max(350, config.scroll_pixels if config.scroll_pixels > 0 else 350)
|
|
timeout = max(30000, config.timeout if config.timeout > 0 else 90000)
|
|
|
|
# Ensure browser is connected and healthy
|
|
if not await ScraperService._ensure_browser_connected():
|
|
logger.error("Failed to establish browser connection")
|
|
return None, ""
|
|
|
|
try:
|
|
context = await ScraperService._create_context(ScraperService._browser)
|
|
page = await context.new_page()
|
|
|
|
try:
|
|
await ScraperService._navigate_and_wait(page, url, timeout)
|
|
await ScraperService._handle_popups(page)
|
|
|
|
if selector:
|
|
await ScraperService._wait_for_selector(page, selector)
|
|
else:
|
|
await ScraperService._auto_detect_price(page)
|
|
|
|
if config.smart_scroll:
|
|
await ScraperService._smart_scroll(page, scroll_pixels)
|
|
|
|
if return_html:
|
|
content_data = await page.content()
|
|
else:
|
|
content_data = await ScraperService._extract_text(page, config.text_length)
|
|
|
|
screenshot_path = await ScraperService._take_screenshot(page, url, item_id)
|
|
|
|
return screenshot_path, content_data
|
|
|
|
finally:
|
|
await context.close()
|
|
|
|
except Exception as e:
|
|
logger.error(f"Error scraping {url}: {e}")
|
|
return None, ""
|
|
|
|
@staticmethod
|
|
async def _connect_browser(p) -> Browser:
|
|
logger.info(f"Connecting to Browserless at {BROWSERLESS_URL}")
|
|
return await p.chromium.connect_over_cdp(BROWSERLESS_URL)
|
|
|
|
@staticmethod
|
|
async def _create_context(browser: Browser) -> BrowserContext:
|
|
"""Create context with stealth and locale (aligned with ImprovedSearchService)"""
|
|
context = await browser.new_context(
|
|
viewport={"width": 1920, "height": 1080},
|
|
user_agent=(
|
|
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
|
|
"AppleWebKit/537.36 (KHTML, like Gecko) "
|
|
"Chrome/131.0.0.0 Safari/537.36"
|
|
),
|
|
locale="fr-FR",
|
|
timezone_id="Europe/Paris",
|
|
)
|
|
|
|
# Stealth mode
|
|
await context.add_init_script("""
|
|
Object.defineProperty(navigator, 'webdriver', { get: () => undefined });
|
|
window.chrome = { runtime: {} };
|
|
""")
|
|
|
|
await context.route("**/*", lambda route: route.continue_())
|
|
return context
|
|
|
|
@staticmethod
|
|
async def _navigate_and_wait(page: Page, url: str, timeout: int):
|
|
logger.info(f"Navigating to {url} (Timeout: {timeout}ms)")
|
|
try:
|
|
await page.goto(url, wait_until="domcontentloaded", timeout=timeout)
|
|
logger.info(f"Page loaded (domcontentloaded): {url}")
|
|
|
|
try:
|
|
await page.wait_for_load_state("networkidle", timeout=5000)
|
|
logger.info("Network idle reached")
|
|
except PlaywrightTimeoutError:
|
|
logger.info("Network idle timed out (non-critical), proceeding...")
|
|
|
|
except Exception as e:
|
|
logger.warning(f"Navigation warning for {url}: {e}")
|
|
|
|
# Wait a bit for dynamic content
|
|
await page.wait_for_timeout(2000)
|
|
|
|
@staticmethod
|
|
async def _handle_popups(page: Page):
|
|
"""Aggressive multi-pass popup and overlay removal"""
|
|
logger.info("Starting aggressive popup removal...")
|
|
|
|
# 1. Multi-pass clicking (some popups appear after others are closed)
|
|
for i in range(2):
|
|
logger.debug(f"Popup removal pass {i+1}")
|
|
for selector in POPUP_SELECTORS:
|
|
try:
|
|
locators = page.locator(selector)
|
|
count = await locators.count()
|
|
if count > 0:
|
|
for j in range(count):
|
|
target = locators.nth(j)
|
|
if await target.is_visible():
|
|
logger.info(f"Closing popup: {selector}")
|
|
await target.click(timeout=1000)
|
|
await page.wait_for_timeout(500)
|
|
except Exception:
|
|
pass
|
|
|
|
# Hammer Escape key
|
|
try:
|
|
await page.keyboard.press("Escape")
|
|
await page.wait_for_timeout(200)
|
|
except: pass
|
|
|
|
# 2. Javascript cleanup (Hide pesky overlays and CMPs that won't close)
|
|
logger.info("Injecting CSS/JS cleanup for persistent overlays...")
|
|
await page.evaluate("""
|
|
() => {
|
|
const selectorsToHide = [
|
|
'#didomi-host', '.didomi-popup-container', '[id*="didomi"]',
|
|
'#onetrust-banner-sdk', '#onetrust-consent-sdk',
|
|
'.cookie-banner', '.cookie-consent', '.qc-cmp2-container',
|
|
'.modal-backdrop', '.modal-overlay', '.fade.show'
|
|
];
|
|
|
|
// Hide specific IDs/classes
|
|
selectorsToHide.forEach(s => {
|
|
const el = document.querySelector(s);
|
|
if (el) el.style.display = 'none';
|
|
});
|
|
|
|
// Clear overlays (fixed/absolute elements with high z-index that cover too much)
|
|
const all = document.getElementsByTagName("*");
|
|
for (let i = 0, max = all.length; i < max; i++) {
|
|
const el = all[i];
|
|
const style = window.getComputedStyle(el);
|
|
if (style.position === 'fixed' || style.position === 'absolute') {
|
|
const zIndex = parseInt(style.zIndex);
|
|
if (zIndex > 100) {
|
|
// Check if it's potentially a popup (covers a lot of area or has modal classes)
|
|
const rect = el.getBoundingClientRect();
|
|
if (rect.width > window.innerWidth * 0.5 && rect.height > window.innerHeight * 0.5) {
|
|
console.log('Hiding potential popup:', el);
|
|
el.style.setProperty('display', 'none', 'important');
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// Unlock scroll if it was locked by a modal
|
|
document.body.style.overflow = 'auto';
|
|
document.documentElement.style.overflow = 'auto';
|
|
}
|
|
""")
|
|
|
|
# Wait for any transitions
|
|
await page.wait_for_timeout(1000)
|
|
|
|
@staticmethod
|
|
async def _wait_for_selector(page: Page, selector: str):
|
|
try:
|
|
logger.info(f"Waiting for selector: {selector}")
|
|
await page.wait_for_selector(selector, timeout=5000)
|
|
element = page.locator(selector).first
|
|
await element.scroll_into_view_if_needed()
|
|
logger.info(f"Scrolled to selector: {selector}")
|
|
except Exception as e:
|
|
logger.warning(f"Selector {selector} not found or timed out: {e}")
|
|
|
|
@staticmethod
|
|
async def _auto_detect_price(page: Page):
|
|
logger.info("No selector provided. Attempting to find price element...")
|
|
try:
|
|
price_locator = page.locator("text=/$[0-9,]+(\\.[0-9]{2})?/")
|
|
if await price_locator.count() > 0:
|
|
await price_locator.first.scroll_into_view_if_needed()
|
|
logger.info("Scrolled to potential price element")
|
|
except Exception as e:
|
|
logger.warning(f"Auto-price detection failed: {e}")
|
|
|
|
@staticmethod
|
|
async def _smart_scroll(page: Page, scroll_pixels: int):
|
|
logger.info(f"Performing smart scroll ({scroll_pixels}px)...")
|
|
try:
|
|
await page.evaluate(f"window.scrollBy(0, {scroll_pixels})")
|
|
await page.wait_for_timeout(1000)
|
|
except Exception as e:
|
|
logger.warning(f"Smart scroll failed: {e}")
|
|
|
|
@staticmethod
|
|
async def _extract_text(page: Page, text_length: int) -> str:
|
|
if text_length <= 0:
|
|
return ""
|
|
|
|
try:
|
|
logger.info(f"Extracting text (limit: {text_length} chars)...")
|
|
raw_text = await page.inner_text("body")
|
|
page_text = raw_text[:text_length]
|
|
logger.info(f"Extracted {len(page_text)} characters")
|
|
return page_text
|
|
except Exception as e:
|
|
logger.error(f"Text extraction failed: {e}")
|
|
return ""
|
|
|
|
@staticmethod
|
|
async def _take_screenshot(page: Page, url: str, item_id: int | None) -> str:
|
|
screenshot_dir = "screenshots"
|
|
os.makedirs(screenshot_dir, exist_ok=True)
|
|
|
|
if item_id:
|
|
# Use timestamp to ensure unique filenames (fixes caching issues)
|
|
timestamp = int(datetime.now().timestamp())
|
|
filename = f"{screenshot_dir}/item_{item_id}_{timestamp}.png"
|
|
else:
|
|
url_part = url.split("//")[-1].replace("/", "_")
|
|
timestamp = datetime.now().timestamp()
|
|
filename = f"{screenshot_dir}/{url_part}_{timestamp}.png"
|
|
|
|
await page.screenshot(path=filename, full_page=False)
|
|
logger.info(f"Screenshot saved to {filename}")
|
|
return filename
|