add helper scripts

dubizzle_scraper/browser/__init__.py

@@ -0,0 +1,6 @@
+from .factory import BrowserFactory
+from .listing import ListingCollector
+from .network import NetworkCapture
+from .pace import HumanPacer
+
+__all__ = ["BrowserFactory", "ListingCollector", "NetworkCapture", "HumanPacer"]

dubizzle_scraper/browser/factory.py

@@ -0,0 +1,214 @@
+import json
+import logging
+import random
+
+from playwright.sync_api import Browser, BrowserContext, Playwright
+
+try:
+    from playwright_stealth import stealth_sync
+except ImportError:
+    stealth_sync = None
+
+from ..core.config import Settings
+
+logger = logging.getLogger("dubizzle_scraper.browser")
+
+
+def _build_init_script() -> str:
+    # Маскировка браузера.
+    hardware_concurrency = random.choice([4, 8, 12, 16])
+    device_memory = random.choice([4, 8, 16])
+    languages = ["en-US", "en"]
+
+    return f"""
+(() => {{
+  const define = (obj, prop, value) => {{
+    try {{
+      Object.defineProperty(obj, prop, {{ get: () => value, configurable: true }});
+    }} catch (e) {{}}
+  }};
+
+  define(navigator, 'webdriver', undefined);
+  define(navigator, 'platform', 'Win32');
+  define(navigator, 'vendor', 'Google Inc.');
+  define(navigator, 'language', '{languages[0]}');
+  define(navigator, 'languages', {json.dumps(languages)});
+  define(navigator, 'hardwareConcurrency', {hardware_concurrency});
+  define(navigator, 'deviceMemory', {device_memory});
+  define(navigator, 'maxTouchPoints', 0);
+
+  if (!window.chrome) {{
+    Object.defineProperty(window, 'chrome', {{
+      value: {{ runtime: {{}}, app: {{}}, csi: () => ({{}}), loadTimes: () => ({{}}) }},
+      configurable: true
+    }});
+  }}
+
+  const originalQuery = navigator.permissions && navigator.permissions.query;
+  if (originalQuery) {{
+    navigator.permissions.query = (params) => (
+      params && params.name === 'notifications'
+        ? Promise.resolve({{ state: Notification.permission }})
+        : originalQuery(params)
+    );
+  }}
+
+  const originalGetParameter = WebGLRenderingContext.prototype.getParameter;
+  WebGLRenderingContext.prototype.getParameter = function(parameter) {{
+    if (parameter === 37445) return 'Intel Inc.';
+    if (parameter === 37446) return 'Intel Iris OpenGL Engine';
+    return originalGetParameter.call(this, parameter);
+  }};
+}})();
+"""
+
+
+class BrowserFactory:
+
+    def __init__(self, settings: Settings) -> None:
+        self.settings = settings
+
+    def _resolve_engine(self) -> str:
+        # Выбор движка.
+        engine = self.settings.browser_engine.strip().lower()
+        if engine == "auto":
+            # Для DUBIZZLE стабильнее Chromium.
+            return "chromium"
+        if engine in ("firefox", "chromium"):
+            return engine
+        logger.warning("Unknown DUBIZZLE_BROWSER_ENGINE=%r, falling back to auto", engine)
+        return "chromium"
+
+    def create_browser(self, playwright: Playwright) -> Browser:
+        engine = self._resolve_engine()
+        proxy_dict = self.settings.proxy.to_playwright_dict()
+        logger.info("Resolved browser engine: requested=%s resolved=%s", self.settings.browser_engine, engine)
+
+        if engine == "firefox":
+            launch_kwargs: dict = {"headless": self.settings.headless}
+            if proxy_dict:
+                launch_kwargs["proxy"] = proxy_dict
+                logger.info("Using proxy: %s", self.settings.proxy.server)
+            # Настройки Firefox.
+            launch_kwargs["firefox_user_prefs"] = {
+                "dom.webdriver.enabled": False,
+                "useAutomationExtension": False,
+                # Базовые оптимизации.
+                "media.autoplay.default": 5,
+                "media.volume_scale": "0.0",
+                "media.audio.playback.standalone": False,
+                "dom.ipc.processCount": 1,
+                "dom.ipc.plugins.enabled": False,
+                "browser.cache.disk.enable": False,
+                "browser.cache.memory.enable": True,
+                "browser.cache.memory.max_entry_size": 8192,
+                "network.prefetch-next": False,
+                "network.dns.disablePrefetch": True,
+                "permissions.default.image": 2,
+                "javascript.options.mem.gc_incremental_mark_slice_ms": 20,
+            }
+            logger.info("Launching Firefox (headless=%s)", self.settings.headless)
+            return playwright.firefox.launch(**launch_kwargs)
+
+        # Запуск Chromium.
+        args = [
+            "--disable-blink-features=AutomationControlled",
+            "--no-default-browser-check",
+            "--disable-dev-shm-usage",
+            "--disable-features=IsolateOrigins,site-per-process",
+        ]
+        if self.settings.headless:
+            args.append("--headless=new")
+            pw_headless = False
+            logger.info("Using Chromium new-headless mode (--headless=new)")
+        else:
+            pw_headless = False
+
+        launch_kwargs = {"headless": pw_headless, "args": args}
+        if proxy_dict:
+            launch_kwargs["proxy"] = proxy_dict
+            logger.info("Using proxy: %s", self.settings.proxy.server)
+        try:
+            logger.info("Trying to launch real Chrome channel")
+            return playwright.chromium.launch(channel="chrome", **launch_kwargs)
+        except Exception:
+            logger.warning("Chrome channel launch failed, falling back to Chromium")
+            return playwright.chromium.launch(**launch_kwargs)
+
+    def create_context(self, browser: Browser) -> BrowserContext:
+        viewport = random.choice(self.settings.fingerprint.viewport_presets)
+        timezone_id = random.choice(self.settings.fingerprint.timezone_candidates)
+        color_scheme = random.choice(["light", "dark"])
+
+        is_firefox = browser.browser_type.name == "firefox"
+
+        ctx_kwargs: dict = {
+            "viewport": viewport,
+            "screen": viewport,
+            "locale": self.settings.fingerprint.locale,
+            "timezone_id": timezone_id,
+            "color_scheme": color_scheme,
+            "java_script_enabled": True,
+            "ignore_https_errors": False,
+        }
+
+        if is_firefox:
+            # Заголовки Firefox.
+            ctx_kwargs["user_agent"] = (
+                "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:128.0) "
+                "Gecko/20100101 Firefox/128.0"
+            )
+            ctx_kwargs["extra_http_headers"] = {
+                "Accept-Language": "en-US,en;q=0.5",
+                "DNT": "1",
+                "Upgrade-Insecure-Requests": "1",
+            }
+        else:
+            ctx_kwargs["user_agent"] = self.settings.fingerprint.user_agent
+            ctx_kwargs["device_scale_factor"] = random.choice([1, 1.25])
+            ctx_kwargs["is_mobile"] = False
+            ctx_kwargs["has_touch"] = False
+            ctx_kwargs["extra_http_headers"] = {
+                "Accept-Language": "en-US,en;q=0.9",
+                "DNT": "1",
+                "Upgrade-Insecure-Requests": "1",
+                "Sec-CH-UA": self.settings.fingerprint.sec_ch_ua,
+                "Sec-CH-UA-Mobile": "?0",
+                "Sec-CH-UA-Platform": '"Windows"',
+            }
+
+        context = browser.new_context(**ctx_kwargs)
+        context.set_default_timeout(self.settings.default_timeout_ms)
+        context.set_default_navigation_timeout(self.settings.default_timeout_ms)
+
+        if not is_firefox:
+            # Маскировка Chromium.
+            context.add_init_script(_build_init_script())
+            if stealth_sync:
+                context.on("page", lambda page: stealth_sync(page))
+                logger.debug("playwright-stealth attached to context")
+
+        return context
+
+    @staticmethod
+    def enable_resource_blocking(page) -> None:
+        # Блокируем тяжёлые ресурсы.
+        BLOCKED_TYPES = {"image", "stylesheet", "font", "media"}
+        BLOCKED_URL_PATTERNS = (
+            "google-analytics", "googletagmanager", "facebook.net",
+            "doubleclick.net", "hotjar", "newrelic", ".woff", ".woff2",
+            "analytics", "tracking", "adservice",
+        )
+
+        def _handle_route(route):
+            req = route.request
+            if req.resource_type in BLOCKED_TYPES:
+                route.abort()
+                return
+            url = req.url.lower()
+            if any(pat in url for pat in BLOCKED_URL_PATTERNS):
+                route.abort()
+                return
+            route.continue_()
+
+        page.route("**/*", _handle_route)

dubizzle_scraper/browser/listing.py

@@ -0,0 +1,725 @@
+import logging
+import re
+import time
+from dataclasses import asdict, dataclass, field
+from typing import Any
+from urllib.parse import urljoin
+
+from playwright.sync_api import Page
+
+from .pace import HumanPacer
+from ..core.config import Settings
+from ..core.utils import first_non_empty
+
+logger = logging.getLogger("dubizzle_scraper.listing")
+VEHICLE_HREF_RE = re.compile(
+	r'(?:/VehicleDetail/(?P<veh_id>\d+)(?:~[A-Z]{2})?)|(?:/motors/used-cars/[^"\s]+?(?:/ad-(?P<ad_id>\d+)/?|---(?P<slug_id>[a-f0-9]{32})/?))|(?:/s/(?P<short_id>[A-Za-z0-9]+))',
+	re.IGNORECASE,
+)
+VEHICLE_LINK_SELECTOR = (
+	"a[href*='/VehicleDetail/'], a[href*='/vehicledetail/'], a[href*='VehicleDetail'], a[href*='vehicledetail'], "
+	"a[href*='/motors/used-cars/'], a[href*='/s/']"
+)
+COOKIE_ACCEPT_SELECTORS: tuple[str, ...] = (
+	"button:has-text('Accept All')",
+	"button:has-text('Accept all')",
+	"button:has-text('I Agree')",
+	"button:has-text('Agree')",
+	"button:has-text('Only necessary')",
+	"button:has-text('Только необходимые')",
+	"button:has-text('Принять все')",
+	"[id*='accept']",
+	"[class*='accept']",
+)
+
+
+@dataclass(slots=True)
+class ListingVehicleLink:
+	href: str
+	title: str = ""
+	lot_number: str | None = None
+
+
+@dataclass(slots=True)
+class ListingPageResult:
+	source_url: str
+	page_number: int
+	vehicle_links: list[ListingVehicleLink] = field(default_factory=list)
+	pagination_available: bool = False
+	next_page_detected: bool = False
+
+
+class ListingCollector:
+	_NEXT_PAGE_SELECTORS: tuple[str, ...] = (
+		"a[aria-label*='Next']",
+		"button[aria-label*='Next']",
+		"a[aria-label*='next']",
+		"button[aria-label*='next']",
+		"a[title*='Next']",
+		"button[title*='Next']",
+		"a[title*='next']",
+		"button[title*='next']",
+		"a[rel='next']",
+		"link[rel='next']",
+		"a.pagination-next",
+		"button.pagination-next",
+		"a.next",
+		"button.next",
+		"a:has-text('Next')",
+		"button:has-text('Next')",
+		"a:has-text('NEXT')",
+		"button:has-text('NEXT')",
+		"a:has-text('›')",
+		"button:has-text('›')",
+		"a:has-text('»')",
+		"button:has-text('»')",
+		"a:has(img[src*='icon-arrow-right'])",
+		"button:has(img[src*='icon-arrow-right'])",
+		"a:has(img[src*='arrow-right'])",
+		"button:has(img[src*='arrow-right'])",
+	)
+
+	def __init__(self, settings: Settings, pacer: HumanPacer) -> None:
+		self.settings = settings
+		self.pacer = pacer
+
+	@staticmethod
+	def _get_current_page_number(page: Page) -> int | None:
+		try:
+			value = page.evaluate(
+				"""
+				() => {
+					const controls = Array.from(document.querySelectorAll('a,button,[role="button"],span,div'));
+					const current = controls.find((el) => {
+						const text = (el.textContent || '').trim();
+						const cls = (el.getAttribute('class') || '').toLowerCase();
+						const ariaCurrent = (el.getAttribute('aria-current') || '').toLowerCase();
+						return /^\d+$/.test(text) && (ariaCurrent === 'page' || cls.includes('active') || cls.includes('current') || cls.includes('selected'));
+					});
+					if (current) {
+						return parseInt((current.textContent || '').trim(), 10);
+					}
+					const match = (document.body?.innerText || '').match(/\b(\d+)\s+of\s+\d+\+?/i);
+					return match ? parseInt(match[1], 10) : null;
+				}
+				"""
+			)
+			return int(value) if value is not None else None
+		except Exception:
+			return None
+
+	@staticmethod
+	def _wait_for_navigation_result(page: Page, old_first_href: str, expected_page_number: int | None) -> bool:
+		if old_first_href:
+			try:
+				page.wait_for_function(
+					f"""() => {{
+						const a = document.querySelector(\"{VEHICLE_LINK_SELECTOR}\");
+						return a && a.getAttribute('href') !== '{old_first_href}';
+					}}""",
+					timeout=12000,
+				)
+				return True
+			except Exception:
+				pass
+
+		if expected_page_number is not None:
+			current_page = ListingCollector._get_current_page_number(page)
+			if current_page == expected_page_number:
+				return True
+
+		try:
+			page.wait_for_selector(VEHICLE_LINK_SELECTOR, timeout=3000)
+		except Exception:
+			pass
+
+		current_page = ListingCollector._get_current_page_number(page)
+		if expected_page_number is not None and current_page == expected_page_number:
+			return True
+
+		return not old_first_href
+
+	@staticmethod
+	def has_page_number(page: Page, target_page_number: int) -> bool:
+		try:
+			return bool(page.evaluate(
+				"""
+				(targetPageNumber) => {
+					const visible = (el) => !!(el && (el.offsetWidth || el.offsetHeight || el.getClientRects().length));
+					const controls = Array.from(document.querySelectorAll('a,button,[role="button"],span,div'));
+					return controls.some((el) => {
+						const text = (el.textContent || '').trim();
+						const cls = (el.getAttribute('class') || '').toLowerCase();
+						const ariaDisabled = (el.getAttribute('aria-disabled') || '').toLowerCase();
+						const disabled = el.hasAttribute('disabled') || ariaDisabled === 'true' || cls.includes('disabled');
+						return visible(el) && !disabled && /^\d+$/.test(text) && parseInt(text, 10) === targetPageNumber;
+					});
+				}
+				""",
+				target_page_number,
+			))
+		except Exception:
+			return False
+
+	def open_cars_listing(self, page: Page, *, url_override: str | None = None) -> None:
+		url = url_override or self.settings.listing.cars_url
+		logger.info("Opening cars listing page: %s", url)
+		last_err = None
+		for attempt in range(3):
+			try:
+				page.goto(url, wait_until="commit", timeout=60_000)
+				last_err = None
+				break
+			except Exception as e:
+				last_err = e
+				logger.warning("goto listing attempt %d failed: %s", attempt + 1, e)
+				# Небольшой backoff при ошибках открытия листинга.
+				time.sleep(5 * (attempt + 1))
+		if last_err:
+			logger.warning("All goto attempts failed, trying JS navigation")
+			try:
+				page.evaluate(f"window.location.href = '{url}'")
+			except Exception:
+				pass
+		# Быстрая проверка готовности страницы.
+		try:
+			page.wait_for_load_state("domcontentloaded", timeout=12_000)
+		except Exception:
+			pass
+		self._accept_cookie_banner(page)
+		self._wait_for_listing_content(page)
+		logger.info("Listing page URL: %s", page.url)
+		self.pacer.after_listing_open()
+
+	def _accept_cookie_banner(self, page: Page) -> None:
+		for selector in COOKIE_ACCEPT_SELECTORS:
+			locator = page.locator(selector).first
+			try:
+				if locator.count() == 0:
+					continue
+				if not locator.is_visible(timeout=500):
+					continue
+				locator.click(timeout=2_000)
+				logger.info("Accepted cookie banner using selector: %s", selector)
+				try:
+					page.wait_for_load_state("domcontentloaded", timeout=3_000)
+				except Exception:
+					pass
+				return
+			except Exception:
+				continue
+
+	def _wait_for_listing_content(self, page: Page) -> None:
+		try:
+			page.wait_for_selector(VEHICLE_LINK_SELECTOR, timeout=12_000)
+			return
+		except Exception:
+			pass
+
+		# Fallback: React/SSR разметка может появиться не сразу, даже если <a> ещё нет в DOM.
+		try:
+			page.wait_for_function(
+				"""() => {
+					const html = document.documentElement?.innerHTML || '';
+					const text = document.body?.innerText || '';
+					return html.includes('/VehicleDetail/') || /\\b\d+\s+VEHICLES\b/i.test(text);
+				}""",
+				timeout=12_000,
+			)
+		except Exception:
+			# Короткая пауза вместо длинного sleep.
+			time.sleep(1.0)
+
+	def apply_filters(
+		self,
+		page: Page,
+		make: str | None = None,
+		model: str | None = None,
+		year_min: int | None = None,
+		year_max: int | None = None,
+	) -> dict[str, str | int | None]:
+		applied: dict[str, str | int | None] = {"make": None, "model": None, "year_min": None, "year_max": None}
+		if make and self._try_fill_filter_input(page, ["input[placeholder*='Make']", "input[aria-label*='Make']"], make):
+			applied["make"] = make
+			self.pacer.after_filter_action()
+		if model and self._try_fill_filter_input(page, ["input[placeholder*='Model']", "input[aria-label*='Model']"], model):
+			applied["model"] = model
+			self.pacer.after_filter_action()
+		if year_min is not None or year_max is not None:
+			if self._apply_year_range(page, year_min, year_max):
+				applied["year_min"] = year_min
+				applied["year_max"] = year_max
+				self.pacer.after_filter_action()
+		return applied
+
+	def _apply_year_range(self, page: Page, year_min: int | None, year_max: int | None) -> bool:
+		"""Заполняет поля фильтра Year и нажимает Apply Year."""
+		if year_min is None and year_max is None:
+			return False
+		try:
+			success = page.evaluate(
+				"""([yearMin, yearMax]) => {
+					const inputs = Array.from(document.querySelectorAll('input'));
+					const yearInputs = inputs.filter(inp => {
+						const v = parseInt(inp.value, 10);
+						return !isNaN(v) && v >= 1900 && v <= 2100;
+					});
+					if (yearInputs.length < 2) return false;
+					yearInputs.sort((a, b) => parseInt(a.value) - parseInt(b.value));
+					const setVal = (el, val) => {
+						const setter = Object.getOwnPropertyDescriptor(
+							HTMLInputElement.prototype, 'value'
+						).set;
+						setter.call(el, String(val));
+						el.dispatchEvent(new Event('input', {bubbles: true}));
+						el.dispatchEvent(new Event('change', {bubbles: true}));
+					};
+					if (yearMin !== null) setVal(yearInputs[0], yearMin);
+					if (yearMax !== null) setVal(yearInputs[yearInputs.length - 1], yearMax);
+					const container = yearInputs[0].closest(
+						'[class*="filter"], [class*="year"], section, fieldset'
+					) || yearInputs[0].parentElement.parentElement;
+					if (container) {
+						const btn = Array.from(container.querySelectorAll(
+							'button, a, [role="button"], span[class*="apply"]'
+						)).find(el => /apply|\u043f\u0440\u0438\u043c\u0435\u043d/i.test(el.textContent));
+						if (btn) { btn.click(); return true; }
+					}
+					yearInputs[yearInputs.length - 1].dispatchEvent(
+						new KeyboardEvent('keydown', {
+							key: 'Enter', code: 'Enter', keyCode: 13, bubbles: true
+						})
+					);
+					return true;
+				}""",
+				[year_min, year_max],
+			)
+			if success:
+				try:
+					page.wait_for_load_state("domcontentloaded", timeout=15_000)
+				except Exception:
+					pass
+				self._wait_for_listing_content(page)
+				logger.info("Applied year range filter: %s — %s", year_min, year_max)
+				return True
+		except Exception as exc:
+			logger.warning("Failed to apply year range filter: %s", exc)
+		return False
+
+	def collect_current_page(self, page: Page, page_number: int = 1) -> ListingPageResult:
+		# Считываем ссылки одним проходом по DOM.
+		self._accept_cookie_banner(page)
+		self._wait_for_listing_content(page)
+		try:
+			raw_items = page.eval_on_selector_all(
+				"a[href], [data-href], [href]",
+				"""
+				(nodes) => nodes.map((node) => ({
+					href:
+						node.getAttribute('href') ||
+						node.getAttribute('data-href') ||
+						node.getAttribute('data-url') ||
+						'',
+					title: node.getAttribute('title') || node.getAttribute('aria-label') || '',
+					text: (node.textContent || '').trim(),
+				}))
+				""",
+			)
+		except Exception as exc:
+			logger.warning("collect_current_page failed on page %d: %s", page_number, exc)
+			raw_items = []
+		total = min(len(raw_items), self.settings.listing.page_link_limit)
+		links: list[ListingVehicleLink] = []
+		seen: set[str] = set()
+		for idx in range(total):
+			item = raw_items[idx] if isinstance(raw_items[idx], dict) else {}
+			href = str(item.get("href") or "")
+			match = VEHICLE_HREF_RE.search(href)
+			if not match:
+				continue
+			lot_number = match.group("veh_id") or match.group("ad_id") or match.group("slug_id") or None
+			if lot_number is None:
+				short = match.group("short_id")
+				lot_number = short if short else None
+			absolute = urljoin(self.settings.home_url, match.group(0))
+			if absolute in seen:
+				continue
+			seen.add(absolute)
+			title = first_non_empty([item.get("title"), item.get("text"), ""]) or ""
+			links.append(ListingVehicleLink(href=absolute, title=str(title).strip(), lot_number=lot_number))
+			if len(links) >= self.settings.listing.max_vehicles_per_run:
+				break
+
+		# Fallback: на DUBIZZLE ссылки иногда не рендерятся как <a>,
+		# но присутствуют в hydration/inline JSON внутри HTML (часто как \/VehicleDetail\/").
+		if not links:
+			try:
+				page.wait_for_timeout(1_500)
+			except Exception:
+				pass
+			try:
+				html = page.content()
+			except Exception as exc:
+				logger.debug("page.content() failed on page %d: %s", page_number, exc)
+				html = ""
+
+			for absolute, lot_number in self._extract_vehicle_links_from_html(html):
+				if absolute in seen:
+					continue
+				seen.add(absolute)
+				links.append(ListingVehicleLink(href=absolute, title="", lot_number=lot_number))
+				if len(links) >= self.settings.listing.max_vehicles_per_run:
+					break
+
+			if links:
+				logger.info(
+					"Page %d: recovered %d vehicle links from HTML fallback",
+					page_number,
+					len(links),
+				)
+		next_page_detected = self._has_next_page(page)
+		return ListingPageResult(source_url=page.url, page_number=page_number, vehicle_links=links, pagination_available=next_page_detected, next_page_detected=next_page_detected)
+
+	def _extract_vehicle_links_from_html(self, html: str) -> list[tuple[str, str]]:
+		if not html:
+			return []
+
+		# Частый формат в JSON внутри HTML: "\/VehicleDetail\/12345678~US"
+		normalized = html.replace("\\/", "/")
+		found: list[tuple[str, str]] = []
+		seen: set[str] = set()
+
+		for match in VEHICLE_HREF_RE.finditer(normalized):
+			lot_number = match.group("veh_id") or match.group("ad_id") or match.group("slug_id") or match.group("short_id") or ""
+			absolute = urljoin(self.settings.home_url, match.group(0))
+			if absolute in seen:
+				continue
+			seen.add(absolute)
+			found.append((absolute, lot_number))
+			if len(found) >= self.settings.listing.page_link_limit:
+				break
+
+		return found
+
+	def go_to_next_page(self, page: Page, expected_page_number: int | None = None) -> bool:
+		# Запоминаем первую ссылку текущей страницы для определения смены контента.
+		old_first_href = ""
+		try:
+			first_link = page.locator(VEHICLE_LINK_SELECTOR).first
+			if first_link.count() > 0:
+				old_first_href = first_link.get_attribute("href") or ""
+		except Exception:
+			pass
+
+		for selector in self._NEXT_PAGE_SELECTORS:
+			locator = page.locator(selector).first
+			if locator.count() == 0:
+				continue
+			try:
+				disabled = (locator.get_attribute("disabled", timeout=1500) or "").lower()
+				aria_disabled = (locator.get_attribute("aria-disabled", timeout=1500) or "").lower()
+				classes = (locator.get_attribute("class", timeout=1500) or "").lower()
+			except Exception:
+				continue
+			if disabled or aria_disabled == "true" or "disabled" in classes:
+				continue
+			try:
+				self.pacer.move_mouse_to(page, locator)
+				locator.click(timeout=8000)
+			except Exception:
+				continue
+
+			if self._wait_for_navigation_result(page, old_first_href, expected_page_number):
+				self.pacer.after_page_change()
+				return True
+
+		# Fallback для DUBIZZLE: пагинация часто рендерится как набор номеров страниц
+		# + стрелка с иконкой, без явного текста Next.
+		try:
+			clicked = bool(page.evaluate(
+				"""
+				() => {
+					const visible = (el) => !!(el && (el.offsetWidth || el.offsetHeight || el.getClientRects().length));
+					const disabled = (el) => {
+						if (!el) return true;
+						const cls = (el.getAttribute('class') || '').toLowerCase();
+						const ariaDisabled = (el.getAttribute('aria-disabled') || '').toLowerCase();
+						return el.hasAttribute('disabled') || ariaDisabled === 'true' || cls.includes('disabled');
+					};
+
+					const controls = Array.from(document.querySelectorAll('a,button,[role="button"]'))
+						.filter((el) => visible(el) && !disabled(el));
+
+					const current = controls.find((el) => {
+						const text = (el.textContent || '').trim();
+						const cls = (el.getAttribute('class') || '').toLowerCase();
+						const ariaCurrent = (el.getAttribute('aria-current') || '').toLowerCase();
+						return /^\d+$/.test(text) && (ariaCurrent === 'page' || cls.includes('active') || cls.includes('current') || cls.includes('selected'));
+					});
+
+					if (current) {
+						const currentPage = parseInt((current.textContent || '').trim(), 10);
+						const nextNumber = controls.find((el) => {
+							const text = (el.textContent || '').trim();
+							return /^\d+$/.test(text) && parseInt(text, 10) === currentPage + 1;
+						});
+						if (nextNumber) {
+							nextNumber.click();
+							return true;
+						}
+					}
+
+					const iconNext = controls.find((el) => {
+						const text = (el.textContent || '').trim().toLowerCase();
+						const aria = (el.getAttribute('aria-label') || '').trim().toLowerCase();
+						const title = (el.getAttribute('title') || '').trim().toLowerCase();
+						const rel = (el.getAttribute('rel') || '').trim().toLowerCase();
+						const cls = (el.getAttribute('class') || '').trim().toLowerCase();
+						const hasRightArrowIcon = !!el.querySelector('img[src*="icon-arrow-right"], img[src*="arrow-right"]');
+						return hasRightArrowIcon || rel === 'next' || aria.includes('next') || title.includes('next') || cls.includes('next') || ['next', '›', '»', '>'].includes(text);
+					});
+
+					if (iconNext) {
+						iconNext.click();
+						return true;
+					}
+
+					return false;
+				}
+				"""
+			))
+			if clicked:
+				if self._wait_for_navigation_result(page, old_first_href, expected_page_number):
+					self.pacer.after_page_change()
+					return True
+		except Exception as exc:
+			logger.debug("Numeric/icon pagination fallback failed: %s", exc)
+
+		# JS fallback: ищем любой видимый pagination-control «next» по атрибутам/тексту.
+		try:
+			clicked = bool(page.evaluate(
+				"""
+				() => {
+					const candidates = Array.from(document.querySelectorAll('a,button,[role="button"]'));
+					for (const el of candidates) {
+						const text = (el.textContent || '').trim().toLowerCase();
+						const aria = (el.getAttribute('aria-label') || '').trim().toLowerCase();
+						const title = (el.getAttribute('title') || '').trim().toLowerCase();
+						const rel = (el.getAttribute('rel') || '').trim().toLowerCase();
+						const cls = (el.getAttribute('class') || '').trim().toLowerCase();
+						const disabled = el.hasAttribute('disabled') || el.getAttribute('aria-disabled') === 'true' || cls.includes('disabled');
+						const visible = !!(el.offsetWidth || el.offsetHeight || el.getClientRects().length);
+						const looksNext = rel === 'next' || aria.includes('next') || title.includes('next') || cls.includes('next') || ['next', '›', '»', '>'].includes(text);
+						if (!disabled && visible && looksNext) {
+							el.click();
+							return true;
+						}
+					}
+					return false;
+				}
+				"""
+			))
+			if clicked:
+				if self._wait_for_navigation_result(page, old_first_href, expected_page_number):
+					self.pacer.after_page_change()
+					return True
+		except Exception as exc:
+			logger.debug("JS next-page fallback failed: %s", exc)
+
+		logger.warning("Could not navigate to next page from %s", page.url)
+		return False
+
+	def collect_listing_links(
+		self,
+		page: Page,
+		*,
+		make: str | None = None,
+		model: str | None = None,
+		known_origin_ids: set[str] | None = None,
+		max_duration_seconds: float | None = None,
+	) -> dict[str, Any]:
+		self.open_cars_listing(page)
+		applied_filters = self.apply_filters(page, make=make, model=model)
+		started_at = time.perf_counter()
+		truncated_by_time_budget = False
+		pages: list[dict[str, object]] = []
+		all_links: list[str] = []
+		early_stopped = False
+		threshold = self.settings.listing.early_stop_threshold
+
+		for page_number in range(1, max(1, self.settings.listing.max_pages_per_run) + 1):
+			if max_duration_seconds is not None and max_duration_seconds > 0:
+				elapsed = time.perf_counter() - started_at
+				if elapsed >= max_duration_seconds:
+					truncated_by_time_budget = True
+					logger.warning(
+						"Listing collection stopped by time budget: page=%d elapsed=%.1fs budget=%.1fs",
+						page_number,
+						elapsed,
+						max_duration_seconds,
+					)
+					break
+			page_result = self.collect_current_page(page, page_number=page_number)
+			pages.append({
+				"page_number": page_result.page_number,
+				"source_url": page_result.source_url,
+				"links_found": len(page_result.vehicle_links),
+				"vehicle_links": [asdict(item) for item in page_result.vehicle_links],
+				"next_page_detected": page_result.next_page_detected,
+			})
+			for item in page_result.vehicle_links:
+				if item.href not in all_links:
+					all_links.append(item.href)
+				if len(all_links) >= self.settings.listing.max_vehicles_per_run:
+					break
+
+			# Ранний останов: если на этой странице много известных И нет новых — дальше нет смысла.
+			# Важно: если есть хоть одна новая машина — продолжаем листать (новые могут быть на любой странице).
+			if (
+				known_origin_ids is not None
+				and threshold > 0.0
+				and page_result.vehicle_links
+			):
+				page_known = sum(
+					1 for item in page_result.vehicle_links
+					if item.lot_number and f"dubizzle:{item.lot_number}" in known_origin_ids
+				)
+				page_new = len(page_result.vehicle_links) - page_known
+				ratio = page_known / len(page_result.vehicle_links)
+				# Останавливаемся только если нет новых И порог превышен
+				if ratio >= threshold and page_new == 0:
+					logger.info(
+						"Early stop on page %d: %.0f%% known (%d/%d), 0 new >= threshold %.0f%%",
+						page_number, ratio * 100, page_known,
+						len(page_result.vehicle_links), threshold * 100,
+					)
+					early_stopped = True
+					break
+				elif page_new > 0 and ratio >= threshold:
+					logger.info(
+						"Page %d: %.0f%% known but %d new found — продолжаем",
+						page_number, ratio * 100, page_new,
+					)
+
+			if (
+				len(all_links) >= self.settings.listing.max_vehicles_per_run
+				or self.settings.listing.collect_current_page_only
+				or not self.settings.listing.include_pagination
+				or not page_result.next_page_detected
+			):
+				break
+			if not self.go_to_next_page(page):
+				break
+
+		return {
+			"listing_url": self.settings.listing.cars_url,
+			"applied_filters": applied_filters,
+			"pages_collected": len(pages),
+			"vehicles_collected": len(all_links),
+			"vehicle_urls": all_links,
+			"early_stopped": early_stopped,
+			"truncated_by_time_budget": truncated_by_time_budget,
+			"pages": pages,
+			"strategy": {
+				"sequential": True,
+				"collect_current_page_only": self.settings.listing.collect_current_page_only,
+				"include_pagination": self.settings.listing.include_pagination,
+				"max_pages_per_run": self.settings.listing.max_pages_per_run,
+				"max_vehicles_per_run": self.settings.listing.max_vehicles_per_run,
+				"early_stop_threshold": threshold,
+			},
+		}
+
+	@staticmethod
+	def _try_fill_filter_input(page: Page, selectors: list[str], value: str) -> bool:
+		for selector in selectors:
+			locator = page.locator(selector).first
+			if locator.count() == 0:
+				continue
+			try:
+				locator.click()
+				locator.fill(value)
+				page.keyboard.press("Enter")
+				try:
+					page.wait_for_load_state("domcontentloaded", timeout=15000)
+				except Exception:
+					pass
+				try:
+					page.wait_for_selector(VEHICLE_LINK_SELECTOR, timeout=3000)
+				except Exception:
+					pass
+				return True
+			except Exception:
+				continue
+		return False
+
+	@staticmethod
+	def _has_next_page(page: Page) -> bool:
+		for selector in ListingCollector._NEXT_PAGE_SELECTORS:
+			locator = page.locator(selector)
+			count = locator.count()
+			if count == 0:
+				continue
+			# Тестовые/fake локаторы могут не поддерживать nth/get_attribute.
+			# В таком случае считаем наличие селектора достаточным признаком next.
+			if not hasattr(locator, "nth"):
+				return True
+			# Проверяем, что хотя бы один элемент не disabled.
+			# Disabled "Next" на последней странице не означает наличия следующей.
+			for i in range(min(count, 3)):
+				try:
+					el = locator.nth(i)
+					disabled_attr = el.get_attribute("disabled", timeout=300)
+					aria_disabled = el.get_attribute("aria-disabled", timeout=300)
+					cls = (el.get_attribute("class", timeout=300) or "").lower()
+					if disabled_attr is None and aria_disabled != "true" and "disabled" not in cls:
+						return True
+				except Exception:
+					continue
+		try:
+			return bool(page.evaluate(
+				"""
+				() => {
+					const visible = (el) => !!(el && (el.offsetWidth || el.offsetHeight || el.getClientRects().length));
+					const controls = Array.from(document.querySelectorAll('a,button,[role="button"]')).filter((el) => {
+						const cls = (el.getAttribute('class') || '').trim().toLowerCase();
+						const disabled = el.hasAttribute('disabled') || el.getAttribute('aria-disabled') === 'true' || cls.includes('disabled');
+						return !disabled && visible(el);
+					});
+
+					const hasExplicitNext = controls.some((el) => {
+						const text = (el.textContent || '').trim().toLowerCase();
+						const aria = (el.getAttribute('aria-label') || '').trim().toLowerCase();
+						const title = (el.getAttribute('title') || '').trim().toLowerCase();
+						const rel = (el.getAttribute('rel') || '').trim().toLowerCase();
+						const cls = (el.getAttribute('class') || '').trim().toLowerCase();
+						const hasRightArrowIcon = !!el.querySelector('img[src*="icon-arrow-right"], img[src*="arrow-right"]');
+						return rel === 'next' || aria.includes('next') || title.includes('next') || cls.includes('next') || hasRightArrowIcon || ['next', '›', '»', '>'].includes(text);
+					});
+
+					if (hasExplicitNext) {
+						return true;
+					}
+
+					const current = controls.find((el) => {
+						const text = (el.textContent || '').trim();
+						const cls = (el.getAttribute('class') || '').toLowerCase();
+						const ariaCurrent = (el.getAttribute('aria-current') || '').toLowerCase();
+						return /^\d+$/.test(text) && (ariaCurrent === 'page' || cls.includes('active') || cls.includes('current') || cls.includes('selected'));
+					});
+
+					if (!current) {
+						return false;
+					}
+
+					const currentPage = parseInt((current.textContent || '').trim(), 10);
+					return controls.some((el) => {
+						const text = (el.textContent || '').trim();
+						return /^\d+$/.test(text) && parseInt(text, 10) === currentPage + 1;
+					});
+				}
+				"""
+			))
+		except Exception:
+			return False
+		return False

dubizzle_scraper/browser/network.py

@@ -0,0 +1,130 @@
+import json
+import logging
+from dataclasses import dataclass, field
+from typing import Any
+from urllib.parse import urlparse
+
+from playwright.sync_api import Page, Request, Response
+
+from ..core.config import Settings
+
+logger = logging.getLogger("dubizzle_scraper.network")
+
+
+@dataclass
+class NetworkCapture:
+    # Перехватчик сетевых запросов.
+
+    settings: Settings
+    requests: list[dict[str, Any]] = field(default_factory=list)
+    json_responses: list[dict[str, Any]] = field(default_factory=list)
+    _seen_req: set[str] = field(default_factory=set)
+    _seen_resp: set[str] = field(default_factory=set)
+    _origin: str | None = None
+    _page: Any = field(default=None)
+
+    def attach(self, page: Page, origin_url: str | None = None) -> None:
+        # Снимаем старые подписки перед повторным attach.
+        try:
+            page.remove_listener("request", self._on_request)
+            page.remove_listener("response", self._on_response)
+        except Exception:
+            pass
+        # Подписка на сетевые события
+        try:
+            self._origin = urlparse(origin_url or page.url).netloc.lower() or None
+        except Exception:
+            self._origin = None
+        self._page = page
+        page.on("request", self._on_request)
+        page.on("response", self._on_response)
+
+    def _is_same_origin(self, url: str) -> bool:
+        # Фильтр по домену
+        if not self.settings.capture.capture_same_origin_only or not self._origin:
+            return True
+        netloc = urlparse(url).netloc.lower()
+        return netloc == self._origin or netloc.endswith(".dubizzle.com")
+
+    def _on_request(self, request: Request) -> None:
+        # Берём только xhr/fetch
+        if request.resource_type not in {"xhr", "fetch"}:
+            return
+        if not self._is_same_origin(request.url):
+            return
+        if len(self.requests) >= self.settings.capture.max_requests:
+            return
+        key = f"{request.method}:{request.url}:{request.post_data or ''}"
+        # Убираем дубли
+        if key in self._seen_req:
+            return
+        self._seen_req.add(key)
+        self.requests.append({
+            "url": request.url, "method": request.method,
+            "resource_type": request.resource_type, "post_data": request.post_data,
+        })
+
+    def _on_response(self, response: Response) -> None:
+        # Сохраняем JSON
+        request = response.request
+        if request.resource_type not in {"xhr", "fetch"}:
+            return
+        if not self._is_same_origin(response.url):
+            return
+        if len(self.json_responses) >= self.settings.capture.max_json_responses:
+            return
+        if not self._is_json(response):
+            return
+        key = f"{request.method}:{response.url}:{response.status}"
+        if key in self._seen_resp:
+            return
+        self._seen_resp.add(key)
+        try:
+            payload = response.json()
+        except Exception:
+            try:
+                payload = json.loads(response.text())
+            except Exception:
+                return
+        self.json_responses.append({
+            "url": response.url, "status": response.status,
+            "request_method": request.method, "post_data": request.post_data,
+            "resource_type": request.resource_type, "payload": payload,
+            "category": self._categorize(response.url),
+        })
+
+    @staticmethod
+    def _is_json(response: Response) -> bool:
+        ct = (response.headers.get("content-type") or "").lower()
+        if "application/json" in ct or "+json" in ct:
+            return True
+        url = response.url.lower()
+        return any(m in url for m in ["/api/", "/graphql", "vehicledetail", "vehicle", "auction", "bid", "images", "media"])
+
+    @staticmethod
+    def _categorize(url: str) -> str:
+        # Простая категория URL
+        low = url.lower()
+        mapping = {
+            "images": ["image", "media", "photos", "gallery"],
+            "bids": ["bid", "offer", "buy-now", "buynow"],
+            "auction": ["auction", "sale", "lane", "branch"],
+            "vehicle": ["vehicle", "detail", "vin", "damage", "runanddrive"],
+            "documents": ["title", "document", "report"],
+        }
+        for cat, markers in mapping.items():
+            if any(m in low for m in markers):
+                return cat
+        return "other"
+
+    def export(self) -> dict[str, Any]:
+        responses = sorted(self.json_responses, key=lambda i: (i["category"] == "other", i["url"]))
+        return {
+            "requests": self.requests,
+            "json_responses": responses,
+            "capture_limits": {
+                "same_origin_only": self.settings.capture.capture_same_origin_only,
+                "max_requests": self.settings.capture.max_requests,
+                "max_json_responses": self.settings.capture.max_json_responses,
+            },
+        }

dubizzle_scraper/browser/pace.py

@@ -0,0 +1,46 @@
+import random
+import time
+
+from playwright.sync_api import Locator, Page
+
+from ..core.config import Settings
+
+
+class HumanPacer:
+    # Случайные паузы между действиями.
+
+    def __init__(self, settings: Settings) -> None:
+        self.settings = settings
+
+    def pause(self, min_s: float, max_s: float) -> None:
+        if self.settings.pace.enabled:
+            time.sleep(random.uniform(min_s, max_s))
+
+    def after_listing_open(self) -> None:
+        self.pause(self.settings.pace.after_listing_open_min_s, self.settings.pace.after_listing_open_max_s)
+
+    def after_filter_action(self) -> None:
+        self.pause(self.settings.pace.after_filter_action_min_s, self.settings.pace.after_filter_action_max_s)
+
+    def before_vehicle_open(self) -> None:
+        self.pause(self.settings.pace.before_vehicle_open_min_s, self.settings.pace.before_vehicle_open_max_s)
+
+    def after_vehicle_open(self) -> None:
+        self.pause(self.settings.pace.after_vehicle_open_min_s, self.settings.pace.after_vehicle_open_max_s)
+
+    def between_vehicles(self) -> None:
+        self.pause(self.settings.pace.between_vehicles_min_s, self.settings.pace.between_vehicles_max_s)
+
+    def after_page_change(self) -> None:
+        self.pause(self.settings.pace.after_page_change_min_s, self.settings.pace.after_page_change_max_s)
+
+    def move_mouse_to(self, page: Page, locator: Locator) -> None:
+        try:
+            box = locator.bounding_box()
+        except Exception:
+            box = None
+        if not box:
+            return
+        x = box["x"] + box["width"] * random.uniform(0.2, 0.8)
+        y = box["y"] + box["height"] * random.uniform(0.2, 0.8)
+        page.mouse.move(x, y, steps=random.randint(8, 18))