datamarket 0.6.0__py3-none-any.whl → 0.10.3__py3-none-any.whl

datamarket/utils/playwright/sync_api.py

@@ -0,0 +1,281 @@
+########################################################################################################################
+# IMPORTS
+
+import json
+import logging
+import time
+from datetime import timedelta
+from random import randint
+from types import TracebackType
+from typing import Optional, Self, Sequence
+
+from bs4 import BeautifulSoup
+from camoufox import Camoufox
+from playwright.sync_api import Browser, BrowserContext, Page, Response
+from playwright.sync_api import (
+    Error as PlaywrightError,
+)
+from playwright.sync_api import (
+    TimeoutError as PlaywrightTimeoutError,
+)
+from requests.exceptions import HTTPError, ProxyError
+from tenacity import (
+    before_sleep_log,
+    retry,
+    retry_if_exception_type,
+    retry_if_not_exception_type,
+    stop_after_attempt,
+    stop_after_delay,
+    wait_exponential,
+)
+
+from datamarket.exceptions import BadRequestError, EmptyResponseError, NotFoundError, RedirectionDetectedError
+from datamarket.exceptions.main import IgnoredHTTPError
+from datamarket.interfaces.proxy import ProxyInterface
+from datamarket.utils.main import ban_sleep
+
+########################################################################################################################
+# SETUP LOGGER
+
+logger = logging.getLogger(__name__)
+
+
+########################################################################################################################
+# HELPER FUNCTIONS
+
+
+def human_type(page: Page, text: str, delay: int = 100):
+    for char in text:
+        page.keyboard.type(char, delay=randint(int(delay * 0.5), int(delay * 1.5)))  # noqa: S311
+
+
+def human_press_key(page: Page, key: str, count: int = 1, delay: int = 100, add_sleep: bool = True) -> None:
+    """Presses a key with a random delay, optionally sleeping between presses."""
+    for _ in range(count):
+        page.keyboard.press(key, delay=randint(int(delay * 0.5), int(delay * 1.5)))  # noqa: S311
+        if add_sleep:
+            time.sleep(randint(int(delay * 1.5), int(delay * 2.5)) / 1000)  # noqa: S311
+
+
+########################################################################################################################
+# CLASSES
+
+
+class PlaywrightCrawler:
+    """A robust, proxy-enabled Playwright crawler with captcha bypass and retry logic."""
+
+    _REDIRECT_STATUS_CODES = set(range(300, 309))
+
+    def __init__(self, proxy_interface: Optional[ProxyInterface] = None):
+        """
+        Initializes the crawler.
+
+        Args:
+            proxy_interface (Optional[ProxyInterface], optional): Provider used to fetch
+                proxy credentials. Defaults to None. When None, no proxy is configured and
+                the browser will run without a proxy.
+        """
+        self.proxy_interface = proxy_interface
+        self.pw: Optional[Camoufox] = None
+        self.browser: Optional[Browser] = None
+        self.context: Optional[BrowserContext] = None
+        self.page: Optional[Page] = None
+
+    def __enter__(self) -> Self:
+        """Initializes the browser context when entering the `with` statement."""
+        self.init_context()
+        return self
+
+    def __exit__(
+        self,
+        exc_type: Optional[type[BaseException]],
+        exc_val: Optional[BaseException],
+        exc_tb: Optional[TracebackType],
+    ) -> None:
+        """Safely closes the browser context upon exit."""
+        if self.pw:
+            try:
+                self.pw.__exit__(exc_type, exc_val, exc_tb)
+            except Exception as e:
+                logger.warning(f"Error closing browser: {e}")
+            finally:
+                self.pw = None
+                self.browser = None
+                self.context = None
+                self.page = None
+
+    def _build_proxy_config(self) -> Optional[dict]:
+        """Builds the proxy configuration dictionary.
+
+        Returns:
+            Optional[dict]: Proxy configuration if a proxy_interface is provided; otherwise None.
+        """
+        if not self.proxy_interface:
+            logger.info("Starting browser without proxy.")
+            return None
+
+        host, port, user, pwd = self.proxy_interface.get_proxies(raw=True, use_auth=True)
+        proxy_url = f"http://{host}:{port}"
+        proxy_cfg: dict = {"server": proxy_url}
+        if user and pwd:
+            proxy_cfg.update({"username": user, "password": pwd})
+
+        logger.info(f"Starting browser with proxy: {proxy_url}")
+        return proxy_cfg
+
+    @retry(
+        wait=wait_exponential(exp_base=2, multiplier=3, max=90),
+        stop=stop_after_delay(timedelta(minutes=10)),
+        before_sleep=before_sleep_log(logger, logging.INFO),
+        reraise=True,
+    )
+    def init_context(self) -> Self:
+        """
+        Initializes a new browser instance and context.
+
+        Behavior:
+        - If a proxy_interface is provided, fetches fresh proxy credentials and starts
+          the browser using that proxy.
+        - If proxy_interface is None, starts the browser without any proxy.
+
+        Returns:
+            Self: The crawler instance with active browser, context, and page.
+        """
+        try:
+            proxy_cfg: Optional[dict] = self._build_proxy_config()
+
+            self.pw = Camoufox(headless=True, geoip=True, humanize=True, proxy=proxy_cfg)
+            self.browser = self.pw.__enter__()
+            self.context = self.browser.new_context()
+            self.page = self.context.new_page()
+        except Exception as e:
+            logger.error(f"Failed to initialize browser context: {e}")
+            if self.pw:
+                self.pw.__exit__(type(e), e, e.__traceback__)
+            raise
+        return self
+
+    def restart_context(self) -> None:
+        """Closes the current browser instance and initializes a new one."""
+        logger.info("Restarting browser context...")
+        if self.pw:
+            self.pw.__exit__(None, None, None)
+        self.init_context()
+
+    @retry(
+        retry=retry_if_exception_type((PlaywrightTimeoutError, PlaywrightError)),
+        wait=wait_exponential(exp_base=2, multiplier=3, max=90),
+        stop=stop_after_delay(timedelta(minutes=10)),
+        before_sleep=before_sleep_log(logger, logging.INFO),
+        before=lambda rs: rs.args[0].restart_context() if rs.attempt_number > 1 else None,
+        reraise=True,
+    )
+    def _goto_with_retry(self, url: str, timeout: int = 30_000) -> Response:
+        """
+        Navigates to a URL with retries for common Playwright errors.
+        Restarts the browser context on repeated failures.
+        """
+        if not (self.page and not self.page.is_closed()):
+            logger.warning("Page is not available or closed. Restarting context.")
+            self.restart_context()
+
+        response = self.page.goto(url, timeout=timeout, wait_until="domcontentloaded")
+        return response
+
+    def goto(self, url: str, max_proxy_delay: timedelta = timedelta(minutes=10), timeout: int = 30_000) -> Response:
+        """
+        Ensures the browser is initialized and navigates to the given URL.
+        Public wrapper for the internal retry-enabled navigation method.
+        """
+        if not self.page:
+            logger.info("Browser context not found, initializing now...")
+            self.init_context()
+        return self._goto_with_retry.retry_with(stop=stop_after_delay(max_proxy_delay))(self, url, timeout)
+
+    def _handle_http_error(self, status_code: int, url: str, response, allow_redirects: bool = True) -> None:
+        """
+        Handle HTTP errors with special handling for redirects.
+
+        Args:
+            status_code: HTTP status code
+            url: Request URL
+            response: Response object
+
+        Raises:
+            RedirectionDetectedError: If a redirect status is received
+            NotFoundError: For 404/410 errors
+            BadRequestError: For 400 errors
+            HTTPError: For other non-2xx status codes
+        """
+
+        if not allow_redirects and response.request.redirected_from:  # noqa: F841
+            raise RedirectionDetectedError(
+                message=f"HTTP redirect detected from {response.request.redirected_from.url} to {response.request.redirected_from.redirected_to.url}",
+                response=response,
+            )
+
+        # Standard error handlers
+        error_handlers = {
+            404: lambda: NotFoundError(message=f"404 Not Found error for {url}", response=response),
+            410: lambda: NotFoundError(message=f"410 Gone error for {url}", response=response),
+            400: lambda: BadRequestError(message=f"400 Bad Request error for {url}", response=response),
+        }
+
+        if status_code in error_handlers:
+            raise error_handlers[status_code]()
+
+        # Raise for any other non-2xx status
+        if status_code >= 400:
+            raise HTTPError(f"Navigation failed: {status_code} - {url}", response=response)
+
+    @retry(
+        retry=retry_if_not_exception_type(
+            (IgnoredHTTPError, NotFoundError, BadRequestError, RedirectionDetectedError, ProxyError)
+        ),
+        wait=wait_exponential(exp_base=3, multiplier=3, max=60),
+        stop=stop_after_attempt(5),
+        before_sleep=before_sleep_log(logger, logging.WARNING),
+        reraise=False,
+        retry_error_callback=lambda rs: None,
+    )
+    def get_data(
+        self,
+        url: str,
+        output: str = "json",
+        sleep: tuple = (6, 3),
+        max_proxy_delay: timedelta = timedelta(minutes=10),
+        ignored_status_codes: Sequence[int] = (),
+        timeout: int = 30_000,
+        **kwargs,
+    ):
+        """Crawls a given URL using Playwright and attempts to parse its body content"""
+
+        params = kwargs.copy()
+
+        allow_redirects = params.get("allow_redirects", True)
+
+        logger.info(f"Fetching data from {url} ...")
+        r = self.goto(url, max_proxy_delay, timeout)
+        ban_sleep(*sleep)
+        body_content = self.page.eval_on_selector("body", "body => body.innerText")
+
+        if r.status in ignored_status_codes:
+            raise IgnoredHTTPError(message=f"Status {r.status} in ignored_status_codes for URL {url}", response=r)
+
+        # Handle HTTP errors with redirect detection
+        self._handle_http_error(r.status, url, r, allow_redirects)
+
+        if not body_content:
+            raise EmptyResponseError(message=f"Empty body received from {url} (status {r.status})", response=r)
+
+        output_format = {
+            "json": lambda: json.loads(body_content),
+            "text": lambda: body_content,
+            "soup": lambda: BeautifulSoup(body_content, "html.parser"),
+            "response": lambda: r,
+        }
+
+        if output in output_format:
+            return output_format[output]()
+
+        raise ValueError(f"Unsupported output format: {output}")