Navigate before initializing page state

crawler/scenarios/fresh_session.py

@@ -1,8 +1,11 @@
 """Scenario 2 — fresh cookie jar per run so the site treats the visitor as a new user."""
 from __future__ import annotations
 
+from typing import Any, cast
+from urllib.parse import urlsplit
+
 import undetected_chromedriver as uc
-from selenium.common.exceptions import TimeoutException
+from selenium.common.exceptions import TimeoutException, WebDriverException
 from selenium.webdriver.support.ui import WebDriverWait
 
 from crawler.driver import human_delay
@@ -10,56 +13,100 @@ from logger import get_logger
 
 log = get_logger(__name__)
 
+_BLANK_URLS = {"", "about:blank", "data:,"}
+
+
+class FreshSessionError(RuntimeError):
+    """The browser could not leave its initial blank page."""
+
+
+def _normalise_url(url: str) -> str:
+    value = url.strip()
+    if value and "://" not in value:
+        value = f"https://{value}"
+    parsed = urlsplit(value)
+    if parsed.scheme not in {"http", "https"} or not parsed.netloc:
+        raise FreshSessionError(
+            "target_url must be a complete HTTP(S) URL, for example https://example.com"
+        )
+    return value
+
 
 class FreshSessionScenario:
     """
     Opens the target URL in a brand-new browser profile (no stored cookies,
     localStorage, or cache) so each run appears as a first-time visitor.
 
-    Because undetected-chromedriver already creates an isolated temp profile
-    per instance, simply spinning up a new driver is sufficient — but we also
-    explicitly delete all cookies after loading to be safe.
+    Undetected-chromedriver creates an isolated temporary profile per instance,
+    so no cookie or storage clearing is needed before navigation.
     """
 
     def __init__(self, driver: uc.Chrome, url: str, timeout: int = 30) -> None:
         self.driver = driver
-        self.url = url
+        self.url = _normalise_url(url)
         self.wait = WebDriverWait(driver, timeout)
 
     def run(self) -> dict[str, object]:
-        """Navigate to the target, clear any residual state, return page info."""
-        self.driver.delete_all_cookies()
-
-        # Clear localStorage / sessionStorage via JS
-        self.driver.execute_script(
-            "try { window.localStorage.clear(); window.sessionStorage.clear(); } catch(e) {}"
-        )
-
-        if not self.url:
-            return {"title": "", "url": "", "cookie_count": 0, "cookies": []}
+        """Navigate to the target and return page information."""
+        # Each driver already owns a brand-new temporary profile. Do not issue
+        # renderer commands on Chrome's initial blank tab: on small servers that
+        # renderer can stall before navigation is ever attempted.
+        log.info("Navigating browser to %s", self.url)
+        try:
+            navigation = cast(
+                dict[str, Any],
+                self.driver.execute_cdp_cmd(  # type: ignore[reportUnknownMemberType]
+                    "Page.navigate", {"url": self.url}
+                ),
+            )
+            error_text = navigation.get("errorText")
+            if error_text:
+                raise FreshSessionError(f"Chrome rejected navigation: {error_text}")
+        except FreshSessionError:
+            raise
+        except WebDriverException:
+            # Older Chrome builds may not expose Page.navigate. Keep a normal
+            # WebDriver fallback, but only after navigation has been attempted.
+            try:
+                self.driver.get(self.url)
+            except TimeoutException:
+                log.warning("Page load timed out; stopping navigation")
+                self.driver.execute_cdp_cmd(  # type: ignore[reportUnknownMemberType]
+                    "Page.stopLoading", {}
+                )
 
         try:
-            self.driver.get(self.url)
-        except TimeoutException:
-            log.warning(
-                "Page load exceeded the browser timeout; stopping navigation and continuing"
-            )
-            self.driver.execute_cdp_cmd("Page.stopLoading", {})
-        human_delay(2.0, 4.0)
+            self.wait.until(lambda d: d.current_url not in _BLANK_URLS)
+        except TimeoutException as exc:
+            raise FreshSessionError(
+                f"navigation to {self.url} never left Chrome's blank page"
+            ) from exc
+
+        human_delay(1.0, 2.0)
 
         # Interactive is sufficient for the dynamic flow and avoids waiting on
         # analytics, ads, or other background resources indefinitely.
-        try:
-            self.wait.until(
-                lambda d: d.execute_script("return document.readyState")
-                in ("interactive", "complete")
+        def document_is_ready(driver: uc.Chrome) -> bool:
+            state = cast(
+                str,
+                driver.execute_script(  # type: ignore[reportUnknownMemberType]
+                    "return document.readyState"
+                ),
             )
+            return state in ("interactive", "complete")
+
+        try:
+            self.wait.until(document_is_ready)
         except TimeoutException:
             log.warning("Document did not report a ready state; continuing with the current DOM")
 
         title = self.driver.title
         current_url = self.driver.current_url
-        cookies = self.driver.get_cookies()
+        cookies = cast(
+            list[dict[str, Any]],
+            self.driver.get_cookies(),  # type: ignore[reportUnknownMemberType]
+        )
+        log.info("Browser reached %s", current_url)
 
         return {
             "title": title,