fix: reset page.redirect_url before browsing to prevent false errors … (#441)

* fix: reset page.redirect_url before browsing to prevent false errors in retry loops

* chore: formatting unit test

* chore: more formatting

* chore: refactor reset of page.redirect_url

Author: adam-miller

brozzler/model.py

@@ -445,6 +445,9 @@ def __str__(self):
     def note_redirect(self, url):
         self.redirect_url = url
 
+    def clear_redirect(self):
+        self.redirect_url = None
+
     def _calc_priority(self):
         if not self.url:
             return None

brozzler/worker.py

@@ -580,6 +580,7 @@ def _on_service_worker_version_updated(chrome_msg):
                 window_width=self._window_width,
                 headless=self._headless,
             )
+        page.clear_redirect()
         final_page_url, outlinks = browser.browse_page(
             page.url,
             extra_headers=site.extra_headers(page),

tests/test_brozzling.py

@@ -398,3 +398,18 @@ def test_network_monitoring(httpd):
         # 3 second request + 2 seconds idle = 5 seconds
         elapsed = time.time() - start
         assert 5 < elapsed < 6
+
+
+def test_chrome_error(httpd):
+    chrome_exe = brozzler.suggest_default_chrome_exe()
+    url = "http://nonexistent.invalid"
+    worker = brozzler.BrozzlerWorker(None)
+    site = brozzler.Site(None, {})
+    page = brozzler.Page(None, {"url": url})
+    with brozzler.Browser(chrome_exe=chrome_exe) as browser:
+        with pytest.raises(brozzler.PageConnectionError):
+            worker.brozzle_page(browser, site, page)
+        assert page.redirect_url and page.redirect_url.startswith("chrome-error:")
+        page.url = "http://localhost:%s/site1/" % httpd.server_port
+        worker.brozzle_page(browser, site, page)
+        assert not (page.redirect_url and page.redirect_url.startswith("chrome-error:"))