switchback 0.6.2__tar.gz → 0.6.3__tar.gz

{switchback-0.6.2 → switchback-0.6.3}/CHANGELOG.md

@@ -4,6 +4,15 @@ All notable changes to this project are documented here. Format loosely follows
 [Keep a Changelog](https://keepachangelog.com/); this project uses semantic-ish
 versioning while pre-1.0.
 
+## [0.6.3] - 2026-10-14
+
+### Fixed
+- **"Page not found" pages returned as articles by the browser tiers.** tier_4
+  and tier_5 ignored the HTTP status of the page they loaded, so a site answering
+  404 (seen on bhaskarenglish.in, which serves its 404 page to some bot traffic)
+  had its "404 - Page not found" page returned as a success. A 404 or 410 now
+  fails the tier, as it already did in tier_2, and the cascade moves on.
+
 ## [0.6.2] - 2026-10-14
 
 ### Fixed

{switchback-0.6.2 → switchback-0.6.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: switchback
-Version: 0.6.2
+Version: 0.6.3
 Summary: One cost-ordered scrape cascade (HTTP → stealth browser → paid), shared by every tool.
 Author-email: Akash Kodavuru <akash@theaklabs.com>
 License: MIT

{switchback-0.6.2 → switchback-0.6.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "switchback"
-version = "0.6.2"
+version = "0.6.3"
 description = "One cost-ordered scrape cascade (HTTP → stealth browser → paid), shared by every tool."
 readme = "README.md"
 requires-python = ">=3.10"

{switchback-0.6.2 → switchback-0.6.3}/switchback/tiers/_browser.py

@@ -34,6 +34,14 @@ def looks_blocked(html: str, url: str) -> bool:
     return _looks_like_botwall(html_to_markdown(html, base_url=url))
 
 
+def raise_if_missing(response) -> None:
+    """Fail the tier on a 404/410 document, as tier_2 does; a browser otherwise
+    renders the site's "Page not found" page and returns it as the article."""
+    status = getattr(response, "status", None)
+    if status in (404, 410):
+        raise RuntimeError(f"HTTP Error {status}: page not found")
+
+
 def response_bytes(responses) -> int:
     """Total wire bytes a render pulled across every resource — the residential-cost
     basis. Reads only the Content-Length header: it's non-blocking. (We deliberately

{switchback-0.6.2 → switchback-0.6.3}/switchback/tiers/tier_4.py

@@ -80,7 +80,7 @@ def fetch(url: str, timeout_ms: int = int(_TIMEOUT_S * 1000)) -> str:
             page = ctx.new_page()
             responses: list = []
             page.on("response", lambda resp: responses.append(resp))
-            page.goto(url, timeout=timeout_ms, wait_until="domcontentloaded")
+            _browser.raise_if_missing(page.goto(url, timeout=timeout_ms, wait_until="domcontentloaded"))
             html = page.content()
             if len(html) < 5000:
                 try:

{switchback-0.6.2 → switchback-0.6.3}/switchback/tiers/tier_5.py

@@ -86,7 +86,7 @@ def fetch(url: str) -> str:
             auth = session_cache.browser_cookies(url)
             if auth:
                 page.context.add_cookies(auth)
-            page.goto(url, wait_until="networkidle", timeout=_TIMEOUT_MS)
+            _browser.raise_if_missing(page.goto(url, wait_until="networkidle", timeout=_TIMEOUT_MS))
             html = page.content()
             # JS bot-manager sensor interstitial → settle + reload once.
             if _browser.looks_blocked(html, url):

{switchback-0.6.2 → switchback-0.6.3}/switchback.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: switchback
-Version: 0.6.2
+Version: 0.6.3
 Summary: One cost-ordered scrape cascade (HTTP → stealth browser → paid), shared by every tool.
 Author-email: Akash Kodavuru <akash@theaklabs.com>
 License: MIT