From 4f5c66fd33263a146609efbfcbfde4a344f0ac2d Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Sat, 17 Oct 2020 17:09:57 -0700
Subject: ingest: handle cookieAbsent and partial SPNv2 URL reponse cases
 better

---
 python/sandcrawler/ia.py | 31 +++++++++++++++++++++++++++++++
 1 file changed, 31 insertions(+)

(limited to 'python/sandcrawler/ia.py')

diff --git a/python/sandcrawler/ia.py b/python/sandcrawler/ia.py
index 2bc52ce..60e3d9a 100644
--- a/python/sandcrawler/ia.py
+++ b/python/sandcrawler/ia.py
@@ -910,6 +910,9 @@ class SavePageNowClient:
                 raise SavePageNowBackoffError("SPNv2 user-session-limit")
             elif status.startswith("error:"):
                 status = "spn2-" + status
+            # despite other errors, call these a failure (so we don't retry)
+            if spn_result.terminal_url and (spn_result.terminal_url.endswith('/cookieAbsent') or spn_result.terminal_url.endswith("cookieSet=1")):
+                status = "blocked-cookie"
             return ResourceResult(
                 start_url=start_url,
                 hit=False,
@@ -923,6 +926,34 @@ class SavePageNowClient:
             )
         #print(spn_result, file=sys.stderr)
 
+        # detect partial URL response (aka, missing full URL)
+        if spn_result.terminal_url.startswith('/'):
+            return ResourceResult(
+                start_url=start_url,
+                hit=False,
+                status="spn2-success-partial-url",
+                terminal_url=spn_result.terminal_url,
+                terminal_dt=spn_result.terminal_dt,
+                terminal_status_code=None,
+                body=None,
+                cdx=None,
+                revisit_cdx=None,
+            )
+
+        # don't try to CDX fetch for this common cookie block terminal
+        if spn_result.terminal_url.endswith('/cookieAbsent') or spn_result.terminal_url.endswith("cookieSet=1"):
+            return ResourceResult(
+                start_url=start_url,
+                hit=False,
+                status="blocked-cookie",
+                terminal_url=spn_result.terminal_url,
+                terminal_dt=spn_result.terminal_dt,
+                terminal_status_code=None,
+                body=None,
+                cdx=None,
+                revisit_cdx=None,
+            )
+
         cdx_row = None
         # hack to work around elsevier weirdness
         if "://pdf.sciencedirectassets.com/" in spn_result.request_url:
-- 
cgit v1.2.3