From f5286affa62c0caf95ff4848626e85aff476737e Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Mon, 12 Oct 2020 13:06:31 -0700 Subject: store no-capture URLs in terminal_url --- python/sandcrawler/ia.py | 2 +- python/sandcrawler/ingest.py | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'python') diff --git a/python/sandcrawler/ia.py b/python/sandcrawler/ia.py index 7b623bc..2bc52ce 100644 --- a/python/sandcrawler/ia.py +++ b/python/sandcrawler/ia.py @@ -589,7 +589,7 @@ class WaybackClient: start_url=start_url, hit=False, status="no-capture", - terminal_url=None, + terminal_url=next_url, terminal_dt=None, terminal_status_code=None, body=None, diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index e8e517a..5ab7e13 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -387,7 +387,7 @@ class IngestFileWorker(SandcrawlerWorker): if not resource.hit: result['status'] = resource.status - if resource.terminal_dt and resource.terminal_status_code: + if resource.terminal_url: result['terminal'] = { "terminal_url": resource.terminal_url, "terminal_dt": resource.terminal_dt, @@ -465,7 +465,7 @@ class IngestFileWorker(SandcrawlerWorker): result['status'] = "max-hops-exceeded" return result - if resource.terminal_dt: + if resource.terminal_url: result['terminal'] = { "terminal_url": resource.terminal_url, "terminal_dt": resource.terminal_dt, -- cgit v1.2.3