mirror of
https://github.com/internetarchive/warcprox.git
synced 2025-01-18 13:22:09 +01:00
Merge branch 'limit_revisits' into qa
This commit is contained in:
commit
8563b95ff6
@ -66,7 +66,6 @@ class LimitRevisitsPGMixin():
|
|||||||
"""
|
"""
|
||||||
Limit revisits recorded to one per revisit_key
|
Limit revisits recorded to one per revisit_key
|
||||||
"""
|
"""
|
||||||
|
|
||||||
def __init__(self):
|
def __init__(self):
|
||||||
self.datasource = "postgresql://archiveit@db.qa-archive-it.org/archiveit" # "postgresql://user@db_host/db_name"
|
self.datasource = "postgresql://archiveit@db.qa-archive-it.org/archiveit" # "postgresql://user@db_host/db_name"
|
||||||
self.datatable = "crawl_revisits" # postgres table in db_name
|
self.datatable = "crawl_revisits" # postgres table in db_name
|
||||||
@ -85,7 +84,7 @@ class LimitRevisitsPGMixin():
|
|||||||
and "metadata" in recorded_url.warcprox_meta
|
and "metadata" in recorded_url.warcprox_meta
|
||||||
and "ait-job-id" in recorded_url.warcprox_meta["metadata"]
|
and "ait-job-id" in recorded_url.warcprox_meta["metadata"]
|
||||||
):
|
):
|
||||||
revisit_key = recorded_url.warcprox_meta["metadata"]["ait-job-id"]
|
revisit_key = str(recorded_url.warcprox_meta["metadata"]["ait-job-id"])
|
||||||
else:
|
else:
|
||||||
revisit_key = '__unspecified__'
|
revisit_key = '__unspecified__'
|
||||||
|
|
||||||
|
Loading…
x
Reference in New Issue
Block a user