mirror of
https://github.com/internetarchive/brozzler.git
synced 2025-02-24 16:49:56 -05:00
fix case where rethinkdb page already has claimed=True
This commit is contained in:
parent
ffcf26b6c9
commit
2e687b65fb
@ -197,12 +197,12 @@ class RethinkDbFrontier:
|
|||||||
1).update({
|
1).update({
|
||||||
"claimed":True,
|
"claimed":True,
|
||||||
"last_claimed_by":worker_id},
|
"last_claimed_by":worker_id},
|
||||||
return_changes=True).run()
|
return_changes="always").run()
|
||||||
self._vet_result(result, replaced=[0,1])
|
self._vet_result(result, unchanged=[0,1], replaced=[0,1])
|
||||||
if result["replaced"] == 1:
|
if result["unchanged"] == 0 and result["replaced"] == 0:
|
||||||
return brozzler.Page(**result["changes"][0]["new_val"])
|
|
||||||
else:
|
|
||||||
raise brozzler.NothingToClaim
|
raise brozzler.NothingToClaim
|
||||||
|
else:
|
||||||
|
return brozzler.Page(**result["changes"][0]["new_val"])
|
||||||
|
|
||||||
def has_outstanding_pages(self, site):
|
def has_outstanding_pages(self, site):
|
||||||
results_iter = self.r.table("pages").between(
|
results_iter = self.r.table("pages").between(
|
||||||
|
2
setup.py
2
setup.py
@ -21,7 +21,7 @@ import setuptools
|
|||||||
|
|
||||||
setuptools.setup(
|
setuptools.setup(
|
||||||
name='brozzler',
|
name='brozzler',
|
||||||
version='1.1.dev34',
|
version='1.1.dev35',
|
||||||
description='Distributed web crawling with browsers',
|
description='Distributed web crawling with browsers',
|
||||||
url='https://github.com/internetarchive/brozzler',
|
url='https://github.com/internetarchive/brozzler',
|
||||||
author='Noah Levitt',
|
author='Noah Levitt',
|
||||||
|
Loading…
x
Reference in New Issue
Block a user