diff options
author | Cédric Bonhomme <cedric@cedricbonhomme.org> | 2016-01-11 15:22:26 +0100 |
---|---|---|
committer | Cédric Bonhomme <cedric@cedricbonhomme.org> | 2016-01-11 15:22:26 +0100 |
commit | 6d0e7c8fe0da4ce96fb9c7df3ab9c970f3a2913d (patch) | |
tree | 36d6ff44e909aa00bcb52adbe1f2fe5ede1f53f3 /src/web/lib/crawler.py | |
parent | Moved the crawler in the parent folder. (diff) | |
parent | fixing stuffs (diff) | |
download | newspipe-6d0e7c8fe0da4ce96fb9c7df3ab9c970f3a2913d.tar.gz newspipe-6d0e7c8fe0da4ce96fb9c7df3ab9c970f3a2913d.tar.bz2 newspipe-6d0e7c8fe0da4ce96fb9c7df3ab9c970f3a2913d.zip |
Merge pull request #30 from jaesivsm/master
misc fixes on crawlers and use of User-Agent
Diffstat (limited to 'src/web/lib/crawler.py')
-rw-r--r-- | src/web/lib/crawler.py | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/src/web/lib/crawler.py b/src/web/lib/crawler.py index 90a268e8..979ccbfc 100644 --- a/src/web/lib/crawler.py +++ b/src/web/lib/crawler.py @@ -52,7 +52,7 @@ class AbstractCrawler: auth=self.auth, data=json.dumps(data, default=default_handler), headers={'Content-Type': 'application/json', - 'User-Agent': 'jarr'}) + 'User-Agent': conf.USER_AGENT}) def wait(self, max_wait=300, checks=5, wait_for=2): checked, second_waited = 0, 0 @@ -217,7 +217,7 @@ class CrawlerScheduler(AbstractCrawler): def prepare_headers(self, feed): """For a known feed, will construct some header dictionnary""" - headers = {'User-Agent': 'jarr/crawler'} + headers = {'User-Agent': conf.USER_AGENT} if feed.get('last_modified'): headers['If-Modified-Since'] = feed['last_modified'] if feed.get('etag') and 'pyagg' not in feed['etag']: |