diff options
| author | François Schmidts <francois.schmidts@gmail.com> | 2016-01-08 16:30:09 +0100 |
|---|---|---|
| committer | François Schmidts <francois.schmidts@gmail.com> | 2016-01-11 09:52:39 +0100 |
| commit | 71b185bf1984080077937a814a0d19d70faf2f77 (patch) | |
| tree | a4a1e9321121de05d0ae13075a34cffad58ca9ee /src/web | |
| parent | Moved the crawler in the parent folder. (diff) | |
| download | newspipe-71b185bf1984080077937a814a0d19d70faf2f77.tar.gz newspipe-71b185bf1984080077937a814a0d19d70faf2f77.tar.bz2 newspipe-71b185bf1984080077937a814a0d19d70faf2f77.zip | |
using user agent in web crawler
Diffstat (limited to 'src/web')
| -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']: |