Exemplo n.º 1
0
 def setUp(self):
     """Set up a mock patcher."""
     self.patcher = mock.patch('requests.post', fake_requests_post)
     self.patcher.start()
     import diffbot
     self.client = diffbot.Client(token=TOKEN)
     self.client_v1 = diffbot.Client(token=TOKEN, version=1)
     self.client_v2 = diffbot.Client(token=TOKEN, version=2)
Exemplo n.º 2
0
    def setUp(self):
        """Set up a mock patcher.

        This will make the `requests` library unavailable in `diffbot`.
        """
        self.import_hook = ImportHook('requests')
        sys.meta_path.insert(0, self.import_hook)
        try:
            self.patcher = mock.patch('urllib2.urlopen', fake_urllib2_urlopen)
            self.patcher.start()
        except ImportError:
            self.patcher = mock.patch('urllib.request.urlopen',
                                      fake_urllib2_urlopen)
            self.patcher.start()
        import diffbot
        diffbot = imp.reload(diffbot)
        self.client = diffbot.Client(token=TOKEN)
Exemplo n.º 3
0
from pymongo import MongoClient
import diffbot
from hackernews import HackerNews
import os

envHost = os.environ['MONGO_DB_HOST']
envPort = os.environ['MONGO_DB_PORT']
envDB = os.environ['MONGO_DB_DATABASE']
dbURL = 'mongodb://' + envHost + ':' + envPort + '/' + envDB
client = MongoClient(dbURL)
db = client.get_database()
hn = HackerNews()
diffbot_token = os.environ['DIFFBOT_TOKEN']
diffbot_client = diffbot.Client(token=diffbot_token)

links = []

# Get new links
for story_id in hn.top_stories(limit=1000):
    item = hn.get_item(story_id)
    url = item.url
    # Check if link is already in database
    if db.unrelatedlinks.find_one({'url': item.url}) is not None:
        continue
    try:
        result = diffbot_client.api('article', url)
        title = result["objects"][0]["title"]
        text = result["objects"][0]["text"]
        date = result["objects"][0]["date"]
        tags = []
        try: