Beispiel #1
0
    def __init__(self, max_size, redis_url, params, name, timeout=30):
        redis_url = res_template(redis_url, params)
        super(RedisPendingCounterTempBuffer, self).__init__(max_size=max_size)
        self.redis, self.key = RedisIndexSource.parse_redis_url(redis_url)
        self.timeout = timeout

        self.redis.incrby(self.key, 1)
        self.redis.expire(self.key, self.timeout)
    def setup_class(cls):
        super(TestIndexSources, cls).setup_class()
        cls.add_cdx_to_redis(TEST_CDX_PATH + 'iana.cdxj', 'test:rediscdx')

        cls.all_sources = {
            'file': FileIndexSource(TEST_CDX_PATH + 'iana.cdxj'),
            'redis': RedisIndexSource('redis://localhost:6379/2/test:rediscdx'),
            'remote_cdx': RemoteIndexSource('https://webenact.rhizome.org/all/cdx?url={url}',
                              'https://webenact.rhizome.org/all/{timestamp}id_/{url}'),

            'memento': MementoIndexSource('https://webenact.rhizome.org/all/{url}',
                               'https://webenact.rhizome.org/all/timemap/link/{url}',
                               'https://webenact.rhizome.org/all/{timestamp}id_/{url}')
        }
Beispiel #3
0
 def _get_source_for_key(self, key):
     return RedisIndexSource(None, self.redis, key)
    def __init__(self):
        init_logging()

        config = load_wr_config()

        app = BaseWarcServer(debug=True)

        redis_base = os.environ['REDIS_BASE_URL'] + '/'

        rec_url = redis_base + config['cdxj_key_templ']
        coll_url = redis_base + config['coll_cdxj_key_templ']
        warc_url = redis_base + config['warc_key_templ']
        rec_list_key = config['rec_list_key_templ']

        redis_resolver = RedisResolver(redis_url=warc_url,
                                       member_key_templ=rec_list_key)
        redis = redis_resolver.redis
        warc_resolvers = [redis_resolver]

        cache_proxy_url = os.environ.get('CACHE_PROXY_URL', '')
        global PROXY_PREFIX
        PROXY_PREFIX = cache_proxy_url

        timeout = 20.0

        rec_redis_source = RedisIndexSource(timeout=timeout,
                                            redis_url=rec_url,
                                            redis=redis)

        coll_redis_source = RedisIndexSource(timeout=timeout,
                                             redis_url=coll_url,
                                             redis=redis)

        live_rec = DefaultResourceHandler(
            SimpleAggregator({'live': LiveIndexSource()}, ), warc_resolvers,
            cache_proxy_url)

        # Extractable archives (all available)
        wam_loader = WAMSourceLoader(memento_cls=ProxyMementoIndexSource,
                                     remote_cls=ProxyRemoteIndexSource,
                                     wb_memento_cls=ProxyWBMementoIndexSource)

        extractable_archives = wam_loader.sources

        # Extract Source
        extractor = GeventTimeoutAggregator(extractable_archives,
                                            timeout=timeout)
        extract_primary = DefaultResourceHandler(extractor, warc_resolvers,
                                                 cache_proxy_url)

        # Patch fallback archives
        fallback_archives = self.filter_archives(
            extractable_archives, config['patch_archives_index'])

        # patch + live
        #patch_archives = fallback_archives.copy()
        patch_archives = fallback_archives
        patch_archives['live'] = LiveIndexSource()

        extractor2 = GeventTimeoutAggregator(patch_archives,
                                             timeout=timeout,
                                             sources_key='inv_sources',
                                             invert_sources=True)

        extract_other = DefaultResourceHandler(extractor2, warc_resolvers,
                                               cache_proxy_url)

        patcher = GeventTimeoutAggregator(patch_archives, timeout=timeout)
        patch_rec = DefaultResourceHandler(patcher, warc_resolvers,
                                           cache_proxy_url)

        # Single Rec Replay
        replay_rec = DefaultResourceHandler(
            SimpleAggregator({'local': rec_redis_source}), warc_resolvers,
            cache_proxy_url)

        # Coll Replay
        replay_coll = DefaultResourceHandler(
            SimpleAggregator({'local': coll_redis_source}), warc_resolvers,
            cache_proxy_url)

        app.add_route('/live', live_rec)
        app.add_route('/extract',
                      HandlerSeq([extract_primary, extract_other, replay_rec]))
        app.add_route('/replay', replay_rec)
        app.add_route('/replay-coll', replay_coll)
        app.add_route('/patch', HandlerSeq([replay_coll, patch_rec]))

        self.app = app