Skip to content

Latest commit

 

History

History
105 lines (92 loc) · 3.57 KB

README.md

File metadata and controls

105 lines (92 loc) · 3.57 KB
Raw

中文文档 | English

Scrpay-Kafka-Redis

In the case of a large number of requests, even using the Bloomfilter algorithm, but using [scrapy-redis] (https://github.com/rmax/scrapy-redis) still consumes a lot of memory. This project refers to scrapy- Redis.

Features

  • Support for distributed
  • Use Redis as a deduplication queue, Simultaneous use of Bloomfilter to reduce the memory footprint, but increased the amount of deduplication
  • Use Kafka as a request queue, Can support a large number of request stacks, capacity and disk size related, rather than running memory
  • Due to the feature of Kafka, priority queues are not supported, only FIFO queues are supported.

Dependencies

  • Python 3.0+
  • Redis >= 2.8
  • Scrapy >= 1.5
  • kafka-python >= 1.4.0
  • kafka <= 1.1.0 (Since kafka-python only supports kafka-1.1.0 version)

How to Use

  • pip install scrapy-kafka-redis
  • Configuration settings.py file Must add param in settings.py file
# Enable Kafka scheduling storage request queue
SCHEDULER = "scrapy_kafka_redis.scheduler.Scheduler"

# Use BloomFilter as a deduplication queue
DUPEFILTER_CLASS = "scrapy_kafka_redis.dupefilter.BloomFilter"

Default values for other optional parameters

# the key of the deduplication queue stored in redis
DUPEFILTER_KEY = 'dupefilter:%(timestamp)s'

REDIS_CLS = redis.StrictRedis
REDIS_ENCODING = 'utf-8'
REDIS_URL = 'redis://localhost:6378/1'

REDIS_PARAMS = {
    'socket_timeout': 30,
    'socket_connect_timeout': 30,
    'retry_on_timeout': True,
    'encoding': REDIS_ENCODING,
}

KAFKA_BOOTSTRAP_SERVERS=['localhost:9092']
# Default TOPIC for the dispatch queue
SCHEDULER_QUEUE_TOPIC = '%(spider)s-requests'
# Scheduled queue used by default
SCHEDULER_QUEUE_CLASS = 'scrapy_kafka_redis.queue.KafkaQueue'
# The name of the key stored in the redis queue in redis
SCHEDULER_DUPEFILTER_KEY = '%(spider)s:dupefilter'
# Deduplication algorithm used by the scheduler
SCHEDULER_DUPEFILTER_CLASS = 'scrapy_kafka_redis.dupefilter.BloomFilter'
# Number of blocks in the BloomFilter algorithm
BLOOM_BLOCK_NUM = 1

# TOPIC used by start urls
START_URLS_TOPIC = '%(name)s-start_urls'

KAFKA_BOOTSTRAP_SERVERS = None
# Kafka producer constructing the request queue
KAFKA_REQUEST_PRODUCER_PARAMS = {
    'api_version': (0, 10, 1),
    'value_serializer': dumps
}
# Constructing a Kafka consumer of the request queue
KAFKA_REQUEST_CONSUMER_PARAMS = {
    'api_version': (0, 10, 1),
    'value_deserializer': loads
}
# Constructing a Kafka consumer in the start queue
KAFKA_START_URLS_CONSUMER_PARAMS = {
    'api_version': (0, 10, 1),
    'value_deserializer': lambda m: m.decode('utf-8'),
}
  • how to use in spiders
import scrapy
from scrapy_kafka_redis.spiders import KafkaSpider

class DemoSpider(KafkaSpider):
    name = "demo"
    def parse(self, response):
        pass
  • Create Kafka Topic Set the number of partitions for the topic based on the distributed scrapy instance you need to create.
./bin/kafka-topics.sh --create --zookeeper localhost:2181 --partitions 3 --replication-factor 1 --topic demo-start_urls

./bin/kafka-topics.sh --create --zookeeper localhost:2181 --partitions 3 --replication-factor 1 --topic demo-requests
  • Send Msg
./bin/kafka-console-producer.sh --broker-list localhost:9092 --topic demo-start_urls

It is recommended to manually create a Topic and specify the number of partitions.

  • run scrapy

Reference:

scrapy-redis Bloomfilter