python - scrapy log handler

Question

I seek your help in the following 2 questions - How do I set the handler for the different log levels like in python. Currently, I have

STATS_ENABLED = True
STATS_DUMP = True 

LOG_FILE = 'crawl.log'

But the debug messages generated by Scrapy are also added into the log files. Those are very long and ideally, I would like the DEBUG level messages to left on standard error and INFO messages to be dump to my LOG_FILE.

Secondly, in the docs, it says The logging service must be explicitly started through the scrapy.log.start() function. My question is, where do I run this scrapy.log.start()? Is it inside my spider?

Answer 1

Secondly, in the docs, it says The logging service must be explicitly started through the scrapy.log.start() function. My question is, where do I run this scrapy.log.start()? Is it inside my spider?

If you run a spider using scrapy crawl my_spider -- the log is started automatically if STATS_ENABLED = True

If you start the crawler process manually, you can do scrapy.log.start() before starting the crawler process.

from scrapy.crawler import CrawlerProcess
from scrapy.conf import settings


settings.overrides.update({}) # your settings

crawlerProcess = CrawlerProcess(settings)
crawlerProcess.install()
crawlerProcess.configure()

crawlerProcess.crawl(spider) # your spider here

log.start() # depends on LOG_ENABLED

print "Starting crawler."
crawlerProcess.start()
print "Crawler stopped."

The little knowledge I have about your first question:

Because you have to start the scrapy log manually, this allows you to use your own logger.

I think you can copy module scrapy/scrapy/log.py in scrapy sources, modify it, import it instead of scrapy.log and run start() - scrapy will use your log. In it there is a line in function start() which says log.startLoggingWithObserver(sflo.emit, setStdout=logstdout).

Make your own observer (http://docs.python.org/howto/logging-cookbook.html#logging-to-multiple-destinations) and use it there.