def cache_writer_service(self, val):
raise_feature_is_deprecated('Cache feature')
def cache_reader_service(self):
raise_feature_is_deprecated('Cache feature')
def setup_cache(self, *args, **kwargs):
raise_feature_is_deprecated('Cache feature')
def __init__(
self,
thread_number=None,
network_try_limit=None, task_try_limit=None,
request_pause=NULL,
priority_mode='random',
meta=None,
config=None,
args=None,
parser_requests_per_process=10000,
parser_pool_size=1,
http_api_port=None,
network_service='threaded',
grab_transport='pycurl',
# Deprecated
transport=None,
only_cache=False,
):
"""
Arguments:
* thread-number - Number of concurrent network streams
* network_try_limit - How many times try to send request
again if network error was occurred, use 0 to disable
* task_try_limit - Limit of tries to execute some task
this is not the same as network_try_limit
network try limit limits the number of tries which
are performed automatically in case of network timeout
of some other physical error
but task_try_limit limits the number of attempts which
are scheduled manually in the spider business logic
* priority_mode - could be "random" or "const"
* meta - arbitrary user data
* retry_rebuild_user_agent - generate new random user-agent for each
network request which is performed again due to network error
* args - command line arguments parsed with `setup_arg_parser` method
"""
self.fatal_error_queue = Queue()
self.task_queue_parameters = None
self.http_api_port = http_api_port
self._started = None
assert grab_transport in ('pycurl', 'urllib3')
self.grab_transport_name = grab_transport
self.parser_requests_per_process = parser_requests_per_process
self.stat = Stat()
self.task_queue = None
if args is None:
self.args = {}
else:
self.args = args
if config is not None:
self.config = config
else:
self.config = {}
if meta:
self.meta = meta
else:
self.meta = {}
self.thread_number = (
thread_number or
int(self.config.get('thread_number',
DEFAULT_NETWORK_STREAM_NUMBER)))
self.task_try_limit = (
task_try_limit or
int(self.config.get('task_try_limit', DEFAULT_TASK_TRY_LIMIT)))
self.network_try_limit = (
network_try_limit or
int(self.config.get('network_try_limit',
DEFAULT_NETWORK_TRY_LIMIT)))
self._grab_config = {}
if priority_mode not in ['random', 'const']:
raise SpiderMisuseError('Value of priority_mode option should be '
'"random" or "const"')
else:
self.priority_mode = priority_mode
if only_cache:
raise_feature_is_deprecated('Cache feature')
self.work_allowed = True
if request_pause is not NULL:
warn('Option `request_pause` is deprecated and is not '
'supported anymore')
self.proxylist_enabled = None
self.proxylist = None
self.proxy = None
self.proxy_auto_change = False
self.interrupted = False
self.parser_pool_size = parser_pool_size
self.parser_service = ParserService(
spider=self,
pool_size=self.parser_pool_size,
)
if transport is not None:
warn('The "transport" argument of Spider constructor is'
' deprecated. Use "network_service" argument.')
network_service = transport
assert network_service in ('threaded',)
if network_service == 'threaded':
# pylint: disable=no-name-in-module, import-error
from grab.spider.network_service.threaded import (
NetworkServiceThreaded
)
self.network_service = NetworkServiceThreaded(
self, self.thread_number
)
self.task_dispatcher = TaskDispatcherService(self)
if self.http_api_port:
self.http_api_service = HttpApiService(self)
else:
self.http_api_service = None
self.task_generator_service = TaskGeneratorService(
self.task_generator(), self,
)
def __init__(
self,
name=None,
url=None,
grab=None,
grab_config=None,
priority=None,
priority_set_explicitly=True,
network_try_count=0,
task_try_count=1,
valid_status=None,
use_proxylist=True,
delay=None,
raw=False,
callback=None,
fallback_name=None,
# deprecated
disable_cache=False,
refresh_cache=False,
cache_timeout=None,
# kwargs
**kwargs):
"""
Create `Task` object.
If more than one of url, grab and grab_config options are non-empty
then they processed in following order:
* grab overwrite grab_config
* grab_config overwrite url
Args:
:param name: name of the task. After successful network operation
task's result will be passed to `task_<name>` method.
:param url: URL of network document. Any task requires `url` or
`grab` option to be specified.
:param grab: configured `Grab` instance. You can use that option in
case when `url` option is not enough. Do not forget to
configure `url` option of `Grab` instance because in this case
the `url` option of `Task` constructor will be overwritten
with `grab.config['url']`.
:param priority: - priority of the Task. Tasks with lower priority
will be processed earlier. By default each new task is assigned
with random priority from (80, 100) range.
:param priority_set_explicitly: - internal flag which tells if that
task priority was assigned manually or generated by spider
according to priority generation rules.
:param network_try_count: you'll probably will not need to use it.
It is used internally to control how many times this task was
restarted due to network errors. The `Spider` instance has
`network_try_limit` option. When `network_try_count` attribute
of the task exceeds the `network_try_limit` attribute then
processing of the task is abandoned.
:param task_try_count: the as `network_try_count` but it increased
only then you use `clone` method. Also you can set it manually.
It is useful if you want to restart the task after it was
cancelled due to multiple network errors. As you might guessed
there is `task_try_limit` option in `Spider` instance. Both
options `network_try_count` and `network_try_limit` guarantee
you that you'll not get infinite loop of restarting some task.
:param valid_status: extra status codes which counts as valid
:param use_proxylist: it means to use proxylist which was
configured via `setup_proxylist` method of spider
:param delay: if specified tells the spider to schedule the task
and execute it after `delay` seconds
:param raw: if `raw` is True then the network response is
forwarding to the corresponding handler without any check of
HTTP status code of network error, if `raw` is False (by
default) then failed response is putting back to task queue or
if tries limit is reached then the processing of this request
is finished.
:param callback: if you pass some function in `callback` option
then the network response will be passed to this callback and
the usual 'task_*' handler will be ignored and no error will be
raised if such 'task_*' handler does not exist.
:param fallback_name: the name of method that is called when spider
gives up to do the task (due to multiple network errors)
Any non-standard named arguments passed to `Task` constructor will
be saved as attributes of the object. You can get their values
later as attributes or with `get` method which allows to use
default value if attribute does not exist.
"""
if disable_cache or refresh_cache or cache_timeout:
raise_feature_is_deprecated('Cache feature')
if name == 'generator':
# The name "generator" is restricted because
# `task_generator` handler could not be created because
# this name is already used for special method which
# generates new tasks
raise SpiderMisuseError('Task name could not be "generator"')
self.name = name
if url is None and grab is None and grab_config is None:
raise SpiderMisuseError('Either url, grab or grab_config argument '
'of Task constructor should not be None')
if url is not None and grab is not None:
raise SpiderMisuseError('Options url and grab could not be used '
'together')
if url is not None and grab_config is not None:
raise SpiderMisuseError('Options url and grab_config could not be '
'used together')
if grab is not None and grab_config is not None:
raise SpiderMisuseError(
'Options grab and grab_config could not be used together')
if grab:
self.setup_grab_config(grab.dump_config())
elif grab_config:
self.setup_grab_config(grab_config)
else:
self.grab_config = None
self.url = url
if valid_status is None:
self.valid_status = []
else:
self.valid_status = valid_status
self.process_delay_option(delay)
self.fallback_name = fallback_name
self.priority_set_explicitly = priority_set_explicitly
self.priority = priority
self.network_try_count = network_try_count
self.task_try_count = task_try_count
self.use_proxylist = use_proxylist
self.raw = raw
self.callback = callback
self.coroutines_stack = []
for key, value in kwargs.items():
setattr(self, key, value)
def main(spider_name,
thread_number=None,
settings_module='settings',
network_logs=False,
disable_proxy=False,
ignore_lock=False,
disable_report=False,
api_port=None,
parser_pool_size=2,
grab_log_file=None,
network_log_file=None,
network_service=None,
grab_transport=None,
**kwargs): # pylint: disable=unused-argument
default_logging(
grab_log=grab_log_file,
network_log=network_log_file,
propagate_network_logger=network_logs,
)
root_config = build_root_config(settings_module)
spider_class = load_spider_class(root_config, spider_name)
spider_config = build_spider_config(spider_class, root_config)
spider_args = None
if hasattr(spider_class, 'setup_arg_parser'):
parser = ArgumentParser()
spider_class.setup_arg_parser(parser)
opts, _ = parser.parse_known_args()
spider_args = vars(opts)
bot = spider_class(
thread_number=thread_number,
config=spider_config,
network_try_limit=None,
task_try_limit=None,
args=spider_args,
http_api_port=api_port,
parser_pool_size=parser_pool_size,
network_service=network_service,
grab_transport=grab_transport,
)
opt_queue = spider_config.get('queue')
if opt_queue:
bot.setup_queue(**opt_queue)
if spider_config.get('cache'):
raise_feature_is_deprecated('Cache feature')
opt_proxy_list = spider_config.get('proxy_list')
if opt_proxy_list:
if disable_proxy:
logger.debug('Proxy servers disabled via command line')
else:
bot.load_proxylist(**opt_proxy_list)
opt_ifaces = spider_config.get('command_interfaces')
if opt_ifaces:
for iface_config in opt_ifaces:
bot.controller.add_interface(**iface_config)
try:
bot.run()
except KeyboardInterrupt:
pass
stats = bot.render_stats()
if spider_config.get('display_stats'):
logger.debug(stats)
pid = os.getpid()
logger.debug('Spider pid is %d', pid)
if not disable_report:
if spider_config.get('save_report'):
for subdir in (str(pid), 'last'):
dir_ = 'var/%s' % subdir
if not os.path.exists(dir_):
os.makedirs(dir_)
else:
clear_directory(dir_)
for key, lst in bot.stat.collections.items():
fname_key = key.replace('-', '_')
save_list(lst, '%s/%s.txt' % (dir_, fname_key))
with open('%s/report.txt' % dir_, 'wb') as out:
out.write(make_str(stats))
return {
'spider_stats': bot.render_stats(),
}