def __parse_response(self, response, spider):
"""处理响应,将新的请求放入请求队列,提取到的item抛给输入管理器处理"""
try:
callback = getattr(spider, response.request.callback)
for each in callback(response):
if isinstance(each, req.Request): # 将新产生的请求添加至队列
RequestQueue.push(each, spider)
elif isinstance(each, dict): # 将提取到的item抛给op_manager进行处理
self.op_manager(each, spider)
except Exception as err:
self.__save_mode(response.request, err)
def __init__(self, name, params, request_runner):
FMSBot.__init__(self, name)
RequestQueue.__init__(self, request_runner)
self.ctx = None
self.applier = None
self.params = params.copy()
self.ui_ = None
self.repo = None
self.trust = None
self.update_sm = None
# Why doesn't the base class ctr do this?
request_runner.add_queue(self)
def __init__(self, name, params, request_runner):
FMSBot.__init__(self, name)
RequestQueue.__init__(self, request_runner)
self.ctx = None
self.applier = None
self.params = params.copy()
self.ui_ = None
self.repo = None
self.trust = None
self.update_sm = None
# Why doesn't the base class ctr do this?
request_runner.add_queue(self)
def __roll_request(self):
"""不断从请求列表中提取请求进行处理(发送请求,解析响应)"""
while 1:
try:
request = RequestQueue.pop()
except EmptyError:
# 请求队列为空, 退出程序
gevent.killall(self.jobs)
try:
# 递出请求
response = self.__throw_request(request, request.spider)
except NoResponseError as err:
logger.debug('No response, reason: %s', err)
if 'duplicate url' in err.message:
RequestQueue.del_requesting(request)
else:
# 解析响应
self.__parse_response(response, request.spider)
RequestQueue.del_requesting(request)
def __init_requests(self):
for spider in self.spiders:
for request in spider.start_requests():
RequestQueue.push(request, spider)
def __init__(self, runner, ctx):
RequestQueue.__init__(self, runner)
StateMachine.__init__(self)
self.ctx = None
self.set_context(ctx) # Do early. States might depend on ctx.
self.states = {
QUIESCENT:Quiescent(self, QUIESCENT),
# Justing inverting
INVERTING_URI:InvertingUri(self, INVERTING_URI,
QUIESCENT,
FAILING),
# Requesting previous graph in order to do insert.
INVERTING_URI_4_INSERT:InvertingUri(self, INVERTING_URI_4_INSERT,
REQUESTING_URI_4_INSERT,
FAILING),
REQUESTING_URI_4_INSERT:RequestingUri(self,
REQUESTING_URI_4_INSERT,
REQUESTING_GRAPH,
FAILING),
REQUESTING_GRAPH:RequestingGraph(self, REQUESTING_GRAPH,
INSERTING_BUNDLES,
FAILING),
# Inserting
INSERTING_BUNDLES:InsertingBundles(self,
INSERTING_BUNDLES),
INSERTING_GRAPH:InsertingGraph(self, INSERTING_GRAPH,
INSERTING_URI,
FAILING),
INSERTING_URI:InsertingUri(self,INSERTING_URI,
FINISHING,
FAILING),
CANCELING:CleaningUp(self, CANCELING, QUIESCENT),
FAILING:CleaningUp(self, FAILING, QUIESCENT),
# Requesting
REQUESTING_URI:RequestingUri(self, REQUESTING_URI,
REQUESTING_BUNDLES,
FAILING),
REQUESTING_BUNDLES:RequestingBundles(self, REQUESTING_BUNDLES,
FINISHING,
FAILING),
FINISHING:CleaningUp(self, FINISHING, QUIESCENT),
# Requesting head info from freenet
REQUESTING_URI_4_HEADS:RequestingUri(self, REQUESTING_URI_4_HEADS,
REQUIRES_GRAPH_4_HEADS,
FAILING),
REQUIRES_GRAPH_4_HEADS:RequiresGraph(self, REQUIRES_GRAPH_4_HEADS,
REQUESTING_GRAPH_4_HEADS,
FINISHING),
REQUESTING_GRAPH_4_HEADS:RequestingGraph(self,
REQUESTING_GRAPH_4_HEADS,
FINISHING,
FAILING),
# Run and arbitrary StatefulRequest.
RUNNING_SINGLE_REQUEST:RunningSingleRequest(self,
RUNNING_SINGLE_REQUEST,
FINISHING,
FAILING),
# Copying.
# This doesn't verify that the graph chk(s) are fetchable.
REQUESTING_URI_4_COPY:RequestingUri(self, REQUESTING_URI_4_COPY,
INSERTING_URI,
FAILING),
}
self.current_state = self.get_state(QUIESCENT)
self.params = {}
# Must not change any state!
self.monitor_callback = lambda parent, client, msg: None
runner.add_queue(self)
def __init__(self, runner, ctx):
RequestQueue.__init__(self, runner)
StateMachine.__init__(self)
self.ctx = None
self.set_context(ctx) # Do early. States might depend on ctx.
self.states = {
QUIESCENT:
Quiescent(self, QUIESCENT),
# Justing inverting
INVERTING_URI:
InvertingUri(self, INVERTING_URI, QUIESCENT, FAILING),
# Requesting previous graph in order to do insert.
INVERTING_URI_4_INSERT:
InvertingUri(self, INVERTING_URI_4_INSERT, REQUESTING_URI_4_INSERT,
FAILING),
REQUESTING_URI_4_INSERT:
RequestingUri(self, REQUESTING_URI_4_INSERT, REQUESTING_GRAPH,
FAILING),
REQUESTING_GRAPH:
RequestingGraph(self, REQUESTING_GRAPH, INSERTING_BUNDLES,
FAILING),
# Inserting
INSERTING_BUNDLES:
InsertingBundles(self, INSERTING_BUNDLES),
INSERTING_GRAPH:
InsertingGraph(self, INSERTING_GRAPH, INSERTING_URI, FAILING),
INSERTING_URI:
InsertingUri(self, INSERTING_URI, FINISHING, FAILING),
CANCELING:
CleaningUp(self, CANCELING, QUIESCENT),
FAILING:
CleaningUp(self, FAILING, QUIESCENT),
# Requesting
REQUESTING_URI:
RequestingUri(self, REQUESTING_URI, REQUESTING_BUNDLES, FAILING),
REQUESTING_BUNDLES:
RequestingBundles(self, REQUESTING_BUNDLES, FINISHING, FAILING),
FINISHING:
CleaningUp(self, FINISHING, QUIESCENT),
# Requesting head info from freenet
REQUESTING_URI_4_HEADS:
RequestingUri(self, REQUESTING_URI_4_HEADS, REQUIRES_GRAPH_4_HEADS,
FAILING),
REQUIRES_GRAPH_4_HEADS:
RequiresGraph(self, REQUIRES_GRAPH_4_HEADS,
REQUESTING_GRAPH_4_HEADS, FINISHING),
REQUESTING_GRAPH_4_HEADS:
RequestingGraph(self, REQUESTING_GRAPH_4_HEADS, FINISHING,
FAILING),
# Run and arbitrary StatefulRequest.
RUNNING_SINGLE_REQUEST:
RunningSingleRequest(self, RUNNING_SINGLE_REQUEST, FINISHING,
FAILING),
# Copying.
# This doesn't verify that the graph chk(s) are fetchable.
REQUESTING_URI_4_COPY:
RequestingUri(self, REQUESTING_URI_4_COPY, INSERTING_URI, FAILING),
}
self.current_state = self.get_state(QUIESCENT)
self.params = {}
# Must not change any state!
self.monitor_callback = lambda parent, client, msg: None
runner.add_queue(self)
