def list_spiders(request):
spiders ={}
result = endpoint.list_spider(OPTS)
spider_names = result["spiders"]
for spider_name in spider_names:
spiders[spider_name]={}
jobs = endpoint.list_jobs(OPTS)
running_jobs= jobs["running"]
pending_jobs= jobs["pending"]
finished_jobs= jobs["finished"]
for job in running_jobs:
job_in_spider = job["spider"]
job_id = job["id"]
pid = job["pid"] if job.has_key("pid") else ""
r_key = "%s:status:%s:latest"%(job_in_spider,job_id)
status_key = "%s:status:%s:status"%(job_in_spider,job_id)
status =rserver.hgetall(r_key)
redis_status =rserver.get(status_key)
if len(status)==0:
status["status"]="pending"
elif redis_status == "stopping":
status["status"]="stopping"
elif redis_status == "finished":
status["status"] == "finished"
else:
status["status"]="running"
status["pid"]=pid
spiders[job_in_spider][job_id]=status
for job in pending_jobs:
job_in_spider = job["spider"]
job_id = job["id"]
pid = job["pid"] if job.has_key("pid") else ""
r_key = "%s:status:%s:latest"%(job_in_spider,job_id)
status =rserver.hgetall(r_key)
status["status"]="pending"
status["pid"]=pid
spiders[job_in_spider][job_id]=status
for job in finished_jobs:
job_in_spider = job["spider"]
job_id = job["id"]
pid = job["pid"] if job.has_key("pid") else ""
r_key = "%s:status:%s:latest"%(job_in_spider,job_id)
status_key = "%s:status:%s:status"%(job_in_spider,job_id)
redis_status = rserver.get(status_key)
#if redis_status != "finished":
# rserver.set(status_key, "finished")
status =rserver.hgetall(r_key)
if len(status)==0:
status["status"]="pending"
else:
status["status"]="finished"
status["pid"]=pid
spiders[job_in_spider][job_id]=status
return HttpResponse(json.dumps(spiders))
def list_spiders(request):
spiders = {}
result = endpoint.list_spider(OPTS)
spider_names = result["spiders"]
for spider_name in spider_names:
spiders[spider_name] = {}
jobs = endpoint.list_jobs(OPTS)
running_jobs = jobs["running"]
pending_jobs = jobs["pending"]
finished_jobs = jobs["finished"]
for job in running_jobs:
job_in_spider = job["spider"]
job_id = job["id"]
pid = job["pid"] if job.has_key("pid") else ""
r_key = "%s:status:%s:latest" % (job_in_spider, job_id)
status_key = "%s:status:%s:status" % (job_in_spider, job_id)
status = rserver.hgetall(r_key)
redis_status = rserver.get(status_key)
if len(status) == 0:
status["status"] = "pending"
elif redis_status == "stopping":
status["status"] = "stopping"
elif redis_status == "finished":
status["status"] == "finished"
else:
status["status"] = "running"
status["pid"] = pid
spiders[job_in_spider][job_id] = status
for job in pending_jobs:
job_in_spider = job["spider"]
job_id = job["id"]
pid = job["pid"] if job.has_key("pid") else ""
r_key = "%s:status:%s:latest" % (job_in_spider, job_id)
status = rserver.hgetall(r_key)
status["status"] = "pending"
status["pid"] = pid
spiders[job_in_spider][job_id] = status
for job in finished_jobs:
job_in_spider = job["spider"]
job_id = job["id"]
pid = job["pid"] if job.has_key("pid") else ""
r_key = "%s:status:%s:latest" % (job_in_spider, job_id)
status_key = "%s:status:%s:status" % (job_in_spider, job_id)
redis_status = rserver.get(status_key)
#if redis_status != "finished":
# rserver.set(status_key, "finished")
status = rserver.hgetall(r_key)
if len(status) == 0:
status["status"] = "pending"
else:
status["status"] = "finished"
status["pid"] = pid
spiders[job_in_spider][job_id] = status
return HttpResponse(json.dumps(spiders))
def new_crawl(request):
name = request.GET.get("name")
if name==None:
status = {"status":"Error","message":"invalid name"}
return HttpResponse(status)
spider_counter= 0
jobs = endpoint.list_jobs(OPTS)
running_jobs= jobs["running"]
for job in running_jobs:
job_in_spider = job["spider"]
if job_in_spider==name:
spider_counter+=1
if spider_counter>=MAX_INST:
status={"status":"Error","message":"maximum instances! can not fork any more!"}
return HttpResponse(json.dumps(status))
result =endpoint.new_crawl(name,OPTS)
return HttpResponse(json.dumps(result))
def new_crawl(request):
name = request.GET.get("name")
if name == None:
status = {"status": "Error", "message": "invalid name"}
return HttpResponse(status)
spider_counter = 0
jobs = endpoint.list_jobs(OPTS)
running_jobs = jobs["running"]
for job in running_jobs:
job_in_spider = job["spider"]
if job_in_spider == name:
spider_counter += 1
if spider_counter >= MAX_INST:
status = {
"status": "Error",
"message": "maximum instances! can not fork any more!"
}
return HttpResponse(json.dumps(status))
result = endpoint.new_crawl(name, OPTS)
return HttpResponse(json.dumps(result))
