def __init__(self, clawer_task, clawer_setting):
from clawer.models import ClawerDownloadLog, RealTimeMonitor
self.clawer_task = clawer_task
self.download_log = ClawerDownloadLog(clawer=clawer_task.clawer,
task=clawer_task,
hostname=socket.gethostname())
self.monitor = RealTimeMonitor()
self.background_queue = BackgroundQueue()
self.headers = {
"user-agent":
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10.10; rv:40.0) Gecko/20100101 Firefox/40.0"
}
self.sentry = SentryClient()
self.clawer_setting = clawer_setting
self.downloader = Download(self.clawer_task.uri,
engine=self.clawer_setting.download_engine,
js=self.clawer_setting.download_js)
if self.clawer_setting.proxy:
self.downloader.add_proxies(
self.clawer_setting.proxy.strip().split("\n"))
if self.clawer_task.cookie:
self.headers["cookie"] = self.clawer_task.cookie
self.downloader.add_cookie(self.clawer_task.cookie)
if self.clawer_setting.cookie:
self.headers["cookie"] = self.clawer_task.cookie
self.downloader.add_cookie(self.clawer_task.cookie)
self.downloader.add_headers(self.headers)
def run():
clawers = Clawer.objects.filter(status=Clawer.STATUS_ON).all()
monitor = RealTimeMonitor()
download_queue = DownloadQueue()
for clawer in clawers:
clawer_settting = clawer.cached_settings()
queue_name = clawer_settting.prior_to_queue_name()
# clawer_tasks = ClawerTask.objects.filter(clawer_id=clawer.id, status=ClawerTask.STATUS_LIVE).order_by("id")[:clawer_settting.dispatch]
# 不按照id排序
clawer_tasks = ClawerTask.objects.filter(
clawer_id=clawer.id,
status=ClawerTask.STATUS_LIVE)[:clawer_settting.dispatch]
for item in clawer_tasks:
if not download_queue.enqueue(queue_name,
download_clawer_task,
args=[item, clawer_settting]):
break
item.status = ClawerTask.STATUS_PROCESS
item.save()
#trace it
monitor.trace_task_status(item)
print "clawer is %d, job count %d, queue name %s" % (
clawer.id, len(download_queue.jobs), queue_name)
return download_queue
def task_stat(request):
result = {"is_ok":True, "status":[], "series":[], "xAxis":[]}
monitor = RealTimeMonitor()
for (status, name) in ClawerTask.STATUS_CHOICES:
result["status"].append(name)
remote_data = monitor.load_task_stat(status)
dts = sorted(remote_data["data"].keys())
if result["xAxis"] == []:
result["xAxis"] = [x.strftime("%d %H:%M") for x in dts]
serie = [remote_data["data"][x]["count"] for x in dts]
result["series"].append(serie)
return result
def __init__(self, clawer, clawer_task):
from clawer.models import RealTimeMonitor, ClawerAnalysisLog
self.clawer = clawer
self.clawer_task = clawer_task
self.monitor = RealTimeMonitor()
self.background_queue = BackgroundQueue()
self.hostname = socket.gethostname()[:16]
self.runing_analysis = self.clawer.runing_analysis()
self.analysis_log = ClawerAnalysisLog(clawer=self.clawer, task=self.clawer_task, hostname=self.hostname, analysis=self.runing_analysis)
def test_task_stat(self):
clawer = Clawer.objects.create(name="hi", info="good")
clawer_generator = ClawerTaskGenerator.objects.create(
clawer=clawer,
code="print hello",
cron="*",
status=ClawerTaskGenerator.STATUS_PRODUCT)
clawer_task = ClawerTask.objects.create(
clawer=clawer,
task_generator=clawer_generator,
uri="http://github.com",
status=ClawerTask.STATUS_FAIL)
monitor = RealTimeMonitor()
monitor.trace_task_status(clawer_task)
monitor.trace_task_status(clawer_task)
monitor.trace_task_status(clawer_task)
url = reverse("clawer.apis.monitor.task_stat")
resp = self.logined_client.get(url)
result = json.loads(resp.content)
self.assertTrue(result["is_ok"])
clawer.delete()
clawer_generator.delete()
clawer_task.delete()
def test_trace(self):
clawer = Clawer.objects.create(name="hi", info="good")
clawer_generator = ClawerTaskGenerator.objects.create(
clawer=clawer,
code="print hello",
cron="*",
status=ClawerTaskGenerator.STATUS_PRODUCT)
clawer_task = ClawerTask.objects.create(
clawer=clawer,
task_generator=clawer_generator,
uri="http://github.com",
status=ClawerTask.STATUS_FAIL)
monitor = RealTimeMonitor()
monitor.trace_task_status(clawer_task)
monitor.trace_task_status(clawer_task)
result = monitor.trace_task_status(clawer_task)
#print result
self.assertEqual(len(result["data"]), monitor.POINT_COUNT)
clawer.delete()
clawer_generator.delete()
clawer_task.delete()
def __init__(self, task_generator):
from clawer.models import ClawerGenerateLog, RealTimeMonitor
self.task_generator = task_generator
self.clawer = self.task_generator.clawer
self.out_path = "/tmp/task_generator_%d" % self.task_generator.id
self.monitor = RealTimeMonitor()
self.hostname = socket.gethostname()[:16]
self.generate_log = ClawerGenerateLog(clawer=self.clawer, task_generator=self.task_generator, hostname=self.hostname)
self.start_time = time.time()
self.end_time = None
self.content_bytes = 0
self.url_cache = UrlCache()
class DownloadClawerTask(object):
def __init__(self, clawer_task, clawer_setting):
from clawer.models import ClawerDownloadLog, RealTimeMonitor
self.clawer_task = clawer_task
self.download_log = ClawerDownloadLog(clawer=clawer_task.clawer,
task=clawer_task,
hostname=socket.gethostname())
self.monitor = RealTimeMonitor()
self.background_queue = BackgroundQueue()
self.headers = {
"user-agent":
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10.10; rv:40.0) Gecko/20100101 Firefox/40.0"
}
self.sentry = SentryClient()
self.clawer_setting = clawer_setting
self.downloader = Download(self.clawer_task.uri,
engine=self.clawer_setting.download_engine,
js=self.clawer_setting.download_js)
if self.clawer_setting.proxy:
self.downloader.add_proxies(
self.clawer_setting.proxy.strip().split("\n"))
if self.clawer_task.cookie:
self.headers["cookie"] = self.clawer_task.cookie
self.downloader.add_cookie(self.clawer_task.cookie)
if self.clawer_setting.cookie:
self.headers["cookie"] = self.clawer_task.cookie
self.downloader.add_cookie(self.clawer_task.cookie)
self.downloader.add_headers(self.headers)
def download(self):
from clawer.models import ClawerTask, ClawerDownloadLog
if not self.clawer_task.status in [
ClawerTask.STATUS_LIVE, ClawerTask.STATUS_PROCESS
]:
return 0
failed = False
self.downloader.download()
if self.downloader.failed:
self.download_failed()
return 0
#save
try:
path = self.clawer_task.store_path()
if os.path.exists(os.path.dirname(path)) is False:
os.makedirs(os.path.dirname(path), 0775)
with open(path, "w") as f:
content = self.downloader.content
if isinstance(content, types.UnicodeType):
content = content.encode("utf-8")
f.write(content)
self.clawer_task.store = path
except:
failed = True
self.download_log.failed_reason = traceback.format_exc(10)
self.sentry.capture()
if failed:
self.download_failed()
return 0
#success handle
self.clawer_task.status = ClawerTask.STATUS_SUCCESS
self.clawer_task.save()
if self.downloader.response_headers.get("content-length"):
self.download_log.content_bytes = self.downloader.response_headers[
"Content-Length"]
else:
self.download_log.content_bytes = len(self.downloader.content)
self.download_log.status = ClawerDownloadLog.STATUS_SUCCESS
self.download_log.content_encoding = self.downloader.content_encoding
self.download_log.spend_time = int(self.downloader.spend_time * 1000)
self.download_log.save()
self.monitor.trace_task_status(self.clawer_task)
return self.clawer_task.id
def download_failed(self):
from clawer.models import ClawerTask, ClawerDownloadLog
self.download_log.status = ClawerDownloadLog.STATUS_FAIL
if self.downloader.failed_exception:
self.download_log.failed_reason = self.downloader.failed_exception
self.download_log.spend_time = int(self.downloader.spend_time * 1000)
self.background_queue.enqueue(clawer_download_log_delay_save,
[self.download_log])
self.clawer_task.status = ClawerTask.STATUS_FAIL
self.background_queue.enqueue(clawer_task_delay_save,
[self.clawer_task])
self.monitor.trace_task_status(self.clawer_task)