def project_create():
project_name = request.form['project_name']
project = Project()
project.project_name = project_name
db.session.add(project)
db.session.commit()
return redirect("/project/%s/spider/deploy" % project.id, code=302)
def post(self):
project_name = request.form['project_name']
project = Project()
project.project_name = project_name
db.session.add(project)
db.session.commit()
return project.to_dict()
def get_project_list(self):
data = request("get", self._scrapyd_url() + "/listprojects.json", return_type="json")
result = []
if data:
for project_name in data['projects']:
project = Project()
project.project_name = project_name
result.append(project)
return result
def job_add(project_id):
project = Project.find_project_by_id(project_id)
job_instance = JobInstance()
job_instance.spider_name = request.form['spider_name']
job_instance.project_id = project_id
job_instance.spider_arguments = request.form['spider_arguments']
job_instance.priority = request.form.get('priority', 0)
job_instance.run_type = request.form['run_type']
# chose daemon manually
if request.form['daemon'] != 'auto':
spider_args = []
if request.form['spider_arguments']:
spider_args = request.form['spider_arguments'].split(",")
spider_args.append("daemon={}".format(request.form['daemon']))
job_instance.spider_arguments = ','.join(spider_args)
if job_instance.run_type == JobRunType.ONETIME:
job_instance.enabled = -1
db.session.add(job_instance)
db.session.commit()
agent.start_spider(job_instance)
if job_instance.run_type == JobRunType.PERIODIC:
job_instance.cron_minutes = request.form.get('cron_minutes') or '0'
job_instance.cron_hour = request.form.get('cron_hour') or '*'
job_instance.cron_day_of_month = request.form.get(
'cron_day_of_month') or '*'
job_instance.cron_day_of_week = request.form.get(
'cron_day_of_week') or '*'
job_instance.cron_month = request.form.get('cron_month') or '*'
# set cron exp manually
if request.form.get('cron_exp'):
job_instance.cron_minutes, job_instance.cron_hour, job_instance.cron_day_of_month, job_instance.cron_day_of_week, job_instance.cron_month = \
request.form['cron_exp'].split(' ')
db.session.add(job_instance)
db.session.commit()
return redirect(request.referrer, code=302)
def get(self, project_id):
project = Project.find_project_by_id(project_id)
return [
spider_instance.to_dict()
for spider_instance in SpiderInstance.query.filter_by(
project_id=project_id).all()
]
def job_periodic(project_id):
project = Project.find_project_by_id(project_id)
job_instance_list = [
job_instance.to_dict() for job_instance in JobInstance.query.filter_by(
run_type="periodic", project_id=project_id).all()
]
return render_template("job_periodic.html",
job_instance_list=job_instance_list)
def log_url(self, job_execution):
job_instance = JobInstance.find_job_instance_by_id(
job_execution.job_instance_id)
project = Project.find_project_by_id(job_instance.project_id)
for spider_service_instance in self.spider_service_instances:
if spider_service_instance.server == job_execution.running_on:
return spider_service_instance.log_url(
project.project_name, job_instance.spider_name,
job_execution.service_job_execution_id)
def cancel_spider(self, job_execution):
job_instance = JobInstance.find_job_instance_by_id(
job_execution.job_instance_id)
project = Project.find_project_by_id(job_instance.project_id)
for spider_service_instance in self.spider_service_instances:
if spider_service_instance.server == job_execution.running_on:
if spider_service_instance.cancel_spider(
project.project_name,
job_execution.service_job_execution_id):
job_execution.end_time = datetime.datetime.now()
job_execution.running_status = SpiderStatus.CANCELED
db.session.commit()
break
def inject_project():
project_context = {}
project_context['project_list'] = Project.query.all()
if project_context['project_list'] and (not session.get('project_id')):
project = Project.query.first()
session['project_id'] = project.id
if session.get('project_id'):
project_context['project'] = Project.find_project_by_id(
session['project_id'])
project_context['spider_list'] = [
spider_instance.to_dict()
for spider_instance in SpiderInstance.query.filter_by(
project_id=session['project_id']).all()
]
else:
project_context['project'] = {}
return project_context
def spider_egg_upload(project_id):
project = Project.find_project_by_id(project_id)
if 'file' not in request.files:
flash('No file part')
return redirect(request.referrer)
file = request.files['file']
# if user does not select file, browser also
# submit a empty part without filename
if file.filename == '':
flash('No selected file')
return redirect(request.referrer)
if file:
filename = secure_filename(file.filename)
dst = os.path.join(tempfile.gettempdir(), filename)
file.save(dst)
agent.deploy(project, dst)
flash('deploy success!')
return redirect(request.referrer)
def start_spider(self, job_instance):
project = Project.find_project_by_id(job_instance.project_id)
spider_name = job_instance.spider_name
#arguments = {}
#if job_instance.spider_arguments:
# arguments = dict(map(lambda x: x.split("="), job_instance.spider_arguments.split(",")))
from collections import defaultdict
arguments = defaultdict(list)
if job_instance.spider_arguments:
for k, v in list(
map(lambda x: x.split('=', 1),
job_instance.spider_arguments.split(','))):
arguments[k].append(v)
threshold = 0
daemon_size = len(self.spider_service_instances)
if job_instance.priority == JobPriority.HIGH:
threshold = int(daemon_size / 2)
if job_instance.priority == JobPriority.HIGHEST:
threshold = int(daemon_size)
threshold = 1 if threshold == 0 else threshold
candidates = self.spider_service_instances
leaders = []
if 'daemon' in arguments:
for candidate in candidates:
if candidate.server == arguments['daemon']:
leaders = [candidate]
else:
# TODO optimize some better func to vote the leader
for i in range(threshold):
leaders.append(random.choice(candidates))
for leader in leaders:
serviec_job_id = leader.start_spider(project.project_name,
spider_name, arguments)
job_execution = JobExecution()
job_execution.project_id = job_instance.project_id
job_execution.service_job_execution_id = serviec_job_id
job_execution.job_instance_id = job_instance.id
job_execution.create_time = datetime.datetime.now()
job_execution.running_on = leader.server
db.session.add(job_execution)
db.session.commit()
def service_stats(project_id):
project = Project.find_project_by_id(project_id)
run_stats = JobExecution.list_run_stats_by_hours(project_id)
return render_template("server_stats.html", run_stats=run_stats)
def spider_deploy(project_id):
project = Project.find_project_by_id(project_id)
return render_template("spider_deploy.html")
def project_delete(project_id):
project = Project.find_project_by_id(project_id)
agent.delete_project(project)
db.session.delete(project)
db.session.commit()
return redirect("/project/manage", code=302)
def get_project_list(self):
project_list = self.spider_service_instances[0].get_project_list()
Project.load_project(project_list)
return [project.to_dict() for project in Project.query.all()]