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_month, job_instance.cron_day_of_week = \
request.form['cron_exp'].split(' ')
db.session.add(job_instance)
db.session.commit()
return redirect(request.referrer, code=302)
def job_back_in_time(project_id):
if not config.BACK_IN_TIME_ENABLED:
return redirect(request.referrer, code=302)
spider_names = request.form.getlist('spider_name')
for spider in spider_names:
job_instance = JobInstance()
job_instance.project_id = project_id
job_instance.spider_name = spider
spider_args = request.form['spider_arguments'].split(",")
spider_args.append("--callback={}".format(request.form['callback']))
spider_args.append("SCRAPY_PROJECT=SCRAPY_PROJECT")
job_instance.spider_arguments = ','.join(spider_args)
job_instance.priority = request.form.get('priority', 0)
job_instance.run_type = JobRunType.ONETIME
job_instance.overlapping = True
# 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)
job_instance.enabled = -1
db.session.add(job_instance)
try:
db.session.commit()
except Exception as e:
db.session.rollback()
raise e
agent.run_back_in_time(job_instance)
return redirect(request.referrer, code=302)
def post(self, project_id):
post_data = request.form
if post_data:
job_instance = JobInstance()
job_instance.spider_name = post_data['spider_name']
job_instance.project_id = project_id
job_instance.spider_arguments = post_data.get('spider_arguments')
job_instance.desc = post_data.get('desc')
job_instance.tags = post_data.get('tags')
job_instance.run_type = post_data['run_type']
job_instance.priority = post_data.get('priority', 0)
if job_instance.run_type == "periodic":
job_instance.cron_minutes = post_data.get(
'cron_minutes') or '0'
job_instance.cron_hour = post_data.get('cron_hour') or '*'
job_instance.cron_day_of_month = post_data.get(
'cron_day_of_month') or '*'
job_instance.cron_day_of_week = post_data.get(
'cron_day_of_week') or '*'
job_instance.cron_month = post_data.get('cron_month') or '*'
job_instance.cron_month = post_data.get('cron_month') or '*'
job_instance.cron_month = post_data.get('cron_month') or '*'
job_instance.start_tasks = post_data.get('start_tasks', 1)
job_instance.max_start_tasks = post_data.get(
'max_start_tasks', 1)
db.session.add(job_instance)
db.session.commit()
return True
def _run_spider(spider_name, project_id):
"""
Run a spider
:param spider_name:
:param project_id:
:return:
"""
job_instance = JobInstance()
job_instance.project_id = project_id
job_instance.spider_name = spider_name
job_instance.priority = JobPriority.NORMAL
job_instance.run_type = JobRunType.ONETIME
job_instance.overlapping = True
job_instance.enabled = -1
# settings for tempering the requests
throttle_value = _get_throttle_value(spider_name, project_id)
job_instance.spider_arguments = "setting=AUTOTHROTTLE_TARGET_CONCURRENCY={}".format(
throttle_value)
job_instance.throttle_concurrency = throttle_value
db.session.add(job_instance)
try:
db.session.commit()
except Exception as e:
db.session.rollback()
raise e
agent.start_spider(job_instance)
def put(self, project_id, spider_id):
spider_instance = SpiderInstance.query.filter_by(project_id=project_id,
id=spider_id).first()
if not spider_instance: abort(404)
job_instance = JobInstance()
job_instance.spider_name = spider_instance.spider_name
job_instance.project_id = project_id
job_instance.spider_arguments = request.form.get('spider_arguments')
job_instance.desc = request.form.get('desc')
job_instance.tags = request.form.get('tags')
job_instance.run_type = JobRunType.ONETIME
job_instance.priority = request.form.get('priority', 0)
job_instance.enabled = -1
db.session.add(job_instance)
db.session.commit()
agent.start_spider(job_instance)
return True
def post(self, project_id):
post_data = request.form
if post_data:
job_instance = JobInstance()
job_instance.spider_name = post_data['spider_name']
job_instance.project_id = project_id
job_instance.spider_arguments = post_data.get('spider_arguments')
job_instance.desc = post_data.get('desc')
job_instance.tags = post_data.get('tags')
job_instance.run_type = post_data['run_type']
job_instance.priority = post_data.get('priority', 0)
if job_instance.run_type == "periodic":
job_instance.cron_minutes = post_data.get('cron_minutes') or '0'
job_instance.cron_hour = post_data.get('cron_hour') or '*'
job_instance.cron_day_of_month = post_data.get('cron_day_of_month') or '*'
job_instance.cron_day_of_week = post_data.get('cron_day_of_week') or '*'
job_instance.cron_month = post_data.get('cron_month') or '*'
db.session.add(job_instance)
try:
db.session.commit()
except Exception as e:
db.session.rollback()
raise e
return True