runScrape.py

#!flask/bin/python

from WebMirror.Runner import PROCESSES
from settings import MAX_DB_SESSIONS
MAX_DB_SESSIONS = PROCESSES + 5

if __name__ == "__main__":
	import logSetup
	logSetup.initLogging()


# This HAS to be included before the app, to prevent circular dependencies.
import WebMirror.runtime_engines

import WebMirror.Runner
import WebMirror.rules
import traceback
import datetime
import config
import WebMirror.database as db
import WebMirror.LogBase as LogBase
import sqlalchemy.exc

from apscheduler.schedulers.background import BackgroundScheduler
from apscheduler.schedulers.blocking import BlockingScheduler
from apscheduler.executors.pool import ProcessPoolExecutor
from apscheduler.jobstores.sqlalchemy import SQLAlchemyJobStore

import activePlugins

executors = {
	'main_jobstore': ProcessPoolExecutor(10),
}
job_defaults = {
	'coalesce': True,
	'max_instances': 3,
}

SQLALCHEMY_DATABASE_URI = 'postgresql://{user}:{passwd}@{host}:5432/{database}'.format(user=config.C_DATABASE_USER, passwd=config.C_DATABASE_PASS, host=config.C_DATABASE_IP, database=config.C_DATABASE_DB_NAME)

jobstores = {
	'main_jobstore'      : SQLAlchemyJobStore(url=SQLALCHEMY_DATABASE_URI)

}


CALLABLE_LUT = {}
for item, dummy_interval in activePlugins.scrapePlugins.values():
	CALLABLE_LUT[item.__name__] = item


class JobNameException(Exception):
	pass

class JobCaller(LogBase.LoggerMixin):

	loggerPath = "Main.PluginRunner"

	def __init__(self, job_name):

		if not job_name in CALLABLE_LUT:
			raise JobNameException("Callable '%s' is not in the class lookup table: '%s'!" % (job_name, CALLABLE_LUT))
		self.runModule = CALLABLE_LUT[job_name]
		self.job_name = job_name


		session = db.get_session()

		try:
			query = session.query(db.PluginStatus).filter(db.PluginStatus.plugin_name==job_name)
			have = query.scalar()
			if not have:
				new = db.PluginStatus(plugin_name=job_name)
				session.add(new)
				session.commit()
		except sqlalchemy.exc.OperationalError:
			session.rollback()
		except sqlalchemy.exc.InvalidRequestError:
			session.rollback()

		finally:
			db.delete_session()

	def doCall(self):

		session = db.get_session()
		item = session.query(db.PluginStatus).filter(db.PluginStatus.plugin_name==self.job_name).one()
		if item.is_running:
			session.commit()
			self.log.error("Plugin %s is already running! Not doing re-entrant call!", self.job_name)
			return

		item.is_running = True
		item.last_run = datetime.datetime.now()
		session.commit()

		try:
			self._doCall()
		except Exception:
			item.last_error      = datetime.datetime.now()
			item.last_error_msg  = traceback.format_exc()
			raise
		finally:
			item.is_running = False
			item.last_run_end = datetime.datetime.now()
			session.commit()

			db.delete_session()

	# Should probably be a lambda? Laaaazy.
	def _doCall(self):
		instance = self.runModule()
		instance._go()

	@classmethod
	def callMod(cls, passMod):
		mod = cls(passMod)
		mod.doCall()

def do_call(job_name):
	caller = JobCaller(job_name)

	while 1:
		try:
			caller.doCall()
			break
		except JobNameException:
			print("Error! Invalid job name: '%s'!" % job_name)
			break
		except AttributeError:
			traceback.print_exc()
			print("Call error!")

def scheduleJobs(sched, timeToStart):

	jobs = []
	offset = 0
	for key, value in activePlugins.scrapePlugins.items():
		baseModule, interval = value
		jobs.append((key, baseModule, interval, timeToStart+datetime.timedelta(seconds=60*offset)))
		offset += 1

	activeJobs = []

	print("JobCaller: ", JobCaller)
	print("JobCaller.callMod: ", JobCaller.callMod)

	for jobId, callee, interval, startWhen in jobs:
		jId = callee.__name__
		print("JobID = ", jId)
		activeJobs.append(jId)
		if not sched.get_job(jId):
			sched.add_job(do_call,
						args               = (callee.__name__, ),
						trigger            = 'interval',
						seconds            = interval,
						start_date         = startWhen,
						id                 = jId,
						max_instances      =  1,
						replace_existing   = True,
						jobstore           = "main_jobstore",
						misfire_grace_time = 2**30)


	for job in sched.get_jobs('main_jobstore'):
		if not job.id in activeJobs:
			sched.remove_job(job.id, 'main_jobstore')

	print("JobSetup call resetting run-states!")
	session = db.get_session()
	session.query(db.PluginStatus).update({db.PluginStatus.is_running : False})
	session.commit()
	db.delete_session()

	print("Run-states reset.")


def go_sched():

	sched = BlockingScheduler(jobstores=jobstores, executors=executors, job_defaults=job_defaults)

	startTime = datetime.datetime.now()+datetime.timedelta(seconds=10)
	scheduleJobs(sched, startTime)
	sched.start()

def go_test():
	import WebMirror.Engine

	engine = WebMirror.Engine.SiteArchiver(None)
	print(engine)
	for x in range(100):
		engine.getTask()

def go():

	rules = WebMirror.rules.load_rules()
	WebMirror.Runner.initializeStartUrls(rules)
	WebMirror.Runner.resetInProgress()
	runner = WebMirror.Runner.Crawler()
	runner.run()

	# print("Thread halted. App exiting.")

def profile():
	import cProfile
	import pstats
	cProfile.run('go()', "run.stats")
	p = pstats.Stats("run.stats")
	p.sort_stats('cumulative')
	p.print_stats(250)

if __name__ == "__main__":
	import sys
	print("Auxilliary modes: 'test', 'scheduler'.")
	if "scheduler" in sys.argv:
		MAX_DB_SESSIONS = 4
		go_sched()
	if "test" in sys.argv:
		go_test()
	else:

		started = False
		if not started:
			started = True

			go()