def start(self):
buildtools.bake_css()
buildtools.bake_js()
buildtools.bake_templates()
buildtools.bake_beta_templates()
buildtools.copy_woff()
# Setup variables for the long poll module
# Bypass Tornado's forking processes for Windows machines if num_processes is set to 1
if config.get("api_num_processes") == 1 or config.get("web_developer_mode"):
self._listen(0)
else:
# The way this works, is that the parent PID is hijacked away from us and everything after this
# is a child process. As of Tornado 2.1, fork() is used, which means we do have a complete
# copy of all execution in memory up until this point and we will have complete separation of
# processes from here on out. Tornado handles child cleanup and zombification.
#
# We can have a config directive for numprocesses but it's entirely optional - a return of
# None from the config option getter (if the config didn't exist) will cause Tornado
# to spawn as many processes as there are cores on the server CPU(s).
tornado.process.fork_processes(config.get("api_num_processes"))
task_id = tornado.process.task_id()
if task_id != None:
self._listen(task_id)
def start(self):
buildtools.bake_css()
buildtools.bake_js()
buildtools.bake_templates()
buildtools.bake_beta_templates()
buildtools.copy_woff()
# Setup variables for the long poll module
# Bypass Tornado's forking processes for Windows machines if num_processes is set to 1
if config.get("api_num_processes") == 1 or config.get("web_developer_mode"):
self._listen(0)
else:
# The way this works, is that the parent PID is hijacked away from us and everything after this
# is a child process. As of Tornado 2.1, fork() is used, which means we do have a complete
# copy of all execution in memory up until this point and we will have complete separation of
# processes from here on out. Tornado handles child cleanup and zombification.
#
# We can have a config directive for numprocesses but it's entirely optional - a return of
# None from the config option getter (if the config didn't exist) will cause Tornado
# to spawn as many processes as there are cores on the server CPU(s).
tornado.process.fork_processes(config.get("api_num_processes"))
task_id = tornado.process.task_id()
if task_id != None:
self._listen(task_id)
def _listen(self, task_id):
import api_requests.sync
api_requests.sync.init()
# task_ids start at zero, so we gobble up ports starting at the base port and work up
port_no = int(config.get("api_base_port")) + task_id
pid = os.getpid()
pid_file = open(
"%s/api_%s.pid" % (config.get_directory("pid_dir"), port_no), 'w')
pid_file.write(str(pid))
pid_file.close()
# Log according to configured directory and port # we're operating on
log_file = "%s/rw_api_%s.log" % (config.get_directory("log_dir"),
port_no)
if config.test_mode and os.path.exists(log_file):
os.remove(log_file)
log.init(log_file, config.get("log_level"))
log.debug("start", "Server booting, port %s." % port_no)
db.connect()
cache.connect()
memory_trace.setup(port_no)
if config.get("web_developer_mode"):
for station_id in config.station_ids:
playlist.prepare_cooldown_algorithm(station_id)
# automatically loads every station ID and fills things in if there's no data
schedule.load()
for station_id in config.station_ids:
schedule.update_memcache(station_id)
rainwave.request.update_line(station_id)
rainwave.request.update_expire_times()
cache.set_station(station_id, "backend_ok", True)
cache.set_station(station_id, "backend_message", "OK")
cache.set_station(station_id, "get_next_socket_timeout", False)
for sid in config.station_ids:
cache.update_local_cache_for_sid(sid)
playlist.prepare_cooldown_algorithm(sid)
playlist.update_num_songs()
# If we're not in developer, remove development-related URLs
if not config.get("developer_mode"):
i = 0
while (i < len(request_classes)):
if request_classes[i][0].find("/test/") != -1:
request_classes.pop(i)
i = i - 1
i = i + 1
# Make sure all other errors get handled in an API-friendly way
request_classes.append((r"/api/.*", api.web.Error404Handler))
request_classes.append((r"/api4/.*", api.web.Error404Handler))
request_classes.append((r".*", api.web.HTMLError404Handler))
# Initialize the help (rather than it scan all URL handlers every time someone hits it)
api.help.sectionize_requests()
# Fire ze missiles!
app = tornado.web.Application(
request_classes,
debug=(config.test_mode or config.get("developer_mode")),
template_path=os.path.join(os.path.dirname(__file__),
"../templates"),
static_path=os.path.join(os.path.dirname(__file__), "../static"),
autoescape=None)
http_server = tornado.httpserver.HTTPServer(app, xheaders=True)
http_server.listen(port_no)
if config.get("api_user") and config.get("api_group"):
chuser.change_user(config.get("api_user"), config.get("api_group"))
if task_id == 0:
buildtools.bake_css()
buildtools.bake_js()
buildtools.bake_beta_js()
for request in request_classes:
log.debug("start", " Handler: %s" % str(request))
log.info("start", "API server on port %s ready to go." % port_no)
self.ioloop = tornado.ioloop.IOLoop.instance()
try:
self.ioloop.start()
finally:
self.ioloop.stop()
http_server.stop()
db.close()
log.info("stop", "Server has been shutdown.")
log.close()
def _listen(self, task_id):
import api_requests.sync
api_requests.sync.init()
# task_ids start at zero, so we gobble up ports starting at the base port and work up
port_no = int(config.get("api_base_port")) + task_id
pid = os.getpid()
pid_file = open("%s/api_%s.pid" % (config.get_directory("pid_dir"), port_no), 'w')
pid_file.write(str(pid))
pid_file.close()
# Log according to configured directory and port # we're operating on
log_file = "%s/rw_api_%s.log" % (config.get_directory("log_dir"), port_no)
if config.test_mode and os.path.exists(log_file):
os.remove(log_file)
log.init(log_file, config.get("log_level"))
log.debug("start", "Server booting, port %s." % port_no)
db.connect()
cache.connect()
memory_trace.setup(port_no)
if config.get("web_developer_mode"):
for station_id in config.station_ids:
playlist.prepare_cooldown_algorithm(station_id)
# automatically loads every station ID and fills things in if there's no data
schedule.load()
for station_id in config.station_ids:
schedule.update_memcache(station_id)
rainwave.request.update_line(station_id)
rainwave.request.update_expire_times()
cache.set_station(station_id, "backend_ok", True)
cache.set_station(station_id, "backend_message", "OK")
cache.set_station(station_id, "get_next_socket_timeout", False)
for sid in config.station_ids:
cache.update_local_cache_for_sid(sid)
playlist.prepare_cooldown_algorithm(sid)
playlist.update_num_songs()
# If we're not in developer, remove development-related URLs
if not config.get("developer_mode"):
i = 0
while (i < len(request_classes)):
if request_classes[i][0].find("/test/") != -1:
request_classes.pop(i)
i = i - 1
i = i + 1
# Make sure all other errors get handled in an API-friendly way
request_classes.append((r"/api/.*", api.web.Error404Handler))
request_classes.append((r"/api4/.*", api.web.Error404Handler))
request_classes.append((r".*", api.web.HTMLError404Handler))
# Initialize the help (rather than it scan all URL handlers every time someone hits it)
api.help.sectionize_requests()
# Fire ze missiles!
app = tornado.web.Application(request_classes,
debug=(config.test_mode or config.get("developer_mode")),
template_path=os.path.join(os.path.dirname(__file__), "../templates"),
static_path=os.path.join(os.path.dirname(__file__), "../static"),
autoescape=None)
http_server = tornado.httpserver.HTTPServer(app, xheaders = True)
http_server.listen(port_no)
if config.get("api_user") and config.get("api_group"):
chuser.change_user(config.get("api_user"), config.get("api_group"))
if task_id == 0:
buildtools.bake_css()
buildtools.bake_js()
buildtools.bake_beta_js()
for request in request_classes:
log.debug("start", " Handler: %s" % str(request))
log.info("start", "API server on port %s ready to go." % port_no)
self.ioloop = tornado.ioloop.IOLoop.instance()
try:
self.ioloop.start()
finally:
self.ioloop.stop()
http_server.stop()
db.close()
log.info("stop", "Server has been shutdown.")
log.close()