from tool.MailReport import MailReporter
if __name__ == "__main__":
mr = MailReporter()
mr.report("hi")
print "done"
def mgrThreadBody(self):
"Management Thread"
print "#init:> start mgr & provider."
getter = HtmlRetriever.getInstance(self.settings.use_proxy)
while self.running or not self.stopped:
# interval seconds passed.
interval_seconds = (datetime.datetime.now() - self.last_report_time).seconds
if interval_seconds == 0: interval_seconds = 1
self.last_report_time = datetime.datetime.now();
try:
self.PersonThreadActive = 0
self.PubThreadActive = 0
for x in self.person_thread_pool:
if x.check_idle():
self.PersonThreadActive += 1
for y in self.pub_thread_pool:
if y.check_idle():
self.PubThreadActive += 1
except Exception:
print "ERROR:count errer"
print Exception
try:
# save pdf link
if self.settings.save_pdflink:
self.pdfcache.flush()
except Exception:
print "ERROR: pdf link"
print Exception
message = None
# 什么时候重启所有线程&进程
reload_all_thread = False
if self.num_report % 1000 == 0:
reload_all_thread = True
message = "Kill & Restart All Thread."
try:
# Maintain Threads and get worker threads status.
(num_persont_alive, num_pubt_alive) = self._maintainThreadPool(reload_all_thread=False)
except Exception:
print "ERROR: maintain threads and worker"
print Exception
try:
# Finish Condition.
if self._checkFinishCondition():
self.running = False # -> tell all threads finish.
message = "MESSAGE! Send terminal signal to all worker thread."
except Exception:
print "ERROR: condition check"
print Exception
# if all worker threads stopped, mgrThread can stop.
if num_persont_alive == 0 and num_pubt_alive == 0:
self.stopped = True
message = "Send terminal signal to mgr_thread."
# check network and count
period_success_connection = getter.success_connection_count - getter.last_success_connection_count
period_bad_connection = getter.bad_connection_count - getter.last_bad_connection_count
total_connections = period_success_connection + period_bad_connection
getter.last_success_connection_count = getter.success_connection_count
getter.last_bad_connection_count = getter.bad_connection_count
average_success_persecond = period_success_connection / float(interval_seconds)
average_bad_persecond = period_bad_connection / float(interval_seconds)
if False: # 是否Block模式，就是暂停整个程序
if getter.detect_mode:
if getter.detect_success_count > 3:
getter.leave_detect_mode()
self.detect_exit_wait = 1 # 刚出来时，下两轮都不要再进入block模式了。
else:
if total_connections * 0.9 < period_bad_connection:
if self.detect_exit_wait > 0:
print "---- waiting %s rounds ----" % self.detect_exit_wait
self.detect_exit_wait -= 1
else:
getter.enter_detect_mode()
################ print interval string ################
try:
# print report
if not getter.detect_mode:
str_report = None
if not self.pause:
self.num_report += 1
str_report = self.num_report
else:
str_report = "paused"
report_strs = []
report_strs.append("-" * 100)
report_strs.append("\n")
report_strs.append("$&mgr:%s(%s):> " % (datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"), str_report))
report_strs.append("Person(%sT on %s), " % (num_persont_alive, self.store.person_queue.qsize()))
report_strs.append("Pub(%sT on %s, %s items), " % (num_pubt_alive, len(self.store.pubmap), len(self.store.person_pub_map)))
report_strs.append("DBCache({{{ %s }}}), " % len(self.store.pub_db_cache))
report_strs.append("T(busy/idle)(%s/%s), " % (self.busy_semaphore, self.settings.max_person_thread + self.settings.max_pub_thread - self.busy_semaphore))
report_strs += '\n'
report_strs.append("Person(busy/idle)(%s/%s), Pub(busy/idle)(%s/%s)" % (self.busy_person_semaphore, self.settings.max_person_thread-self.busy_person_semaphore, self.busy_pub_semaphore, self.settings.max_pub_thread-self.busy_pub_semaphore))
g = getter.success_connection_count
b = getter.bad_connection_count
t = g + b
rate = 0
if(t > 0):
rate = g / float(t)
report_strs.append("network(g+b=t)=(%s+%s=%s),rate=%.2f " % (g, b , t, rate))
report_strs.append("interval-network(g+b=t)=(%s+%s=%s), " % (period_success_connection, period_bad_connection, total_connections))
report_strs.append("avg:(g%.1f b%.1f in %s seconds.), " % (average_success_persecond, average_bad_persecond, interval_seconds))
report_strs.append("\n")
report_strs.append("now have %s child threads, " % self.threadChildren)
report_strs.append("active threads (%s person, %s pub) , " % (self.PersonThreadActive, self.PubThreadActive))
report_strs.append("\n")
report_strs.append("time:(wait=%.2f, getlock=%.2f, get=%.2f)" % (self.store.ppt_wait, self.store.ppt_getlock, self.store.ppt_get))
if message is not None:
report_strs.append("\n")
report_strs.append(message)
report_strs.append("\n")
report_strs.append(" * Process NA Persons : %s.\n" % self.reportPersonProgress(self.generation))
report_strs.append(" * Process Publication: %s.\n" % self.reportPublicationProgress(self.generation))
report_strs.append("-" * 100)
report_strs.append("\n")
print "".join(report_strs)
if (self.num_report%100 == 0):
mr = MailReporter()
mr.report(report_strs)
except Exception:
print "ERROR: report error"
print Exception
try:
self.store.flushDBCache() # last flush cache to db.
self.store.running = self.running # pass main running thread to Store object.
except Exception:
print "ERROR: flush db cache"
print Exception
time.sleep(self.mgr_interval) # interval
print "$mgr:> exit."
def mgrThreadBody(self):
"Management Thread"
print "#init:> start mgr & provider."
getter = HtmlRetriever.getInstance(self.settings.use_proxy)
while self.running or not self.stopped:
# interval seconds passed.
interval_seconds = (datetime.datetime.now() -
self.last_report_time).seconds
if interval_seconds == 0: interval_seconds = 1
self.last_report_time = datetime.datetime.now()
try:
self.PersonThreadActive = 0
self.PubThreadActive = 0
for x in self.person_thread_pool:
if x.check_idle():
self.PersonThreadActive += 1
for y in self.pub_thread_pool:
if y.check_idle():
self.PubThreadActive += 1
except Exception:
print "ERROR:count errer"
print Exception
try:
# save pdf link
if self.settings.save_pdflink:
self.pdfcache.flush()
except Exception:
print "ERROR: pdf link"
print Exception
message = None
# 什么时候重启所有线程&进程
reload_all_thread = False
if self.num_report % 1000 == 0:
reload_all_thread = True
message = "Kill & Restart All Thread."
try:
# Maintain Threads and get worker threads status.
(num_persont_alive, num_pubt_alive) = self._maintainThreadPool(
reload_all_thread=False)
except Exception:
print "ERROR: maintain threads and worker"
print Exception
try:
# Finish Condition.
if self._checkFinishCondition():
self.running = False # -> tell all threads finish.
message = "MESSAGE! Send terminal signal to all worker thread."
except Exception:
print "ERROR: condition check"
print Exception
# if all worker threads stopped, mgrThread can stop.
if num_persont_alive == 0 and num_pubt_alive == 0:
self.stopped = True
message = "Send terminal signal to mgr_thread."
# check network and count
period_success_connection = getter.success_connection_count - getter.last_success_connection_count
period_bad_connection = getter.bad_connection_count - getter.last_bad_connection_count
total_connections = period_success_connection + period_bad_connection
getter.last_success_connection_count = getter.success_connection_count
getter.last_bad_connection_count = getter.bad_connection_count
average_success_persecond = period_success_connection / float(
interval_seconds)
average_bad_persecond = period_bad_connection / float(
interval_seconds)
if False: # 是否Block模式，就是暂停整个程序
if getter.detect_mode:
if getter.detect_success_count > 3:
getter.leave_detect_mode()
self.detect_exit_wait = 1 # 刚出来时，下两轮都不要再进入block模式了。
else:
if total_connections * 0.9 < period_bad_connection:
if self.detect_exit_wait > 0:
print "---- waiting %s rounds ----" % self.detect_exit_wait
self.detect_exit_wait -= 1
else:
getter.enter_detect_mode()
################ print interval string ################
try:
# print report
if not getter.detect_mode:
str_report = None
if not self.pause:
self.num_report += 1
str_report = self.num_report
else:
str_report = "paused"
report_strs = []
report_strs.append("-" * 100)
report_strs.append("\n")
report_strs.append(
"$&mgr:%s(%s):> " %
(datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
str_report))
report_strs.append(
"Person(%sT on %s), " %
(num_persont_alive, self.store.person_queue.qsize()))
report_strs.append("Pub(%sT on %s, %s items), " %
(num_pubt_alive, len(self.store.pubmap),
len(self.store.person_pub_map)))
report_strs.append("DBCache({{{ %s }}}), " %
len(self.store.pub_db_cache))
report_strs.append(
"T(busy/idle)(%s/%s), " %
(self.busy_semaphore, self.settings.max_person_thread +
self.settings.max_pub_thread - self.busy_semaphore))
report_strs += '\n'
report_strs.append(
"Person(busy/idle)(%s/%s), Pub(busy/idle)(%s/%s)" %
(self.busy_person_semaphore,
self.settings.max_person_thread -
self.busy_person_semaphore, self.busy_pub_semaphore,
self.settings.max_pub_thread -
self.busy_pub_semaphore))
g = getter.success_connection_count
b = getter.bad_connection_count
t = g + b
rate = 0
if (t > 0):
rate = g / float(t)
report_strs.append("network(g+b=t)=(%s+%s=%s),rate=%.2f " %
(g, b, t, rate))
report_strs.append(
"interval-network(g+b=t)=(%s+%s=%s), " %
(period_success_connection, period_bad_connection,
total_connections))
report_strs.append(
"avg:(g%.1f b%.1f in %s seconds.), " %
(average_success_persecond, average_bad_persecond,
interval_seconds))
report_strs.append("\n")
report_strs.append("now have %s child threads, " %
self.threadChildren)
report_strs.append(
"active threads (%s person, %s pub) , " %
(self.PersonThreadActive, self.PubThreadActive))
report_strs.append("\n")
report_strs.append(
"time:(wait=%.2f, getlock=%.2f, get=%.2f)" %
(self.store.ppt_wait, self.store.ppt_getlock,
self.store.ppt_get))
if message is not None:
report_strs.append("\n")
report_strs.append(message)
report_strs.append("\n")
report_strs.append(
" * Process NA Persons : %s.\n" %
self.reportPersonProgress(self.generation))
report_strs.append(
" * Process Publication: %s.\n" %
self.reportPublicationProgress(self.generation))
report_strs.append("-" * 100)
report_strs.append("\n")
print "".join(report_strs)
if (self.num_report % 100 == 0):
mr = MailReporter()
mr.report(report_strs)
except Exception:
print "ERROR: report error"
print Exception
try:
self.store.flushDBCache() # last flush cache to db.
self.store.running = self.running # pass main running thread to Store object.
except Exception:
print "ERROR: flush db cache"
print Exception
time.sleep(self.mgr_interval) # interval
print "$mgr:> exit."