1
0
mirror of https://github.com/moparisthebest/SickRage synced 2024-11-11 20:05:04 -05:00
SickRage/cherrypy/lib/profiler.py
echel0n 0d9fbc1ad7 Welcome to our SickBeard-TVRage Edition ...
This version of SickBeard uses both TVDB and TVRage to search and gather it's series data from allowing you to now have access to and download shows that you couldn't before because of being locked into only what TheTVDB had to offer.

Also this edition is based off the code we used in our XEM editon so it does come with scene numbering support as well as all the other features our XEM edition has to offer.

Please before using this with your existing database (sickbeard.db) please make a backup copy of it and delete any other database files such as cache.db and failed.db if present, we HIGHLY recommend starting out with no database files at all to make this a fresh start but the choice is at your own risk!

Enjoy!
2014-03-09 22:39:12 -07:00

206 lines
6.3 KiB
Python

"""Profiler tools for CherryPy.
CherryPy users
==============
You can profile any of your pages as follows:
from cherrypy.lib import profiler
class Root:
p = profile.Profiler("/path/to/profile/dir")
def index(self):
self.p.run(self._index)
index.exposed = True
def _index(self):
return "Hello, world!"
cherrypy.tree.mount(Root())
You can also turn on profiling for all requests
using the make_app function as WSGI middleware.
CherryPy developers
===================
This module can be used whenever you make changes to CherryPy,
to get a quick sanity-check on overall CP performance. Use the
"--profile" flag when running the test suite. Then, use the serve()
function to browse the results in a web browser. If you run this
module from the command line, it will call serve() for you.
"""
# Make profiler output more readable by adding __init__ modules' parents.
def new_func_strip_path(func_name):
filename, line, name = func_name
if filename.endswith("__init__.py"):
return os.path.basename(filename[:-12]) + filename[-12:], line, name
return os.path.basename(filename), line, name
try:
import profile
import pstats
pstats.func_strip_path = new_func_strip_path
except ImportError:
profile = None
pstats = None
import os, os.path
import sys
import warnings
try:
from cStringIO import StringIO
except ImportError:
from StringIO import StringIO
_count = 0
class Profiler(object):
def __init__(self, path=None):
if not path:
path = os.path.join(os.path.dirname(__file__), "profile")
self.path = path
if not os.path.exists(path):
os.makedirs(path)
def run(self, func, *args, **params):
"""Dump profile data into self.path."""
global _count
c = _count = _count + 1
path = os.path.join(self.path, "cp_%04d.prof" % c)
prof = profile.Profile()
result = prof.runcall(func, *args, **params)
prof.dump_stats(path)
return result
def statfiles(self):
"""statfiles() -> list of available profiles."""
return [f for f in os.listdir(self.path)
if f.startswith("cp_") and f.endswith(".prof")]
def stats(self, filename, sortby='cumulative'):
"""stats(index) -> output of print_stats() for the given profile."""
sio = StringIO()
if sys.version_info >= (2, 5):
s = pstats.Stats(os.path.join(self.path, filename), stream=sio)
s.strip_dirs()
s.sort_stats(sortby)
s.print_stats()
else:
# pstats.Stats before Python 2.5 didn't take a 'stream' arg,
# but just printed to stdout. So re-route stdout.
s = pstats.Stats(os.path.join(self.path, filename))
s.strip_dirs()
s.sort_stats(sortby)
oldout = sys.stdout
try:
sys.stdout = sio
s.print_stats()
finally:
sys.stdout = oldout
response = sio.getvalue()
sio.close()
return response
def index(self):
return """<html>
<head><title>CherryPy profile data</title></head>
<frameset cols='200, 1*'>
<frame src='menu' />
<frame name='main' src='' />
</frameset>
</html>
"""
index.exposed = True
def menu(self):
yield "<h2>Profiling runs</h2>"
yield "<p>Click on one of the runs below to see profiling data.</p>"
runs = self.statfiles()
runs.sort()
for i in runs:
yield "<a href='report?filename=%s' target='main'>%s</a><br />" % (i, i)
menu.exposed = True
def report(self, filename):
import cherrypy
cherrypy.response.headers['Content-Type'] = 'text/plain'
return self.stats(filename)
report.exposed = True
class ProfileAggregator(Profiler):
def __init__(self, path=None):
Profiler.__init__(self, path)
global _count
self.count = _count = _count + 1
self.profiler = profile.Profile()
def run(self, func, *args):
path = os.path.join(self.path, "cp_%04d.prof" % self.count)
result = self.profiler.runcall(func, *args)
self.profiler.dump_stats(path)
return result
class make_app:
def __init__(self, nextapp, path=None, aggregate=False):
"""Make a WSGI middleware app which wraps 'nextapp' with profiling.
nextapp: the WSGI application to wrap, usually an instance of
cherrypy.Application.
path: where to dump the profiling output.
aggregate: if True, profile data for all HTTP requests will go in
a single file. If False (the default), each HTTP request will
dump its profile data into a separate file.
"""
if profile is None or pstats is None:
msg = ("Your installation of Python does not have a profile module. "
"If you're on Debian, try `sudo apt-get install python-profiler`. "
"See http://www.cherrypy.org/wiki/ProfilingOnDebian for details.")
warnings.warn(msg)
self.nextapp = nextapp
self.aggregate = aggregate
if aggregate:
self.profiler = ProfileAggregator(path)
else:
self.profiler = Profiler(path)
def __call__(self, environ, start_response):
def gather():
result = []
for line in self.nextapp(environ, start_response):
result.append(line)
return result
return self.profiler.run(gather)
def serve(path=None, port=8080):
if profile is None or pstats is None:
msg = ("Your installation of Python does not have a profile module. "
"If you're on Debian, try `sudo apt-get install python-profiler`. "
"See http://www.cherrypy.org/wiki/ProfilingOnDebian for details.")
warnings.warn(msg)
import cherrypy
cherrypy.config.update({'server.socket_port': int(port),
'server.thread_pool': 10,
'environment': "production",
})
cherrypy.quickstart(Profiler(path))
if __name__ == "__main__":
serve(*tuple(sys.argv[1:]))