Clean up har_extractor example

- Fix spacing, line length, unused imports, unusual import idioms
- Prevent it from barfing into our test output
This commit is contained in:
Aldo Cortesi 2015-01-02 13:41:40 +13:00
parent 1b5f5021dc
commit bb5fb2dbe0
1 changed files with 76 additions and 54 deletions

View File

@ -1,22 +1,18 @@
"""
This inline script utilizes harparser.HAR from https://github.com/JustusW/harparser
to generate a HAR log object.
"""
try:
from harparser import HAR
from pytz import UTC
except ImportError as e:
import sys
print >> sys.stderr, "\r\nMissing dependencies: please run `pip install mitmproxy[examples]`.\r\n"
raise
from datetime import datetime, timedelta, tzinfo
This inline script utilizes harparser.HAR from
https://github.com/JustusW/harparser to generate a HAR log object.
"""
from harparser import HAR
from datetime import datetime
class _HARLog(HAR.log):
# The attributes need to be registered here for them to actually be available later via self. This is
# due to HAREncodable linking __getattr__ to __getitem__. Anything that is set only in __init__ will
# just be added as key/value pair to self.__classes__.
# The attributes need to be registered here for them to actually be
# available later via self. This is due to HAREncodable linking __getattr__
# to __getitem__. Anything that is set only in __init__ will just be added
# as key/value pair to self.__classes__.
__page_list__ = []
__page_count__ = 0
__page_ref__ = {}
@ -58,55 +54,66 @@ class _HARLog(HAR.log):
def start(context, argv):
"""
On start we create a HARLog instance. You will have to adapt this to suit your actual needs
of HAR generation. As it will probably be necessary to cluster logs by IPs or reset them
from time to time.
On start we create a HARLog instance. You will have to adapt this to
suit your actual needs of HAR generation. As it will probably be
necessary to cluster logs by IPs or reset them from time to time.
"""
context.dump_file = None
if len(argv) > 1:
context.dump_file = argv[1]
else:
raise ValueError('Usage: -s "har_extractor.py filename" '
'(- will output to stdout, filenames ending with .zhar will result in compressed har)')
raise ValueError(
'Usage: -s "har_extractor.py filename" '
'(- will output to stdout, filenames ending with .zhar '
'will result in compressed har)'
)
context.HARLog = _HARLog(['https://github.com'])
context.seen_server = set()
def response(context, flow):
"""
Called when a server response has been received. At the time of this message both
a request and a response are present and completely done.
Called when a server response has been received. At the time of this
message both a request and a response are present and completely done.
"""
# Values are converted from float seconds to int milliseconds later.
ssl_time = -.001
connect_time = -.001
if flow.server_conn not in context.seen_server:
# Calculate the connect_time for this server_conn. Afterwards add it to seen list, in
# order to avoid the connect_time being present in entries that use an existing connection.
# Calculate the connect_time for this server_conn. Afterwards add it to
# seen list, in order to avoid the connect_time being present in entries
# that use an existing connection.
connect_time = flow.server_conn.timestamp_tcp_setup - flow.server_conn.timestamp_start
context.seen_server.add(flow.server_conn)
if flow.server_conn.timestamp_ssl_setup is not None:
# Get the ssl_time for this server_conn as the difference between the start of the successful
# tcp setup and the successful ssl setup. If no ssl setup has been made it is left as -1 since
# it doesn't apply to this connection.
# Get the ssl_time for this server_conn as the difference between
# the start of the successful tcp setup and the successful ssl
# setup. If no ssl setup has been made it is left as -1 since it
# doesn't apply to this connection.
ssl_time = flow.server_conn.timestamp_ssl_setup - flow.server_conn.timestamp_tcp_setup
# Calculate the raw timings from the different timestamps present in the request and response object.
# For lack of a way to measure it dns timings can not be calculated. The same goes for HAR blocked:
# MITMProxy will open a server connection as soon as it receives the host and port from the client
# connection. So the time spent waiting is actually spent waiting between request.timestamp_end and
# response.timestamp_start thus it correlates to HAR wait instead.
timings_raw = {'send': flow.request.timestamp_end - flow.request.timestamp_start,
'wait': flow.response.timestamp_start - flow.request.timestamp_end,
'receive': flow.response.timestamp_end - flow.response.timestamp_start,
'connect': connect_time,
'ssl': ssl_time}
# Calculate the raw timings from the different timestamps present in the
# request and response object. For lack of a way to measure it dns timings
# can not be calculated. The same goes for HAR blocked: MITMProxy will open
# a server connection as soon as it receives the host and port from the
# client connection. So the time spent waiting is actually spent waiting
# between request.timestamp_end and response.timestamp_start thus it
# correlates to HAR wait instead.
timings_raw = {
'send': flow.request.timestamp_end - flow.request.timestamp_start,
'wait': flow.response.timestamp_start - flow.request.timestamp_end,
'receive': flow.response.timestamp_end - flow.response.timestamp_start,
'connect': connect_time,
'ssl': ssl_time
}
# HAR timings are integers in ms, so we have to re-encode the raw timings to that format.
# HAR timings are integers in ms, so we have to re-encode the raw timings to
# that format.
timings = dict([(key, int(1000 * value)) for key, value in timings_raw.iteritems()])
# The full_time is the sum of all timings. Timings set to -1 will be ignored as per spec.
# The full_time is the sum of all timings. Timings set to -1 will be ignored
# as per spec.
full_time = 0
for item in timings.values():
if item > -1:
@ -157,21 +164,30 @@ def response(context, flow):
"cache": {},
"timings": timings, })
# If the current url is in the page list of context.HARLog or does not have a referrer we add it as a new
# pages object.
# If the current url is in the page list of context.HARLog or does not have
# a referrer we add it as a new pages object.
if flow.request.url in context.HARLog.get_page_list() or flow.request.headers.get('Referer', None) is None:
page_id = context.HARLog.create_page_id()
context.HARLog.add(HAR.pages({"startedDateTime": entry['startedDateTime'],
"id": page_id,
"title": flow.request.url, }))
context.HARLog.add(
HAR.pages({
"startedDateTime": entry['startedDateTime'],
"id": page_id,
"title": flow.request.url,
})
)
context.HARLog.set_page_ref(flow.request.url, page_id)
entry['pageref'] = page_id
# Lookup the referer in the page_ref of context.HARLog to point this entries pageref attribute to the right
# pages object, then set it as a new reference to build a reference tree.
# Lookup the referer in the page_ref of context.HARLog to point this entries
# pageref attribute to the right pages object, then set it as a new
# reference to build a reference tree.
elif context.HARLog.get_page_ref(flow.request.headers.get('Referer', (None, ))[0]) is not None:
entry['pageref'] = context.HARLog.get_page_ref(flow.request.headers['Referer'][0])
context.HARLog.set_page_ref(flow.request.headers['Referer'][0], entry['pageref'])
entry['pageref'] = context.HARLog.get_page_ref(
flow.request.headers['Referer'][0]
)
context.HARLog.set_page_ref(
flow.request.headers['Referer'][0], entry['pageref']
)
context.HARLog.add(entry)
@ -186,22 +202,28 @@ def done(context):
json_dump = context.HARLog.json()
compressed_json_dump = context.HARLog.compress()
print "=" * 100
if context.dump_file == '-':
pprint(json.loads(json_dump))
context.log(pprint.pformat(json.loads(json_dump)))
elif context.dump_file.endswith('.zhar'):
file(context.dump_file, "w").write(compressed_json_dump)
else:
file(context.dump_file, "w").write(json_dump)
print "=" * 100
print "HAR log finished with %s bytes (%s bytes compressed)" % (len(json_dump), len(compressed_json_dump))
print "Compression rate is %s%%" % str(100. * len(compressed_json_dump) / len(json_dump))
print "=" * 100
context.log(
"HAR log finished with %s bytes (%s bytes compressed)" % (
len(json_dump), len(compressed_json_dump)
)
)
context.log(
"Compression rate is %s%%" % str(
100. * len(compressed_json_dump) / len(json_dump)
)
)
def print_attributes(obj, filter_string=None, hide_privates=False):
"""
Useful helper method to quickly get all attributes of an object and its values.
Useful helper method to quickly get all attributes of an object and its
values.
"""
for attr in dir(obj):
if hide_privates and "__" in attr: