parser.py

import re
import sys
import json
import itertools
from collections import Counter
from collections import defaultdict

SPLIT_MESSAGES = re.compile(r'\n.?\w{3} \w+ +\d+ \d+:\d+:\d+ \d+[\r\s]*\n')
MATCH_SIGNAL = re.compile(r'signal sender=(?P<sender>.*) -> dest=(?P<dest>.*) serial=(?P<serial>.*) path=(?P<path>.*) interface=(?P<interface>.*) member=(?P<member>.*)\n\s+string (?P<function>.*)\n\s+string (?P<params>.*)')
MATCH_METHOD_CALL = re.compile(r'method call sender=(?P<sender>.*) -> dest=(?P<dest>.*) serial=(?P<serial>.*) path=(?P<path>.*) interface=(?P<interface>.*) member=(?P<member>.*)\n\s+string (?P<function>.*)\n\s+string (?P<params>.*)')
MATCH_METHOD_RETURN = re.compile(r'method return sender=(?P<sender>.*) -> dest=(?P<dest>.*) reply_serial=(?P<reply_serial>.*)')


def parse_message(text):
    signal = MATCH_SIGNAL.search(text)
    method_call = MATCH_METHOD_CALL.search(text)
    method_return = MATCH_METHOD_RETURN.search(text)

    if signal:
        return signal.groupdict()
    if method_call:
        return method_call.groupdict()
    if method_return:
        return method_return.groupdict()
    raise Exception('Cannot parse message:\n{}'.format(text))


def parse_messages(text):
    messages = SPLIT_MESSAGES.split('\n' + text)
    return [parse_message(message) for message in messages if message.strip()]


if __name__ == '__main__':
    data = open(sys.argv[1], 'r').read()
    messages = parse_messages(data)
    function_messages = [x for x in messages if 'function' in x]

    def groupby_key(message):
        return message['function']

    """
    Signal printout
    """

    print '{0: <30}{1: <15}{2:}'.format('Signals', 'Occurrence', 'Sender')
    # this prints out all the signal names, and their respective occurences
    grouped = itertools.groupby(sorted(function_messages, key=groupby_key), groupby_key)
    for group, items in grouped:
        items = list(items)
        print '{func:<30}{occurences:<15}{path}'.format(
            func=items[0]['function'],
            occurences=len(items),
            path=items[0]['path']
        )

    """
    Alternative was return both function & path to the counter
    print '-----'
    c = Counter((x['function'], x['path']) for x in messages if 'function' in x and 'path' in x)
    for (function, path), count in c.items():
        print '{func}\t{count}\t{path}'.format(
            func=function,
            count=count,
            path=path
        )
    """

    """
    Methods printout
    """
    print '\n{0: <35}{1: <15}{2: <10}{3}'.format('Methods: (Member=Invoke)', 'Occurrence', 'Sender', 'Destination')
    grouped = itertools.groupby(sorted(function_messages, key=groupby_key), groupby_key)
    for group, items in grouped:
        items = list(items)
        print '{func:<35}{occurences:<15}{sender:<10}{destination}'.format(
            func=items[0]['function'],
            occurences=len(items),
            sender=items[0]['sender'],
            destination=items[0]['dest']
        )

    """
    SERVICES printout
    """
    print '\n{0: <45}{1: <15}'.format('Services', 'Occurrence')
    services = Counter((x['path']) for x in messages if 'path' in x)  # give me a Counter with all the serives from path=
    for service, occurrence in services.iteritems():
        print '{key:<45}{value}'.format(key=service, value=occurrence)