#
# Generate trace event data from MongoDB test logs.
# Can easily view this data in Google Chrome's chrome://tracing tab.
#

import re
import dateutil.parser
import json
import sys

def parse_ts(ts):
    dtime = dateutil.parser.parse(ts)
    return ((dtime.minute * 60 * 1000) + (dtime.second * 1000) + dtime.microsecond/1000) * 1000

event_matchers = [
    # (BeginEventRegex, EndEventRegex, EventName)
    ("ReplSetTest starting set", "ReplSetTest finished starting set", "ReplSetTest.startSet"),
    ("ReplSetTest stopSet stopping all", "ReplSetTest stopSet stopped all replica set nodes", "ReplSetTest.stopSet.shutdown")
]

# Matches a log line output by a JS test, not a mongo server node.
# timestamp: 2019-10-02T10:59:34.291-0400
ts_match = "\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}.\d{3}-\d{4}"
jsTestLineMatch = ".*\] (?P<ts>%s) (?P<msg>.*)" % ts_match

def line_to_event(line):
    js = re.search(jsTestLineMatch, line)
    # js test log events.
    if js and "----" not in js.group("msg") and "New session" not in js.group("msg"):
        msg = js.group("msg")
        ts = parse_ts(js.group("ts"))
        # Check against each matcher.
        for matcher in event_matchers:
            begin, end, name = matcher
            if re.search(begin, msg):
                return {"name": name, "pid": "test", "cat": "test", "ph": "B", "ts": ts}
            if re.search(end, msg):
                return {"name": name, "pid": "test", "cat": "test", "ph": "E", "ts": ts}

        return None

if __name__=="__main__":
    # Parse all lines from the given log file.
    logfile = sys.argv[1]
    lines = open(logfile).read().splitlines()
    events = [l for l in map(line_to_event, lines) if l]
    # Print all the events.
    for e in events:
        print e

    # Save the output to a JSON for viewing in Chrome tracing tab.
    json.dump(events, open("events.json", 'w'))

    # Also print out the duration of all spans that took place.
    begins = [e for e in events if e["ph"] == "B"]
    for begin in begins:
        name = begin["name"]
        end = [e for e in events if e["ph"] == "E" and e["name"] == name][0]
        dur = end["ts"] - begin["ts"]
        print name, ("%dms" % (dur/1000.0))