linux - using awk to check between two dates

Question

I have a file with multiple data structures in it like so:

eventTimestamp: 2010-03-23T07:56:19.166
result: Allowed
protocol: SMS
payload: RCOMM_SMS

eventTimestamp: 2010-03-23T07:56:19.167
result: Allowed
protocol: SMS
payload: RCOMM_SMS

eventTimestamp: 2010-03-23T07:56:19.186
result: Allowed
protocol: SMS
payload: SMS-MO-FSM

eventTimestamp: 2010-03-23T07:56:19.197
result: Allowed
protocol: SMS
payload: COPS

eventTimestamp: 2010-03-23T07:56:29.519
result: Blocked
protocol: SMS
payload: COPS
type: URL_IWF
result: Blocked

I want to find all of the events that are payload: SMS-MO-FSM or payload: SMS-MO-FSM-INFO that occurred between the times 2010-03-23 12:56:47 and 2010-03-23 13:56:47. When querying this file so far I have used awk in the following manner:

cat checkThis.txt |
awk 'BEGIN{FS="
"; RS=""; OFS=";"; ORS="
"}
     $1~/eventTimestamp: 2010-03-23T14:16:35/ && $4~/SMS-MO-FSM-INFO|SMS-MO-FSM$/ {$1=$1 ""; print $0}'

Which will give me all of the events that occurred on the second of 14:16:35 in 2010-03-23. I am struggling, however, to think of how I could put the date range into my query. I could use the following to put the dates into epoch time but how can I use the following in my awk to check whether the date is between the times needed:

python -c "import time; ENGINE_TIME_FORMAT='%Y-%m-%dT%H:%M:%S'; print int(time.mktime(time.strptime('2010-03-23T12:52:52', ENGINE_TIME_FORMAT)))"

I know this could done in Python but I have written a parser in Python for this and I want this method as an alternative checker so I want to use awk if at all possible.

I took this a little further and created a python script for time conversion:

#!/usr/local/bin/python
import time, sys
ENGINE_TIME_FORMAT='%Y-%m-%dT%H:%M:%S'
testTime = sys.argv[1]
try:
    print int(time.mktime(time.strptime(testTime, ENGINE_TIME_FORMAT)))
except:
    print "Time to convert %s" % testTime
    raise

I then tried to use getline to assign the conversion to a variable for comparison:

cat checkThis.txt| awk 'BEGIN {FS="
"; RS=""; OFS=";"; ORS="
"; "./firstDate '2010-03-23T12:56:47'" | getline start_time; close("firstDate"); "./firstDate '2010-03-23T13:56:47'" | getline end_time; close("firstDate");} ("./firstDate $1" | getline) > start_time {$1=$1 ""; print $0}'
Traceback (most recent call last):
  File "./firstDate", line 4, in <module>
testTime = sys.argv[1]
IndexError: list index out of range

The getline works in the BEGIN and I checked it in the final print but I seem to have problems in the comparison part of the script.

Answer 1

The key observation is that you can compare your timestamps using alphanumeric comparisons and get the correct answer - that is the beauty of ISO 8601 notation.

Thus, adapting your code slightly - and formatting to avoid scroll bars:

awk 'BEGIN {
        FS  = "
"
        RS  = ""
        OFS = ";"
        ORS = "
"
        t1  = "2010-03-23T07:45:00"
        t2  = "2010-03-23T08:00:00"
        m1  = "eventTimestamp: " t1
        m2  = "eventTimestamp: " t2
        }
$1 ~ /eventTimestamp:/ && $4 ~ /SMS-MO-FSM(-INFO)?$/ {
    if ($1 >= m1 && $1 <= m2) print $1, $2, $3, $4;
}' "$@"

Obviously, you could put this into a script file - you wouldn't want to type it often. And getting the date range entered accurately and conveniently is one of the hard parts. Note that I've adjusted the time range to match the data.

When run on the sample data, it outputs one record:

eventTimestamp: 2010-03-23T07:56:19.186;result: Allowed;protocol: SMS;payload: SMS-MO-FSM