regexp match within a log file, return dynamic content above and below match

1

I have some catchall log files in a format as follows:

timestamp event summary
foo details
account name: userA
bar more details
timestamp event summary
baz details
account name: userB
qux more details
timestamp etc.

I would like to search the log file for userB, and if found, echo from the preceding timestamp down to (but not including) the following timestamp. There will likely be several events matching my search. It would be nice to echo some sort of --- start --- and --- end --- surrounding each match.

This would be perfect for pcregrep -M, right? Problem is, GnuWin32's pcregrep crashes with multiline regexps searching large files, and these catch-all logs can be 100 megs or more.

What I've tried

My hackish workaround thus far involves using grep -B15 -A30 to find matching lines and print surrounding content, then piping the now more manageable chunk into pcregrep for polishing. Problem is that some events are less than ten lines, while others are 30 or more; and I'm getting some unexpected results where the shorter events are encountered.

:parselog <username> <logfile>

set silent=1
set count=0
set deez=20\d\d-\d\d-\d\d \d\d:\d\d:\d\d
echo Searching %~2 for records containing %~1...

for /f "delims=" %%I in (
    'grep -P -i -B15 -A30 ":\s+\b%~1\b(@mydomain\.ext)?$" "%~2" ^| pcregrep -M -i "^%deez%(.|\n)+?\b%~1\b(@mydomain\.ext|\r?\n)(.|\n)+?\n%deez%" 2^>NUL'
) do (
    echo(%%I| findstr "^20[0-9][0-9]-[0-9][0-9]-[0-9][0-9].[0-9][0-9]:[0-9][0-9]:[0-9][0-9]" >NUL && (
        if defined silent (
            set silent=
            set found=1
            set /a "count+=1"
            echo;
            echo ---------------start of record !count!-------------
        ) else (
            set silent=1
            echo ----------------end of record !count!--------------
            echo;
        )
    )
    if not defined silent echo(%%I
)

goto :EOF

Is there a better way to do this? I've come across an awk command that looked interesting, something like:

awk "/start pattern/,/end pattern/" logfile

... but it would need to match a middle pattern as well. Unfortunately, I'm not that familiar with awk syntax. Any suggestions?


Ed Morton suggested that I supply some example logging and expected output.

Example catch-all

2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security    11730158    Mon Mar 25 08:02:28 2013    529 Security    NT AUTHORITY\SYSTEM N/A Audit Failure   dc3 2   Logon Failure:

    Reason:     Unknown user name or bad password

    User Name:  user5f

    Domain:     MYDOMAIN

    Logon Type: 3

    Logon Process:  Advapi  

    Authentication Package: Negotiate

    Workstation Name:   dc3

    Caller User Name:   dc3$

    Caller Domain:  MYDOMAIN

    Caller Logon ID:    (0x0,0x3E7)

    Caller Process ID:  400

    Transited Services: -

    Source Network Address: 169.254.7.86

    Source Port:    40838
2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security    11730159    Mon Mar 25 08:02:29 2013    680 Security    NT AUTHORITY\SYSTEM N/A Audit Failure   dc3 9   Logon attempt by:   MICROSOFT_AUTHENTICATION_PACKAGE_V1_0

Logon account:  USER6Q

Source Workstation: dc3

Error Code: 0xC0000234
2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security    11730160    Mon Mar 25 08:02:29 2013    539 Security    NT AUTHORITY\SYSTEM N/A Audit Failure   dc3 2   Logon Failure:

    Reason:     Account locked out

    User Name:  USER6Q@MYDOMAIN.TLD

    Domain: MYDOMAIN

    Logon Type: 3

    Logon Process:  Advapi  

    Authentication Package: Negotiate

    Workstation Name:   dc3

    Caller User Name:   dc3$

    Caller Domain:  MYDOMAIN

    Caller Logon ID:    (0x0,0x3E7)

    Caller Process ID: 400

    Transited Services: -

    Source Network Address: 169.254.7.89

    Source Port:    55314
2013-03-25 08:02:32 Auth.Notice 169.254.5.62    Mar 25 08:36:38 DC4.mydomain.tld MSWinEventLog  5   Security    201326798   Mon Mar 25 08:36:37 2013    4624    Microsoft-Windows-Security-Auditing     N/A Audit Success   DC4.mydomain.tld    12544   An account was successfully logged on.

Subject:
    Security ID:        S-1-0-0
    Account Name:       -
    Account Domain:     -
    Logon ID:       0x0

Logon Type:         3

New Logon:
    Security ID:        S-1-5-21-606747145-1409082233-725345543-160838
    Account Name:       DEPTACCT16$
    Account Domain:     MYDOMAIN
    Logon ID:       0x1158e6012c
    Logon GUID:     {BCC72986-82A0-4EE9-3729-847BA6FA3A98}

Process Information:
    Process ID:     0x0
    Process Name:       -

Network Information:
    Workstation Name:   
    Source Network Address: 169.254.114.62
    Source Port:        42183

Detailed Authentication Information:
    Logon Process:      Kerberos
    Authentication Package: Kerberos
    Transited Services: -
    Package Name (NTLM only):   -
    Key Length:     0

This event is generated when a logon session is created. It is generated on the computer that was accessed.

The subject fields indicate...
2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security    11730162    Mon Mar 25 08:02:30 2013    675 Security    NT AUTHORITY\SYSTEM N/A Audit Failure   dc3 9   Pre-authentication failed:

    User Name:  USER8Y

    User ID:        %{S-1-5-21-606747145-1409082233-725345543-3904}

    Service Name:   krbtgt/MYDOMAIN

    Pre-Authentication Type:    0x0

    Failure Code:   0x19

    Client Address: 169.254.87.158
2013-03-25 08:02:32 Auth.Critical   etc.

Example command

call :parselog user6q \\path\to\catch-all.log

Expected result

---------------start of record 1-------------
2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security    11730159    Mon Mar 25 08:02:29 2013    680 Security    NT AUTHORITY\SYSTEM N/A Audit Failure   dc3 9   Logon attempt by:   MICROSOFT_AUTHENTICATION_PACKAGE_V1_0

Logon account:  USER6Q

Source Workstation: dc3

Error Code: 0xC0000234
---------------end of record 1-------------


---------------start of record 2-------------
2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security    11730160    Mon Mar 25 08:02:29 2013    539 Security    NT AUTHORITY\SYSTEM N/A Audit Failure   dc3 2   Logon Failure:

    Reason:     Account locked out

    User Name:  USER6Q@MYDOMAIN.TLD

    Domain: MYDOMAIN

    Logon Type: 3

    Logon Process:  Advapi  

    Authentication Package: Negotiate

    Workstation Name:   dc3

    Caller User Name:   dc3$

    Caller Domain:  MYDOMAIN

    Caller Logon ID:    (0x0,0x3E7)

    Caller Process ID: 400

    Transited Services: -

    Source Network Address: 169.254.7.89

    Source Port:    55314
---------------end of record 2-------------
windows
batch-file
awk
grep
pcre
asked on Stack Overflow Mar 26, 2013 by rojo • edited Mar 26, 2013 by rojo

4 Answers

1

This is all you need with GNU awk (for IGNORECASE):

$ cat tst.awk
function prtRecord() {
    if (record ~ regexp) {
        printf "-------- start of record %d --------%s", ++numRecords, ORS
        printf "%s", record
        printf "--------- end of record %d ---------%s%s", numRecords, ORS, ORS
    }
    record = ""
}
BEGIN{ IGNORECASE=1 }
/^[[:digit:]]+-[[:digit:]]+-[[:digit:]]+/ { prtRecord() }
{ record = record $0 ORS }
END { prtRecord() }

or with any awk:

$ cat tst.awk
function prtRecord() {
    if (tolower(record) ~ tolower(regexp)) {
        printf "-------- start of record %d --------%s", ++numRecords, ORS
        printf "%s", record
        printf "--------- end of record %d ---------%s%s", numRecords, ORS, ORS
    }
    record = ""
}
/^[[:digit:]]+-[[:digit:]]+-[[:digit:]]+/ { prtRecord() }
{ record = record $0 ORS }
END { prtRecord() }

Either way you'd run it on UNIX as:

$ awk -v regexp=user6q -f tst.awk file

I don't know the Windows syntax but I expect it's very similar if not identical.

Note the use of tolower() in the script to make both sides of the comparison lower case so the match is case-insensitive. If you can instead pass in a search regexp that's the correct case, then you don't need to call tolower() on either side of the comparison. nbd, it might just speed the script up slightly.

$ awk -v regexp=user6q -f tst.awk file
-------- start of record 1 --------
2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security
    11730159    Mon Mar 25 08:02:29 2013    680 Security    NT AUTHORITY\SYSTEM N/A Audit Failure
dc3 9   Logon attempt by:   MICROSOFT_AUTHENTICATION_PACKAGE_V1_0

Logon account:  USER6Q

Source Workstation: dc3

Error Code: 0xC0000234
--------- end of record 1 ---------

-------- start of record 2 --------
2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security
    11730160    Mon Mar 25 08:02:29 2013    539 Security    NT AUTHORITY\SYSTEM N/A Audit Failure
dc3 2   Logon Failure:

    Reason:     Account locked out

    User Name:  USER6Q@MYDOMAIN.TLD

    Domain: MYDOMAIN

    Logon Type: 3

    Logon Process:  Advapi

    Authentication Package: Negotiate

    Workstation Name:   dc3

    Caller User Name:   dc3$

    Caller Domain:  MYDOMAIN

    Caller Logon ID:    (0x0,0x3E7)

    Caller Process ID: 400

    Transited Services: -

    Source Network Address: 169.254.7.89

    Source Port:    55314
--------- end of record 2 ---------
answered on Stack Overflow Mar 26, 2013 by Ed Morton • edited Mar 26, 2013 by Ed Morton
1

Here's my effort:

@ECHO OFF
SETLOCAL
::
:: Target username
::
SET target=%1
CALL :zaplines
SET count=0
FOR /f "delims=" %%I IN (rojoslog.txt) DO (
  ECHO.%%I| findstr /r "^20[0-9][0-9]-[0-9][0-9]-[0-9][0-9].[0-9][0-9]:[0-9][0-9]:[0-9][0-9]" >NUL
  IF NOT ERRORLEVEL 1 (
    IF DEFINED founduser CALL :report
    CALL :zaplines
  )
  (SET stored=)
  FOR /l %%L IN (1000,1,1200) DO IF NOT DEFINED stored IF NOT DEFINED line%%L (
    SET line%%L=%%I
    SET stored=Y
   )
  ECHO.%%I|FINDSTR /b /e /i /c:"account name: %target%" >NUL
  IF NOT ERRORLEVEL 1 (SET founduser=Y)
)
IF DEFINED founduser CALL :report
GOTO :eof

::
:: remove all envvars starting 'line'
:: Set 'not found user' at same time
::
 :zaplines
(SET founduser=)
FOR /f "delims==" %%L IN ('set line 2^>nul') DO (SET %%L=)
GOTO :eof

:report
IF NOT DEFINED line1000 GOTO :EOF 
SET /a count+=1
ECHO.
ECHO.---------- START of record %count% ----------
FOR /l %%L IN (1000,1,1200) DO IF DEFINED line%%L CALL ECHO.%%line%%L%%
ECHO.----------- END of record %count% -----------
GOTO :eof
answered on Stack Overflow Mar 26, 2013 by Magoo • edited Mar 26, 2013 by Magoo
1

Below there is a pure Batch solution that does not use grep. It locates timestamp lines because the "summary" word that must not exist in other lines, but this word may be changed for another one if needed.

EDIT: I changed the word that identify timestamp lines to "Auth."; I also changed FINDSTR seek to ignore case. This is the new version:

@echo off
setlocal EnableDelayedExpansion

:parselog <username> <logfile>
echo Searching %~2 for records containing %~1...

set n=0
set previousMatch=Auth.
for /F "tokens=1* delims=:" %%a in ('findstr /I /N "Auth\. %~1" %2') do (
   set currentMatch=%%b
   if "!previousMatch:Auth.=!" neq "!previousMatch!" (
      if "!currentMatch:Auth.=!" equ "!currentMatch!" (
         set /A n+=1
         set /A skip[!n!]=!previousLine!-1
      )
   ) else (
      set /A end[!n!]=%%a-1
   )
   set previousLine=%%a
   set previousMatch=%%b
)
if %n% equ 0 (
   echo No records found
   goto :EOF
)

if not defined end[%n%] set end[%n%]=-1
set i=1
:nextRecord
   echo/
   echo ---------------start of record %i%-------------
   if !skip[%i%]! equ 0 (
      set skip=
   ) else (
      set skip=skip=!skip[%i%]!
   )
   set end=!end[%i%]!
   for /F "%skip% tokens=1* delims=:" %%a in ('findstr /N "^" %2') do (
      echo(%%b
      if %%a equ %end% goto endOfRecord
   )
   :endOfRecord
   echo ---------------end of record %i%-------------
   set /A i+=1
if %i% leq %n% goto nextRecord

Example command:

C:>test user6q catch-all.log

Result:

Searching catch-all.log for records containing user6q...

---------------start of record 1-------------
2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security    11730159    Mon Mar 25 08:02:29 2013    680 Security    NT AUTHORITY\SYSTEM N/A Audit Failure   dc3 9   Logon attempt by:   MICROSOFT_AUTHENTICATION_PACKAGE_V1_0

Logon account:  USER6Q

Source Workstation: dc3

Error Code: 0xC0000234
---------------end of record 1-------------

---------------start of record 2-------------
2013-03-25 08:02:32 Auth.Critical   169.254.8.110   Mar 25 08:02:32 dc3 MSWinEventLog   2   Security    11730160    Mon Mar 25 08:02:29 2013    539 Security    NT AUTHORITY\SYSTEM N/A Audit Failure   dc3 2   Logon Failure:

    Reason:     Account locked out

    User Name:  USER6Q@MYDOMAIN.TLD

    Domain: MYDOMAIN

    Logon Type: 3

    Logon Process:  Advapi  

    Authentication Package: Negotiate

    Workstation Name:   dc3

    Caller User Name:   dc3$

    Caller Domain:  MYDOMAIN

    Caller Logon ID:    (0x0,0x3E7)

    Caller Process ID: 400

    Transited Services: -

    Source Network Address: 169.254.7.89

    Source Port:    55314
---------------end of record 2-------------

This method use just one execution of findstr command to locate all matching records, and then one additional findstr command to show each record. Note that first for /F ... command works over findstr "Auth. user.." results, and the second for /F command have a "skip=N" option and a GOTO that break the loop as soon as the record was displayed. This mean that FOR commands does not slow down the program; the speed of this program depends on the speed of FINDSTR command.

However, it is possible that the second for /F "%skip% ... in ('findstr /N "^" %2') command take too long because the size of FINDSTR output result before it is processed by the FOR. If this happen, we could modify the second FOR by another faster method (an asynchronous pipe that will be break, for example). Please, report the result.

Antonio

answered on Stack Overflow Mar 26, 2013 by Aacini • edited Mar 26, 2013 by Aacini
0

I think awk is all you need:

awk "/---start of record---/,/---end of record---/ {print}" logfile

That's all you need if the first line indicator is:

---start of record---

and the last is:

---end of record---

Notice that there is no middle-pattern matching, that "," is just a separator for both regexps.

answered on Stack Overflow Mar 26, 2013 by Daniel

User contributions licensed under CC BY-SA 3.0