Parsing a job name out of a server log

Question

The log output lines look like follows

HLGAPL65.HOU150.CHEVRONTEXACO.NET/UPSTREAM_MDM_D2/Jobs/Keystone Release 2.0.2.0/0.0. Loading_SOR_to_Landing/EGI/EGI_WV_WELLHDR.pjb

I need to extract the name that appears right after the last "/" and before the ".pjb"

In this particular case - the needed name is EGI_WV_WELLHDR.

What is the most efficient and simple way to do that?

score 2 · Answer 1 · edited May 23 '17 at 12:20

You can use this awk:

$ awk -F"[/.]" '{print $(NF-1)}' file
EGI_WV_WELLHDR

Explanation

-F"[/.]" sets delimiter as dot or slash.
{print $(NF-1)} prints the penultimate field based on those field separators.

If what you want is the filename without extension, then you can do the following:

Get the last field in the line, which is the full path of the file.
Get the file name as described in Extract filename and extension in bash

See:

$ awk '{print $NF}' file
Loading_SOR_to_Landing/EGI/EGI_WV_WELLHDR.pjb
$ t=$(basename $(awk '{print $NF}' a))
$ echo "$t"
EGI_WV_WELLHDR.pjb
$ echo ${t%.*}
EGI_WV_WELLHDR

Avinash Raj · Answer 2 · 2014-05-26T10:44:33.647

2

Try this sed command,

$ sed -r 's/^.*\/([^.]*)\.pjb$/\1/g' file
EGI_WV_WELLHDR

-r --> Extended regex.

^.*\/([^.]*)\.pjb$

The above regex fetches the characters that are inbetween the last / and .pjb. Then the fetched characters in group are printed through backreference.

edited May 26 '14 at 10:44

answered May 26 '14 at 10:33

Avinash Raj

172,303
28
230
274

How do I make the regexp start one slash earlier? – Denys Jun 16 '14 at 10:56
did you want this `EGI/EGI_WV_WELLHDR` as output? – Avinash Raj Jun 16 '14 at 10:58
1

then try `sed -r 's/^.*\/([^.]*\/.*)\.pjb$/\1/g' file` – Avinash Raj Jun 16 '14 at 11:01

score 2 · Accepted Answer · answered May 26 '14 at 10:34

2

Using sed, run the following:

sed -r 's/.*\/([^\/]+)\.pjb/\1/g' logfile

answered May 26 '14 at 10:34

David Rabinowitz

29,904
14
93
125

1

Hold on. The initial answer fitted my purpose better :) – Denys May 26 '14 at 10:36
How do I make the regexp start one slash earlier? – Denys Jun 16 '14 at 10:56
1

sed -r 's/.*\/([^\/]+\/[^\/]+)\.pjb/\1/g' logfile – David Rabinowitz Jun 16 '14 at 11:02
Thanks! One lost plea :) Is it possible to make this regexp start from the 3rd slash to the left? – Denys Jun 16 '14 at 12:49

Farvardin · Answer 4 · 2014-05-26T10:59:47.920

1

using cut and rev:

rev | cut -d'/' -f1 | cut -d'.' -f2 | rev

edited May 26 '14 at 10:59

answered May 26 '14 at 10:38

Farvardin

5,336
5
33
54

score 0 · Answer 5 · answered May 27 '14 at 07:39

0

As your separator is a slash, you can use basename :

$ basename "HLGAPL65.HOU150.CHEVRONTEXACO.NET/UPSTREAM_MDM_D2/Jobs/Keystone Release 2.0.2.0/0.0. Loading_SOR_to_Landing/EGI/EGI_WV_WELLHDR.pjb" .pjb
EGI_WV_WELLHDR

answered May 27 '14 at 07:39

Mickaël Bucas

683
5
20

Parsing a job name out of a server log

5 Answers5

Explanation