Unix/Linux Help for the below command

Hi, it is a 'one liner' that processed the file HD_sync.log.10 with awk and sed and then sorts the output leaving only the unique entries.

Here's a description of what is happening. More details can be given if can post a sample HD_sync.log file.

awk'/STARTED/,/FINISHED/{if($0~/IMAGES[/]$/){print $2}}'
1. Perform this awk command - for lines that match pattern STARTED or FINISHED - check if pattern /IMAGES[/]$/ is on this line, if so output the 2nd field of that line

HD_sync.log.10
2. on this file (your input file)

|sed -ne 's/^[^/]*[/]\(IMD.*\)\/IMAGES\//\1/p'
3. search beginning (^) of each line for pattern [^/]*[/]\(IMD.*\)\/IMAGES\/ and print the first matching pattern

|sort
4. sort the output

|uniq
5. keep only unique lines

Is this enough explanation for you? The above is really not 'a command' but a series of commands on the input file. You can view the output steps in between by doing these separate steps:

awk  '/STARTED/,/FINISHED/{if($0~/IMAGES[/]$/){print $2}}' HD_sync.log.10
awk  '/STARTED/,/FINISHED/{if($0~/IMAGES[/]$/){print $2}}' HD_sync.log.10|sed -ne 's/^[^/]*[/]\(IMD.*\)\/IMAGES\//\1/p'
awk  '/STARTED/,/FINISHED/{if($0~/IMAGES[/]$/){print $2}}' HD_sync.log.10|sed -ne 's/^[^/]*[/]\(IMD.*\)\/IMAGES\//\1/p'|sort
awk  '/STARTED/,/FINISHED/{if($0~/IMAGES[/]$/){print $2}}' HD_sync.log.10|sed -ne 's/^[^/]*[/]\(IMD.*\)\/IMAGES\//\1/p'|sort |uniq

Hi,
Thank you so much for the above explanation it is crystal clear. Also need a small help.  how can we tweak the above series of commands  to grab  data based on date(it should get yesterday's data from the file)  and write it to a log file.

> awk'/STARTED/,/FINISHED/{if($0~/IMAGES[/]$/){print $2}}'
> 1. Perform this awk command - for lines that match pattern STARTED or FINISHED
No
In HD_sync.log.10, search the content between "STARTED" and "FINISHED", If any line contain "IMAGES/" or "IMAGES" at the end of line, then print the
second field of that line

> grab  data based on date(it should get yesterday
You need to provide the date format and the location of date in  a line, like
2011/09/29 or
2011-09-29

I can't hardcode the date. so is there a way to  tell it  to look for data sysdate -1

> can't hardcode the date.
You just need to look at the HD_sync.log.10 and post the date/time format in that file (more than one for comparison)

the date format in the file is:'Mon Sep 26 05:10:05 ' and this log file  gets data every hour  365 days . So everday I need to run the above command throught script and grab the data for yesterday . will it be possible, if so  can you please let me know.

Thanks,

@wesley_chen
>> search the content between "STARTED" and "FINISHED"
You are absolutely correct, sorry about that.

About adding the date filter: I would add this:

grep "`date -d 'yesterday' +"%a %b %d"`"

then add the (somewhat modified) original line to this:

grep "`date -d 'yesterday' +"%a %b %d"`" | awk  '/STARTED/,/FINISHED/{if($0~/IMAGES[/]$/){print $2}}' | sed -ne 's/^[^/]*[/]\(IMD.*\)\/IMAGES\//\1/p' | sort | uniq

Select allOpen in new window

The 'grep' line will filter the lines matching 'yesterday' in this format:
Mon Sep 26

oh my bad.  Yesterday in the sense I eas telling that it should get previous day data, but not grep for yestarday. It has to grep only date for previous day.

For example if I am executing the command tomorrow  it has to read the file and get today's data  as output.