Having Problems with AWK

Shell Programming and Scripting
1

So, I'm having a lot of crazy problems with Awk that I cannot understand. This one in particular is driving me nuts. Here is one section of my Awk script:

    print $0
    
    sub(/Jan/,"",$2)
    sub(/Feb/,"",$2)
    sub(/Mar/,"",$2)
    sub(/Apr/,"",$2)
    sub("May","",$2)
    sub(/Jun/,"",$2)
    sub(/Jul/,"",$2)
    sub(/Aug/,"",$2)
    sub(/Sep/,"",$2)
    sub(/Oct/,"",$2)
    sub(/Nov/,"",$2)
    sub(/Dec/,"",$2)

    print $0

and here is the single line that is in the datafile I am working with:

Thursday, May 29, 2014,4,11,20 PM GMT-0700,1,Hard to get mario and pokemon gamecube cases ,http,//www.listia.com/auction/17243312-hard-to-get-mario-and-pokemon-gamecube

I am trying to simply delete the word, "May", and do nothing else. I have tried this a few different ways, using a double-quoted string instead of RegExp, and the output was exactly the same. Here is the output I am getting:

Thursday, May 29, 2014,4,11,20 PM GMT-0700,1,Hard to get mario and pokemon gamecube cases ,http,//www.listia.com/auction/17243312-hard-to-get-mario-and-pokemon-gamecube
Thursday   29  2014 4 11 20 PM GMT-0700 1 Hard to get mario and pokemon gamecube cases  http //www.listia.com/auction/17243312-hard-to-get-mario-and-pokemon-gamecube

You can see that after running the script, I am somehow getting a very weird result. The word "May" has been stripped from the second field, but somehow all of the commas have also been stripped from the output.

For the record I am using Gawk under Ubuntu. Any help is greatly appreciated.

---------- Post updated at 05:55 AM ---------- Previous update was at 05:43 AM ----------

Here is a more complete explanation. This is the datafile:

"Thursday, May 29, 2014","4:11:20 PM GMT-0700","1",                      "Hard to get mario and pokemon gamecube cases"           ,"http://www.listia.com/auction/17243312-hard-to-get-mario-and-pokemon-gamecube"

This is the AWK script:

{
    gsub(/\ \ /, "")
    gsub(/\"/, "")
    
    gsub(/\:/, "\,")
       
    print $0
    
    sub(/Jan/,"",$2)
    sub(/Feb/,"",$2)
    sub(/Mar/,"",$2)
    sub(/Apr/,"",$2)
    sub("May","",$2)
    sub(/Jun/,"",$2)
    sub(/Jul/,"",$2)
    sub(/Aug/,"",$2)
    sub(/Sep/,"",$2)
    sub(/Oct/,"",$2)
    sub(/Nov/,"",$2)
    sub(/Dec/,"",$2)

    print $0
    
    gsub(/" AM"/, "\,AM", $4)
    gsub(/" PM"/, "\,PM", $4)
    gsub(/" GMT"/, "\,GMT", $4)
    
    gsub(/\ /,"", $2)

    gsub(/\ /,"", $3)

}

And, here are the two output lines, printed before and after the block of code pertaining to the months above.

Thursday, May 29, 2014,4,11,20 PM GMT-0700,1,Hard to get mario and pokemon gamecube cases ,http,//www.listia.com/auction/17243312-hard-to-get-mario-and-pokemon-gamecube
Thursday   29  2014 4 11 20 PM GMT-0700 1 Hard to get mario and pokemon gamecube cases  http //www.listia.com/auction/17243312-hard-to-get-mario-and-pokemon-gamecube

The command I am using to launch awk is:

gawk -F, -f <awkfile> <datafile>

hope that helps

2

Since you are modifying a field ( $2 ), the record gets recomputed and all input field separators FS ( a comma) get replaced by the default output field separators OFS (a single space). Try it like this:

gawk -F, -v OFS=, -f <awkfile> <datafile>
3

Just curious, I have the book, "Effective Awk Programming". Suffice it to say that is does not to a very effective job at actually teaching Awk. This is all it really has to say about OFS:

"It is output between the fields output by a print statement. Its default value is " ", a string consisting of a single space.

I am just wondering if you can tell me why the first print statement in my code leaves the commas, while the second one removes them? Thank you.

4

Because in the first case none of the fields have been altered ( $1 .. $NF ) . The preceding gsubs operate on the record itself ( $0 ) and then the record does not get recomputed and thus FS does not get replaced by OFS .

5

Hi/

My copy says:

EAP 2nd Edition, page 42, A Robbins, SSC.

I think the real issue is changing the field, although I can see your point perhaps desiring that references to OFS would be more completely covered in the index.

Best wishes ... cheers, drl

6

Thank you guys for your help. I believe drl hit the nail on the head with the index. I originally read the Aho/Weinberger book many years ago and found it very simple and easy to follow. I misplaced the book over the years and thought I could save a buck with the Robbins book (2nd Edition), but I have also noticed that the index is very sparse, and even does not provide info for many topics that are shown prominently in the ToC.

I have company today, but as soon as I get a chance I will try these solutions and give some more thanks.