I have a file that is a mapping between filename and the corresponding transcripts. The file name and transcripts is separated by a TAB character.

The transcription of the file may contain one or more words that are separated by a single blank space. Following is a layout of the file,

[filename] [tab space] [trancription]

In some lines, the transcript column is empty. Such lines will be of the form


i.e. there is no transcript available for that filename.

Now, my job is to make sure that I select only those lines that have a filename and transcription (that is the files whose transcript column is not empty).

I tried the following commands

(1) awk 'NF>2' filename

(2) awk 'NF==2' filename

(3) awk 'NF>1' filename

but did not get the results

In addition, when I used the command

(4) awk ' NF==2 {print $0} '   myfile  > newfile

I was also getting those lines with had only one column namely the "filename" field.

When I write NF<1 there is no output (as expected) When I write NF<2 again there is no output (strange, it should have displayed lines with only one column) When I write NF ==3 I am getting those lines which have exactly two columns (again confusing)

What's the catch? Its really confusing.

Now I am sending you the input sample

M07UP36A0821I410.wav    gaajara <bn>
M07UP36A0821I411.wav    tiina sau <pau> taintaaliisa
M07UP36A0821I412.wav    geehuun anya <bn>
M07UP36A0821I413.wav    geehuun daraa <babble>

Now I use the command

grep '^[^[:blank:]]\+[[:blank:]]\+[^[:blank:]]\+$' cll

This command is giving no output (neither on the terminal nor in the redirected file).

Now there is an INTERESTING thing to note:

When the input file contains

M07UP36A0822I413.wav    <bn> geehuun daraa <horn> <babble>
M07UP36A0822I415.wav    gudxqa piilaa <horn> <babble>
M07UP36A0822I416.wav    <vn> gudxqa
M07UP36A0822I417.wav    gudxqa
M07UP36A0822I418.wav    gudxqa anya <babble>
M07UP36A0822I419.wav    harii matxara <bn> <babble>

Again, on using the same command

grep '^[^[:blank:]]\+[[:blank:]]\+[^[:blank:]]\+$' foo

terminal has STARTED showing the output. The output in this was obtained as

M07UP36A0822I417.wav    gudxqa

My desired output for the file foo would be those lines that are complete (both first column and second column must be there) Here is the required output

M07UP36A0822I413.wav    <bn> geehuun daraa <horn> <babble>
M07UP36A0822I415.wav    gudxqa piilaa <horn> <babble>
M07UP36A0822I416.wav    <vn> gudxqa
M07UP36A0822I417.wav    gudxqa
M07UP36A0822I418.wav    gudxqa anya <babble>
M07UP36A0822I419.wav    harii matxara <bn> <babble>

I used the following command on the file cll (the first sample in my question)

awk -F'\t' '(NF !=2) { print "line: " NR " does not have 2 columns: " $0 ;}' cll

The result was displayed on the terminal. The result is

line: 1 does not have 2 columns: M07UP36A0821I40.wav 
line: 2 does not have 2 columns: M07UP36A0821I41.wav 
line: 3 does not have 2 columns: M07UP36A0821I410.wav    gaajara <bn>
line: 4 does not have 2 columns: M07UP36A0821I411.wav    tiina sau <pau> taintaaliisa
line: 5 does not have 2 columns: M07UP36A0821I412.wav    geehuun anya <bn>
line: 6 does not have 2 columns: M07UP36A0821I413.wav    geehuun daraa <babble>
  • 1
    "I tried ___ but did not get the results." What is that supposed to mean? What results did you get, exactly? The best thing to do on these types of questions is to give a sample of input and expected output.
    – Wildcard
    Feb 8 '16 at 7:46
  • 1
    Please edit with a sample of your real input file. awk 'NF==2' ought to be a correct solution to your problem as described, so something else is going on here. Feb 8 '16 at 8:00
  • "The best thing to do on these types of questions is to give a sample of input and expected output" . Is it possible to upload input file and the output file on this forum so that you could have a better look at it?
    – Upendra01
    Feb 8 '16 at 8:06
  • 1
    Just copy and paste a representative extract from the input file into the question, and put it in a code block (indented four spaces, like what I did in my edit). Feb 8 '16 at 8:10
  • I have given one small portion of the input file (see the modified question). This produced nothing on the output
    – Upendra01
    Feb 8 '16 at 9:15

Using grep:

grep -E '^[^\s]+\s+[^\s]+$' file.txt
  • [^\s]+ will contain the first column, then any number of white spaces \s+, followed by the second (last) column

  • ^ indicates the start and $ indicates the end of the line


grep '^[^[:blank:]]\+[[:blank:]]\+[^[:blank:]]\+$' file.txt
  • When I use this command on the terminal, there is no output(no text at all on the terminal). However doing wc -l on the output gives the output 539. Then I tried to redirect the output of this command (grep command that you mentioned) to some text file, I got the results. Why was there nothing to display on the terminal? Also, when I redirected the output to a file, I got getting some lines with empty second column.
    – Upendra01
    Feb 8 '16 at 7:59
  • @Snowbell check the second one..
    – heemayl
    Feb 8 '16 at 8:06
  • grep '^[^[:blank:]]\+[[:blank:]]\+[^[:blank:]]\+$' file.txt. This command is behaving in the same manner as the first one(no output at terminal; output is displayed only when command output is redirected to some file). Moreover, this command produces a files with (116) lines only. Also, there were only few lines with second column that was not empty.
    – Upendra01
    Feb 8 '16 at 8:10
  • @Snowbell Please edit your question to add some example text so that we can verify this..
    – heemayl
    Feb 8 '16 at 8:14
  • I have added two samples of my input file and the corresponding output that I received...
    – Upendra01
    Feb 8 '16 at 9:24

Other answers, using \s+ are over-matching! \s matches spaces, too, which occur in the data. What you are looking for is "Field_not_containing_TAB" followed by "Tab", followed by "Field_not_containing_TAB". or, in regexp-speak,

egrep '^[^Tab]+Tab[^Tab]+$' file.txt

Note: there should be no Space characters between the ''s in my egrep rexexp. You see them there due to my use of the kbd formatting tag, which I had to use to make Tab visible.

The regexp matches (or with the -v option to egrep, does not match) the supplied data.

  • In this case, whatever is the input file, the same is the output file. NO changes at all.....I used this command on the 'cll' file (as described in my question)
    – Upendra01
    Feb 8 '16 at 10:52

For you, 2 columns is: something[TAB]something, something can have a space in it. So just tell awk your Field Separator is a tab:

awk -F'\t' '(NF==2)'

to find out the others :

awk -F'\t' '(NF !=2) { print "line: " NR " does not have 2 columns: " $0 ;}'

Edit : you seem to "mislead" about the file's structure : you said it will contain either 1 word, or 1 word, a TAB, and 1 or several words (separated with spaces). If so, the above should have worked. However, given the result of the above command on your file, I guess it is instead: 1 word, [spaces or tabs], and maybe words[and spaces or tabs in between]

Thus the solution becomes:

awk -F'[\t ]+'  '(NF>=2)'    #words are separated by space&tabs, and we want at least 2 words [a filename, and its description]

a variant: outputs any line that have the 2nd word non empty:

awk -F'[\t ]+'  '(length($2)>0)'    #words are separated by space&tabs, and we want athe 2nd word to be non empty
#or, one that also verifies the 1st word is non empty:
awk -F'[\t ]+'  '(length($1)>0 && length($2)>0)'    #1st and 2nd words are non-empty
  • I used the command. I have put the results that I got in the question...kindly see it
    – Upendra01
    Feb 8 '16 at 10:09
  • if it didn't work, it may be because you don't have just a tab ? or no tab at all? Feb 8 '16 at 10:52
  • When you will copy the content of cll file in a gedit file, the tabs that are there no longer appear in the gedit file. You have to manually put the tabs before executing any command. I tried this also.....The command awk -F'\t' '(NF !=2) { print "line: " NR " does not have 2 columns: " $0 ;}' shows no output at all
    – Upendra01
    Feb 8 '16 at 11:10
  • @Snowbell : It didn't show any output because it contained always 2 fields separated by 1 tab. However, try the latest version : it should work if you have TABs or if you have SPACES, in both cases. Feb 8 '16 at 12:01
  • awk -F'[\t ]+' '(length($2)>0)' This command is working. Thanks. could you please explain "[\t ]+" in the command
    – Upendra01
    Feb 9 '16 at 6:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.