awk/grep in bash script

[screechy]

There is a file xyz.txt, the third line of which can be something like one of the following cases:

a) random word, random word , word1, word2, word3, word4, random word
or
b) random word, random word, word1, random word
or
c) random word, word2, word3, random word, random word
or
d) random word, random word, random word

I want to do the following using awk or grep in bash script.

{
If word1, word2, word3, and word4 all are absent
print 'a' in file output.txt

if only word1 is present
print 'b' in file output.txt

if word1 && word2 are present
print 'c' in file output.txt


}

Is there a very efficient and fast method of doing it? :confused:

[ofnuts]

There is a file xyz.txt, the third line of which can be something like one of the following cases:

a) random word, random word , word1, word2, word3, word4, random word
or
b) random word, random word, word1, random word
or
c) random word, word2, word3, random word, random word
or
d) random word, random word, random word

I want to do the following using awk or grep in bash script.

{
If word1, word2, word3, and word4 all are absent
print 'a' in file output.txt

if only word1 is present
print 'b' in file output.txt

if word1 && word2 are present
print 'c' in file output.txt


}

Is there a very efficient and fast method of doing it? :confused:
Yes, there are two efficient ways, that both involve regular expressions:

Awk: The typical awk process is "if this line matches the filter, then do this" where the filter can be a regular expression and "do this" can be printing something: some arbitrary string, or the line itself (or part of it).

Sed: Using very similar regexps to those used with awk, you substitute the matching lines with a, b, c, or d.

So your only problem with both methods is to come up with regular expressions that match each case (and don't match the others of course).

[binouche22]

Hi,

Here is a proposal :


awk 'NR == 3 {if(/.*word.*word2.*word3.*/) print "a"; else if(/.*word.*word2.*/) print "b"; else if(/.*word.*word2.*word3.*/) print "c"}' input.txt > output.txt

hth,
binouche22

[screechy]

thank you binouche22! But, does it suffice the first condition of the words being absent?

[binouche22]

oops! I answered too hastly ;)!
There's probably a more elegant way to do it but this should work:



awk 'NR == 3 {
if (/.*word1.*/ && $0 !~ /.*word2.*word3.*word4.*/ ) print "b";
else if ($0 !~ /.*word1.*word2.*word3.*word4.*/) print "a";
else if (/.*word1.*word2.*/) print "c";
}' input.txt > output.txt

[screechy]

I am trying to test the code! I still have a question! What if instead of the third line, it is the last line of the input file that the operation needs to be done? What can we use instead of NR == 3 ? Thanks a lot!

[screechy]

and what if the word1, word2 and word3 are not necessarily in the same order. i.e. if it is word3, word2, word1, does the regular expression change? :s I think i solved the last line problem. awk 'END{print}' should solve it, I think!

[binouche22]

If the order is not predictable, you can use the index() function to check whether a word is present or not:


awk 'END{
if (!index($0,"word1") > 0 &&
!index($0,"word2") > 0 &&
!index($0,"word3") > 0 &&
!index($0,"word4") > 0 ){
print "a";
} else if (index($0,"word1") > 0 &&
!index($0,"word2") > 0 &&
!index($0,"word3") > 0 &&
!index($0,"word4") > 0 ){
print "b";
} else if (index($0,"word1") > 0 &&
index($0,"word2") > 0 &&
!index($0,"word3") > 0 &&
!index($0,"word4") > 0 ){
print "c";
}
}' input.txt > output.txt

[screechy]

I removed the .* part from /.*word/ and it worked too! But, this is more elegant! Thank you! :)

[ofnuts]

If the order is not predictable, you can use the index() function to check whether a word is present or not:


awk 'END{
if (!index($0,"word1") > 0 &&
!index($0,"word2") > 0 &&
!index($0,"word3") > 0 &&
!index($0,"word4") > 0 ){
print "a";
} else if (index($0,"word1") > 0 &&
!index($0,"word2") > 0 &&
!index($0,"word3") > 0 &&
!index($0,"word4") > 0 ){
print "b";
} else if (index($0,"word1") > 0 &&
index($0,"word2") > 0 &&
!index($0,"word3") > 0 &&
!index($0,"word4") > 0 ){
print "c";
}
}' input.txt > output.txt

If you want to go this way, wouldn't


if ($3 == "word1" && $4 == "word2" && $5 == "word3" && $6 == "word4")
print "a";

cover all bases for "a") (once you add a check for the presence of a 7th word) ?

[binouche22]

If you want to go this way, wouldn't
Code:
if ($3 == "word1" && $4 == "word2" && $5 == "word3" && $6 == "word4")
print "a";
cover all bases for "a") (once you add a check for the presence of a 7th word) ?

This only works if "word1" is in 3rd position, "word2" in 4th position, etc... but afaiu screechy needs the script to work whatever the positions of the words are.