Use awk to remove lines based on a column from another file - awk

I have the following code that works to extract lines from the multiple-column file_1 that have a value in the first column that appears in the single-column file_2:
awk 'NR==FNR{a[$1][$0];next} $0 in a {for (i in a[$0]) print i}' file_1 file_2
I got this code from the answer to this question: AWK to filter a file based upon columns of another file
I want to change to code to do the opposite, namely to remove every line from file_1 where the first column matches any value that appears in the single-column file_2. How do I do this?


awk 'NR==FNR { arr[$0]="1";next } arr[$1]!="1" { print $0 }' file2 file1
Process the second file first (NR==FNR) and create an array called arr, with the line ($0) as the key and 1 the value. Then when processing the next file (file1), check if the first space delimited field ($1) exists as a key in the array arr and if it doesn't, print the line.
Direct the output to a file if you want to store the results:
awk 'NR==FNR { arr[$0]="1";next } arr[$1]!="1" { print $0 }' file2 file1 > file3

Related

Countif like function in AWK with field headers

I am looking for a way of counting the number of times a value in a field appears in a range of fields in a csv file much the same as countif in excel although I would like to use an awk command if possible.
So column 6 should have the range of values and column 7 would have the times the value appears in column 7, as per below
>awk -F, '{print $0}' file3
f1,f2,f3,f4,f5,test
row1_1,row1_2,row1_3,SBCDE,row1_5,SBCD
row2_1,row2_2,row2_3,AWERF,row2_5,AWER
row3_1,row3_2,row3_3,ASDFG,row3_5,ASDF
row4_1,row4_2,row4_3,PRE-ASDQG,row4_5,ASDQ
row4_1,row4_2,row4_3,PRE-ASDQF,row4_5,ASDQ
>awk -F, '{print $6}' file3
test
SBCD
AWER
ASDF
ASDQ
ASDQ
What i want is:
f1,f2,f3,f4,f5,test,count
row1_1,row1_2,row1_3,SBCDE,row1_5,SBCD,1
row2_1,row2_2,row2_3,AWERF,row2_5,AWER,1
row3_1,row3_2,row3_3,ASDFG,row3_5,ASDF,1
row4_1,row4_2,row4_3,PRE-ASDQG,row4_5,ASDQ,2
row4_1,row4_2,row4_3,PRE-ASDQF,row4_5,ASDQ,2
#adds field name count that I want:
awk -F, -v OFS=, 'NR==1{ print $0, "count"}
NR>1{ print $0}' file3
Ho do I get the output I want?
I have tried this from previous/similar question but no joy,
>awk -F, 'NR>1{c[$6]++;l[NR>1]=$0}END{for(i=0;i++<NR;){split(l[i],s,",");print l[i]","c[s[1]]}}' file3
row4_1,row4_2,row4_3,PRE-ASDQF,row4_5,ASDQ,
,
,
,
,
,
very similar question to this one
similar python related Q, for my ref

I would harness GNU AWK for this task following way, let file.txt content be
f1,f2,f3,f4,f5,test
row1_1,row1_2,row1_3,SBCDE,row1_5,SBCD
row2_1,row2_2,row2_3,AWERF,row2_5,AWER
row3_1,row3_2,row3_3,ASDFG,row3_5,ASDF
row4_1,row4_2,row4_3,PRE-ASDQG,row4_5,ASDQ
row4_1,row4_2,row4_3,PRE-ASDQF,row4_5,ASDQ
then
awk 'BEGIN{FS=OFS=","}NR==1{print $0,"count";next}FNR==NR{arr[$6]+=1;next}FNR>1{print $0,arr[$6]}' file.txt file.txt
gives output
f1,f2,f3,f4,f5,test,count
row1_1,row1_2,row1_3,SBCDE,row1_5,SBCD,1
row2_1,row2_2,row2_3,AWERF,row2_5,AWER,1
row3_1,row3_2,row3_3,ASDFG,row3_5,ASDF,1
row4_1,row4_2,row4_3,PRE-ASDQG,row4_5,ASDQ,2
row4_1,row4_2,row4_3,PRE-ASDQF,row4_5,ASDQ,2
Explanation: this is two-pass approach, hence file.txt appears twice. I inform GNU AWK that , is both field separator (FS) and output field separator (OFS), then for first line (header) I print it followed by count and instruct GNU AWK to go to next line, so nothing other is done regarding 1st line, then for first pass, i.e. where global number of line (NR) is equal to number of line in file (FNR) I count number of occurences of values in 6th field and store them as values in array arr, then instruct GNU AWK to get to next line, so onthing other is done in this pass. During second pass for all lines after 1st (FNR>1) I print whole line ($0) followed by corresponding value from array arr
(tested in GNU Awk 5.0.1)

You did not copy the code from the linked question properly. Why change l[NR] to l[NR>1] at all? On the other hand, you should change s[1] to s[6] since it's the sixth field that has the key you're counting:
awk -F, 'NR>1{c[$6]++;l[NR]=$0}END{for(i=0;i++<NR;){split(l[i],s,",");print l[i]","c[s[6]]}}'
You can also output the header with the new field name:
awk -F, -vOFS=, 'NR==1{print $0,"count"}NR>1{c[$6]++;l[NR]=$0}END{for(i=0;i++<NR;){split(l[i],s,",");print l[i],c[s[6]]}}'

One awk idea:
awk '
BEGIN { FS=OFS="," } # define input/output field delimiters as comma
{ lines[NR]=$0
if (NR==1) next
col6[NR]=$6 # copy field 6 so we do not have to parse the contents of lines[] in the END block
cnt[$6]++
}
END { for (i=1;i<=NR;i++)
print lines[i], (i==1 ? "count" : cnt[col6[i]] )
}
' file3
This generates:
f1,f2,f3,f4,f5,test,count
row1_1,row1_2,row1_3,SBCDE,row1_5,SBCD,1
row2_1,row2_2,row2_3,AWERF,row2_5,AWER,1
row3_1,row3_2,row3_3,ASDFG,row3_5,ASDF,1
row4_1,row4_2,row4_3,PRE-ASDQG,row4_5,ASDQ,2
row4_1,row4_2,row4_3,PRE-ASDQF,row4_5,ASDQ,2

Awk Remove lines if one column matches another column, and keep line if max value from another column

I have a file of ~8,000 lines. I am trying to remove the lines where when the 5th column matches (in this case ga2016mldlzd), but keep only the line with the max value in the 6th column. For example, if given this:
-25.559,129.8529,6674.560547,2.0,ga2016mldlzd,6
-25.5596,129.8565,6902.750651,2.0,ga2016mldlzd,7
-25.5450,129.830,969.8079427,2.0,ga2016mldlzd,8
-25.5450,129.834,57.04752604,2.0,ga2016mldlzd,9
-25.57067,129.856,7929.60612,2.0,ga2016mldlzd,10
remove all lines except the final line with 10 as the max value, to get this. I'm stumped as to how this could be done either in awk or sed?
-25.57067,129.856,7929.60612,2.0,ga2016mldlzd,10
If tried this:
awk -F, '!a[$5]++'
but I want to keep last column e.g., the column with '10', rather than the column with '6'. Thanks

Keep track of the max and line associated with that max and print at the end:
awk -F, '
{
if ($6>max[$5]) {
max[$5]=$6
tl[$5]=$0
}
}
END{
for (l in tl) print tl[l]
}' file
Prints:
-25.57067,129.856,7929.60612,2.0,ga2016mldlzd,10
The order of the file will be lost; ie, the groups may be reordered compared to the original file.
If you are dealing with a file with many different keys for $5 and not all of them could fit in memory, you could sort into blocks grouped by the fifth field and then by the numeric value of the sixth. Then have awk print the last line every time the fifth field changes. Since it is sorted, that will be the max:
sort -t , -k 5,5 -k 6n file |
awk -F, '
FNR==1{lf=$5;ll=$0}
lf!=$5{print ll}
{ll=$0; lf=$5}
END{print $0}'
# same print out
The second there will be way slower but way less memory for a large number of $5 uniq values.

If you want to maintain original order of lines then use this awk:
awk -F, 'NR==FNR {if ($6 > max[$5]) max[$5] = $6; next} $5 in max && max[$5] == $6' file file
-25.57067,129.856,7929.60612,2.0,ga2016mldlzd,10
If you want to filter for ga2016mldlzd while maintaining original order of lines then use this awk:
awk -F, '
NR==FNR {
if ($5 == "ga2016mldlzd" && $6 > max[$5]) {
max[$5] = $6
n = FNR
}
next
}
FNR == n' file file
-25.57067,129.856,7929.60612,2.0,ga2016mldlzd,10

Awk command has unexpected results when comparing two files

I am using an awk command to compare the first column in two file.
I want to take col1 of file1 and if there is a match in col1 of file2, update the "date updated" in the last column. If there is no match, I want to append the entire line of file1 to file2 and append a "date updated" value to that line as well. Here is the command I'm currently using:
awk 'FNR == NR { f1[$1] = $0; next }
$1 in f1 { print; delete f1[$1] }
END { for (user in f1) print f1[user] }' file1 file2
File1:
tnash,172.2,plasma-de+,serv01,datetimeNEW
jhwagner,169.4,plasma-de+,serv01,datetimeNEW
steadmah,161.1,plasma-de+,serv01,datetimeNEW
guillwt,158.3,plasma-de+,serv01,datetimeNEW
mwinebra,122.2,plasma-de+,serv01,datetimeNEW
jbomba,100.0,plasma-de+,serv01,datetimeNEW
mschrode,161.1,plasma-de+,serv01,datetimeNEW
File2:
jbomba,114.0,plasma-de+,serv01,datetimeOLD
mschrode,104.0,plasma-de+,serv01,datetimeOLD
deleteme,192.0,random,serv01,datetimeOLD #<---- Needs to be removed: WORKS!
Expected Output:(order does not matter)
tnash,172.2,plasma-de+,serv01,datetimeNEW
jhwagner,169.4,plasma-de+,serv01,datetimeNEW
steadmah,161.1,plasma-de+,serv01,datetimeNEW
guillwt,158.3,plasma-de+,serv01,datetimeNEW
mwinebra,122.2,plasma-de+,serv01,datetimeNEW
jbomba,100.0,plasma-de+,serv01,datetimeOLD #<---- NEED THIS VALUE
mschrode,161.1,plasma-de+,serv01,datetimeOLD #<---- NEED THIS VALUE
Current Output:(order does not matter)
tnash,172.2,plasma-de+,serv01,datetimeNEW
jhwagner,169.4,plasma-de+,serv01,datetimeNEW
steadmah,161.1,plasma-de+,serv01,datetimeNEW
guillwt,158.3,plasma-de+,serv01,datetimeNEW
mwinebra,122.2,plasma-de+,serv01,datetimeNEW
jbomba,100.0,plasma-de+,serv01,datetimeNEW #<----WRONG OUTPUT
mschrode,161.1,plasma-de+,serv01,datetimeNEW #<----WRONG OUTPUT
The Logic Broken Down:
If $usr/col1 in file2 does NOT exist in file1
remove entire line from file2
(ex: line3 in file2, user: deleteme)
If $usr/col1 in file1 does NOT exist in file2
append entire line to file2
(ex: lines 1-5 in file1)
So the issue is, when there IS a match between the two files, I need to keep the information from file2, not the information from file1. In the output examples above you'll see I need to keep the datetimeOLD from file2 along with the new information from file1.

Set field separator to comma, and read file2 first:
$ awk -F',' 'FNR==NR{a[$1]=$0;next} $1 in a{print a[$1];next} 1' file2 file1
tnash,172.2,plasma-de+,serv01,datetimeNEW
jhwagner,169.4,plasma-de+,serv01,datetimeNEW
steadmah,161.1,plasma-de+,serv01,datetimeNEW
guillwt,158.3,plasma-de+,serv01,datetimeNEW
mwinebra,122.2,plasma-de+,serv01,datetimeNEW
jbomba,114.0,plasma-de+,serv01,datetimeOLD
mschrode,104.0,plasma-de+,serv01,datetimeOLD

printing multiple NR from one file based on the value from other file using awk

I want to print out multiple rows from one file based on the input values from the other.
Following is the representation of file 1:
2
4
1
Following is the representation of file 2:
MANCHKLGO
kflgklfdg
fhgjpiqog
fkfjdkfdg
fghjshdjs
jgfkgjfdk
ghftrysba
gfkgfdkgj
jfkjfdkgj
Based on the first column of the first file, the code should first print the second row of the second file followed by fourth row and then the first row of the second file. Hence, the output should be following:
kflgklfdg
fkfjdkfdg
MANCHKLGO
Following are the codes that I tried:
awk 'NR==FNR{a[$1];next}FNR in a{print $0}' file1.txt file2.txt
However, as expected, the output is not in the order as it first printed the first row then the second and fourth row is the last. How can I print the NR from the second file as exactly in the order given in the first file?

Try:
$ awk 'NR==FNR{a[NR]=$0;next} {print a[$1]}' file2 file1
kflgklfdg
fkfjdkfdg
MANCHKLGO
How it works
NR==FNR{a[NR]=$0;next}
This saves the contents of file2 in array a.
print a[$1]
For each number in file1, we print the desired line of file2.
Solution to earlier version of question
$ awk 'NR==FNR{a[NR]=$0;next} {print a[2*$1];print a[2*$1+1]}' file2 file1
fkfjdkfdg
fghjshdjs
gfkgfdkgj
jfkjfdkgj
kflgklfdg
fhgjpiqog

Another take:
awk '
NR==FNR {a[$1]; order[n++] = $1; next}
FNR in a {lines[FNR] = $0}
END {for (i=0; i<n; i++) print lines[order[i]]}
' file1.txt file2.txt
This version stores fewer lines in memory, if your files are huge.

searching multiple patterns in awk

I've a text file of thousands of lines
:ABC:xyz:1234:200:some text:xxx:yyyy:11818:AAA:BBB
:ABC:xyz:6789:200:some text:xxx:yyyy:203450:AAA:BBB
:EFG:xyz:11818:200:some text:xxx:yyyy:154678:AAA:BBB
:HIJ:xyz:203450:200:some text:xxx:yyyy:154678:AAA:BBB
:KLM:xyz:7777:200:some text:xxx:yyyy:11818:AAA:BBB
.....
....
:DEL:xyz:1234:200:some text:xxx:yyyy:203450:AAA:BBB
I need to find more than one occurrence of the 9th column i.e the o/p should show
:ABC:xyz:1234:200:some text:xxx:yyyy:11818:AAA:BBB
:KLM:xyz:7777:200:some text:xxx:yyyy:11818:AAA:BBB
:ABC:xyz:6789:200:some text:xxx:yyyy:203450:AAA:BBB
:DEL:xyz:1234:200:some text:xxx:yyyy:203450:AAA:BBB
I tried:
awk -F ":" '$9 > 2 {split($0,a,":"); print $0}'
this prints all the records.

awk -F':' 'NR==FNR{cnt[$9]++;next} cnt[$9]>1' file file
or if you don't want to parse the file twice:
awk -F':' 'cnt[$9]++{printf "%s", prev[$9]; delete prev[$9]; print; next} {prev[$9]=$0 ORS}' file

This should do it in pure awk:
awk -F":" '{if( s[$9] ){ print } else if( f[$9] ){ print f[$9]; s[$9]=1; print }; f[$9]=$0 }'
Explanation:
The "f" array stores values of the 9th column that have occurred at least once.
The "s" array stores values of the 9th column that have occurred twice or more.
If the 9th column has occurred before, print the first occurrence, and this line.
If the 9th column has occurred twice or more before, print this line.

Here is another awk
awk -F: '{++a[$9];b[NR]=$0} END {for (i=1;i<=NR;i++) {split(b[i],c,":");if (a[c[9]]>1) print b[i]}}' file