sort | uniq -c equivalent

[alperyilmaz]

Is your feature request related to a problem? Please describe.
I want to count rows, just like sort | uniq -c does.. whatever in the row, maybe single value, or multiple values. Neither qsv count nor qsv frequency does this.

Describe the solution you'd like
Maybe qsv count might have --rows option.

Describe alternatives you've considered
sqlp might be used for SQL based approach but I was looking for a practical sort | uniq -c alternative

Additional context
One of the qsv functions should achieve the same output

$ echo -e "a,1\nb,2\na,1" | sort | uniq -c
      2 a,1
      1 b,2

[jqnatividad]

Hi @alperyilmaz ,

You can achieve this with qsv frequency by specifying --no-headers and a --delimiter that doesn't occur in the file:

$ echo -e "a,1\nb,2\na,1" | qsv frequency --no-headers --delimiter '|'
field,value,count
1,"a,1",2
1,"b,2",1

# to get exactly the output you want
$ echo -e "a,1\nb,2\na,1" | qsv frequency --no-headers --delimiter '|' \
  | qsv select count,value | qsv behead \
  | qsv fmt --quote-never --out-delimiter ' '
2 a,1
1 b,2

[ondohotola]

RTFM :-)-O

[alperyilmaz]

come on! I was just asking for a practical solution.. I tried frequency and it needed clean up. I usually use awk, and it's not practical. I was hoping for some practical counting with one or two commands.. I read the FM..

[13minutes-yt]

try this in awk (or mawk for faster executions)
echo -e "a,1\nb,2\na,1" |awk -F, '{a[$1","$2]++} END {for (i in a) print a[i],i }'

[aborruso]

Using awk it's great in this way

echo -e "a,1\nb,2\na,1" | awk '!seen[$0]++'

[jqnatividad]

Moved it to Discussions as its the closest thing we have to a "Fine Manual" 😉

[aborruso]

Hi @alperyilmaz one tool that I use and find convenient is Miller.

My sample input is this

a,1
b,2
a,1

If I run

mlr --nidx --ifs '|' count-distinct -f 1 input.txt

I get

a,1 2
b,2 1

I use "|" as separator, so long as it is not in the csv file, to have only one field (one field that contains the entire row).

I know, I am off-topic, but I am pleased to share a related experience.

[ggrothendieck]

Not as simple as you wanted, certainly what you propose is simpler and I know you specifically said you did not want an SQL solution but nevertheless with SQL it is not that difficult either and it is only one line of code. Suppose file1.csv contains:

Name,Value
a,1
b,2
a,1

Then

qsv sqlp file1.csv "select Name, Value, count(*) as n from file1 group by Name, Value order by Name, Value"

giving

Name,Value,n
a,1,2
b,2,1
(2, 3)

[ggrothendieck]

One other comment. There is an huniq that supports huniq -c and is written in rust which would make it easier to incorporate into qsv or alternately it could be used independently of qsv.

https://crates.io/crates/huniq
https://github.com/koraa/huniq

Example run

>qsv behead file1.csv | huniq -c
2 a,1
1 b,2