Horizontal file concatenation

[aborruso]

Hi,
is there a way to concatenate two files horizontally?

Starting from

field1,field2
0,a
1,b
3,a

and

field3,field4
0,a
1,b
3,a

obtain

field1,field2,field3,field4
0,a,0,a
1,b,1,b
3,a,3,a

Thank you

[shenwei356]

csvtk join? or shell paste?

I'm not sure whether the two files have the same number of rows or share common fields values.

[tseemann]

I had a similar problem last night which I could not solve using csvtk

I want to merge two CSV files
They share an ID column (key)
The do NOT have any IDs in common
They have some columns in common
The have some columns different
Columns are in different order

join/merge --outer is not correct
concat i hoped would work but it does not

Example:

# FILE 1
ID,A,B,C
1,a,b,c
2,d,e,f

# FILE 2
ID,C,A,D,E
3,g,h,i,j
4,k,l,m,n
5,,,,

# "MERGE"
ID,A,B,C,D,E
1,a,b,c,,
2,d,e,f,,
3,h,,g,i,j
4,l,,k,m,n
5,,,,,

[aborruso]

I'm not sure whether the two files have the same number of rows or share common fields values.

In my example they have the same number of rows and they not share any fields.
I want only to know if it's possible to do in csvtk whate paste does (but it's not aware of CSV format).

Thank you

[shenwei356]

In my example they have the same number of rows and they not share any fields.
I want only to know if it's possible to do in csvtk what paste does (but it's not aware of CSV format).

Sorry, it does not support this.

[shenwei356]

I had a similar problem last night which I could not solve using csvtk

Again, let's use transpose

csvtk transpose f1.csv > f1.t.csv
csvtk transpose f2.csv > f2.t.csv

csvtk join f1.t.csv f2.t.csv -O | csvtk transpose | csvtk pretty 
ID   A   B   C    D   E
--   -   -   --   -   -
1    a   b   c        
2    d   e   f        
3    h       g    i   j
4    l       k    m   n
5

[shenwei356]

In my example they have the same number of rows and they not share any fields.
I want only to know if it's possible to do in csvtk what paste does (but it's not aware of CSV format).

There's a dirty solution by adding a column of row number, which can be removed after joining.

$ csvtk mutate f1.csv -n n | csvtk replace -f n -p '.+' -r '{nr}'
field1,field2,n
0,a,1
1,b,2
3,a,3

$ csvtk mutate f1.csv -n n | csvtk replace -f n -p '.+' -r '{nr}' > f1.n.csv
$ csvtk mutate f2.csv -n n | csvtk replace -f n -p '.+' -r '{nr}' > f2.n.csv

$ csvtk join -f n f1.n.csv f2.n.csv | csvtk cut -f '-n'
field1,field2,field3,field4
0,a,0,a
1,b,1,b
3,a,3,a

[aborruso]

@shenwei356 could I add the "paste" command as csvtk feature request?

[shenwei356]

You can just use shell paste

$ paste -d , a.csv b.csv 
field1,field2,field3,field4
0,a,0,a
1,b,1,b
3,a,3,a

Or csvtk transpose + concat

$ csvtk concat -H <(csvtk transpose a.csv) <(csvtk transpose b.csv) | csvtk transpose 
field1,field2,field3,field4
0,a,0,a
1,b,1,b
3,a,3,a

[aborruso]

You can just use shell paste

I think no, because paste is no CSV aware and if I have CSV that has a comma inside a field "a classic, content" I think I will have problems using paste.

I'm asking it as a feature proposal because it does not exist a cli tool that has paste command CSV aware.