sep output inconsistent when using -N vs -n and the -H flag

[davised]

Prerequisites

• [X] make sure you're are using the latest version by csvtk version
• [X] read the usage

$ csvtk version
csvtk v0.25.0

$ csvtk sep -h
separate column into multiple columns

Usage:
  csvtk sep [flags]

Aliases:
  sep, separate

Flags:
...
      --merge           only splits at most N times, exclusive with --drop
...
  -n, --names strings   new column names
  -N, --num-cols int    preset number of new created columns
...
  -s, --sep string      separator
...

Global Flags:
...
  -H, --no-header-row          specifies that the input CSV file does not have header row
...

Describe your issue

• [x] describe the problem
• [x] provide a reproducible example

csvtk sep does not provide the same output when headers are disabled with -H option.

See examples below:

$ cat 96_well_sample.txt | csvtk sep -H -s '_' -N 3 --merge | csvtk pretty -H | head
A1_S1_L001_R1_001.fastq.gz     A1    S1    L001   R1   001.fastq.gz
A1_S1_L001_R2_001.fastq.gz     A1    S1    L001   R2   001.fastq.gz
B1_S2_L001_R1_001.fastq.gz     B1    S2    L001   R1   001.fastq.gz
B1_S2_L001_R2_001.fastq.gz     B1    S2    L001   R2   001.fastq.gz
C1_S3_L001_R1_001.fastq.gz     C1    S3    L001   R1   001.fastq.gz
C1_S3_L001_R2_001.fastq.gz     C1    S3    L001   R2   001.fastq.gz
D1_S4_L001_R1_001.fastq.gz     D1    S4    L001   R1   001.fastq.gz
D1_S4_L001_R2_001.fastq.gz     D1    S4    L001   R2   001.fastq.gz
E1_S5_L001_R1_001.fastq.gz     E1    S5    L001   R1   001.fastq.gz
E1_S5_L001_R2_001.fastq.gz     E1    S5    L001   R2   001.fastq.gz

As you can see, the -N 3 --merge is not being respected. However, if headers are added:

$ cat 96_well_sample.txt | csvtk add-header -n fastq | csvtk sep -s '_' -n 'sample,sequence,suffix' --merge | csvtk pretty | head
fastq                          sample   sequence   suffix
----------------------------   ------   --------   --------------------
A1_S1_L001_R1_001.fastq.gz     A1       S1         L001_R1_001.fastq.gz
A1_S1_L001_R2_001.fastq.gz     A1       S1         L001_R2_001.fastq.gz
B1_S2_L001_R1_001.fastq.gz     B1       S2         L001_R1_001.fastq.gz
B1_S2_L001_R2_001.fastq.gz     B1       S2         L001_R2_001.fastq.gz
C1_S3_L001_R1_001.fastq.gz     C1       S3         L001_R1_001.fastq.gz
C1_S3_L001_R2_001.fastq.gz     C1       S3         L001_R2_001.fastq.gz
D1_S4_L001_R1_001.fastq.gz     D1       S4         L001_R1_001.fastq.gz
D1_S4_L001_R2_001.fastq.gz     D1       S4         L001_R2_001.fastq.gz

96_well_sample.txt

[shenwei356]

Hi Ed, that's a bug, I've fixed it. Thanks for reporting this.

$  cat 96_well_sample.txt | csvtk sep -H -s '_' -N 3 --merge | csvtk pretty -H | head
A1_S1_L001_R1_001.fastq.gz     A1    S1    L001_R1_001.fastq.gz
A1_S1_L001_R2_001.fastq.gz     A1    S1    L001_R2_001.fastq.gz
B1_S2_L001_R1_001.fastq.gz     B1    S2    L001_R1_001.fastq.gz

$ cat 96_well_sample.txt | csvtk add-header -n fastq | csvtk sep -s '_' -n 'sample,sequence,suffix' --merge | csvtk pretty | head
fastq                          sample   sequence   suffix
----------------------------   ------   --------   --------------------
A1_S1_L001_R1_001.fastq.gz     A1       S1         L001_R1_001.fastq.gz
A1_S1_L001_R2_001.fastq.gz     A1       S1         L001_R2_001.fastq.gz
B1_S2_L001_R1_001.fastq.gz     B1       S2         L001_R1_001.fastq.gz

Please use the new binaries.

• csvtk_darwin_amd64.tar.gz
• csvtk_linux_amd64.tar.gz
• csvtk_darwin_arm64.tar.gz