Underscore nullifies/ignored in replacement between matched groups

[ammaraziz]

Hi Shenwei,

Thanks for the awesome. I came across a weird issue. I am trying to replace this pattern:

Input:

1234_OC43_S17_R1.fastq.gz
1234_OC43_S17_R2.fastq.gz
1235_OC43_S24_R1.fastq.gz
...
etc

Expected output:

S17_1234_OC43_R1.fastq.gz
S17_1234_OC43_R2.fastq.gz
S24_1235_OC43_R1.fastq.gz
...
etc

I want to move S\d\d pattern from the middle to the first of the string. eg this: XXX_XXX_S17_R1.fastq.gz becomes S17_XXX_XXX_R1.fastq.gz.

Command I used:

brename -p "(.*)_(S\d+)_(.*)" -r '$2_$1_$3' -d

Output:

  [OK] 1234_OC43_S17_R1.fastq.gz -> R1.fastq.gz
  [OK] 1234_OC43_S17_R2.fastq.gz  -> R2.fastq.gz
  [overwriting newly renamed path] 1235_OC43_S24_R1.fastq.gz -> R1.fastq.gz

Odd! _ seems to completely nullify the group string eg $1

Replacing underscores with - fixes this issue:

brename -p "(.*)_(S\d+)_(.*)" -r '$2-$1-$3' -d

To me the issues seems to be replaced to the regex groups. Because when I run this to achieve the desired result it works:

brename -p "-" -r '_' -d

Thanks!

Versions:

brename v2.14.0
LSB Version:    core-11.1.0ubuntu4-noarch:security-11.1.0ubuntu4-noarch
Distributor ID: Linuxmint
Description:    Linux Mint 21
Release:        21
Codename:       vanessa

[shenwei356]

It's a common issue. see #33

[ammaraziz]

Apologies I missed that issue when I did a quick search.

[shenwei356]

No worries, it happens a lot. I really have to add this to attention or FAQs.

[shenwei356]

Added:

Special cases of replacement string:
 *1. Capture variables better be in the format of '${1}'.
    a). If the capture variable is followed with space or other simple, it's OK:
            -r '$1 abc'
    b). If followed by numbers, characters, or underscore. That is ambiguous:
            -r '$1abc' actually refers to the variable '1abc', please use '${1}abc'.
            -r '$2_$1' actually refers to the variable '2_', please use '${2}_${1}'.
  2. Want to replace with a character '$',
    a). If using '{kv}', you need use '$$$$' instead of a single '$':
            -r '{kv}' -k <(sed 's/\$/$$$$/' kv.txt)
    b). If not, use '$$'. e.g., adding '$' to all numbers:
            -p '(\d+)' -d -r '$$${1}'

[ammaraziz]

Thank you @shenwei356 !

If I may offer a suggestion on the wording:

Capture variables better be in the format of '${1}'.

[it] better be is an idiom in English that conveys an angry/threatening tone. I suggest changing the wording to something along the lines of:

Capture variables should be in the format of '${1}' to reduce errors

I hope you don't mind the correction.

Thank you again for taking the time to respond to my issue so quickly.

[shenwei356]

Thank you Ammar very much!
I never realized that, because the tones of these two phrases are reversed in Chinese ...
I'll correct that.

[ammaraziz]

That's super interesting. I can see how the word 'should' can be a bit forward.

I generally avoid the word as it's ambiguous but I couldn't think of another way to phrase it.

Thanks again!

[shenwei356]

"recommended to be" might be a good alternative.