What does mv do when changing names?

Question

I have a very peculiar problem, following an answer in Rename multiples files using Bash scripting I am trying to rename some files. This is the code I am using

#!/bin/bash
for file in *.HHZ_00; do mv $file ${file/${file:14:2}/00}; done

What I understand this to do is to replace the 2 characters following character index 14 (starting from zero) to 00.

I am using 2 subsets of files to test it, one is this

2018113154211.25.AGCC.HHZ_00
2018113154211.25.APAC.HHZ_00

When running the code these successfully rename to

2018113154211.00.AGCC.HHZ_00
2018113154211.00.APAC.HHZ_00

The second subset fails at this, this subset is

2018070220829.28.AGCC.HHZ_00
2018070220829.29.APAC.HHZ_00

These rename unsuccessfully to

2018070220829.00.AGCC.HHZ_00
2018070220800.29.APAC.HHZ_00

Notice how the indices 11 and 12 are changed for the second file, not the 14 and 15 as I want it to, very curiously now the SECOND time I run it, using the last files as input, I get

2018070220829.00.AGCC.HHZ_00
2018070220800.00.APAC.HHZ_00

This is partially successful, but now the second filename is ruined...what am I doing wrong? Is it because the 29 is repeating in these files?

If someone has a solution using the rename command that would be great too, I tried using it but I am unfamiliar with the syntax.

Yes it's because the characters repeat; ${file/${file:14:2}/00} replaces the first instance of whatever characters occur in positions 14-15 — steeldriver, Jun 12 '20 at 19:18
... you could simply use "${file:0:14}00${file:16}" I think? — steeldriver, Jun 12 '20 at 19:22
You might find https://stackoverflow.com/questions/11809666/rename-files-using-regular-expression-in-linux helpful if you're comfortable with regex — einfeyn496, Jun 12 '20 at 19:27
@steeldriver That's an answer in the comments. Let's turn it into a proper answer instead of just comment — Sergiy Kolodyazhnyy, Jun 12 '20 at 22:13
The ${…} substitution is done by the shell. It has nothing to do with mv. First work out what you want to tell mv to do. Then workout how to tell the shell, to tell mv to do it. — ctrl-alt-delor, Jun 13 '20 at 21:59

score 1 · Accepted Answer · answered Jun 13 '20 at 00:30

Yes it's because the characters repeat; ${file/${file:14:2}/00} replaces the first instance of whatever characters occur in positions 14-15.

A simple way to do it in bash, without the nested parameter expansion, would be

$ for file in *.HHZ_00; do 
    echo mv "$file" "${file:0:14}00${file:16}"
  done
mv 2018070220829.28.AGCC.HHZ_00 2018070220829.00.AGCC.HHZ_00
mv 2018070220829.29.APAC.HHZ_00 2018070220829.00.APAC.HHZ_00
mv 2018113154211.25.AGCC.HHZ_00 2018113154211.00.AGCC.HHZ_00
mv 2018113154211.25.APAC.HHZ_00 2018113154211.00.APAC.HHZ_00

POSIXly, you could use %% and # (twice) to remove the longest suffix and shortest prefix respectively:

$ for file in *.HHZ_00; do 
    pfx="${file%%.*}"; rem="${file#*.}"; sfx="${rem#*.}" 
    echo mv "$file" "${pfx}.00.${sfx}"
  done
mv 2018070220829.28.AGCC.HHZ_00 2018070220829.00.AGCC.HHZ_00
mv 2018070220829.29.APAC.HHZ_00 2018070220829.00.APAC.HHZ_00
mv 2018113154211.25.AGCC.HHZ_00 2018113154211.00.AGCC.HHZ_00
mv 2018113154211.25.APAC.HHZ_00 2018113154211.00.APAC.HHZ_00

Remove the echo once you are satisfied that it is doing the right thing.

score 0 · Answer 2 · edited Jun 12 '20 at 21:09

0

@steeldriver explained it wonderfully, I corrected this by using

#!/bin/bash
for file in *.HHZ_00; do mv $file ${file/${file:13:4}/.00.}; done

edited Jun 12 '20 at 21:09

slm

369,824

answered Jun 12 '20 at 19:24

Pablo

3

What does mv do when changing names?

2 Answers2