renaming files without the **rename** command

Question

I have a directory of files,

Measurements Panama 2009年03月22日 Session1.xml
Measurements Panama 2009年03月22日 Session2.xml
Measurements Panama 2009年03月22日 Session3.xml
Measurements Panama 2009年03月22日 Session4.xml
...
Measurements Panama 2009年03月22日 Session10.xml

and I want to remove the blank spaces and the - dash characters. I don't want to use the rename command because I don't know perl or regular expressions. From another post a recommendation for removing part of a file name was:

for f in *.png; do mv $f ${f#image}; done

this looks nice. I can't figure out what the # character does (I assume it removes the string "image" from the file names?), anyways -- how do I remove the spaces and dashes?

I get an error that the files are not a directory when doing

[jings]s0128943: for f in *.xml; do mv $f ${f//[ -]/};done
mv: target `MeasurementsPanama20090322Session10.xml' is not a directory
mv: target `MeasurementsPanama20090322Session1.xml' is not a directory

etc for all the files.

• Falmarri: Why do you think there will be an easy way without perl or regex?
• Vass: I don't know, maybe im wrong. Why doesn't the attempted script work?? plz...

Answer 1

You can find information the ${...} substitutions in your shell's man page, for example bash(1). The most common:

• ${var#word} remove shortest prefix: the value of variable var with the word prefix (if any) removed;

• ${var%word} remove shortest suffix: the value of variable var with the word suffix removed (if any).

• ${var/pattern/replacement} remove first occurrence of pattern

• ${var//pattern/replacment} remove all occurrences of pattern

So, in your example ${f#image} expands to the value of f (e.g., image01.png) removing the image prefix, so it yields the value 01.png.

The word and replacement part in the ${...} expansions are subject to the same wildcard expansions as filenames; therefore, if you want to remove spaces and -, you could use ${f//[ -]/} (replaces any occurrence of characters and - with a null string.

All details on the man page.

• Vass: I get an error that the files are not a directory[jings]s0128943: for f in *.xml; do mv $f ${f//[ -]/};done mv: target MeasurementsPanama20090322Session10.xml' is not a directory mv: target MeasurementsPanama20090322Session1.xml' is not a directory
• Riccardo Murri: @Vass Since your file names contain spaces you need to quote them, otherwise the shell will expand $f to separate words Measurements, Panama, etc. and thus make mv think that you want to move files Measuremens, Panama, etc. into a singleMeasurementsPanama... which must then be a directory. Use for f in *.xml; do mv "$f" "${f//[ -]/}"; done instead. (See Gilles' answer as well.)

Answer 2

If you have the perl rename (e.g. because you're on Debian or Ubuntu), it is the simplest way of skinning this particular cat (as in an individual cat — there are plenty of tools for mass renaming, and searching the archives of this site and Super User should find all the major ones).

rename 's/[- ]//g' *.xml

That being said, your script would have worked if you had followed the most important shell programming principle: all variable substitutions must be double-quoted. (Why do you need this extra bit of syntax fluff? Because there are cases where you actually want the unquoted behavior. But these cases are rather rare.)

for f in *.xml; do mv "$f" "${f//[ -]/}"; done

With zsh, you wouldn't need to write a loop: you could use the convenient zmv function.

autoload zmv
zmv '(*.xml)' '${1//[- ]/}'

• Vass: thanks alot. perl is a bit beyond me as well as zsh, but the other part is great.

Answer 3

You might also use sed to build the new name. There it would be

for i in *.xml; do mv "$i" `echo "$i" | sed -e "s/[ -]//g"`; done

• user unknown: Backticks are deprecated. Recommendet is $(...) instead, which is easy to nest, and not so easy to be overseen or confused with apostrophs: for i in *.xml; do mv "$i" $(echo "$i" | sed -e "s/[ -]//g"); done