Re: Help needed

[Bernhard Voelker]

On 09/08/2015 01:18 PM, Carlos Lopes wrote:
> Hi,
>
> I'm trying to use findutils to find files based on their name, but since
> there are many patterns to search, I would like to know if it is possible to
> pass a file with the patterns:
>
> Ex:
> This works fine:
>
> find w:\ -iname "*33*"
>
> (it returns a list of files that have "33" on the name )
>
> But I wanted to use a file instead of "*33*", is it possible?
> For example, if I have a test.txt file with this data:
>
> 33
> 43435
> 343434
> 756565
>
> How could I pass the information so that the program searches all
> information in the file with just one command?

GNU find doesn't have an option to read the patterns from a file,
and it would maybe be problematic to say if such patterns should
be treated like as for the -name, -iname, -regex etc. options.

Therefore, I'd go with a more practical approach e.g. based on

  find . | grep -f test.txt

But again, you need to know what the patterns should match.
E.g., if it should only match the basename of each hit, then
you should wrap the patterns in test.txt to match only the basename
like maybe:

  $ find . | grep -f <( sed 's#^#/[^/]*#; s#$#[^/]*#' test.txt)
  ./333

If the number of patterns is not that big, then I'd maybe generate
the command line for 'find' like this:

  $ sed "s/^/-iname '*/; 2,$ s/^/-o /; s/$/*'/" test.txt | tr '\n' ' '
  -iname '*33*' -o -iname '*43435*' -o -iname '*343434*' -o -iname '*756565*'

> Note: I'm working on a windows7 environment.

There exist several implementations of GNU find for Windows,
e.g. the one in Cygwin.  That in particular comes with a nice
shell which allows the same syntax as used in the examples
above.

However, this is all out of scope of the upstream find project.

Have a nice day,
Berny