Search pattern from my file?

[razorxtr]

Could use some help with a project im working on:

As an example, I created the following folder structure on my system:
/test/NA/123tst/some files
/test/PM/145test/some files
/test/GN/1236test/some files
/test/RG/234test/some files
...


I created a file named "pattern" with the following directories that i know will always exist in the path i am searching. (Normally i would extract this list from a database)

123tst
145test
1236test
234test
...

I want to run a search (find, grep?) in order to search each line my above pattern file, locate the complete directory path for every file that matches the directory and append it to a "result" file like this:

/test/NA/123tst/some files
/test/NA/123tst/some files
/test/NA/123tst/some files
/test/NA/123tst/some files
/test/PM/145test/some files
/test/PM/145test/some files
/test/PM/145test/some files
/test/GN/1236test/some files
/test/GN/1236test/some files
/test/GN/1236test/some files
/test/RG/234test/some files
/test/RG/234test/some files
/test/RG/234test/some files


What would be the best way to complete this?


I am playing around with:
find . -type d -iname "directory for every line in pattern file" but this only returns the path without any files under it and i cant figure out how to make it loop for every line in my pattern file. Maybe you guys have some ideas using a loop statement in Bash to accomplish this?


Thanks

[nmset]

If you don't have unwanted folders in /test, the following should do the job :

find /test -type f > out.file

[Cabhan]

So you want to find every file under /test such that the absolute pathname contains a line from your pattern file.

This is a bit complicated: find can't canonicalize paths, and it doesn't support taking regular expressions from a file. Therefore, we need to use a bit of a script here.

For the following, I will assume that the pattern file contains no "strange" characters (that is, it's all alphanumeric).

#!/bin/bash

IFS=$'\012'
for file in $(find /test -type f); do
    fullpath=$(readlink -m "$file")

    exec 3< "pattern"

    while read -u 3 pattern; do
        if echo "$fullpath" | grep -q "$pattern"; then
            echo "$fullpath"
            break
        fi
    done

    exec 3<&-
done

So what does this do? Well, it starts by using find to get every file in the tree starting at /test. For each of these, it gets the absolute path, and then it opens the "pattern" file and looks at each pattern in it to match against the file path.

If you could create a single regular expression of the patterns ("123tst|145test|1236test|234test"), and were careful about always using absolute paths to run find, this could be dramatically simplified to:

find /path/to/search -path "$singleregex" -type f