cp select files recursively?

[xSauronx]

just had this on a test and couldnt seem to figure it out

the goal is to selectively go through a directory "dir1/" and its subdirectories and copy all files starting in a,e,i,o,u and ending in ".py" to dir2/

is there a way to do that with cp? googling around gave a couple of suggestions that didnt seem to work either (one using grep, another using xargs)

any ideas? cant go back and change anything but im curious

thank

 

[Gooberlx2]

probably could use find with -exec

Here's a thread about it

 

[LCTSI]

use `find` with the -regex option coupled with the -exec option

ie:
find /path/dir1 -type f -regex 'regexpattern' -exec cp {} /path/dir2 \;

the regex pattern is the hard part.
use -iregex if you want it to be case insensitive

 

[Gooberlx2]

Here're a set of commands that worked for me

I had files <a,e,i,o,u>test.py, mixed in with others, in various nested subdirs

mkdir /dir2/
cd /dir1/
find . -type d -exec mkdir -p /dir2/{} \;
find . -type f -regex '.+/[aeiou].*\.py' -exec cp {} /dir2/{} \;

edits: had to adjust regex to filter unwanted patterns (ta.py, ablahpy, etc...)

 

[xSauronx]

yeah, thats all way more involved than anything im used to doing (im familiar with linux to a degree, but never got into playing with regex or xargs or anything similar)

thanks for the replies

 

[Gooberlx2]

yeah, thats all way more involved than anything im used to doing (im familiar with linux to a degree, but never got into playing with regex or xargs or anything similar)

thanks for the replies

I like this regex tutorial. Regex syntax makes things look cryptic and complicated, but it's really not (and incredibly useful) once you understand how it works.

 

[xSauronx]

bookmarked and part of my todo list!

 

[LCTSI]

mkdir /dir2/
cd /dir1/
find . -type d -exec mkdir -p /dir2/{} \;
>-exec cp {} /dir2/{} \;

edits: had to adjust regex to filter unwanted patterns (ta.py, ablahpy, etc...)

Fails if a directory is named alpha.py, then it returns everything listed under that directory.

 

[Gooberlx2]

Fails if a directory is named alpha.py, then it returns everything listed under that directory.

:hmm: Not everything, but yes, it does match stuff it shouldn't.

sample directory/file structure [ (#) where it should match]:

/test/temp1/
	./alpha.py/
		./a.py (1)
		./apy
		./atest.py (2)
		./atestpy
		./etest.py (3)
		./etest.txt
	./temp/
		./alpha.py/
			./o.py (4)
			./o.py.txt
			./otesteapy.py (5)
			./otesteapy.txt
			./test.py
		./alphapy/
			./otesteapy.py (6)
			./otesteapy.txt
			./test.py
		./otets.py (7)
		./totest.pypy
	./atest.py (8)
	./ta.py
	./test.txt
	./testa.py

From within temp1, output of find . -type f -regex '.+/[aeiou].*\.py':
(#) with valid match, ** with invalid match

find . -type f -regex '.+/[aeiou].*\.py'
./atest.py (8)
./alpha.py/atest.py (2)
./alpha.py/etest.py (3)
./alpha.py/a.py (1)
./temp/alphapy/otesteapy.py (6)
./temp/alphapy/test.py **
./temp/alpha.py/otesteapy.py (5)
./temp/alpha.py/o.py (4)
./temp/alpha.py/test.py **
./temp/otets.py (7)

Good exercise. Maybe someone better versed with regex can explain how to filter for such an issue.

OP could also just use multiple -name parameters instead, like the following, which gives a properly matched set:

> find . -type f /( -name 'a*.py' -o -name 'e*.py' -o -name 'i*.py' -o -name 'o*.py' -o -name 'u*.py' -o /)

./atest.py (8)
./alpha.py/atest.py (2)
./alpha.py/etest.py (3)
./alpha.py/a.py (1)
./temp/alphapy/otesteapy.py (6)
./temp/alpha.py/otesteapy.py (5)
./temp/alpha.py/o.py (4)
./temp/otets.py (7)

 

[VinDSL]

I like this regex tutorial. Regex syntax makes things look cryptic and complicated, but it's really not (and incredibly useful) once you understand how it works.

Here's another one for you.

I've used this site for years -- one stop shopping!

 

[LCTSI]

gooberlx2 I will tinker with it at work a bit on Tuesday and let you know.

I know that past the first / you will need to negate that / with [^/]... find from GNU Findutils by default uses emacs regex so it throws me off as I'm used to perl regex or awk's regex. (Its regex engine can be selected with a runtime option though)

 

[LCTSI]

So I've screwed with it some and the negation didn't work the way I thought it would... If you say [^/] all you're saying is "any character that is not /" ... so you'd have to do a negative lookahead, or a lookbehind assertion.

I'm going to go through the GNU findutils regex engines and see which ones do negative lookahead/lookbehind assertions. There may already be an answer on the interwebs!