Copy to current path files searched with Perl regular expressions (regex) recursively under subdirectories, in linux and Windows, with UTF-8 support.
Copied files retain the directories from which they were copied, separated with '_' chars.
If folder name is too long it is abreviated inserting '...' in the middle and erasing chars to just 17 chars per directory.
If a filename already exists in current path when copying, a number (starting at 0001) will be appended to the copied filename, and so cpr never will overwrite any file.

Files can also be moved, instead of copied (with -M parameter).

cpr [-#hMnNpryY] [ '/FILTER_REGEX/' [PATH] ]

where '/FILTER_REGEX/' is a regular expression match pattern. If '/FILTER_REGEX' is not indicated, '/./' is used (all files). If PATH is not indicated, '.' is used (current path).

Please note that in linux simple quotation marks are preferred: '' whilst in Windows double quotation marks are needed: ""

-#: a number indicating how many files at most will be copied from each subdirectory. For example: -21

-h: show full help with examples.

-M: MOVE files: they'll be removed from origin and moved to pwd. Note that, depending on -p and -r, files will be renamed. Default action (without -M) is COPY.

-n: when asked for confirmation a "NO" is automatically answered.
-N: as -n, but output is also written to a log file in current path.

-p: preserve original filename. Without -p the path to each file is prepended to the filename.

-r: remove path passed as parameted from file prefixes. Note that -p makes -r redundant.

-y: when asked for confirmation a "YES" is automatically answered.
-Y: as -y, but output is also written to a log file in current path.

The command shows a list of changes to be made, which must be confirmed with 'y' (yes). If capital 'Y' is entered, the results are written to a log file.

Parameters can also be indicated using stdin.

The stdin input for parameters is needed in Windows in order to use UTF-8 characters on regex, FILTER and/or PATH. Note that a cmd console with UTF-8 support must be used. CMD [/U] isn't appropriate in general, even with UTF-8 fonts (see below). In Windows, if stdin is used, -[yY] would be needed to apply changes, because input cannot be read from the keyboard.

As you can see, the command shows a list of actions to execute, and requires previous confirmation:

C:\temp> cpr "/LOG$/"

Using:
        path:                   '.'
        search pattern:         /LOG$/

.\subdir\file.20180702004423.LOG       ->      subdir_file.20180702004423.LOG
.\subdir3\subdir3.4\file.LOG       ->      subdir3_subdir3.4_file.LOG

--- (files =~ search pattern: 2)
--- files to copy (listed): 2   (<=2)
Do you want to copy here this file list? (y/ Y(=>log)/ *):

$ cpr -1 '/\.jpg$/'

This copies one jpg image from each and every subdirectory under the current path, to the current directory.

$ cpr '/log$/' /home

This copies all files whose name end with 'log' from each and every subdirectory under '/home', to the current path.

$ cpr -N '/report/i' /works

This creates a text file containing the path and filename of all files under '/works' containing the word 'report' on their name. (The file also contains the filename they would have in case they were copied to current path.) With -n the results are just printed on screen.

In order to use the script in Windows you need:

• a Perl installation (for example ActivePerl or Strawberry Perl). The script has been tested with ActivePerl.
• copy the cpr script and cpr.bat to a folder in your execution PATH

The cpr.bat batch script is provided as an example: it supposes that perl.exe is in the PATH, and cpr must be preceded by its complete path.

If your are planning to manage file names with UTF-8 chars (mostly outside Latin-1), a console with proprer UTF-8 I/O must be used. Windows' CMD.EXE isn't 100% adequate, even after installing UTF-8 fonts (See serganov fonts and metacpan UI-KeyboardLayout), because some visual artifacts remain. Instead, ConEmu is suitable.

Also note that in order to use most UTF-8 chars directly in the regex or filter parameters, the stdin with a | must be used. Note also that in Windows the use of stdin for input parameters to cpr command requires the use of -[yYnN] modifiers, because unfortunately input cannot be redirected to the keyboard. In Windows, to copy all files containing the japanese hiragana UTF-8 "の" char under 'C:\temp2': (Note though that -[yY] is needed in order to actually take actions):

C:\temp> echo  -y  "/の/"  "C:\temp2" |  cpr

2018-Nov amendment: In Windows perl.exe binary can be manually patched and recompiled in order to accept Unicode chars in command line: Please take a look at Perl-with-Unicode-for-Windows.

Subroutine ListContent() can be extracted from code and used indepently in other Windows or linux Perl projects. It just needs some global stuff, commented as "# (...) ListContent() to run properly" in the code. Input parameters are ListContent( $parameters_ref, $pattern, $path ) where:

• $parameters_ref is a reference to an array of boolean ( 0 | 1 ) numeric values representing, in order, the options: ( $LS_STYLE, $TREAT_DIRECTORIES, $RECURSIVE, $NEGATED_MATCH, $MSDOS_SHORT_NAMES )
• $pattern is a regex filter to apply to file names (or ls style pattern, even prepended with options (and even piped throu another command)). If $NEGATED_MATCH is 1, is the regex filter that filenames do not comply with ($LS_STYLE must be zero, or otherwise $NEGATED_MATCH will be ignored!).
• $path is the filesystem path to be examinated

Return parameter is a reference to an array containing:

• \@FILE_LIST, a reference to an array of filenames (each path is in the same index on @PATH_LIST array)
• \@PATH_LIST, a reference to an array of paths corresponding to the filename of the same index on @FILE_LIST array
• number of directories returned, only to inform the number. Has sense only if $TREAT_DIRECTORIES call option was 1
• \@SHORT_FILENAME, a reference to an array of path+filename in Windows' MSDOS short name format. Is not () only if OS is Windows and $MSDOS_SHORT_NAMES call parameter was 1 (or the number of returned files is 0)

• The difference between "files =~ search pattern" and "files to copy" numbers, if such, is due to the fact that files in current path (not under any subdirectory) are initially counted, though they're later discarded as there is no sense in copying them to pwd... (they'd be overwritten by themselves). Future versions may stop counting files in current directory.

Written by circulosmeos based on rnm

2018-11.04

GPL v3

This code is provided "as is", without warranty of any kind, express or implied.