findstr not working with string containing dash (-)?

Question

I've been working on a BAT file which will delete old files based on creation date. To do this I've generated a list of all files and paths, then a list of files names to be protected. FINDSTR is then used to remove these files from the list of files and paths.

This system works fine until I encounter a file with a dash (or so it seems!)

Here's an example:

cleaner_protect.txt contains:

New File.txt 
New File - Copy.txt

cleaner_fullpath.txt contains:

P:\New File.txt
P:\New File - Copy.txt
P:\Old File.txt

I want to remove the New Files stored in cleaner_protect.txt from the cleaner_fullpath.txt, leaving the Old Files behind which I will later delete (not up to that bit yet lol). Here is my code so far:

:: Remove protected files from list to be deleted (fullpath)
:RemoveFile
:: load string into variable
set /p target= <cleaner_protect.txt
:: remove protected file from full path list
echo -----------------------------
echo Searching for: "%target%"
echo -----------------------------
pause
findstr /v ".\<%target%\>." cleaner_fullpath.txt > cleaner_temp.txt
echo -----------------------------
type cleaner_temp.txt
echo -----------------------------
pause
del cleaner_fullpath.txt
ren cleaner_temp.txt cleaner_fullpath.txt
:: Count remaining lines in list
Set target=cleaner_protect.txt
Set /a lines=0
For /f %%j in ('Find "" /v /c ^< %target%') Do Set /a lines=%%j
Echo %target% has %lines% lines.
pause
:: Loop until completed
IF %lines% GTR 0 (
:: Remove line from protected list
    more +1 cleaner_protect.txt > cleaner_temp.txt
    del cleaner_protect.txt
    ren cleaner_temp.txt cleaner_protect.txt
    set /a lines-=1
    GOTO :RemoveFile
)

Pauses and echos are for debugging purposes... I want this to run almost invisibly.

Can anyone shed some light on this? I need this code to repeatedly go through a dropbox and delete old files which may be in various levels of structure.

Answer 1

Maybe this simple line does all you want:

findstr /E /V /G:cleaner_protect.txt cleaner_fullpath.txt > cleaner_temp.txt

Sample output:

P:\Old File.txt

Answer 2

I would do it as follows:

@echo off
setlocal EnableExtensions DisableDelayedExpansion

rem /* Prepend each line of `cleaner_protect.txt` with `\\`, remove a trailing space,
rem    if applicable, and write result to `cleaner_temp.txt`: */
> "cleaner_temp.txt" (
    for /F "usebackq delims= eol=|" %%E in ("cleaner_protect.txt") do (
        set "ITEM=\\%%E|"
        setlocal EnableDelayedExpansion
        echo !ITEM: ^|=!
        endlocal
    )
)

rem /* Search `cleaner_fullpath.txt` for lines that do not end in any of the lines of
rem    `cleaner_temp.txt` and process the returned items only: */
for /F "delims= eol=|" %%F in ('
    findstr /V /I /E /L /G:"cleaner_temp.txt" "cleaner_fullpath.txt"
') do (
    ECHO del "%%F"
)

rem // Clean up temporary file `cleaner_temp.txt`:
del "cleaner_temp.txt"

endlocal
exit /B

After having tested the script, remove the upper-case ECHO command.

Supposing cleaner_protect.txt contains this:

New File.txt
New File - Copy.txt

The temporary file cleaner_temp.txt is going to contain this:

\\New File.txt
\\New File - Copy.txt

So having the text file cleaner_fullpath.txt:

P:\New File.txt
P:\New File - Copy.txt
P:\Old File.txt

Only the following items are processed:

P:\Old File.txt

The leading \\ is taken as one literal \ by findstr (as it uses the \ as an escape character).

The prefix \\ is implemented in order to cover also the following situations:

1. Let us assume cleaner_protect.txt holds this:

   New File.txt
   

   And cleaner_fullpath.txt holds this:

   P:\New File.txt
   P:\Very New File.txt
   

   Without the prefix, P:\Very New File.txt would also match New File.txt at the end, hence it would not become deleted erroneously.

2. Then let us assume cleaner_protect.txt holds this:

   .File.txt
   

   And cleaner_fullpath.txt holds this:

   P:\.File.txt
   P:\Other.File.txt
   

   With a \ prefix, P:\Other.File.txt would also match \.File.txt at the end, because . is a meta character to findstr (even though literal search strings are defined by /L, but escaping like \. still applies and results in a literal .), hence it would also not become deleted erroneously. However, with a \\ prefix, escaping applies to the second \, so a literal \ is the result; the . does not need to be escaped with the /L option.