I am running a batch file on Windows 7 and running into this error (I have narrowed down the error to the following line):
FOR /F "delims=" %%I in ('echo %RegVal%') do set sasroot=%%~sI
Where Regval is the file path of a given software, which in this case (on my Win7 machine) is:
RegVal = C:\Program Files\SAS 9.2_M3_10w37\SASFoundation\9.2(32-bit)
This same script used to work on Windows Vista, although I suspect it may be that there a parenthesis in RegVal now as it was previousy C :\Program Files\SAS 9.2_M3_10w37\SASFoundation\ on my previous Vista machine.
You suspection is correct.
To get around it, enclose your variable into doublequotes (You remove them again with the ~ in the setcommand)
FOR /F "delims=" %%I in ('echo "%RegVal%"') do set sasroot=%%~sI
I suggest you create a file with the value of RegVal in it, then parse it using the FOR loop:
echo %RegVal%>C:\SomeFile.txt
FOR /F "delims=" %%I in (C:\SomeFile.txt) do set sasroot=%%~sI
This should help you get around your problem.
Stephan's solution is much simpler, but I'll explain my solution anyway, which might prove useful in some cases.
When the FOR command parses the data specified in the IN part using a command, it replaces the command with the result of the command, then runs the FOR command. For example, with the question above, the FOR command that will be executed after expanding echo %RegVal% is:
FOR /F "delims=" %%I in (C:\Program Files\SAS 9.2_M3_10w37\SASFoundation\9.2(32-bit)) do set sasroot=%%~sI
Thus, when the parser hits the first closing parenthesis, it will stop, thinking that everything it read before is the text to work on. However, in this case this is wrong, as the first closing parenthesis is part of the string to read; it doesn't indicate the end of the string.
When parsing a file with the FOR command, it will read each line, assign the predefined tokens with the correct values, then execute the code block that follows. Rinse and repeat for every line in the file. But in this case, it will not replace the IN part with each line; it will only parse it and assign values to the tokens. This is the reason why special characters (such as parenthesis) do not create parsing errors in this case.
Related
Edit 3 (SOLUTION):
As michael_heath described in his answer, the issue came down to two things: how exactly windows builds commands when executing something through a shortcut, and the very specific (and frankly ridiculous) consequences of the \C switch on cmd.exe. For future reference, if any other poor soul stumbles into this StackOverflow question, the problem was fixed by changing the "Target" property in the shortcut to a slightly edited version of Michael's answer, specifically C:\Windows\System32\cmd.exe /C #"C:\{path-to-script}\link.bat". Here's a screenshot too, if necessary, although you unfortunately can't see the whole Target line.
Huge thanks again to Michael.
I am attempting to make a personal batch utility to create a symlink on the desktop, much in the same way "Send to... > Desktop" works with shortcuts. I use symbolic links frequently to allow things like my bash configuration files (.bashrc and .bash_profile, etc) to be version controlled elsewhere for portability, and for several other things on my computer.
For ease of use, my idea was to create a simple batch file to do this for me, and place the symlink on the desktop. Then, I would put a shortcut to this file in the Send To folder so it appears in Send To in the context menu (I am aware that mklink requires admin privileges, so the shortcut is set to run as administrator also).
The following is the file I have written:
#echo off
set f=%~1
set switch=
if exist "%f%\*" set switch=/D
for /F "delims=" %%i in ("%f%") do set name=%%~nxi
mklink %switch% "%USERPROFILE%\Desktop\%name%" "%f%"
if not %ERRORLEVEL%==0 pause
Here is the general idea of what I'm trying to do:
Strip the quotes on the input, if any (%~1)
Check if the input is a directory
Get the base name and file extension of the input (%%~nxi)
Make the link
If an error occurred, pause so it can be seen rather than exiting (because the batch file is called from the shortcut)
It works perfectly fine until I give it an input that contains spaces in the name of the file or directory. I actually haven't tested what happens if there is a directory with a space in the path, but not in the base name of the actual file or directory, but I assume the same problem will be present.
I have made several changes to attempt to get it to work with files with spaces, including stripping the quotes on the input in that first line so that the quotes aren't doubled later, and that "delims=" thing on the for loop. Those two solutions I found here, actually.
But despite my best efforts, no matter what I do, the file closes immediately when given an input with a space. I have littered every line with pauses, run the script from the command line with a manually entered input so it would not exit, and run each individual command (where possible) from the command line.
Infuriatingly, when I run it from the command line or run the individual commands, it all works perfectly even with spaces in the input. I even created another batch file that does nothing but output the input it receives and ran that from Send To, and confirmed that the input is the same as I entered from the command line.
What on Earth is going wrong then when called from that shortcut in Send To?
Edit 1: The properties of the shortcut itself are as follows:
Target: C:\Users{username}\vc\git\util-scripts\bat\link.bat
Start in: C:\Users{username}\vc\git\util-scripts\bat
Here is a screenshot as well:
Because I just made my account here I can't embed the picture but here it is
Edit 2: This is the current code I am using, as suggested by Gerhard Barnard, however the problem still persists:
#echo off
set "fname=%~1"
set switch=
if exist "%fname%\.*" set "switch=/D"
for /F "delims=" %%i in ("%fname%") do (
mklink %switch% "%USERPROFILE%\Desktop\%%~nxi" "%fname%"
if errorlevel 1 pause
)
#echo off
setlocal
rem Set the path for the created symlink.
set "linkdir=%USERPROFILE%\Desktop"
for %%A in (%*) do call :link "%%~A"
rem Check results.
echo: & dir /A:L "%linkdir%" & pause
exit /b
:link
set "switch="
if exist "%~1\*" set "switch=/D"
for /F "delims=" %%A in ("%~1") do set "name=%%~nxA"
if not defined name echo Variable "name" not defined.& exit /b 1
mklink %switch% "%linkdir%\%name%" "%~1"
exit /b 0
Your batch-file code is working.
This code does multiple file or folders.
This code also helped to test all at once,
files, folders, symlinked files and symlinked
folders as targets.
If you only want 1 target to be processed,
then just change the %* to "%~1".
The main issue is the command string in the shortcut.
C:\Users\{username}\vc\git\util-scripts\bat\link.bat
The file type of .bat is going to build a command such as:
C:\Windows\System32\cmd.exe /C "C:\Users\{username}\vc\git\util-scripts\bat\link.bat" %*
%* is substituted with the passed arguments.
If you have the command with an argument with double quotes,
it may look like:
C:\Windows\System32\cmd.exe /C "C:\Users\{username}\vc\git\util-scripts\bat\link.bat" "C:\Users\a file.txt"
The command string after /C has 4 double quotes
and double quotes are at both ends.
The behavior changes due to the double quotes.
A quote from cmd /?:
If /C or /K is specified, then the remainder of the command line after
the switch is processed as a command line, where the following logic is
used to process quote (") characters:
1. If all of the following conditions are met, then quote characters
on the command line are preserved:
- no /S switch
- exactly two quote characters
- no special characters between the two quote characters,
where special is one of: &<>()#^|
- there are one or more whitespace characters between the
two quote characters
- the string between the two quote characters is the name
of an executable file.
2. Otherwise, old behavior is to see if the first character is
a quote character and if so, strip the leading character and
remove the last quote character on the command line, preserving
any text after the last quote character.
In section 1, "no /S switch" is true, then the next is
"exactly two quote characters" which is false.
This now applies section 2, which can make the
command string after /C with stripped double quotes:
C:\Users\{username}\vc\git\util-scripts\bat\link.bat" "C:\Users\a file.txt
The space is quoted though the rest is exposed,
which is an invalid command string in this case.
Note that C:\Users\a file.txt is an example
passed argument that was double quoted as
"C:\Users\a file.txt".
A change in the command string in the shortcut to:
C:\Windows\System32\cmd.exe /C echo: & "C:\Users\{username}\vc\git\util-scripts\bat\link.cmd"
The command is now C:\Windows\System32\cmd.exe which
gives some more control with the command string.
After the /C, echo: is used to avoid the command
string beginning with a double quote, which helps to
prevent double quotes being stripped at both ends because
the command string no longer starts with a double quote.
After the & is the command that is important and will
now work even if the arguments end with a double quote.
You can replace the echo: & with another initial command.
You can also use # before the command string, so the shortcut
command string would be:
C:\Windows\System32\cmd.exe /C #"C:\Users\{username}\vc\git\util-scripts\bat\link.cmd"
So whatever works the best for your use case.
Try without the /F option which is not needed here, we also do not need to set a variable as it will work perfectly fine with the expanded meta variable:
#echo off
set "fname=%~1"
set switch=
if exist "%fname%\.*" set "switch=/D"
for %%i in ("%fname%") do (
echo mklink %switch% "%USERPROFILE%\Desktop\%%~nxi" "%fname%"
if errorlevel 1 pause
)
Sorry, for bothering you for the (n+1)th time about search & replace with batch scripts.
I have text files (actually PS-files) (approx. 10kB-3MB) where I need to replace just a few numbers.
This should be easy, I thought.
I found quite a few scripts here on Stackoverflow but none of them worked properly so far. If I have overlooked THE "working one" please let me know.
The last one I tried:
#echo off
setlocal DisableDelayedExpansion
set OutputFile=%1
set OutputFile=%OutputFile:"=%
set InputFile=%OutputFile%.tmp
set SearchString=636170656C6C6133
set ReplaceString=636170656C6C6134
rem write empty file
type NUL > %OutputFile%
for /f "tokens=1,* delims=ΒΆ" %%A in ( '"type %InputFile%"') do (
SET string=%%A
setlocal EnableDelayedExpansion
SET modified=!string:%SearchString%=%ReplaceString%!
echo !modified!>>%OutputFile%
endlocal
)
del %InputFile%
First of all, it seems to be pretty(!) slow. I can see on disk how the file size increases.
The occurrences of the numbers seem to be replaced. However, the file is altered, which I easily can see from the different file size. As far as I can see, empty lines, exclamation marks and lines beginning with semicolon are skipped. This is messing up my file completely.
How to avoid this?
If I do the same thing with Perl I really get only the numbers altered, nothing else. However, I don't want to and cannot use Perl. I also don't want to use other extra programs or Windows-Powershell, since it should work on older systems too.
Is there any way to achieve this with a simple Windows batch script?
Thanks!
I believe the following is a working bat script that should not make any changes other than the desired number change:
#echo off
setlocal DisableDelayedExpansion
set "out=%~1"
set "in=%out%.tmp"
set "find=636170656C6C6133"
set "repl=636170656C6C6134"
>"%out%" (
for /f "delims=" %%A in ('findstr /n "^" "%in%"') do (
set "str=%%A"
setlocal EnableDelayedExpansion
set "str=!str:*:=!"
if defined str set "str=!str:%find%=%repl%!"
echo(!str!
endlocal
)
)
del "%in%"
Changes I have made:
Use %~1 to remove enclosing parentheses. Though technically, that is not necessary. Something like echo test >"someName.txt".new will work just fine.
FOR /F strips empty lines. I used FINDSTR to prefix each line with the line number, followed by a colon. Now there are no empty lines.
I use an extra variable expansion find/replace with * to remove the line number prefix.
Variable expansion find/replace will fail if a string is empty (undefined variable). So I verify the variable is defined before doing find/replace.
ECHOing an empty line, or line containing only white space, will result in ECHO is off. output. This is solved by using echo(
It takes time to initialize redirection, and your loop does this every iteration, which slows things down. I improved performance by enclosing the entire FOR loop in parentheses and redirecting only once.
You still may see a slight file size change for any of the following reasons
If the input has \n line terminators instead of \r\n.
If the last line of input is not terminated by \r\n. The script terminates all lines with \r\n, regardless what the input had.
The script will fail if any line contains a null byte, or if any line is >~8k length.
I hate editing text files with batch - it is complicated code, slow, and even the best possible solution still has significant limitations.
I recommend you use JREPL.BAT - a command line regular expression text processing utility. JREPL is pure script (hybrid batch/JScript) that runs natively on any Windows machine from XP onward - no 3rd party exe file or special configuration is needed.
The tool is very powerful, with many options. Full documentation is available from the command line via jrepl /?, or jrepl /?? for paged help.
Solving your problem with JREPL is trivial - you don't even need another script. The following command will work right from a command prompt:
jrepl 636170656C6C6133 636170656C6C6134 /f input.txt /o output.txt
Use CALL JREPL if you put the command within another batch script.
JREPL is way more powerful than what you need for this simple problem. But it is incredibly convenient, and once you have the utility, I suspect you will find many uses for it. Especially if you learn to use regular expressions, as well as the many JREPL options.
A VBS script
Set Inp = WScript.Stdin
Set Outp = Wscript.Stdout
Text = Inp.readall
Text = Replace(Text, "636170656C6C6133", "636170656C6C6134")
outp.write Text
To use
cscript //nologo script.vbs < input.txt > Output.txt
You use the right tool for the job. Batch is for starting programs and copying files.
The above is suited to the file sizes given. However if we are getting up to 100s of MB then this code is better.
Set Inp = WScript.Stdin
Set Outp = Wscript.Stdout
Do Until Inp.AtEndOfStream
Text = Inp.readline
Text = Replace(Text, "636170656C6C6133", "636170656C6C6134")
outp.writeline Text
Loop
I'm pretty sure this is a simple command but I just couldn't find it anywhere.
Example file content:
A-VERY-LONG-LINE-OF-GARBAGE-VERSION123-CONTINUE-LONG-LINE-OF-GARBAGE
1.)Assume the line is really really long
2.)I need to find out what version it is. I know it contains Version but I wouldn't know it is version123.
3.)What I want is a command that would go through the file looking for the sub-string "VERSION" and if it finds it prints out VERSION123 instead of the super duper long line that would most probably causes the system to freeze.
Thank you
Assuming that version is purely numeric and does not start with zero, the following should do it:
set VAR=A-VERY-LONG-LINE-OF-GARBAGE-VERSION123-CONTINUE-LONG-LINE-OF-GARBAGE
set /A VAR=%VAR:*VERSION=%
echo VERSION%VAR%
If there occur multiple VERSION portions, the first one is taken.
Note, that this works for Windows command prompt (cmd.exe) only, it will not work for MS-DOS (command.com) due to set /A which is not supported there (I'm even not sure whether the string substitution syntax works there)!
In case the version code is not purely numeric, you might use the following:
set VAR=A-VERY-LONG-LINE-OF-GARBAGE-VERSION123-CONTINUE-LONG-LINE-OF-GARBAGE
set VAR=%VAR:*VERSION=%
for /F "tokens=1 delims=- eol=-" %%L in ("%VAR%") do (set VAR=%%L)
echo VERSION%VAR%
This relies on the fact that the - character delimits the version code.
If you want to try this in the command prompt directly rather than in a batch file, replace %%L by %L (twice).
My problem is pretty simple, but I'm unsure why I am seeing this behaviour. I want to get a list of parameters down to a single entry at a time so I can do some processing on them. The list of jar files I am processing is separated by a ; delimiter.
set JARS=this.jar;that.jar;and.jar;the.jar;other.jar
for /f "delims=;" %%a in ("%JARS%") do echo.%%a
I'm expecting the script to exit listing as follows.
this.jar
that.jar
and.jar
the.jar
other.jar
C:\>
But the script is instead exiting as follows.
this.jar
C:\>
I'm clearly missing something obvious, but I can't seem to see it.
I'm using Windows 7.
Try this: The semicolon is a separator on a command line so it will delimit the filenames.
set JARS=this.jar;that.jar;and.jar;the.jar;other.jar
for %%a in (%JARS%) do echo.%%a
What does this bat code do?
for /f %%i in ('dir /b Client\Javascript\*_min.js') do (
set n=%%~ni
set t=!n:~0,-4!
cp Client\Javascript\%%i build\Client\Javascript\!t!.js
)
What does %%~ni,~n:~0,-4!,%%i,!t! mean?
Keep in mind that in batch files, you need to escape percentage signs unless you're referring to arguments given to the batch file. Once you remove those, you get
for /f %i in ('dir /b Client\Javascript\*_min.js') do (
set n=%~ni
set t=!n:~0,-4!
cp Client\Javascript\%i build\Client\Javascript\!t!.js
)
%i is the declaration of a variable used to place the current file for has found. %~ni extracts the filename portion of %i. !n:~0,-4! uses delayed expansion to remove the last four characters from %n% (set in the previous line) !t! is simply delayed expansion of the %t% variable set in the previous line.
Delayed expansion is used because otherwise, the variables will be substituted as soon as the line is encountered, and future iterations will not re-expand the variable.
for /f %%i in ('dir /b Client\Javascript\*_min.js') do (
Iterate over every file in the Client\Javascript folder that match "*_min.js". Thedircommand andfor /f` are totally unneeded here, though and only complicate things, especially when file names contain spaces, commas and the like. A more robust and simpler alternative would be
for %%i in (Client\Javascript\*_min.js) do (
But that's just beside the point. People tend to write unelegant batch files sometimes, ignoring the pitfalls and common errors. That's just one example of that.
set n=%%~ni
Creates a variable n, containing the file name (without any directory information or extension) of the file currently processed. We remember that the for statement iterates over every file it finds. With this line starts what it does with those files.
set t=!n:~0,-4!
Creates a second variable, t, containing everything but the last four characters of the file name. This essentially strips away the "_min"
cp Client\Javascript\%%i build\Client\Javascript\!t!.js
Finally, this copies the original file to the directory build\Client\Javascript with the new name, just constructed. So a file like Client\Javascript\foo_min.js will be copied to Client\Javascript\foo.js. The !t! here is just a delayed-evaluated environment variable. More on that below. Here it should suffice that it just inserts the contents of said variable at that point in the line.
Again, bad practice here that will break in numerous interesting ways:
cp is not a command on Windows so this batch will assume cygwin, GNUWin32 or similar things installed. I tend to avoid having too many unneeded dependencies and stick to what Windows provides; in this case the copy command. Two bytes won't kill anyone here, I think.
No quotes are around either argument. Leads to interesting results when spaces start appearing in the file name. Not good, either.
As for why delayed expansion was used (! instead of % surrounding the variables: The for command consists of everything in the block delimited by parentheses here as well. The entire block is parsed at once and normal variable expansion takes place when a line/command is parsed. That would mean that every variable in the block would be evaluated before the loop even runs, leaving just the following:
for /f %%i in ('dir /b Client\Javascript\*_min.js') do (
set n=%%~ni
set t=
cp Client\Javascript\%%i build\Client\Javascript\.js
)
which is certainly not what you want in this case.
Delayed expansion is always needed when creating and using variables in a loop such as this. A workaround not needing delayed expansion would be to offload the loop interior into a subroutine:
for /f %%i in ('dir /b Client\Javascript\*_min.js') do call :process "%%i"
goto :eof
:process
set n=%~n1
set t=%n:0,-4%
copy "Client\Javascript\%~1" "build\Client\Javascript\%t%.js"
goto :eof
Since the subroutine is not a single "block" (something delimited by parentheses) it will be parsed line by line as usual. Therefore it's safe to use normal expansion instead of delayed expansion here.
A complete help for the FOR command can be found on the Microsoft TechNet site. See here for more information on delayed expansion :
// Pseudo code
for each file named *_min.js in the specified directory
n is set to the file name (*_min)
t is set to the file name, excluding the last 4 characters (*)
the file is copied and renamed t.js to the specified directory
%~ni expands to just the filename part of i.
!n:~0,-4! expands to all but the last four characters of n.
In general, help for at the command prompt will give an overview of the multitude of ways for can expand variables these days.