Whats's going on?
helper.bat
#echo off
echo %1
call:foo %1
goto:eof
:foo
echo %1
goto:eof
Run our script like the following
helper "^^^^"
Output
"^^^^"
"^^^^^^^^"
Why? I know that '^' symbol is smth special in case of cmd.exe, but what's going on here? How the function call affect on it?
CALL is very special in this case!
The batch parser has different phases, in the special character phase unquoted carets are used to escape the next character, the caret itself is removed.
In your case, the carets are quoted, so they will not be affected.
Then the carets can be affected again in the delayed expansion phase, but quotes havn't special meaning there, the carets are used only to escape exclamation marks.
Normally after the delayed phase all is done, BUT if you use CALL all carets are doubled.
Normally this is invisible, as the CALL also restarts the parser and carets are removed in the special character phase again.
But in your case they are quoted, therefore they stay doubled.
Try this
call call call call echo a^^ "b^"
Output
a^ "b^^^^^^^^^^^^^^^^"
The parser is explained at How does the Windows Command Interpreter (CMD.EXE) parse scripts?
Related
I've got a batch script (app1.bat) calling another batch script (app2.bat) which itself calls a program in windows (program.exe).
app2.bat calls program.exe with a parameter after a flag in this way:
program.exe -f Parameter with whitespaces coming into the program
What I want to do is to pass the phrase that comes to program.exe from app1.bat into app2.bat but i don't know how to properly handle the doublequotes. Currently I am passing the phrase from app1.bat to app2.bat in double quotes and inside an app2.bat (prior to executing program.exe) I get rid of the quotes like that:
inside app1.bat
call app2.bat "Parameter with whitespaces coming into the program"
inside app2.bat
set old_phrase=%1%
set new_phrase=%old_phrase:"=%
program.exe -f %new_phrase%
old_phrase is
"Parameter with whitespaces coming into the program"
and new_phrase I end up with is
Parameter with whitespaces coming into the program
Is there any standard way to handle such a situation (being passing a string to an external program which expects a tring without quotes and being ok with whitespaces, whereas batch does not allow for no-quotes-and-whitespaces strings)
When you execute call /? from cmd to launch the help you will see quite a bit around expansion of %n
The first one states:
%~1 - expands %1 removing any surrounding quotes (")
You can therefore dump all the other set commands and simply run this in your batch file:
program.exe -f %~1
I'm in the unfortunate position to be forced to invoke a program via echo <input> | program.exe. Of course, I wondered how to escape <input> and found:
How does the Windows Command Interpreter (CMD.EXE) parse scripts?
Escape angle brackets in a Windows command prompt
In essence, it seems sufficient to escape all special chars with ^. Out of curiosity I still would like to know, why echo ingores double-quote escaping in the first place:
C:\>echo "foo"
"foo"
C:\>
Is there any normative reference?
Bonus question: How to echo the strings on and off with the echo command?
Edit: I just found this. I states that only |<> need to be escaped. However, expansion like %FOO% still work.
Special characters like ^, &, (, ), <, >, %, ! and " may cause problems with echo, also when trying to echo a string into a pipe; odd numbers of " are particularly difficult to handle.
Building escape sequences can be very complicated particularly with pipes, because such initiates new cmd instances for either side, so multi-escaping might become necessary.
The only reliable way to pipe the output of echo into a program is to use a variable holding the string to return and to apply delayed expansion, but within the left side of the pipe, like this:
cmd /V /C echo(^^!VARIABLE^^!| program.exe
Note the double-escaping of ! like ^^!, which makes this code even work when delayed expansion is also enabled in the parent cmd instance. There must not be a SPACE in front of the |, because this was echoed too otherwise. Note that echo terminates the output by a line-break.
When I run a command like this:
someprogram.exe --param="Value!"
the Windows shell will treat the exclamation mark in my value as a special character. Is there a way to suppress that or is escaping the only way around this? (Exclamation mark is escaped by ^^! if I'm not mistaken.)
EDIT: Weird thing that cmd.exe does not do the expansion but cmder does. I thought they are just two different GUIs to the same underlying shell? Maybe there's a switch for the expansion behavior? I tried SETLOCAL DisableDelayedExpansion but with no effect.
The Windows command prompt (cmd.exe) has an optional /s parameter, which modifies the behavior of /c (run a particular command and then exit) or /k (run a particular command and then show a shell prompt). This /s parameter evidently has something to do with some arcane quote handling.
The docs are confusing, but as far as I can tell, when you do cmd /csomething, and the something contains quotation marks, then by default cmd will sometimes strip off those quotes, and /s tells it to leave them alone.
What I don't understand is when the quote removal would break anything, because that's the only time /s ("suppress the default quote-removal behavior") would be necessary. It only removes quotes under a certain arcane set of conditions, and one of those conditions is that the first character after the /c must be a quotation mark. So it's not removing quotes around arguments; it's either removing quotes around the path to the EXE you're running, or around the entire command line (or possibly around the first half of the command line, which would be bizarre).
If the path to the EXE is quoted, e.g. cmd /c "c:\tools\foo.exe" arg1 arg2, then quotes are unnecessary, and if cmd wants to remove them, fine. (It won't remove them if the path has a space in the name -- that's another of the arcane rules.) I can't imagine any reason to suppress the quote removal, so /s seems unnecessary.
If the entire command line is quoted, e.g. cmd /c "foo.exe arg1 arg2", then it seems like quote removal would be a necessity, since there's no EXE named foo.exe arg1 arg2 on the system; so it seems like opting out of quote removal using /s would actually break things. (In actual fact, however, it does not break things: cmd /s /c "foo.exe arg1 arg2" works just fine.)
Is there some subtlety to /s that's eluding me? When would it ever be necessary? When would it even make any difference?
Cmd /S is very useful as it saves you having to worry about "quoting quotes". Recall that the /C argument means "execute this command as if I had typed it at the prompt, then quit".
So if you have a complicated command which you want to pass to CMD.exe you either have to remember CMD's argument quoting rules, and properly escape all of the quotes, or use /S, which triggers a special non-parsing rule of "Strip first and last " and treat all other characters as the command to execute unchanged".
You would use it where you want to take advantage of the capabilities of the CMD shell, rather than directly calling another program. For example environment variable expansion, output or input redirection, or using CMD.exe built-ins.
Example:
Use a shell built-in: This executes as-if you had typed DEL /Q/S "%TMP%\TestFile" at the prompt:
CMD.exe /S /C " DEL /Q/S "%TMP%\TestFile" "
This executes SomeCommand.exe redirecting standard output to a temp file and standard error to the same place:
CMD.exe /S /C " "%UserProfile%\SomeCommand.exe" > "%TMP%\TestOutput.txt" 2>&1 "
So what does /S give you extra? Mainly it saves you from having to worry about quoting the quotes. It also helps where you are unsure whether for example an environtment variable contains quote characters. Just say /S and put an extra quote at the beginning and end.
Vaguely Related: $* in Bourne Shell.
Some background
Recall that the list of arguments to main() is a C-ism and Unix-ism. The Unix/Linux shell (e.g. Bourne Shell etc) interprets the command line, un-quotes the arguments, expands wildcards like * to lists of files, and passes a list of arguments to the called program.
So if you say:
$ vi *.txt
The vi command sees for example these arguments:
vi
a.txt
b.txt
c.txt
d.txt
This is because unix/linux operates internally on the basis of "list of arguments".
Windows, which derives ultimately from CP/M and VAX, does not use this system internally. To the operating system, the command line is just a single string of characters. It is the responsibility of the called program to interpret the command line, expand file globs (* etc) and deal with unquoting quoted arguments.
So the arguments expected by C, have to be hacked up by the C runtime library. The operating system only supplies a single string with the arguments in, and if your language is not C (or even if it is) it may not be interpreted as space-separated arguments quoted according to shell rules, but as something completely different.
Here's an example of how it can make a difference.
Suppose you have two executables: c:\Program.exe and c:\Program Files\foo.exe.
If you say
cmd /c "c:\Program Files\foo"
you'll run foo.exe (with no arguments) whereas if you say
cmd /s /c "c:\Program Files\foo"
you'll run Program.exe with Files\foo as the argument.
(Oddly enough, in the first example, if foo.exe didn't exist, Program.exe would run instead.)
Addendum: if you were to type
c:\Program Files\foo
at the command prompt, you would run Program.exe (as happens with cmd /s /c) rather than foo.exe (as happens with just cmd /c). So one reason for using /s would be if you want to make sure a command is parsed in exactly the same way as if it were being typed at the command prompt. This is probably more likely to be desirable in the scenario in the question Michael Burr linked to, where cmd.exe is being launched by CreateProcess rather than from a batch file or the command line itself..
That is, if you say
CreateProcess("cmd.exe", "cmd /s /c \"" MY_COMMAND "\"", ...)
then the string MY_COMMAND will be parsed exactly as if it were typed at the command prompt. If you're taking command-line input from the user, or if you're a library processing a command line provided by an application, that's probably a good idea. For example, the C runtime library system() function might be implemented in this way.
In all but one specific case, the /S won't actually make any difference.
The help for cmd.exe is accurate, if a bit complicated:
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:
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.
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.
I'd summarize as follows:
Normal behavior:
If the rest of the command line after /K or /C starts with a quote, both that quote and the final quote are removed. (See exception below.) Other than that, no quotes are removed.
Exception:
If the rest of the command line after /K or /C starts with a quote, followed by the name of an executable file, followed by another quote, AND if those are the only two quotes, AND if the file name contains spaces but contains no special characters, then the quotes are not removed (even though they normally would have been removed according to the rule above).
The only effect of /S is to override this one exception, so that the two quote characters are still removed in that case.
If you always use /S, you can forget about the exception and just remember the "normal" case. The downside is that cmd.exe /S /C "file name with spaces.exe" argument1 won't work without adding an extra set of quotes, whereas without /S it would have worked... until you decide to replace argument1 with "argument1".
In a Windows batch file, when you do the following:
set myvar="c:\my music & videos"
the variable myvar is stored with the quotes included. Honestly I find that very stupid. The quotes are just to tell where the string begins and ends, not to be stored as part of the value itself.
How can I prevent this from happening?
Thanks.
set "myvar=c:\my music & videos"
Notice the quotes start before myvar. It's actually that simple.
Side note: myvar can't be echoed afterwards unless it's wrapped in quotes because & will be read as a command separator, but it'll still work as a path.
http://ss64.com/nt/set.html under "Variable names can include Spaces"
This is the correct way to do it:
set "myvar=c:\my music & videos"
The quotes will not be included in the variable value.
It depends on how you want to use the variable. If you just want to use the value of the variable without the quotes you can use either delayed expansion and string substitution, or the for command:
#echo OFF
SETLOCAL enabledelayedexpansion
set myvar="C:\my music & videos"
As andynormancx states, the quotes are needed since the string contains the &. Or you can escape it with the ^, but I think the quotes are a little cleaner.
If you use delayed expansion with string substitution, you get the value of the variable without the quotes:
#echo !myvar:"=!
>>> C:\my music & videos
You can also use the for command:
for /f "tokens=* delims=" %%P in (%myvar%) do (
#echo %%P
)
>>> C:\my music & videos
However, if you want to use the variable in a command, you must use the quoted value or enclose the value of the variable in quotes:
Using string substitution and delayed expansion to use value of the variable without quotes, but use the variable in a command:
#echo OFF
SETLOCAL enabledelayedexpansion
set myvar="C:\my music & videos"
md %myvar%
#echo !myvar:"=! created.
Using the for command to use the value of the variable without quotes, but you'll have to surround the variable with quotes when using it in commands:
#echo OFF
set myvar="C:\my music & videos"
for /f "tokens=* delims=" %%P in (%myvar%) do (
md "%%P"
#echo %%P created.
)
Long story short, there's really no clean way to use a path or filename that contains embedded spaces and/or &s in a batch file.
Use jscript.
Many moons ago (i.e. about 8 years give or take) I was working on a large C++/VB6 project, and I had various bits of Batch Script to do parts of the build.
Then someone pointed me at the Joel Test, I was particularly enamoured of point 2, and set about bringing all my little build scripts into one single build script . . .
and it nearly broke my heart, getting all those little scripts working together, on different machines, with slightly different setups, ye Gods it was dreadful - particularly setting variables and parameter passing. It was really brittle, the slightest thing would break it and require 30 minutes of tweaking to get going again.
Eventually - I can be stubborn me - I chucked the whole lot in and in about a day re-wrote it all in JavaScript, running it from the command prompt with CScript.
I haven't looked back. Although these days it's MSBuild and Cruise Control, if I need to do something even slightly involved with a batch script, I use jscript.
The Windows command interpreter allows you to use the quotes around the entire set command (valid in every version of windows NT from NT 4.0 to Windows 2012 R2)
Your script should just be written as follows:
#echo OFF
set "myvar=C:\my music & videos"
Then you may put quotes around the variables as needed.
Working with the CMD prompt can seem esoteric at times, but the command interpreter actually behaves pretty solidly in obeying it's internal logic, you just need to re-think things.
In fact, the set command does not require you to use quotes at all, but both the way you are doing your variable assignment and the way the ,method of using no quotes can cause you to have extra spaces around your variable which are hard to notice when debugging your script.
e.g. Both of the below are technically Valid, but you can have trailing spaces, so it's not a good practice:
set myvar=some text
set myvar="some text"
e.g. Both of the below are good methods for setting variables in Windows Command interpreter, however the double quote method is superior:
set "myvar=Some text"
(set myvar=Some value)
Both of these leave nothing to interpretation the variable will have exactly the data you are looking for.
strong text However, for your purposes, only the quoted method will work validly because you are using a reserved character
Thus, you would use:
set "myvar=c:\my music & videos"
However, even though the variable IS correctly set to this string, when you ECHO the sting the command interpreter will interpret the ampersand as the keyword to indicate another statement follows.
SO if you want to echo the string from the variable the CMD interpreter still needs to be told it's a text string, or if you do not want the quotes to show you have to do one of the following:
echo the variable WITH Quotes:
Echo."%myvar%"
echo the variable WITHOUT Quotes:
Echo.%myvar:&=^&%
<nul SET /P="%myvar%"
In the above two scenarios you can echo the string with no quotes just fine. Example output below:
C:\Admin> Echo.%myvar:&=^&%
C:\my music & videos
C:\Admin> <nul SET /P="%myvar%"
C:\my music & videos
C:\Admin>
Try using the escape character '^', e.g.
set myvar=c:\my music ^& videos
You'll have you be careful when you expand myvar because the shell might not treat the & as a literal. If the above doesn't work, try inserting a caret into the string too:
set myvar=c:\my music ^^^& videos
Two solutions:
Don't use spaces or other characters that are special to the command interpreter in path names (directory or file names). If you use only letters, numbers, underscores, and hyphens (and a period before the extension to identify the file type), your scripting life will become immeasurably simpler.
I have written and otherwise collected a plethora of tools over the years, including a DOS utility that will rename files. (It began as something that just removed spaces from filenames, but morphed into something that will replace characters or strings within the filenames, even recursively.) If anyone's interested, I will get my long-neglected web site up and running and post this and others.
That said, variables aren't just for holding pathnames, so...
As others have already pointed out, SET "myvar=c:\my music & videos" is the correct workaround for such a variable value. (Yes, I said workaround. I agree that your initial inclination to just quote the value ("my music & videos") is far more intuitive, but it is what it is, as they say.