How to run a PowerShell script within a Windows batch file

Question

How do I have a PowerShell script embedded within the same file as a Windows batch script?

I know this kind of thing is possible in other scenarios:

• Em

Answer 1

My current preference for this task is a polyglot header that works much the same way as mklement0's first solution:

<#  :cmd header for PowerShell script
@   set dir=%~dp0
@   set ps1="%TMP%\%~n0-%RANDOM%-%RANDOM%-%RANDOM%-%RANDOM%.ps1"
@   copy /b /y "%~f0" %ps1% >nul
@   powershell -NoProfile -ExecutionPolicy Bypass -File %ps1% %*
@   del /f %ps1%
@   goto :eof
#>

# Paste arbitrary PowerShell code here.
# In this example, all arguments are echoed.
$Args | % { 'arg #{0}: [{1}]' -f ++$i, $_ }

I prefer to lay the cmd header out as multiple lines with a single command on each one, for a number of reasons. First, I think it's easier to see what's going on: the command lines are short enough not to run off the right of my edit windows, and the column of punctuation on the left marks it visually as the header block that the horribly abused label on the first line says it is. Second, the del and goto commands are on their own lines, so they will still run even if something really funky gets passed as a script argument.

I have come to prefer solutions that make a temporary .ps1 file to those that rely on Invoke-Expression, purely because PowerShell's inscrutable error messages will then at least include meaningful line numbers.

The time it takes to make the temp file is usually completely swamped by the time it takes PowerShell itself to lumber into action, and 128 bits worth of %RANDOM% embedded in the temp file's name pretty much guarantees that multiple concurrent scripts won't ever stomp each other's temp files. The only real downside to the temp file approach is possible loss of information about the directory the original cmd script was invoked from, which is the rationale for the dir environment variable created on the second line.

Obviously it would be far less annoying for PowerShell not to be so anal about the filename extensions it will accept on script files, but you go to war with the shell you have, not the shell you wish you had.

Speaking of which: as mklement0 observes,

# BREAKS, due to the `&` inside \"...\"
sample.cmd "A \"rock & roll\" life style"

This does indeed break, due to cmd.exe's completely worthless argument parsing. I've generally found that the less work I do to try to hide cmd's many limitations, the fewer unanticipated bugs I cause myself down the line (I am sure I could come up with arguments containing parentheses that would break mklement0's otherwise impeccable ampersand escaping logic, for example). Less painful, in my view, just to bite the bullet and use something like

sample.cmd "A \"rock ^^^& roll\" life style"

The first and third ^ escapes get eaten when that command line is initially parsed; the second one survives to escape the & embedded in the command line passed to powershell.exe. Yes, this is ugly. Yes, it does make it harder to pretend that cmd.exe isn't what gets first crack at the script. Don't worry about it. Document it if it matters.

In most real-world applications, the & issue is moot anyway. Most of what's going to get passed as arguments to a script like this will be pathnames that arrive via drag and drop. Windows will quote those, which is enough to protect spaces and ampersands and in fact anything other than quotes, which aren't allowed in Windows pathnames anyway.

Don't even get me started on Vinyl LP's, 12" turning up in a CSV file.