Using PowerShell in a .bat file, replace a string with multiple strings - powershell

Question

I'm using a .baat to move several files into another folder, but before the actual move part, I want to replace the LAST line (it is a known line), for example I have a file output.txt like this:
HEADER
BODY
FOOTER
Using this snippet of code:
powershell -Command "(gc output.txt) -replace 'FOOTER', 'ONE_MORE_LINE `r`n FOOTER' | Out-File output.txt"
The return that I expected was
HEADER
BODY
ONE_MORE_LINE
FOOTER
But what I got was:
HEADER
BODY
ONE_MORE_LINE `r`n FOOTER
I've tried:
\n
<br>
"`r`n"
"`n"
echo ONE_MORE_LINE >> output.txt; echo. >> output.txt; echo FOOTER >> output.txt"
This last one got close, but the result was some broken characters.
Other suggestions besides the PowerShell are welcome. I'm only using it because it was an easy get way to do the adding lines and replace it.
EDIT :
Tried this command
powershell -Command "(gc output.txt) -replace 'FOOTER;', ""ONE_MORE_LINE `r`n FOOTER"" | Out-File output.txt "
And returned this error:
A cadeia de caracteres não tem o terminador: ".
+ CategoryInfo : ParserError: (:) [], ParentContainsErrorRecordException
+ FullyQualifiedErrorId : TerminatorExpectedAtEndOfString
EDIT2 - Possible Solution:
I realized that using the PowerShell command altered the encoding of the file, breaking the echo ONE_MORE_LINE, and using the suggestion from #AnsgarWiechers, I made this code
findstr /v "FOOTER" output.sql > new_output.sql
TYPE new_output.sql > output.sql
del new_output.sql
ECHO. >> %%f
ECHO ONE_MORE_LINE >> %%f
ECHO FOOTER >> %%f
ECHO. >> %%f
What it does is using the commant findstr /v "FOOTER" I look for all lines that are not FOOTER in the file output.sql and Write it on new_output.sql
Then I TYPE it back to the original file, and DEL the new_output.sql
Then I Echo all the lines I need right under it.
It works BUT, for big files I think that re-writing it twice will take a lot of time, but I can't figure an other solution.

Answer 1

When working with big files it's best to use a file stream. More typical methods of reading a file line-by-line using a Batch for /f loop or using Get-Content in PowerShell to read the entire file into memory can slow the process to a crawl with large files. Using a file stream on the other hand, you can nearly instantly seek back from the end of the file to the beginning of the last line, insert your desired data, and then reassemble the bytes you overwrote.
The following example will use PowerShell's access to .NET methods to open a file as a byte stream for rapid reading and writing. See inline comments for details. File encoding will hopefully be preserved. Save this with a .bat extension and give it a shot.
<# : batch portion
#echo off & setlocal
set "file=test.txt"
set "line=Line to insert!"
powershell -noprofile "iex (${%~f0} | out-string)"
goto :EOF
: end batch / begin PowerShell hybrid #>
# construct a file stream for reading and writing $env:file
$IOstream = new-object IO.FileStream((gi $env:file).FullName,
[IO.FileMode]::OpenOrCreate, [IO.FileAccess]::ReadWrite)
# read BOM to determine file encoding
$reader = new-object IO.StreamReader($IOstream)
[void]$reader.Read((new-object byte[] 3), 0, 3)
$encoding = $reader.CurrentEncoding
$reader.DiscardBufferedData()
# convert line-to-insert to file's native encoding
$utf8line = [Text.Encoding]::UTF8.GetBytes("`r`n$env:line")
$line = [Text.Encoding]::Convert([Text.Encoding]::UTF8, $encoding, $utf8line)
$charSize = [math]::ceiling($line.length / $utf8line.length)
# move pointer to the end of the stream
$pos = $IOstream.Seek(0, [IO.SeekOrigin]::End)
# walk back pointer while stream returns no error
while ($char -gt -1) {
$IOstream.Position = --$pos
$char = $reader.Peek()
$reader.DiscardBufferedData()
# break out of loop when line feed preceding non-whitespace is found
if ($foundPrintable) { if ($char -eq 10) { break } }
else { if ([char]$char -match "\S") { $foundPrintable++ } }
}
# step pointer back to carriage return and read to end into $buffer with $line prepended
$pos -= $charSize
$IOstream.Position = $pos
$buffer = $encoding.GetBytes($encoding.GetString($line) + $reader.ReadToEnd())
$IOStream.Position = $pos
"Inserting data at byte $pos"
$IOstream.Write($buffer, 0, $buffer.Length)
# Garbage collection
$reader.Dispose()
$IOstream.Dispose()
This method should be much more efficient than reading the file from the beginning, or copying the entire file into memory or on disk with a new line inserted. In my testing, it inserts the line into a hundred meg file in about 1/3 of a second.