Bdtyktl

I have this batch file below, which is a simple emulation of Unix’s head command – it will read the first COUNT rows from the input .csv file (named as a command-line parameter). However, unlike the Unix command, it writes to a hard-coded output file. The only issue now is that this batch file will be called several times to create this output file and will keep appending the results to the output file. I can't change the >> to > since this is a for loop and will just keep overwriting a single line into my output file, which is incorrect. Does anyone have any ideas on how to overwrite the output .csv file each time the batch file is run?

@echo off



if [%1] ==  goto usage

if [%2] ==  goto usage



call :print_head %1 %2

goto :eof



REM

REM print_head

REM Prints the first non-blank %1 lines in the file %2.

REM

:print_head

setlocal EnableDelayedExpansion

set /a counter=0



for /f ^"usebackq^ eol^=^



^ delims^=^" %%a in (%2) do (

        if "!counter!"=="%1" goto :eof

        @echo>>trimmed.csv %%a

        set /a counter+=1

)



goto :eof



:usage

echo Usage: head.bat COUNT FILENAME

Rather than deleting the file, I would personally prefer to overwrite the file with 'nothing':

type NUL>trimmed.csv

Put it somewhere above the loop.

This creates a blank file, or overwrites an existing file with an empty file, with the name trimmed.csv. Since it is overwriting, and not deleting and recreating, it should keep the original file creation date.

Perhaps just delete it at the start of the batch (or at the top of the :print_head section)?

Something like del trimmed.csv /Q.

That way it's deleted on the next run, but still gets created and appended to while it's running the loop.

I know it's an old question; However,as I didn't see this potential answer given, I'm adding it nonetheless as it may be a preferable answer for some folks...

Instead of just:

@echo>>trimmed.csv %%a

Which, as mentioned, will just append the existing output file, try using:

if [[ "$counter" -eq 0 ]]; then 

    @echo > trimmed.csv %%a

else

    @echo >> trimmed.csv %%a        

fi

That way, you don't have to worry about checking if the file exists and deleting it if it does. And additionally, at the first line of output automatically overwrite an existing file as a new empty file or start a new file if it doesn't and then subsequently, it will append the remainder of the lines.

Although, personally, I would probably use some formatted date appended or prepended to the file basename (i.e. 2018-11-17_trimmed.csv or trimmed_2018-11-17.csv) using the output from the 'date' command stuffed into a variable (so that 'date' isn't run for each loop iteration.) or even do both.

Here’s a (IMO) simpler, more direct way:
put the entire for loop into parentheses,
and move the I/O redirection outside the parentheses —
with > instead of >>:

(

    for /f ^"usebackq^ eol^=^



    ^ delims^=^" %%a in (%2) do (

            if "!counter!"=="%1" goto :eof

            @echo %%a

            set /a counter+=1

    )

) > trimmed.csv

So, when the loop is started, the file is opened for writing
(clobbering the previous contents, if the file already / still exists). 
The batch file then keeps the CSV file open as standard output
for the entire duration of the loop,
so anything written to standard output from within the loop
will go into the CSV file.

This might even be infinitesimally more efficient
than the original version of the batch file,
because that one (potentially)
has to open and close the output file COUNT times,
whereas my version needs to do it only once.