How to count and replace text string in file?

[pstein]

Assume I want to count the number of occurencies of a certain text string in a file
and replace all these occurencies by another string. How can I do this?

Obviously this is easy with an GUI based editor like Notepad++.
But how can I achieve this from a DOS batchfile?

If necessary a non-dos-batch command but an external tool can be used.

So at the end a script like the following should be available:

set numocc = countocc("D:\work\myfile123.log","mystring")
Echo Number of occurencies = %numocc%
replace("D:\work\myfile123.log","mystring","newstringAAA",all)

Thank you
Peter

[dbenham]

This is quite simple if you use my REPL.BAT utility.

Code: Select all

@echo off
setlocal
set "file=D:\work\myfile123.log"
for /f %%N in ('^<"%file%" repl "mystring" "\n$&\n" LX ^| find /c "mystring"') do set numocc=%%N
echo Number of occurrences = %numocc%
type "%file%" | repl "mystring" "newstring" L >"%file%.new"
move /y "%file%.new" "%file%" >nul

Dave Benham

[foxidrive]

Maybe the L switch in repl would be useful, to stop regular expression terms from interfering with the task.
Plus the X switch in the first repl term.

[dbenham]

Both good points, thanks.
I've edited my prior post.


Dave Benham

[Squashman]

We do have the old fashion substitute in the Batch Library. But this won't give you the count of occurrences but could be modified to do so.
http://www.dostips.com/DtCodeCmdLib.php ... substitute

[pstein]

Code: Select all

@echo off
setlocal
set "file=D:\work\myfile123.log"
for /f %%N in ('<"%file%" repl "mystring" "\n$&\n" LX | find /c "mystring"') do set numocc=%%N
echo Number of occurrences = %numocc%
type "%file%" | repl "mystring" "newstring" L >"%file%.new"
move /y "%file%.new" "%file%" >nul

Thank you for this suggestion Dave but when I run this the following error occurs (under 64bit Win7):

"< was unexpected at this time."

I guess a modification is necessary.

[Squashman]

I believe you need to escape the PIPE in the FOR arguments.

[dbenham]

Yes, that was not my best day answering questions

I forgot to escape the redirection and the pipe in the command string.

My original post is fixed now.


Dave Benham

[pstein]

Ok, thank you.

However meanwhile a fried told me about the famous "sed" tool as a unix port to windows freely available from here:

http://gnuwin32.sourceforge.net/packages/sed.htm

Since I prefer solutions with short code I give it a try.
For a slight modification of the initial task I want to replace not a string but bytes by other byte(s).

When I code therefore in a DOS batch file:

echo start
sed 's/\xc3\xbc/\xtc/g' <test.txt >testnew.txt
pause

....and run it then I got the following error:

sed: -e expression #1, char 1: unknown command: `''

Why?

Does it matter if the DOS batch file is in ANSI or UTF-8 encoding?

[ShadowThief]

Ok, thank you.

However meanwhile a fried told me about the famous "sed" tool as a unix port to windows freely available from here:

http://gnuwin32.sourceforge.net/packages/sed.htm

Since I prefer solutions with short code I give it a try.
For a slight modification of the initial task I want to replace not a string but bytes by other byte(s).

When I code therefore in a DOS batch file:

echo start
sed 's/\xc3\xbc/\xtc/g' <test.txt >testnew.txt
pause

....and run it then I got the following error:

sed: -e expression #1, char 1: unknown command: `''

Why?

Does it matter if the DOS batch file is in ANSI or UTF-8 encoding?

Wait, did you use ` instead of '? Because that actually matters. You need to be using '.

[foxidrive]

GnuSed under Windows uses double quotes.

Code: Select all

sed "s/\xc3\xbc/\xtc/g" <test.txt >testnew.txt

\xtc isn't a valid term though...

[pstein]

Ok, thank you.

But how else can I specify a certain byte value as replacement?

How can I specify a german Umlaut (e.g. ü = &uuml;) as replacement?

[foxidrive]

But how else can I specify a certain byte value as replacement?

How can I specify a german Umlaut (e.g. ü = &uuml;) as replacement?

\xtc is specifying a hex byte using \x but tc is not a hex code.