I need to remove a trailing backslash=dot "\." and a trailing
backslash-dot-dot "\.." from output created by LogParser 2.2.

It's a listing of directories. For some reason, the output of a single
directory can be shown as follows:
c:\alldata\
c:\alldata\.
c:\alldata\..

I just need the first occurence.

---

Request for Question Clarification by efn-ga on 17 Oct 2006 19:21 PDT

What kind of help are you looking for here?

---

Request for Question Clarification by pafalafa-ga on 17 Oct 2006 19:25 PDT

Can't you just put the text into a text editor, and use Find/Replace
to remove any unwanted characters?

Like efn-ga says:  What sort of help do you want?

pafalafa-ga

---

Clarification of Question by lcid-ga on 18 Oct 2006 07:04 PDT

I use Microsoft's Logparser to list the all the filenames from a
computer system. The output from Logparser becomes a file. Each line
of the file lists one row of data, the name of the file on the
operating system.

For some reason, when a directory is listed, LogParser sometimes
appends two types of additional characters.

If somebody could write this in Perl, I can handle it. 

Original Output
c:\alldata  <<I want to keep this row
c:\alldata\. <<I want to delete this row
c:\alldata\.. << I want to delete this row
c:\cygwin\ << Keep
c:\cygwin\. << Delete
c:\cygwin\.. << Delete
c:\windows   <<keep
c:\temp      <<keep
c:\temp\.    <<delete
c:\temp\..   <<delete

If the string ends with a trailing "\.", then delete the entire row.
If the string ends with a trailing "\..", then delete the entire row.
Keep all other rows.

New Output after transformation
c:\alldata
c:\cygwin
c:\windows

Save the output to a file.

I am going to stuff the output into a database, but I know how to do that.

try the following using simple c++
----------------------------------

char *str = "c:\\alldata\\xyz\\..";
char drive[_MAX_DRIVE];
char dir[_MAX_DIR];
char fname[_MAX_FNAME];
char ext[_MAX_EXT];

_splitpath( str, drive, dir, fname, ext );

in drive & dir u can get path..

1. Install the Free Software Foundation GNU coreutilities:
http://gnuwin32.sourceforge.net/packages/coreutils.htm

2. Set your path environment variable to point to the utilities.

3. grep -v "\\\." output.txt > new.txt

Assuming your input from above is in output.txt:
c:\alldata  <<I want to keep this row
c:\alldata\. <<I want to delete this row
c:\alldata\.. << I want to delete this row
c:\cygwin\ << Keep
c:\cygwin\. << Delete
c:\cygwin\.. << Delete
c:\windows   <<keep
c:\temp      <<keep
c:\temp\.    <<delete
c:\temp\..   <<delete

new.txt will now contain:
c:\alldata  <<I want to keep this row
c:\cygwin\ << Keep
c:\windows   <<keep
c:\temp      <<keep

There are numerous other ways you could do this with awk or sed as
well. Note that the reason you are seeing this is that the directory
output is adding the shortcuts "." and ".." which are the current and
parent directories respectively. EG try typing "dir ." and "dir .." to
see what you get.

Note I'm not a researcher so this won't cost you anything if it works.

Sometimes within a row is an embedded "\.". I think these are hidden directories

Hypothetical example
c:\alldata\nextdir\.anotherdir      <<-"Real" Directory
c:\alldata\nextdir\.anotherdir\.    <<-Current Directory with the single dot
c:\alldata\nextdir\.anotherdir\..   <<-Parent directory with double dot.

I would like to keep the first instance, not the second and third. A
simple GREP finds the second and third instance too.

I suppose I could just read the book called "Mastering Regular
Expressions" and figure this out.

You can tell grep to only match at the end of a line using the "$" character:
grep -v "\\\.\.$" output.txt | grep -v "\\\.$" > new.txt

Note this won't work with the sample where you have added comments
since it is no longer at the end of the line. You could do this with a
single grep to match one or more "." chars at the end but I'm not
skilled enough at regex to figure out how. The sample above takes
output.txt containing:

c:\alldata\nextdir\.anotherdir
c:\alldata\nextdir\.anotherdir\.
c:\alldata\nextdir\.anotherdir\..

and the result in new.txt is:
c:\alldata\nextdir\.anotherdir

Following up on my previous comment, you can probably do a single grep
searching for a "." at the end of a line, as I don't think Windows
will allow a filename to end with a period:

grep -v "\.$" output.txt > new.txt

if you already got perl installed in your server then no need to
install any other thirdparty programs to do this... simply go to the
directory where you've got your logs stored. then type this:

perl -pi~ -e "s/\\(\.|\.\.)$//" [FILENAME]

(this presumes perl is already in your environment variables.)

for example you might have a file called "log.txt" go to dir where
log.txt is saved (eg c:\logs)

then type: perl -pi~ -e "s/\\(\.|\.\.)$//" log.txt

perl will create a log.txt~ as a backup of original file and then
remove the dots according the regexp i wrote there in your original
file (log.txt)

so at the end you'll have a log.txt (the edited file) and a log.txt~
backup of the edited file!

infact i just used this command this evening for renaming some text in
10,000 files ;)