[SOLVED] Need solution to solve the formatting issue

flamingo_l · 03-11-2011, 12:11 AM

Hi All,

I have written a grep command to grep the files containing Windows directories.
I need to output this to a csv.

So here is the code.

Code:

OUTPUT_FILE="Sample-format.txt"
ACT_OUTPUT_FILE="Sample-format.csv"

grep -wHrnI --exclude=*.log '[A-Za-z]:[/\]\{1,2\}[A-Za-z]*[/\]\{1,2\}*' .>$OUTPUT_DIR/$OUTPUT_FILE
paste -s -d"\n" $OUTPUT_DIR/$OUTPUT_FILE|tr ":" "|" >>$OUTPUT_DIR/$ACT_OUTPUT_FILE

Now the problem is, when i do

Code:

tr ":" "|"

, then the search result which contains windows path like for eg: C:\jfj\ff is getting divided into two columns.

Can anybody please give me some inputs to solve this problem.

Disillusionist · 03-11-2011, 12:19 AM

It would help to see a sample of the input and the desired output.

EDIT: If you want to keep the : where it is followed by a \ then you could replace the tr statement:

Old code:

Code:

tr ":" "|"

New code:

Code:

sed "s/:\([^\\]\)/|\1/g"

Second Edit: Just curious

Why are you using a Pipe character as your delimiter in a comma seperated values file?

flamingo_l · 03-11-2011, 01:07 AM

hi Disillusionist,

Thank you for the reply.
The input are all java files so i can give u the input.
Expected Ouput is:

Quote:

Filename|Line no|String Found

Regardinig your second question,
Am using "|" as a delimiter because my search files can contain ",".

Your given code is working for windows directory containg c:\.. in some files i have directories with c:/...
Also taking a lot of time..

Quote:

sed "s/:\([^\\]\)/|\1/g"

The above code would not replace with | if it encounters a \. and you have put another \ to escape the meaning of \.(Please do correct if my understanging is wrong)

Can you please tell me how to include / as well.??

Also, please tell me the usage of

Quote:

,\1

in sed command.(since am new to sed)

Disillusionist · 03-11-2011, 01:29 AM

Just add the / character in the list of characters to ignore following the :

Code:

sed "s/:\([^\\/]\)/|\1/g"

There are a number of elements in play in this sed

:[^\\/] will match the : followed by any character other than a \ or /

Because we want to keep the second character in the output, we need to tag it for use later with \( and \) opening and closing brackets.

the \1 is stating that we want the tagged element (whatever it was) after the | which replaces the :

Example:

echo "this is an example"|sed "s:\([^ ]* [^ ]*\) \([^ ]* [^ ]*\):\2 \1:"

would produce "an example this is"

flamingo_l · 03-11-2011, 05:08 AM

hi Disillusionist,

It solved my problem. Thanka a lot.