Shell Programming and Scripting

Hi

1st problem
--------------
i have this sed command in my unix script which replaces new line and carriage return in a line with the string "&#xA"
the script works fine in Linux 3.0.101-0.5, but not in AIX 1 7 , the "s/\r/\&#xA/g" replacement, replaces
all the character "r" in the file.

Code:

sed "s/^M/\&#xA/g;s/\r/\&#xA\/g;s/\r^M/\&#xA/g"  < filename.txt > filename 2.txt

2nd problem
--------------
and then i have a awk command that works on the file filename2.txt (output of the above command), below is the awk command i'm using..

Code:

awk -v W="$1" '$0 ~ W{ORS="";if(NR!=1)print "\n";}1' filename2.txt > filename3.txt

instead of writing the sed output to a file can i send the output to the awk command, so that it doesn't have to take too much time writing and reading to two files.

any help is appreciated

Thanks
Mj

Moderator's Comments:

Please use CODE and ICODE tags to display sample code, input, and output.

---------- Post updated 04-22-14 at 08:06 AM ---------- Previous update was 04-21-14 at 06:17 PM ----------

Hi All
Any response would be appreciated....

thanks
MJ

Looks like your shell and/or sed version don't take \r to represent the carriage return character. I think you need to resort to other representations, e.g <ctrl>v<ctrl>m or $'\r', or even a variable assigned from the output of a printf command.
For your second problem, why don't you just pipe sed's output to awk?

You could use awk or perl to catch the \r etc. by it's octal value, which is 15 or \015 in this case.

An example:

Code:

$ od -c infile
0000000    s   t   a   r   t  \r  \n   s   t   o   p  \n
0000014
$ awk '{gsub(/\015/,"\&\#xA"); print}' infile > outfile
$ od -c outfile
0000000    s   t   a   r   t   &   #   x   A  \n   s   t   o   p  \n
0000017

There should be plenty of ascii tables with the octal values on the net.
Maybe you can do it all together with awk instead of sed. I assume there is a way with sed, but I don't know it.

thanks rudic, i replaced /r with contorl M