I am a Linux newbie.
I wrote a shell script to handle several html files.

I have several questions:

1 - why this work
cat file.txt |sed -e 's/#/\n/'
(substitute every occurrence of '#'
with a LF)

... and this doesn't work?
cat file2.txt | sed -e 's/\//'
My goal was to get rid off all LF's...

I was obliged to revert to a vim macro:
vim -es file2.txt<m.vim

The content of the macro m.vim was:
:% s/\n//
:w
:q
This worked but now I need a further macro
(the big shell macro and the vim macro...)

2 - the output of the shell macro needs some
formatting...

Is it possible to load the output into
a variable ? or to be more detailed...

is there any mean to get the line length
of each line of a text file?

If it was possible to load a variable with
for example the 'cat file' output (for
a text file of only 1 line using head)...

from a string variable you can obtain the
length and then conditionally output one
or more tabs to format purposes...

3 -When I start working I open a console.
I adjust the shape of this console
running a script:

konsole --vt_sz 72x60 --nohist

Is there any method not so rude?
by means of configuration files, perhaps...?

Hey fiomba,

Good to hear your learning scripting. Unfortunately, I can only help with some of your question, as some are in areas I haven't worked with, and I have to leave soon.
I can give you an idea of how I would trim newlines, but it doesn't use sed (I'm sure there's a way, but I find tr faster).

cat file.txt | tr -d "\n"

Check out the man page on tr, it's a pretty standard command, so it should be fairly portable.

If you wanted to put this into a variable, all you would need to do is set it to execute in a subshell, and have the variable equal that.

textvariable=$(cat file.txt | tr -d "\n")

By using the $( ) form, commands inside will be executed as shell commands before they are used in the script.

Hope these tidbits help you on your way, I know there is a command to count characters, but I've never used it in a script so I forgot it

Thanks, scottman, you answered to 2 of my 3 questions
(the more important ones).

Your suggestions worked perfectly.

count characters:
wc -c filename
or you can use stdin
wc -c < filename
etc...
echo $STRING | wc -c

Thanks also to you, 95se;
your is an alternative method to
count lines.

In effect it has one step less, in
comparison to scottman's method
(load a variable with the line content and
obtain the string length).

To better thank for the help you all gave me,
I am considering to share my findings
(I am a Linux newbie but
not a programmer newbie).

The object of my findings is the manipulation
of html files but you can generalize.
Of course all is turned to newbies (like me),
and not to experts...
But I need some encouragement,
if someone is interested...

Actually it won't work: it will substitute the first occurence only for each pattern. you have to replace 's/#/\n/' with 's/#/\n/g'.
sed can't do that. \n is not recognized by sed in patterns... Side-note: the 'n' is missing from your pattern. But scottman's suggestion to use tr is excellent. I'm glad I learned it!
You could try that:
vim -es file2.txt <<-END
:% s/\\n//
:w
:q
END
Maybe that:
cat file.txt | while read line; do echo "$line" | wc -L; done >result.txt

Yves.

I know the use of sed subsitution.
In my case there is only one '#' per line and so I don't
need the "global " g.

Instead your example of vim macro without a macro file
is very interesting (I did not know this technique!).
Also you made a mistake (:% s/\\n//) instead of ":% s/\n//"
but in your case it works the same!

For the last point (length of each line) in my case I don't need
a loop because only one line pass through the filter:
cat file.txt | grep 'something' | wc -c

Thank you 'theYinYeti', especially for the vim part !

P.S. As far as my question about the size of the consolle (n.3)
is concerned, I have already found the answer (in Internet).

Oh? I thought \\ would be needed because bash interprets what gets written between <<-KEYWORD and KEYWORD. So I feared that "\n" would simply become "n". For example, if you want to put "$PATH" in your stdin, and not "/bin:/usr/bin:...", the you need to write "\$PATH".

Yves.

To get lengths of the lines, compact version: