delete leading blanks in a file

[Clemens Kleibusch]

How can I delete multiple leading blanks in a file?
This site ignores leading blanks so I substitute all blanks with "B" in my example:
#cat file
BBline1
line2
Bline3
BBBBBline4BwithBblank
#
should lead to
#cat file2
line1
line2
line3
line4BwithBblank
#
I found a solution with sed that deletes me only one leading blank :
sed -e 's/^ //g'
but there may be 20 leading blanks in a line...

TIA Clemens

[James R. Ferguson]

Hi Clemens:

# sed -e 's/^[ \t]*//' filename

Note: Type a TAB character in lieu of \t. Some versions of 'sed' simply don't recognize the '\t'.

Regards!

...JRF...

[Jochen Heuer]

Hi Clemens,

you can also try this:

sed -e 's/^[: :]*//'

This should remove all white space (tab or spaces) from the beginning of the line.

Regards,

Jochen

[Jochen Heuer]

Note: there is a blank between
[: and :] ...

[Clemens Kleibusch]

Hi James,
it's great! It works.
But can you explain why?

Clemens

[Jochen Heuer]

Hi Clemens,

it goes somehow like this:

^ --> beginning of line (as you probably already knew)

[\t]* --> followed by zero or more characters like \t (although I could not find an explanation for \t or tab in 'man 5 regexp' it probably means the same as [:SPACE:] whitespace -- space or tab)

Regards,

Jochen

[H.Merijn Brand (procura]

Finally got through.

The perl version might be easier to read:

# perl -pe 's/^\s+//' infile >outfile

-p = print every line
-e = using the following expression

s/// = substitute in the current line
^ = start of line
\s = white space (blanks, tabs, spaces)
+ = one or more

Enjoy, have FUN! H.Merijn

[Jose Mosquera]

Hi,

To remove any space:
#cat |tr -d " " >

To remove tabs:
#cat |tr -d "\t" >

To remove both of them once:
#cat |tr -d " "|tr -d "\t" >

Rgds.

[Justo Exposito]

Hi,

You can do:

sed -e 's/\ \{1,\}/ /g' file

This change one or more blank by one.

Regards,

Justo.

[Clemens Kleibusch]

Thanks to all!
I'll try to resume:
Jochen, your first answer
sed -e 's/^[: :]*//' doesn't
work although I recognized the blank.
The explanation of \t is what I don't understand. I knew \t only for a tab. But here I works only for a blank -- that's what I want!
[:SPACE:] or [:BLANK:] doesn't work.
Procura, your perl is easy to understand but I never used perl before... Now I think about it!
Thanks to Jose Maria and Justo. You didn't recognize that I only want to process leading blanks.

[Laurie Gellatly · Accepted solution]

True blanks?
how about sed -e 's/^ *//'

Cheers ...Laurie :{)

[Clemens Kleibusch]

Hi Laurie,
that's it! It works just like James' 's/^[ \t]*//' but it seems to be logical.
Thanks
Clemens

[Jannik]

sed - stream text editor
-e - script
s/^ *// - s/regular expression/replacement/flags

BR,
Jannik

[Laurie Gellatly]

Your welcome.

...Laurie :{)

[James R. Ferguson]

Hi (again) Clemens:

An easier read and understood variation of my original post is this:

# sed -e 's/^[[:space:]]*//' filename

The [:space:] character class or characters which produce white space (blanks, tabs) in displayed text.

I choose to match *either* spaces or tabs, or a mixture in any combination, in order to provide the most general match. Had you originally used only space characters between the fields of your file, but then used 'unexpand' to change spaces to tabs, a regular expression consisting of only the space character would not have achieved your goal.

Your original 'sed' substitution failed to consider anything other than blank (space) characters (i.e. not TABS), More importantly, your regular expression (RE) consisted only of *one* character. By adding the asterisk following the character class we constructed a RE that matched zero or more occurrences of the characters found in the character class that preceded the asterisk. Thus, the matching string was represented by the maximum number of characters which defined the RE.

Regards!

...JRF...