awk equivalent of LTRIM function in C

Question

I need to delete leading 0s only from a string. I found that there is no in-built function like LTRIM as in C.

I'm thinking of the below AWK script to do that:

awk -F"," 'BEGIN { a[$1] }
   for (v in a) {
       {if ($v == 0) {delete a[$v]; print a;} else exit;}
   }'

But guess I'm not declaring the array correctly, and it throws error. Sorry new to AWK programming. Can you please help me to put it together.

Possible duplicate of [How can I trim white space from a variable in awk?](http://stackoverflow.com/questions/9985528/how-can-i-trim-white-space-from-a-variable-in-awk) — jas, Mar 05 '16 at 20:49
There isn't a function `LTRIM()` in standard C (nor `ltrim()`). If you're used to using said function, it is a non-standard extension for the environment you work in. — Jonathan Leffler, Mar 05 '16 at 21:58
You're asking us to help you do something that may not need to be done at all. See http://xyproblem.info/. Post sample input and expected output so we can see what you want to do, as opposed to how you think you need to do it, so we can start to help you. — Ed Morton, Mar 06 '16 at 04:42
@EdMorton Probably used the word "thinking" in the wrong context. I did already try the above and it threw an error. Will be more cautious to put proper "inputs" and "outputs" while posting queries in future. Thanks a lot for the pointer — dig_123, Mar 06 '16 at 16:07
No, the word `thinking` was fine, it's just that there's a LOT of different ways to trim leading zeros and they all make sense in different contexts so unless you tell us WHY you want to trim leading zeros we're just throwing out a bunch of code, any/all of which could be completely wrong for your application even though they all do trim leading zeros. — Ed Morton, Mar 06 '16 at 16:14

peak · Answer 1 · 2016-03-06T06:23:30.260

2

Here is a somewhat generic ltrim function that can be called as ltrim(s) or ltrim(s,c), where c is the character to be trimmed (assuming it is not a special regex character) and where c defaults to " ":

function ltrim(s,c) {if (c==""){c=" "} sub("^" c "*","",s); return s}

This can be called with 0, e.g. ltrim($0,0)

NOTE:

This will work for some special characters (e.g. "*"), but if you want to trim special characters, it would probably be simplest to call the appropriate sub() function directly.

edited Mar 06 '16 at 06:23

answered Mar 06 '16 at 04:36

peak

105,803
17
152
177

You can't use that when `c` is a some regexp metachars, e.g. `.`. You need `sub("^[" c "]+",...`. – Ed Morton Mar 06 '16 at 04:40
1

... unless c is "^" ... but I've revised to clarify. Thanks for your vigilance. – peak Mar 06 '16 at 06:25
Yeah, I'm not sure it'd work for "\" either but it gets you closer. Maybe `"^\" c "+"`? idk... – Ed Morton Mar 06 '16 at 16:17

Thomas Dickey · Accepted Answer · 2016-03-06T13:24:28.853

2

Using awk, as requested:

#!/usr/bin/awk -f
/^0$/   { print; next; }
/^0*[^0-9]/ { print; next; }
/^0/    { sub("^0+", "", $0); print; next; }
        { print $0; }

This provides for not trimming a plain "0" to an empty string, as well as avoiding the (probably) unwanted trimming of non-numeric fields. If the latter is actually desired behavior, the second pattern/action can be commented out. In either case, substitution is the way to go, since adding a number to a non-numeric field will generate an error.

Input:

Output:

Output trimming non-numeric fields:

0
x
x
12

edited Mar 06 '16 at 13:24

answered Mar 06 '16 at 11:59

Thomas Dickey

51,086
7
70
105

That could be written much more concisely as `awk '/^0[0-9]+$/{sub(/^0+/,"")}1' file`. – Ed Morton Mar 06 '16 at 16:24
I was allowing for the possibility of input like `0123,456` – Thomas Dickey Mar 06 '16 at 16:25
Ah, I see. As good a guess as any at the problem the OP is trying to solve! Adding a number to a non-numeric field won't generate an error, btw, the non-numeric field will just be treated as a string with numeric value zero. – Ed Morton Mar 06 '16 at 16:31

score 2 · Answer 3 · answered Mar 06 '16 at 14:36

Based on other recent questions you posted, you appear to be struggling with the basics of the awk language.

I will not attempt to answer your original question, but instead try to get you on the way in your investigation of the awk language.

It is true that the syntax of awk expressions is similar to c. However there are some important differences.

I would recommend that you spend some time reading a primer on awk and find some exercises. Try for instance the Gnu Awk Getting Started.

That said, there are two major differences with C that I will highlight here:

Types

Awk only uses strings and numbers -- it decides based on context whether it needs to treat input as text or as a number. In some cases you may need to force conversion to string or to a number.

Structure

An Awk program always follows the same structure of a series of patterns, each followed by an action, enclosed in curly braces: pattern { action }:

pattern { action }
pattern { action }
.
.
.
pattern { action }

Patterns can be regular expressions or comparisons of strings or numbers. If a pattern evaluates as true, the associated action is executed.

An empty pattern always triggers an action. The { action } part is optional and is equivalent to { print }.

An empty pattern with no action will do nothing.

Some patterns like BEGIN and END get special treatment. Before reading stdin or opening any files, awk will first collect all BEGIN statements in the program and execute their associated actions in order.

It will then start processing stdin or any files given and subject each line to all other pattern/action pairs in order.

Once all input is exhausted, all files are closed, and awk will process the actions belonging to all END patterns, again in order of appearance.

You can use BEGIN action to initialize variables. END actions are typically used to report summaries.

A warning: Quite often we see people trying to pass data from the shell by partially unquoting the awk script, or by using double quotes. Don't do this; instead, use the awk -v option to pass on parameters into the program:

a="two"
b="strings"
awk -v a=$a \
    -v b=$b \
'BEGIN {
    print a, b
}'
two strings

Thanks a lot. You got my problem right. Lately doing a lot of file content manipulations ( and pretty big files) at my work. Came to know awk is the fastest in these scenarios. Thanks again for getting me started...with AWK. — dig_123, Mar 06 '16 at 16:17
Just trying to address your *need* - which was the underlying theme of your questions. — Henk Langeveld, Mar 06 '16 at 16:23

score 1 · Answer 4 · answered Mar 05 '16 at 21:51

1

you can force awk to convert the field to a number and leading zeros by default will be eliminated.

e.g.

$ echo 0001 | awk '{print $1+0}'

1

answered Mar 05 '16 at 21:51

karakfa

66,216
7
41
56

score 1 · Answer 5 · answered Mar 06 '16 at 08:58

1

If I understand correctly, and you just want to trim the leading '0's from a value in bash, you can use sed to provide precise regex control, or a simple loop works well -- and eliminates spawning a subshell with the external utility call. For example:

var=00104

Using sed:

$ echo "$var" | sed 's/^0*//'
104

or using a herestring to eliminate the pipe and additional subshell (bash only)

$ sed 's/^0*//' <<<$var
104

Using a simple loop with string indexes:

while [ "${var:0:1}" = '0' ]; do
    var="${var:1}"
done

var will contain 104 following 2 iterations of the loop.

answered Mar 06 '16 at 08:58

David C. Rankin

81,885
6
58
85

This doesn't handle the special case where the value is zero. – Thomas Dickey Mar 06 '16 at 11:34
It was never intended to handle the *trivial* case. An additional test of the result would be all that is required. – David C. Rankin Mar 06 '16 at 11:37

awk equivalent of LTRIM function in C

5 Answers5

Linked