how to remove a line from a txt file

graziano1968 · 11-06-2009, 04:56 AM

Hello

how to remove (from console) a line from a txt file

suppose I have

housea
bhouse
house
bhousex

and I would remove the line house to have this

housea
bhouse
bhousex

I tried using

replace "house" "" -- file.txt

however it's not good because I receive this
a
b
bx

How to do ?

Thank you

indiajoe · 11-06-2009, 05:06 AM

Hi
Use sed as follows

Code:

sed '/^house$/d' file.txt

-Cheers
indiajoe

ghostdog74 · 11-06-2009, 05:10 AM

Code:

awk '$1!="house"' file

vikas027 · 11-06-2009, 05:32 AM

Quote:

Originally Posted by ghostdog74

Code:

awk '$1!="house"' file

Could you please explain how this code works ? Thanks in adv.

vikas027 · 11-06-2009, 05:38 AM

Even I struggles in these kinda problems in my starting days,

then I came across sed/awk. These two tools are God for shell scripting people.

For your case, if you want to delete any specific line (if you know the line no.)

Like in you case it is line 3

Code:

sed -i '3 d' file

Other than this you can also use grep command "-w" switch.

Code:

# cat file | grep -vw house
housea
bhouse
bhousex

Hope this helps.

ghostdog74 · 11-06-2009, 06:08 AM

Quote:

Originally Posted by vikas027

Could you please explain how this code works ? Thanks in adv.

what is it that you don't understand? because its very simple. check field 1 against the whole word "house" using inequality. If field1 not equal "house", print.

ghostdog74 · 11-06-2009, 06:10 AM

Quote:

Originally Posted by vikas027

Even I struggles in these kinda problems in my starting days,

then I came across sed/awk. These two tools are God for shell scripting people.

only 1 good tool is needed, awk. If you know awk, no need to use sed. their functions overlap. I don't really understand why people like to combine them together.

Quote:

Other than this you can also use grep command "-w" switch.

Code:

# cat file | grep -vw house
housea
bhouse
bhousex

Hope this helps.

no need cat.

Code:

grep -vw house file

trey85stang · 11-06-2009, 09:19 AM

Quote:

Originally Posted by ghostdog74

only 1 good tool is needed, awk. If you know awk, no need to use sed. their functions overlap. I don't really understand why people like to combine them together.

no need cat.

Code:

grep -vw house file

you can't make a direct edit to a file with just awk... That's where sed comes in.

pixellany · 11-06-2009, 09:28 AM

Quote:

Originally Posted by trey85stang

you can't make a direct edit to a file with just awk... That's where sed comes in.

You mean, I assume: sed -i
Be aware of the risk in this if your sed regex is not correct

But you can of course redirect to a file from any utility--n'est-ce pas?

ghostdog74 · 11-06-2009, 10:04 AM

Quote:

Originally Posted by trey85stang

you can't make a direct edit to a file with just awk... That's where sed comes in.

that's only trivial. at the back end, sed is creating "temp file" as well, just that its not visible to you.

Code:

awk '{....}' file > temp;mv temp file

since you want to compare, i give you this: how about using sed to calculate sine of a number? or logarithm?

Code:

awk 'BEGIN{print sin(90)}' file
awk 'BEGIN{print log(10)}' file

vikas027 · 11-06-2009, 01:09 PM

Quote:

Originally Posted by ghostdog74

what is it that you don't understand? because its very simple. check field 1 against the whole word "house" using inequality. If field1 not equal "house", print.

Got it

pixellany · 11-06-2009, 07:39 PM

For the record, I want to acknowledge that AWK is better than SED...........sometimes

ghostdog74 · 11-06-2009, 07:51 PM

Quote:

Originally Posted by pixellany

For the record, I want to acknowledge that AWK is better than SED...........sometimes

just curious, can you tell me when is the time that sed is better than awk, considering what they both can and cannot do?

pixellany · 11-07-2009, 09:45 AM

Quote:

Originally Posted by ghostdog74

just curious, can you tell me when is the time that sed is better than awk, considering what they both can and cannot do?

First, I was making a feeble attempt at humor....

You have me at a disadvantage--since I do not know AWK as well as I do SED. My impression is that SED will do certain kind of things faster and with less code.

Here's one thing I was just playing with (not sure if it is optimized for SED)

Code:

sed '=' filename | sed 'N;s/\n/\t/' > newfilename

Numbers the lines, with the number on the same line, with a tab.

What is the equivalent in AWK?

ghostdog74 · 11-07-2009, 11:06 AM

Quote:

Originally Posted by pixellany

Code:

sed '=' filename | sed 'N;s/\n/\t/' > newfilename

Numbers the lines, with the number on the same line, with a tab.

What is the equivalent in AWK?

The equivalent, with 1 invocation of awk process.

Code:

$ awk '{print NR"\t"$0}' file

Now let's see how they perform on big files

Code:

$ wc -l <file
3168000
$ head -5 file
this is line
this is line
this is line
this is line
this is line

$ time sed '=' file | sed 'N;s/\n/\t/' > /dev/null

real    0m10.224s
user    0m9.922s
sys     0m0.291s

$ time awk '{print NR"\t"$0}' file > /dev/null

real    0m7.997s
user    0m7.921s
sys     0m0.071s

$ time sed '=' file | sed 'N;s/\n/\t/' > /dev/null

real    0m10.233s
user    0m9.889s
sys     0m0.322s

$ time awk '{print NR"\t"$0}' file > /dev/null

real    0m7.963s
user    0m7.867s
sys     0m0.079s

awk runs faster than sed. I believe there should be a way to only invoke one sed process...maybe it could speed things up a bit, but overall in this test, awk=1,sed=0