Move a line in a file

[Gen2ly]

I'd like to so a simple thing at least it sound like it is but have no idea just how to go about it. I thought about using sed but from what I know of sed it's a line editor and just doesn't seem right for the task. First a line would have to be read then deleted then moved, hmm... yeah just thinking that this isn't the best way to go about it. Or is it? Off the bat I can't think of any tools that do this. Currently I got a line that begins with:


build...

And I'd like to move it earlier in the file that follows the line:


source...

How would you go about doing this?

[benj1]

how about this


#! /usr/bin/awk -f
/build/ {build = NR}
/source/ {source = NR}
//{array[NR] = $0}

END{
for (i=1 ; i<=NR;i++){
if (i == source){print array[build]}
if (i == build){continue}
print array[i]
}
}

you could then pipe it out to a file
ie
awkscript file.txt > outputfile.txt

you may have to modify the regex for /build/ and /source/ to be more specific and eliminate false positives

[ghostdog74]

the above method leaves a blank line as first line (though a minor issue), and you forgot to include <= in the END block for loop. That will omit out the last line.

another way, probably better with large files as I don't have to store every lines in array and print out at the END block.


awk '/build/{
f=0
print $0
for(i=1;i<=d;i++){ print a[i] }
g=0
delete a # remove array after found
next
}
/source/{ f=1; g=1 }
f{ a[++d]=$0 }
!g' file

[benj1]

amended.
except the blank line, will have a think

[ghostdog74]

amended.
except the blank line, will have a think

hint: since you build the array using NR, NR doesn't start with 0.

[benj1]

amended again thanks, again :)

[Gen2ly]

how about this


#! /usr/bin/awk -f
/build/ {build = NR}
/source/ {source = NR}
//{array[NR] = $0}

END{
for (i=1 ; i<=NR;i++){
if (i == source){print array[build]}
if (i == build){continue}
print array[i]
}
}


Hey thanks. Unfortunately, it didn't work:


awk: /home/todd/.bin/bs
awk: ^ syntax error

I'm new to awk so I don't know how to fix this, but am beginning to realize that I'm going to have to learn this. :)


...as I don't have to store every lines in array and print out at the END block.


awk '/build/{
f=0
print $0
for(i=1;i<=d;i++){ print a[i] }
g=0
delete a # remove array after found
next
}
/source/{ f=1; g=1 }
f{ a[++d]=$0 }
!g' file


This is beyond me to understand but I have seen something like 'for(i=1;i<=d;i++)' in bash scripts before that created lists from discovered contents. Again, getting to understand that I'll have to learn awk as a couple things I want to do with sed now just aren't possibly. But this works great, thanks for the help.

[A_Fiachra]

Damn -- I would do it in Perl -- but you guys are handy with awk (a language I've always had a secret envy for).

[ghostdog74]

This is beyond me to understand but I have seen something like 'for(i=1;i<=d;i++)' in bash scripts before that created lists from discovered contents. Again, getting to understand that I'll have to learn awk as a couple things I want to do with sed now just aren't possibly. But this works great, thanks for the help.



awk '/build/{
# if build is found
f=0 #remove flag
print $0 # print the current line
for(i=1;i<=d;i++){ print a[i] } #print the array of lines after source and before build
g=0 #because build already found, set g flag to 0 so the below !g part can print other lines
delete a # remove array after found
next
}
/source/{ f=1; g=1 } #if source found, set flag f and g.
f{ a[++d]=$0 } # if source is found, store lines after source into array, then continue to search for build
!g' # print the rest of the lines before source is found and after build is found...
file

[benj1]

Hey thanks. Unfortunately, it didn't work:


awk: /home/todd/.bin/bs
awk: ^ syntax error



how did you call it ?, it should be

thisscriptfile.awk foofile.txt

not a huge problem ghostdogs script should work fine, and as he points out should be more efficient.

@A_fiachra
you should give it a try, its not especially hard to learn (easier than perl :)) and its amazing at text processing, the only problem is, is that you will be forever looking at shellscripts like:

grep foo | sed bar | awk '{print $1}'
and thinking it could all have been done, more clearly and concisely in awk.
heres a link (http://awk.info/?index) to some useful stuff

[geirha]

I believe this ed should do it. Note that it edits the file in-place.

printf "%s\n" "/^build/d" "/^source/x" "w" | ed -s file.txt

It sends three commands to ed.


/^build/d - find the first line that mathces "^build" and delete it (The line's content is stored in the cut buffer)

/^source/x - find the first line that matches "^source" and paste the contents of the cut buffer below it.

w - write the changes

[Gen2ly]

awk '/build/{
# if build is found
f=0 #remove flag
print $0 # print the current line
for(i=1;i<=d;i++){ print a[i] } #print the array of lines after source and before build...

Not sure how the array understands to the source line, but thanks for the explanation. I'm getting to like how awk works.


how did you call it ?, it should be

thisscriptfile.awk foofile.txt
...

Yeah, I called this from my script path which is a hidden folder. I'm thinking your script is interpretting the . as a regexp variable???


I believe this ed should do it. Note that it edits the file in-place.

printf "%s\n" "/^build/d" "/^source/x" "w" | ed -s file.txt

It sends three commands to ed...[/code]

Yeah, I originally wrote the thread trying to remember this command. I had seen 'ed' before but didn't remember it. +1 for helping me remember this and for ease of use.

I've run into a bit of a problem though. At times, (I've discovered) the source and build lines expand over more lines. For example:


source=('78905905f5a4ed82160c327f3fd34cba'
'5277a9164001a4276837b59dade26af2'
'3f8b60b6fbb993c18442b62ea661aa6b')

build=('a-file-name'
'another-file-name')

I've tried using the ed command and defining a range:


printf "%s\n" "/^build/,/)/d" "/^source/,/)/x" "w" | ed -s file.txt

with no luck. Can I do this with 'ed'??? Or anyone know a way to get awk to do it??

[ghostdog74]

source=('78905905f5a4ed82160c327f3fd34cba'
'5277a9164001a4276837b59dade26af2'
'3f8b60b6fbb993c18442b62ea661aa6b')

build=('a-file-name'
'another-file-name')
... Or anyone know a way to get awk to do it??
you will need extra check, since your build ends with ), you can check whether it spans another line.


/build/ && !/\)$/ {
# if build found but does not end in )
s=$0
getline #get the next line
s=s$0 # join them
# if you are sure those statements don't span more than 2 lines
print s
}
/build/ && /\)$/{
.......
}