在Linux中将文件排序为注释

Question

Say I have a sort_me.txt file:

a
d
b
c
f
g

// dont mix the two sections

a
c
d 
b

at the moment, I do the obvious sort sort_me.txt and I get:

a
a
b
b
c
c
d
d
// dont mix the two sections 
f
g

Which of course is not what I want, what I want is for it to sort the section before the comment and then the section after the comment separately.

With the desired result as:

a
b
c
d
f
g
// dont mix the two sections
a
b
c
d

Answer 1

Perl to the rescue:

perl -007 -nE '
    @sections = map [ split /\n/ ], split m{^(?=//)}m;
    say join "\n", sort @$_ for @sections;
' -- file

• -007 reads the whole file instead of processing it line by line (only works if the file isn't huge)
• @sections is an array of arrays, outer arrays correspond to sections, inner arrays to individual lines

If the file is too large to fit into the memory, you need to process it line by line, storing only the current section:

perl -ne '
    sub out { print sort @lines; @lines = $_ }
    if (m{^//}) { out() }
    else { push @lines, $_ }
    END { out() }
' -- file

Answer 2

Without perl you can do it with a script like this:

#!/bin/bash

FILE_NAME=$1

SEPARATOR='//'

LINE_NUMBER=`grep -n $SEPARATOR $FILE_NAME  | cut -f1 -d:`

FILE_LENGTH=`wc -l $FILE_NAME | cut -f1 -d\s`

head -$(($LINE_NUMBER-1)) $FILE_NAME | sort

grep $SEPARATOR $FILE_NAME

tail -$(($FILE_LENGTH-$LINE_NUMBER-1)) $FILE_NAME | sort

It searches for the separator line and sort the sections one-by-one. Of course if you have more than two sections it won't work.

Answer 3

I was thinking about using csplit to split the sections into separate files, but of course there should be easier ways to accomplish this:

#!/bin/bash

linenum=`csplit -z $1 /^$/ {*}`
count=0
output=''
for line in $linenum
  do
    file=`printf "xx%.2d" $count`
    sorted=`cat $file | sort`
    output="$output$sorted"
    ((count++))
  done
echo "$output"

Notice that csplit will create a temporary file for each section, so you might update the above script to unlink each of these ie unlink $file .