0

I have the output from my diff command between two directories store in a file called differenceOutput.txt.

At present there are only two kinds of lines in differenceOutput.txt So all lines have either format A or B, where A and B look like this:

A)

Only in /tmp/__tmp_comp206_alex/test_files/: file_1.txt

B)

Files /tmp/__tmp_comp206_alex/test_files/file_3.conf and /tmp/__tmp_comp206_alex/diff_dir/file_3.conf differ

Using sed, I want to change all lines in differenceOutput.txt with format A to format C, and all lines with format B to format D, where C and D look like this:

C)

/tmp/__tmp_comp206_alex/test_files/file_1.txt is missing

D)

/tmp/__tmp_comp206_alex/diff_dir/file_3.conf differs

How can I do this with sed? Super confused with the sed syntax. I've spent hours trying to figure this out but can't make heads or tails of this. Can someone help me out please?

Improve this question
4
  • 1
    This is very easy with the specific examples you show, but can get extremely complicated for complex file names. Can your file names contain spaces? Can they contain newlines? Can your file names contain a :? Can they contain an and surrounded by spaces? Commented Oct 9, 2020 at 17:31
  • i can assume that my filenames contain none of things for this particular case Commented Oct 9, 2020 at 17:40
  • 1
    Then roaima's answer should be perfect for you. Do read his final note about using cmp though. Commented Oct 9, 2020 at 17:44
  • it doesn't work for me. Can you help me implement it? Commented Oct 9, 2020 at 22:47

2 Answers 2

Reset to default
2

Here you go. Two straightforward sed substitutions

a='Only in /tmp/__tmp_comp206_alex/test_files/: file_1.txt'
b='Files /tmp/__tmp_comp206_alex/test_files/file_3.conf and /tmp/__tmp_comp206_alex/diff_dir/file_3.conf differ'

printf "%s\n%s\n" "$a" "$b" |
sed -e 's!^Only in \([^:]*\)/: \(.*\)!\1/\2 is missing!' -e 's!^Files .* and \(.*\) differ$!\1 differs!'

Output

/tmp/__tmp_comp206_alex/test_files/file_1.txt is missing
/tmp/__tmp_comp206_alex/diff_dir/file_3.conf differs

Explanation

  • I've used ! instead of / as the separator in the sed recipe s/match/replacement/ since otherwise we'd have to escape every occurrence of / in the match and replacement strings
  • The \1 and \2 in the replacement match escape-bracketed expressions (ie \(...\)) in the match part

The big assumption is that you don't have filenames containing colons and other matching words from the diff output. The diff output is fragile at best and you might be better off rolling your own loop with find and cmp -s that produced the desired output directly. (That would be my preference for a robust solution.)

#!/bin/bash
src='/tmp/__tmp_comp206_alex/test_files'
dst='/tmp/__tmp_comp206_alex/diff_dir'

( cd "$src" && find -type f -print0 ) |
    while IFS= read -r -d '' item
    do
        if [[ ! -f "$dst/$item" ]]
        then
            printf "%s is missing\n" "$dst/$item"

        elif ! cmp -s "$src/$item" "$dst/$item"
        then
            printf "%s differs\n" "$dst/$item"
        fi
    done
Improve this answer
11
  • Thank you!! Sorry if this is a silly question, but how do I use this code to update the differenceOutput.txt, and then print that to screen? Commented Oct 9, 2020 at 17:39
  • Create a new file, then replace the old one with the new Commented Oct 9, 2020 at 17:40
  • i'm still pretty confused about how to actually do this. One of my friends did a one liner in their script like this and it did the job: sed -e "s/Only in $ogdir: /\/$1\//g" -e "s/Only in $compdir: /\/$2\//g" -e '/^\// s/$/ is missing/' -e 's/^F.*and //' -e '/differ/ s/$/s/' < output_diff.txt Commented Oct 9, 2020 at 17:59
  • 1
    @user140161 this is something similar. Almost identical, really. Just take roaima's sed command and run it on your file: sed -e 's!^Only in \([^:]*\)/: \(.*\)!\1/\2 is missing!' -e 's!^Files .* and \(.*\) differ$!\1 differs!' yourFile.txt Commented Oct 9, 2020 at 18:13
  • 1
    @user140161 you don't need to edit it as long as your list is in the format you described Commented Oct 9, 2020 at 21:58
1 
$ awk '
    sub(/^Only in /,"") { sub(/: /,""); $0=$0 " is missing" }
    sub(/^Files /,"")   { sub(/ and .*/,""); $0=$0 " differs" }
1' differenceOutput.txt
/tmp/__tmp_comp206_alex/test_files/file_1.txt is missing
/tmp/__tmp_comp206_alex/test_files/file_3.conf differs

That assumes your directory names don't contain :<blank> or <blank> and <blank> and none of your file/dir names contain newlines.

The above was tested using this file created from the sample input you provided in your question:

$ cat differenceOutput.txt
Only in /tmp/__tmp_comp206_alex/test_files/: file_1.txt
Files /tmp/__tmp_comp206_alex/test_files/file_3.conf and /tmp/__tmp_comp206_alex/diff_dir/file_3.conf differ
Improve this answer
3
  • hey ed, thanks for your comment. I feel like everybody has misinterpreted the questions a little bit, which is why it only partially works, and it seems impossible to explain through the comments here given how slow it is. Can I show you my code if u 5 have minutes? codeshare.io/2pB8p4 we can chat in real time there and i can actually exlain what it going on with my code there. Would appreciate it tons!! Commented Oct 9, 2020 at 23:12
  • 1
    Just edit your question to show a few lines of sample input and expected output that the posted solutions don't work for - that should be all that's needed. I personally am happy to help with a couple minutes of my time here and there when someone asks a reasonably clear question but I'm not interested in discussions or code walkthroughs to try to understand the problem (not for free anyway!), maybe others will be able to help if that's what you feel it'll take. Commented Oct 9, 2020 at 23:14
  • I got the code to work!!! YOU THE REAL MVP ED!!! All SED and DONE, awk > sed CONFIRMED Commented Oct 9, 2020 at 23:32

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.