Bash - 使用新行替换字符串的Sed / Awk

Question

I have a file with strings like this below and want replace everthing between the "steps:" and "-" with a new line:

- name: Change It uid: Change It_TRANS_App - Web type: TRANSACTION steps: 02 - name: Change It 1 type: BSTEP uid: Step 1_STEP_Change It_TRANS_App -strong text Web rules:

I've tried using awk like:

echo $string | awk {'gsub(/steps:.*?-/ , 'steps:\n-' )'; print $0}

However, I get the error below:

awk: {gsub(/steps:.*?-/ , steps:n- )
awk:                          ^ syntax error
awk: fatal: 0 is invalid as number of arguments for gsub

I've tried as well with sed:

sed -r 's:/\bsteps:\b.*?-/\n/' stringfile.txt > output.txt

and removing the ":":

sed -r 's/\bsteps.*?-/\n/' stringfile.txt > output.txt

In the first case, I get the error:

sed: -e expression #1, char 28: unterminated `s' command

And with the second sed I removes a lot of other things that it shouldn't.

[Edit] As you guys told me, I forgot to put the expected output, I've been thinking that it would be simple, however I'm not good. I've want to "break" as a new line cuting out just the word right after "step", in that case, would be "02". It would be to the first occurence of "-":

- name: Change It uid: Change It_TRANS_App - Web type: TRANSACTION steps:
- name: Change It 1 type: BSTEP uid: Step 1_STEP_Change It_TRANS_App -strong text Web rules:

Answer 1

You were close:

awk '{gsub(/steps:.*-/,"steps:\n-");print $0}' <<< "$sting"
- name: Change It uid: Change It_TRANS_App - Web type: TRANSACTION steps:
-strong text Web rules:

---

Can be shorted to:

awk '{gsub(/steps:.*-/,"steps:\n-")}1' <<< "$sting"

---

Ref Eds comments:

awk '{gsub(/steps:[^-]*-/,"steps:\n-")}1' t
- name: Change It uid: Change It_TRANS_App - Web type: TRANSACTION steps:
- name: Change It 1 type: BSTEP uid: Step 1_STEP_Change It_TRANS_App -strong text Web rules:

Answer 2

Since sed won't support non-greedy patterns *? , you could use a negated class character class.

sed -r 's/\bsteps[^-]*-/\n-/' file

Example:

$ sed -r 's/\bsteps[^-]*/\n/' file
- name: Change It uid: Change It_TRANS_App - Web type: TRANSACTION 
- name: Change It 1 type: BSTEP uid: Step 1_STEP_Change It_TRANS_App -strong text Web rules:

Answer 3

While awk and sed gurus are providing the solution, may I suggest to consider using Perl.

To substitute for the new line everything until the first occurrence of - in the string:

echo $string | perl -pe 's/(.*steps:).*?(-.*)/$1\n$2/'
- name: Change It uid: Change It_TRANS_App - Web type: TRANSACTION steps:
- name: Change It 1 type: BSTEP uid: Step 1_STEP_Change It_TRANS_App -strong text Web rules:

To substitute for the new line everything until the last occurrence of - in the string:

echo $string | perl -pe 's/(.*steps:).*(-.*)/$1\n$2/'
- name: Change It uid: Change It_TRANS_App - Web type: TRANSACTION steps:
-strong text Web rules:

Two expressions inside parenthesis are refereed to as $1 and $2 in the second half of the substitution expression. And then ? in the middle part regulates whether .* works greedy or not greedy.