How can I search a file for a specific string until a blank line and paste that into the first line-CodePudding

I have tried sed -n '/- name: string/,/^$/p' dire1/main.yml >> dire2/README.md

This does not add anything, I also want it to add it to the first line of dire2/readme.md if readme.md does not already include '- name: string' in the first line so as not to get duplicates.

For example if a file includes

- name: first step 
  uses: message

- name: second step

I'd want

- name: first step 
  uses: message

to be included in the first line of readme.md. But if readme.md already has this in the first line, it should not be added.

Another example The file is a longer file

name: build
another line here
yet another line

- name: title here
   another line 
   another line

- name: another title
   more lines here
   more lines 
 
- name: third title
  more lines
  more lines

In this example, i'd want it to populate the readme.md file with

- name: title here
   another line 
   another line

or any one of the other paragraphs beginning with '- name ' and ending at a blank line.

This all has to be searched from one file and added to the beginning of readme file if it doesn't already exist

Thank you for any help

CodePudding user response：

I build a quick awk solution, works for your example. First, the 3 example files:

kent$  head *.yml                                                                        
==> main.yml <==
- name: first step 
  uses: message

- name: second step

==> readme1.yml <==
key: 
  whatever2: 200

==> readme2.yml <==
- name: exists
  users: whatever

- key: 
  whatever2: 100

As you can see, the readme1.yml doesn't contain the first key (- name: xxx) in the main.yml. While the readme2.yml contains the key already.

Now the awk command: (the two commands are the same, only the input params are different):

$ awk -F':' -v RS="" 'NR==FNR{if(NR==1){k=$1;a[k]=$0;}next;} FNR==1 && !($1 in a){print a[k]}1' main.yml readme1.yml
- name: first step 
  uses: message
key: 
  whatever2: 200

$ awk -F':' -v RS="" 'NR==FNR{if(NR==1){k=$1;a[k]=$0;}next;} FNR==1 && !($1 in a){print a[k]}1' main.yml readme2.yml
- name: exists
  users: whatever
- key: 
  whatever2: 100

So, if you check the outputs above, it seems they are what you expected.

CodePudding user response：

With GNU sed you can extract the blocks and quit after the first:

sed -n '/^- name.*$/,/^$/{p;/^$/q}' file
- name: title here
    another line 
    another line