Parse yaml file without using outside tool

Question

I am new to bash, dragged into a problem where I need to update some dynamic parameters in a yaml file

Based on query (search4.subsearch1.name), i need to update

name: old_name

into

name: new_name

• Here is the sample file:

  search1:
      name: name1
      pass: pass1
      date: date1
  #these are just commented texts
  
  search2:
    #Adding few more commented lines
      myname: somename
      name: name2
      pass: pass2
      someotherparam: param1
  
  search3:
      nameAndSurname: NS
      namingConv: true
      varSet: P3
  
  search4:
      #Again adding some special name: strings here
          myname: somename1
          name: new_name
          pass: new_pass
          ptherparam1: T3
  
      subsearch1:
          #Again adding some special name strings here
              myname: somename2
              name: old_name
              pass: pass2
              ptherparam1: param3
  search5:
      #Again adding some special name strings here
          var1: value1
          name: some_name
          pass: some_pass
  

So based on query (search4.subsearch1.name), I need to update name: old_name into name: new_name

• Expeced output:

  search4:
      #Again adding some special name: strings here
          myname: somename1
          name: new_name
          pass: new_pass
          ptherparam1: T3
  
      subsearch1:
          #Again adding some special name strings here
              myname: somename2
              name: new_name
              pass: pass2
              ptherparam1: param3
  

Tried with awk and sed but that did not work:

lineNR=$(awk "/\<search4:/{f=1} f && /\<name:/ {print NR; exit}" testfile.txt);sed "${lineNR}s/name.*/name: new_name/" testfile.txt

Answer 1

Assuming you can have Mike Farah's yq downloaded and installed, the process becomes trivial

yq eval '.search4.subsearch1.name = "new_name"' file.yml

I note that you say in a comment that "we do not have internet connectivity, and any outside package is not allowed in those linux boxes". Even if you yourself are not allowed to add packages, there will be an administrative process for asking for software to be added to these systems. A "Change Request", maybe. Sometimes it's worth going through the apparent pain of that to deliver more reliable software that in turn is easier to support

---

^{For testing, I had to edit the content to ensure that indendation was consistent under search4. I have not edited the content in the question itself}

Answer 2

Using any awk in any shell on every Unix box:

$ cat tst.awk
NF && !/^[[:space:]]*#/ && match($0,/^[[:space:]]*/) {
    prevIndent = currIndent
    currIndent = RLENGTH

    name = $1
    sub(/:.*/,"",name)

    if ( currIndent == 0 ) {
        path = name
    }
    else {
        if ( currIndent <= prevIndent ) {
            sub(/\.[^.]+$/,"",path)
        }
        path = path "." name
    }

    if ( path == tgt ) {
        $0 = substr($0,1,currIndent) $1 " " val
    }
}
{ print }

$ awk -v tgt='search4.subsearch1.name' -v val='new_name' -f tst.awk file
search1:
    name: name1
    pass: pass1
    date: date1
#these are just commented texts

search2:
  #Adding few more commented lines
    myname: somename
    name: name2
    pass: pass2
    someotherparam: param1

search3:
    nameAndSurname: NS
    namingConv: true
    varSet: P3

search4:
    #Again adding some special name: strings here
        myname: somename1
        name: new_name
        pass: new_pass
        ptherparam1: T3

    subsearch1:
        #Again adding some special name strings here
            myname: somename2
            name: new_name
            pass: pass2
            ptherparam1: param3
search5:
    #Again adding some special name strings here
        var1: value1
        name: some_name
        pass: some_pass

The above is not a YAML parser and is only intended to work with files structured in the same way as the examples shown and where all indentation is either tabs or blanks but not both.

See https://stackoverflow.com/questions/66173552/update-yaml-value-on-cli#comment116993514_66173552 for another application of the above.

Answer 3

You haven't given us any idea of what system you are using or what's already installed. So this is just an educated guess. A lot of linux distributions ship with "python" and the rise of yaml means that the library "pyaml" it is also often installed. The very fact you are interacting with yaml files yourself makes this more likely.

You can test this really easily. Make a file foo.py containing just:

import yaml
print("I have yaml")

Then on the command line run:

python3 foo.py

If it just prints I have yaml then you have both python and pyaml installed and can use it in a trivial script (update_script.py):

import yaml
import sys

# Load the file into a data structure
with open(sys.argv[1], "r") as file:
    content = yaml.safe_load(file)

# Update the key you want to change
content["search4"]["subsearch1"]["name"] = "new_name"

# Write the data structure back to your file in YAML
with open(sys.argv[1], "w") as file:
    yaml.safe_dump(content, file)

Then just run:

python3 update_script.py config_file.yaml