Replacing matching text after two matching lines

Question

I have a YAML file which includes a stanza looks this:

admin::common::passwords: alice: password: '$6$oTQhLvN/4VFJPscD$8LYwUMSFi' bob: password: '$6$JKOtLF0wHeZfIskt$W/M5.ugDS' 

If the shell variable $ACCOUNT contains the account name (alice, bob, etc), and $YAML is the filename of the YAML file, I can add a new entry with sed, like this. It inserts the new account directly under the "admin::common::passwords:" line, so I know it's in the correct stanza:

 sed -i /'^admin::common::passwords:'/a"\ ${ACCOUNT}:\n password: '${ENCRYPTED_PASS}'" $YAML 

I can find an existing account like this with awk. Again, this finds the stanza, then finds the account within that stanza (it's probably possible to do this all in awk, but I'm not good enough at awk, and this will do the job):

 awk "BEGIN{RS=ORS="\n\n";FS=OFS="\n"}/admin::common::passwords:/" $YAML | grep -A1 "^[ ]*${ACCOUNT}:" 

But to delete a line, the best I can come up with is this:

sed -i "/^ ${ACCOUNT}:/,+1d" $YAML 

Which would delete any line "^ alice:" anywhere in the file, whether it's in the admin::common::passwords stanza or elsewhere.

I can't make any assumptions about the content of the file, except that it does contain exactly one admin::common:passwords stanza (if it did not, then it would have been created by another part of my script; that bit is easy).

So, my question is this: How can I improve this sed to say 'Search for "^ ${ACCOUNT}:" within the admin::common::passwords stanza, and delete that matching line as well as the line below it'?

(it doesn't have to be sed, of course; it may be that awk or even Perl are better suited to the task)

Answer 1

sed " /^admin::common::passwords:$/,/^[^ ].*:$/ { /^ $ACCOUNT:$/ { N;d } }" < "$YAML" 

That is, enclose your deleting stanza in a section that's matched between admin::common... and the next whatever-section:.

Beware that the . character, common in user names is also a regular expression operator, so john.doe would match john.doe but also johnWdoe for instance.

Note that the above won't work if you have two consecutive admin::common::passwords sections and the account to delete is in the second one.

If, as Otheus indicates, sections run until the next line that has the same or less amount of leading space characters and your admin::common::passwords: section may have some leading spaces, then it's probably time to switch to another language like awk:

awk -v account="$ACCOUNT" ' match($0, /[^ ]/) && RSTART <= n {n = 0} n && NF == 1 && $1 == account ":" {getline; next} !n && /^ *admin::common::passwords:$/ { match($0, /[^ ]/) n = RSTART } {print}' < "$YAML" 

Answer 2

It would be far safer to use pre-built libraries to parse the whole file (like this, for example https://stackoverflow.com/questions/1773805/how-can-i-parse-a-yaml-file) make your modifications and then re-save the modified file. Doing somewhat blind text modifications on a structured text file like this is just asking for something to go wrong.