How to grep cloud functions in a multiline format separated by comma?-CodePudding

I have been trying to fetch the cloud functions using grep/awk/sed however the current command is not fetching the cloud functions listed in a multiline format.

This is the current command that is being passed as a variable during runtime:

deploy.sh

functions_list=$(grep 'export' src/index.ts | awk -F '[}{]' '{print $2}' | sed 's/^ *//; s/ *\$//; /^\$/d' | grep -v somefunction || echo '')

index.ts

export { functionA } from './functions';
export { functionB } from './functions';
export {
    functionC,
    functionD,
    functionE
} from './functions';
export { functionF } from './functions';
export { functionG } from './functions';

Then the output will be like this:

functionA
functionB

functionF
functionG

Anyone who has any idea on how to have the output to be like this instead?

functionA
functionB
functionC
functionD
functionE
functionF
functionG

CodePudding user response：

See final command:

functions_list=$(awk 1 ORS=' ' src/index.ts | sed -e 's/    / /g ; s/;/\n/g' | awk -F '[}{]' '{print $2}' | sed 's/^ *//; s/ *\$//; /^\$/d; s/,  /\n/g' | grep -v somefunction || echo '')

awk reads/scans input line by line; hence skips the functionC, functionD and functionE as they do not fall under the provided pattern:

'[}{]' '{print $2}'

I've added awk/sed commands to initially replace new line with whitespaces.

awk 1 ORS=' ' src/index.ts | sed -e 's/    / /g ; s/;/\n/g'

Then, finally, to replace commas with newline.

s/,  /\n/g

CodePudding user response：

The parsing here is very brittle. Without complicating things too much, try switching to Perl, which lets you examine the whole input file in one go.

perl -n0777 -e 'while (m/export \{([^{}] )\}/mg) {
  $s = $1; print "$1\n" while $s =~ m/(\w )/mg }' src/index.ts

The first regex loops over the export { ... } expressions and pulls out the text between the braces. The second picks out the symbols, skipping any whitespace or punctuation, and prints them.