Stuck at passing JSON as argument in bash

You could use jq to transform your file directly. No need to use mapfile or generate a shell array:

jq -Rn '[ inputs, "Infra" ]' </path/to/Dependencies

Example:

printf '%s\n' apple banana cherry |
    jq -Rn '[ inputs, "Infra" ]'

output:

[
  "apple",
  "banana",
  "cherry",
  "Infra" 
]

The -R tells jq to read raw (unquoted) data, and -n tells jq not to read any input "before" the pipeline - leaving it for the inputs operator to process itself at the appropriate point.

You can compress the output into a single line by adding the -c flag:

["apple","banana","cherry","Infra"]

jq is usually my go-to tool for any JSON processing, and Chris Davies is giving a good example of jq applied to this particular issue in their answer.

Another tool, called jo, is made specifically for constructing JSON on the command line. With jo, the JSON array can be had as the string $deps from the following short shell code:

deps=$( jo -a <Dependencies )

Adding an Infra as a final element in the array can be done like so:

deps=$( jo -a <Dependencies | jo -f - "Infra" )

With the data given in the question, this would produce the following JSON array as a string in $deps:

["Accounts","Aggregates","Blog","Configuration","Contacts","Contents","DataTypes","Forms","Geo","Globalization","Media","Modules","Navigation","Podcast","Seo","Social","Taxonomy","Tracking","Vlog","Infra"]

Add -p to the last invocation of jo to generate a pretty-printed array.

Please also be aware of some of the limitations of this tool. The following is a comment from Stéphane Chazelas:

Beware it treats number-looking strings as numbers, empty strings as null, [...] as arrays, @file would load the contents of a file, etc (see for instance printf '%s\n' 1e002 @/etc/shadow @/dev/watchdog 0xDEADBEEF '' '{}' | jo -a). It can't be used that way for arbitrary strings.

If you're in a nushell:

open dependencies.txt | lines | to json

If you want to remain in bash/zsh, you can also use nushell instead of jq:

nu -c 'open dependencies.txt | lines | to json'

With perl:

$ perl -CiI -MJSON::PP -le '
    chomp(@lines = <>);
    print encode_json [@lines, "Infra"]' Dependencies
["Accounts","Aggregates","Blog","Configuration","Contacts","Contents","DataTypes","Forms","Geo","Globalization","Media","Modules","Navigation","Podcast","Seo","Social","Taxonomy","Tracking","Vlog","Infra"]

With the input decoded in UTF-8, and encoded by encode_json (not the STDOUT output layer which should remain in byte/latin1 mode) to produce UTF-8 encoded JSON.

Or if using the OO interface, you can skip those decoding and encoding:

perl -MJSON::PP -le '
  chomp(@lines = <>);
  print JSON::PP->new->encode([@lines, "Infra"])' Dependencies

Which will pass the bytes > 127 as-is. It won't choke on non-UTF-8 input, but would produce non-UTF-8 JSON in those cases.

To decode the input as per the locale's character encoding and output UTF-8 encoded JSON:

perl -MJSON::PP -Mopen=IN,locale -le '
  chomp(@lines = <>);
  print encode_json [@lines, "Infra"]' Dependencies

You can also produce JSON where non-ASCII characters are represented as \uXXXX (or \uDXXXX\uDXXXX for those above U+FFFF as usual):

perl -MJSON::PP -Mopen=IN,locale -le '
  chomp(@lines = <>);
  print JSON::PP->new->ascii->encode([@lines, "Infra"])' Dependencies

If the input contains only ASCII characters like in your sample, that won't make any difference, but that's something you may have to take into account if there may be non-English dependency names.