If you just run the command you ran initially it will print out the packages it wants to pull in. You can then cancel it and install whichever subset you choose manually.

We’re talking about different halves. The regex \w+\s+ matches "The " (“The” followed by a space), not “The MCU”.

Why do we replace the commas again with new lines?

Consider this two-line output:

$ echo 'a\nb'
a
b
$

We convert the newlines to commas. Now there is a comma at the end of the last line as well, and because of no newline, the next prompt is at the end of the output:

$ echo 'a\nb' | tr '\n' ,
a,b,$

Substituting only the last comma ($ means end of line) allows us to get the output we expected:

$ echo 'a\nb' | tr '\n' , | sed 's/,$/\n/'
a,b
$

Or is there a way to combine them

These two commands have equivalent output:

tr '\n' ',' | tr ';' ',' 
tr '\n;' ',,'

What tr does is take a list of characters in parameter 1 and converts them to the equivalent position character in parameter 2. There’s a little more to it (it supports ranges, for example), but this will do the job. To learn more you can run man tr to get the documentation for it.

I tried What \w+\s+ Says About

\w+\s+ matches “at least one word character and then at least one whitespace character”, and that’s not what you want. “The MCU” is one or more word characters, then a space, and then one or more word characters again, and that second part you’re not matching at all. In this case, you’re probably better off making a negative matching group where you make sure you don’t match across separators. What [^,;]+ Says About would match anything that’s not a comma or semicolon, for instance.

The other problem with regex is that every implementation does things differently. For example, sed would interpret that plus as a literal +, so for sed syntax you’d need to use \+ instead. It also does not support \w and \s, and whether to use ( or \( for a literal parenthesis also varies between implementations. I often switch to Perl if I need to do some more complex regex shenanigans.

If you can’t install a dedicated tool like yq but don’t mind creating a standalone script, python would be able to do this out of the box on pretty much any computer, calculator or toaster you can get your hands on in 2026:

#! /usr/bin/env python3

import yaml
import sys

def parse_yaml(filename):
    with open(filename) as fd:
        return yaml.safe_load(fd)

def get_leaf_nodes(data_iterable):
    output = []
    for v in data_iterable:
        if isinstance(v, dict):
            output += get_leaf_nodes(v.values())
        elif isinstance(v, list):
            output += get_leaf_nodes(v)
        else:
            output.append(v)
    return output

print(",".join(get_leaf_nodes(parse_yaml(sys.argv[1]))))

$ /tmp/foo.py /tmp/foo.txt
Harry potter,Perfect Blue,Jurassic world,Jurassic Park,Jedi,Star wars,The clone wars,MCU,Gumball,Flapjack,Steven Universe,Stars vs. the forces of Evil,Wordgril,Flapjack

This takes the first argument on the command line, parses it as yaml, finds all leaf nodes recursively, and prints a comma-separated list of the results.

If you’re feeling a little old school (and some might say masochistic), you could so a similar crude parser with a perl oneliner. This would be more efficient compute wise, but it’s a bit of an acquired taste readability wise:

$ perl -ne 'chomp; push @a, $1 if /^\s*-\s*(.*[^:\s])\s*$/; END{print join(",", @a), "\n"}' /tmp/foo.txt
Harry potter,Perfect Blue,Jurassic world,Jurassic Park,Jedi,Star wars,The clone wars,MCU,Gumball,Flapjack,Steven Universe,Stars vs. the forces of Evil,Wordgril,Flapjack

Here perl -n makes perl look at each line individually, chomp strips off the trailing newline, we match for /^\s*-\s*(.*[^:\s])\s*$/ (a string starting with a dash and ending with something not a colon) and append the content of the matching parenthesis to an implicitly declared array @a. Then we add an END{} block which will be executed after all lines are parsed, where we print the array joined on ,.

If you wanted a somewhat cruder approach using basically ubiquitous tools, you could do something like this:

$ grep '^ *-' /tmp/foo.txt | grep -v ': *$' | sed 's/ *- //' | tr '\n' ',' | sed s'/,$/\n/'
Harry potter,Perfect Blue,Jurassic world,Jurassic Park,Jedi,Star wars,The clone wars,MCU,Gumball ,Flapjack,Steven Universe,Stars vs. the forces of Evil,Wordgril,Flapjack 

Here I’m first using grep '^ *-' to get all lines starting with any amount of whitespace and a leading dash, then piping that to grep -v ': *$' to remove anything with a colon at the end (including those with whitespace after the colon), then using tr '\n' ',' to replace all newlines with commas, and then sed s'/,$/\n/' to replace the trailing comma with a newline again (although sed is finicky across platforms wrt newlines, so you may want to just replace it with an empty string instead).

The above is hardly an efficient approach, but it does the job.

If you can stick to valid YAML like your example is, you can use a reasonably short yq command to get a comma-separated string of all scalar values:

$ yq -r '[.. | scalars] | join(",")' /tmp/foo.txt                
Harry potter,Perfect Blue,Jurassic world,Jurassic Park,Jedi,Star wars,The clone wars,MCU,Gumball,Flapjack,Steven Universe,Stars vs. the forces of Evil,Wordgril,Flapjack

.. goes down the tree recursively, scalars filters out only scalar values, [] around those two makes them an array, and piping it all to join(",") makes it into a comma-separated string.

Your description is too vague to really get a good answer. In general, if you’re doing complex string manipulation, you’ll use a full-fledged programming language with regex support, like Python, Perl or Awk, possibly piped into each other and/or other tools like Sed or Cut. I can’t be more specific than that without a more specific description where you describe the actual data and criteria.

Are you starting with the first or second example? Why do the prefix numbers change between examples? How do you tell text and title/subtitle apart?