Duct Tape a Regex Decider

Question

Your task is to create a program that determines whether a given string is a valid regular expression or not using code snippets sourced from sites on the StackExchange network.

For the purposes of this challenge, the regular expression dialect will be a stripped down and mostly minimal set of meta-characters: ()*?|\. As such, you will not be able to use built-in regex parsers.

• \ is used to escape meta-characters. It must be followed by a meta-character.
• Unescaped parentheses must be balanced
• * and ? must be preceded by either a non-meta-character, a parenthesized group, or an escaped meta-character.
• All other printable ASCII characters plus newline, tab, and space must be supported as non-meta characters. What happens with a string containing other characters is undefined.
• The actual meaning of the regex is not important for this challenge.

Examples

Truthy:
  abc
  a?
  (a|)*
  ()
  a|b*
  \*
  \\
  \\*
  a*b?(cd|e)
  +
  [
  }
  (123\))*
  \|
  (a(b(c|d)*e)*f)*
  (|\)*)
  (abc)+*
  (abc)+
  +abc

^ last test case is an actual newline

Falsy:
  ?abc
  *
  **
  \
  (
  a*?
  a?*
  ?
  a)
  (\)
  (|\)*
  \()
  |*
  (?:abc)
  \\**
  \n

Scoring

Your overall score is the number of snippets taken from questions and answers around StackExchange.

• Repeated snippets count for as many times as they are used.
• Whitespace can be added and removed freely (because of Python, Haskell, and other whitespace-sensitive languages) and does not count toward your snippet count.
  • The exception would be if your code is actually written in Whitespace.
• Snippets are allowed from any StackExchange site as long as they come from questions, answers, and comments that are older (Including by edit time - use older revisions if necessary) than this challenge. (Sep 24, 2019 @ 3:30 PM UTC)
• Snippets can come from anywhere in a question, answer, or comment body, whether it's in a preformatted code block or not.
• Splicing a snippet into the middle of another causes the outer snippet to count as two snippets

Lowest score wins!

Answer 1

Perl 6, 20 snippets

{$_ eq m/[[<-[()*?|\\]>|\\<[()*?|\\]>|'(' <~~>* ')']<[*?]>?|\|]+/}

Try it online!

The snippets are taken from:

{$_ eq, m/[, <-[, ()*?, |\\, ]>, |\\, <[, ()*?, |\\, ]>, |, '(' <~~>* ')', <[, *?, ]>, ?|, \|, ]+/, }.

This is mostly the greedy approach (made obvious by all the one or two character snippets). I used SymbolHound to search for the individual characters, and the only real optimisation was the '(' <~~>* ')' snippet, which is taken from my own answer on recursive Perl 6 regexes.

Explanation:

This basically checks if the input is equal to a greedy match of a valid regex. The reason we can't just use the regex itself and add ^$ to mark the ends is because we are using a recursive regex, which wouldn't work if there were ^$ markers. The regex itself is:

m/[                             ]+/   # Match one or more times
   [              ]  # Any of 
    <-[()*?|\\]> |     # Not a metacharacter
    \\<[()*?|\\]>      # A metacharacter preceded by a \
    '(' <~~>* ')'      # Brackets surrounding a valid regex
                   <[*?]>?  # Optionally followed by a ? or *
                           | \|    # Or just the | metacharacter