Function parse_choice 

pub fn parse_choice<'a>(
    response: &str,
    valid_choices: &[&'a str],
) -> Result<&'a str, ParseError>

Extract a single choice from a set of valid options.

Handles common LLM response patterns:

• Direct match: "positive"
• Bold: "**positive**"
• Quoted: "'positive'" or "\"positive\""
• In prose: "I would classify this as positive because..."
• Parenthesized: "(positive)"

Matching is case-insensitive. If multiple valid choices appear, returns the first one found in the text.

Examples

use llm_output_parser::parse_choice;

let result = parse_choice("I'd classify this as positive", &["positive", "negative"]).unwrap();
assert_eq!(result, "positive");