query-highlighting-system

query-highlighting-system.rs

// TODO: latency optimize this so it can be uncommented
        if phrases.is_empty() {
            let potential_query_split_whitespace = potential_query.split_whitespace().collect_vec();
            if potential_query_split_whitespace.len() > 5 {
                continue;
            }
            let query_without_stop_words = potential_query
                .split_whitespace()
                .filter(|word| !stop_words.contains(&word.to_lowercase()))
                .collect::<Vec<&str>>();
            if query_without_stop_words.len() < 2
                || (potential_query_split_whitespace.len() - query_without_stop_words.len() < 1)
            {
                continue;
            }

            // \b(?:word1\W+(?:\w+\W+){0,3}?word2|word2\W+(?:\w+\W+){0,3}?word1)\b
            let query_regex = format!(
                "\\b(?:{}|{})\\b",
                query_without_stop_words.join("\\W+(?:\\w+\\W+){0,6}?"),
                query_without_stop_words
                    .iter()
                    .rev()
                    .join("\\W+(?i:\\w+\\W+){0,6}?")
            );
            if let Ok(re) = regex::Regex::new(&query_regex) {
                let matched_idxs: Vec<(usize, usize)> = re
                    .find_iter(&content)
                    .map(|x| (x.start(), x.as_str().len()))
                    .collect();

                phrases = matched_idxs
                    .iter()
                    .map(|(index, length)| content[*index..*index + length].to_string())
                    .collect_vec();
            }
        }