phomnemic/lib/src/sys_major/decoder.rs

use crate::common::{entities::DecodedValue, errors::CodecError, traits::SystemDecoder};

#[derive(Debug, Default, Clone)]
pub struct Rule {
    pub phoneme_in: String,
    pub phoneme_out: String,

    pub not_before: Vec<String>,
    pub not_after: Vec<String>,

    pub only_before: Vec<String>,
    pub only_after: Vec<String>,
}

impl Rule {
    pub fn into_lowercase(self) -> Self {
        Rule {
            phoneme_in: self.phoneme_in.to_lowercase(),
            phoneme_out: self.phoneme_out.to_lowercase(),
            not_before: Self::lower_vec(self.not_before),
            not_after: Self::lower_vec(self.not_after),
            only_before: Self::lower_vec(self.only_before),
            only_after: Self::lower_vec(self.only_after),
        }
    }

    fn lower_vec(vec: Vec<String>) -> Vec<String> {
        vec.into_iter().map(|s| s.to_lowercase()).collect()
    }
}

pub type Rules = Vec<Rule>;
// pub struct rules {
//     name: String,
//     entries: Rules,
// }

/// (index, decoded value)
type RuleMatches = Vec<(usize, String)>;

pub struct Decoder {
    rules: Rules,
}

impl Decoder {
    pub fn new(rules: Rules) -> Self {
        Decoder {
            rules: Decoder::to_lower_rules(rules),
        }
    }

    fn to_lower_rules(rules: Rules) -> Rules {
        rules
            .into_iter()
            .map(|entry| entry.into_lowercase())
            .collect()
    }

    fn match_entry(&self, entry: &Rule, word: &str) -> RuleMatches {
        word.match_indices(&entry.phoneme_in)
            .filter(|(index, _)| self.is_context_matched(&entry, &word, *index))
            .map(|(index, _)| (index, entry.phoneme_out.clone()))
            .collect()
    }

    fn is_context_matched(&self, entry: &Rule, word: &str, index: usize) -> bool {
        let before_context = &word[..index];
        let after_context = &word[index + entry.phoneme_in.len()..];
        // dbg!(&before_context);
        // dbg!(&after_context);

        if entry
            .not_after
            .iter()
            .any(|prefix| before_context.ends_with(prefix))
        {
            return false;
        }

        if entry
            .not_before
            .iter()
            .any(|suffix| after_context.starts_with(suffix))
        {
            return false;
        }

        if !entry.only_after.is_empty()
            && entry
                .only_after
                .iter()
                .all(|prefix| !before_context.ends_with(prefix))
        {
            return false;
        }

        if !entry.only_before.is_empty()
            && entry
                .only_before
                .iter()
                .all(|suffix| !after_context.starts_with(suffix))
        {
            return false;
        }

        true
    }
}

impl SystemDecoder for Decoder {
    fn decode(&self, word: &str) -> Result<DecodedValue, CodecError> {
        let mut matches: RuleMatches = self
            .rules
            .iter()
            .flat_map(|entry| self.match_entry(&entry, &word.to_lowercase()))
            .collect();

        matches.sort_by_key(|&(pos, _)| pos);
        let num_str: String = matches.into_iter().map(|(_, value)| value).collect();
        DecodedValue::new(num_str)
    }
}