phomnemic/lib/src/core/major/encoder.rs

use crate::core::entities::{DictEntries, DictEntry};
use crate::core::traits::SystemEncoder;

/// (index, encoded value)
type DictMatches = Vec<(usize, String)>;

pub struct Encoder {
    dict: DictEntries,
}

impl Encoder {
    pub fn new(dict: DictEntries) -> Self {
        Encoder {
            dict: Encoder::to_lower_dict(dict),
        }
    }

    fn to_lower_dict(dict: DictEntries) -> DictEntries {
        dict.into_iter()
            .map(|entry| entry.into_lowercase())
            .collect()
    }

    fn match_entry(&self, entry: &DictEntry, word: &str) -> DictMatches {
        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: &DictEntry, 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 SystemEncoder for Encoder {
    fn encode(&self, word: &str) -> String {
        let mut matches: DictMatches = self
            .dict
            .iter()
            .flat_map(|entry| self.match_entry(&entry, &word.to_lowercase()))
            .collect();

        matches.sort_by_key(|&(pos, _)| pos);
        dbg!(&matches);
        matches.into_iter().map(|(_, value)| value).collect()
    }
}