harper_core/spell/

merged_dictionary.rs

use std::borrow::Cow;
use std::hash::{BuildHasher, Hasher};
use std::sync::Arc;

use foldhash::quality::FixedState;
use itertools::Itertools;

use super::{FstDictionary, FuzzyMatchResult, WordId, dictionary::Dictionary};
use crate::{CharString, DictWordMetadata};

/// A simple wrapper over [`Dictionary`] that allows
/// one to merge multiple dictionaries without copying.
///
/// In cases where more than one dictionary contains a word, data in the first
/// dictionary inserted will be returned.
#[derive(Clone)]
pub struct MergedDictionary {
    children: Vec<Arc<dyn Dictionary>>,
    hasher_builder: FixedState,
    child_hashes: Vec<u64>,
}

impl MergedDictionary {
    pub fn new() -> Self {
        Self {
            children: Vec::new(),
            hasher_builder: FixedState::default(),
            child_hashes: Vec::new(),
        }
    }

    pub fn add_dictionary(&mut self, dictionary: Arc<dyn Dictionary>) {
        self.child_hashes.push(self.hash_dictionary(&dictionary));
        self.children.push(dictionary);
    }

    fn hash_dictionary(&self, dictionary: &Arc<dyn Dictionary>) -> u64 {
        // Hashing the curated dictionary isn't super helpful and takes a long time.
        if Arc::ptr_eq(
            dictionary,
            &(FstDictionary::curated() as Arc<dyn Dictionary>),
        ) {
            return 1;
        }

        let mut hasher = self.hasher_builder.build_hasher();

        dictionary
            .words_iter()
            .for_each(|w| w.iter().for_each(|c| hasher.write_u32(*c as u32)));

        hasher.finish()
    }
}

impl PartialEq for MergedDictionary {
    fn eq(&self, other: &Self) -> bool {
        self.child_hashes == other.child_hashes
    }
}

impl Default for MergedDictionary {
    fn default() -> Self {
        Self::new()
    }
}

impl Dictionary for MergedDictionary {
    fn get_correct_capitalization_of(&self, word: &[char]) -> Option<&'_ [char]> {
        for child in &self.children {
            if let Some(word) = child.get_correct_capitalization_of(word) {
                return Some(word);
            }
        }
        None
    }

    fn contains_word(&self, word: &[char]) -> bool {
        for child in &self.children {
            if child.contains_word(word) {
                return true;
            }
        }
        false
    }

    fn contains_exact_word(&self, word: &[char]) -> bool {
        for child in &self.children {
            if child.contains_exact_word(word) {
                return true;
            }
        }
        false
    }

    fn get_word_metadata(&self, word: &[char]) -> Option<Cow<'_, DictWordMetadata>> {
        let mut meta_iter = self
            .children
            .iter()
            .filter_map(|d| d.get_word_metadata(word));

        let first = meta_iter.next()?;

        // Check if multiple entries were found for the word.
        if let Some(second) = meta_iter.next() {
            // If so, merge them.
            let mut first = first.into_owned();
            first.merge(&second);
            meta_iter.for_each(|additional_md| {
                first.merge(&additional_md);
            });

            Some(Cow::Owned(first))
        } else {
            // If not, return the sole found entry.
            Some(first)
        }
    }

    fn words_iter(&self) -> Box<dyn Iterator<Item = &'_ [char]> + Send + '_> {
        Box::new(self.children.iter().flat_map(|c| c.words_iter()))
    }

    fn contains_word_str(&self, word: &str) -> bool {
        let chars: CharString = word.chars().collect();
        self.contains_word(&chars)
    }

    fn contains_exact_word_str(&self, word: &str) -> bool {
        let chars: CharString = word.chars().collect();
        self.contains_exact_word(&chars)
    }

    fn get_word_metadata_str(&self, word: &str) -> Option<Cow<'_, DictWordMetadata>> {
        let chars: CharString = word.chars().collect();
        self.get_word_metadata(&chars)
    }

    fn fuzzy_match(
        &'_ self,
        word: &[char],
        max_distance: u8,
        max_results: usize,
    ) -> Vec<FuzzyMatchResult<'_>> {
        self.children
            .iter()
            .flat_map(|d| d.fuzzy_match(word, max_distance, max_results))
            .sorted_by_key(|r| r.word)
            .dedup_by(|a, b| a.word == b.word)
            .sorted_by_key(|r| r.edit_distance)
            .take(max_results)
            .collect()
    }

    fn fuzzy_match_str(
        &'_ self,
        word: &str,
        max_distance: u8,
        max_results: usize,
    ) -> Vec<FuzzyMatchResult<'_>> {
        self.children
            .iter()
            .flat_map(|d| d.fuzzy_match_str(word, max_distance, max_results))
            .sorted_by_key(|r| r.word)
            .dedup_by(|a, b| a.word == b.word)
            .sorted_by_key(|r| r.edit_distance)
            .take(max_results)
            .collect()
    }

    fn word_count(&self) -> usize {
        self.children.iter().map(|d| d.word_count()).sum()
    }

    fn get_word_from_id(&self, id: &WordId) -> Option<&[char]> {
        self.children
            .iter()
            .find_map(|dict| dict.get_word_from_id(id))
    }

    fn find_words_with_prefix(&self, prefix: &[char]) -> Vec<Cow<'_, [char]>> {
        self.children
            .iter()
            .flat_map(|dict| dict.find_words_with_prefix(prefix))
            .sorted()
            .dedup()
            .collect()
    }

    fn find_words_with_common_prefix(&self, word: &[char]) -> Vec<Cow<'_, [char]>> {
        self.children
            .iter()
            .flat_map(|dict| dict.find_words_with_common_prefix(word))
            .sorted()
            .dedup()
            .collect()
    }
}

#[cfg(test)]
mod tests {
    use std::sync::Arc;

    use crate::DictWordMetadata;
    use crate::spell::{Dictionary, MergedDictionary, MutableDictionary};

    #[test]
    fn merged_contains_exact_word_str_is_case_sensitive() {
        let mut user_dict = MutableDictionary::new();
        user_dict.append_word_str("Foo", DictWordMetadata::default());

        let mut merged = MergedDictionary::new();
        merged.add_dictionary(Arc::new(user_dict));

        assert!(merged.contains_word_str("Foo"));
        assert!(merged.contains_word_str("foo"));

        assert!(merged.contains_exact_word(&['F', 'o', 'o']));
        assert!(!merged.contains_exact_word(&['f', 'o', 'o']));

        assert!(merged.contains_exact_word_str("Foo"));
        assert!(!merged.contains_exact_word_str("foo"));
    }
}