esmorph

Fast Spanish morphological analyzer written in Rust, the core of the esMorph library for Python.

• The analyses of a word form by a dictionary built from Wiktionary and treebanks, the most probable first: the lemma, the part of speech and the Universal Dependencies features.
• Case does not matter. A word absent as written is looked for without its accents and with a lost diaeresis.
• A verb with enclitic pronouns is analyzed as its verb, with the words of the token: dámelo is da + me + lo.
• The forms of a lexeme in the order of its paradigm and the form with given features: conjugation, gender and number.
• Tokens out of the dictionary by rules: punctuation, symbols, numbers, Roman numerals, names and acronyms.
• Unknown lowercase words guessed by their last part after a hyphen, a known prefix or their ending, with their forms.
• The dictionary is mapped into memory or read from the data of the program without copies.

Installation

[dependencies]
esmorph = "0.1"

The dictionary is in the crate esmorph-dict-es, used through the feature bundled (on by default). Without it the dictionary is read from a file:

let dictionary = esmorph::Dictionary::open("es.dict")?;

Example

use esmorph::Dictionary;

fn main() -> Result<(), esmorph::Error> {
    let dictionary = Dictionary::bundled()?;
    for analysis in dictionary.analyze("camiones") {
        let tag = dictionary.ud_tag(analysis.ud).unwrap();
        println!("{} {} {}", analysis.lemma, tag.pos(), tag.feats_string());
    }
    let hablar = &dictionary.analyze("hablar")[0];
    let past = [("Tense", "Past"), ("Person", "3"), ("Number", "Plur")];
    assert_eq!(dictionary.inflect(hablar, &past).unwrap().word, "hablaron");
    Ok(())
}

License

The code is distributed under the MIT license. The dictionary is under CC BY-SA 4.0.