taipa/lib/dparser/parser.js

Taiwanese morphological parsing library

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.DependencyParser = void 0;
const configuration_1 = require("./configuration");
const guide_1 = require("./guide");
const document_1 = require("../document");
const symbols_1 = require("./symbols");
const relation_1 = require("./relation");
class DependencyParser {
    c = this.getInitialConfiguration();
    s1 = new document_1.Node('');
    s2 = new document_1.Node('');
    b1 = new document_1.Node('');
    s1B1RightRelations = new Map()
        .set(symbols_1.Tagset.ppv + symbols_1.Tagset.psub, symbols_1.DepRelations.compoundPrt)
        .set(symbols_1.Tagset.ppv + symbols_1.Tagset.npr, symbols_1.DepRelations.compoundPrt);
    s1B1LeftRelations = new Map();
    s2S1RightRelations = new Map()
        .set(symbols_1.Tagset.vb + symbols_1.Tagset.ppv, symbols_1.DepRelations.compoundPrt)
        .set(symbols_1.Tagset.vb + symbols_1.Tagset.psub, symbols_1.DepRelations.prt)
        .set(symbols_1.Tagset.vb + symbols_1.Tagset.vb, symbols_1.DepRelations.compound)
        .set(symbols_1.Tagset.vb + symbols_1.Tagset.npr, symbols_1.DepRelations.obj);
    s2S1LeftRelations = new Map()
        .set(symbols_1.Tagset.aux + symbols_1.Tagset.vb, symbols_1.DepRelations.aux)
        .set(symbols_1.Tagset.padv + symbols_1.Tagset.vb, symbols_1.DepRelations.advmod)
        .set(symbols_1.Tagset.appr + symbols_1.Tagset.npr, symbols_1.DepRelations.case);
    getInitialConfiguration() {
        return new configuration_1.Configuration();
    }
    apply(t, c) {
        return t.do(c);
    }
    isQueueEmpty() {
        if (this.c.queue.length === 0)
            return true;
        return false;
    }
    isTwoNodesInStack() {
        if (this.c.stack.length === 2)
            return true;
        return false;
    }
    rightRelation(label) {
        this.s1.dep = label;
        this.s1.head = this.s2.token;
        return new relation_1.Relation(label, this.s2, this.s1);
    }
    leftRelation(label) {
        this.s2.dep = label;
        this.s2.head = this.s1.token;
        return new relation_1.Relation(label, this.s1, this.s2);
    }
    setS1S2B1() {
        this.s1 = new document_1.Node('');
        if (this.c.stack.length > 0)
            this.s1 = this.c.stack[this.c.stack.length - 1];
        this.s2 = new document_1.Node('');
        if (this.c.stack.length > 1)
            this.s2 = this.c.stack[this.c.stack.length - 2];
        this.b1 = new document_1.Node('');
        if (this.c.queue.length > 0)
            this.b1 = this.c.queue[0];
    }
    setS1B1Relation(t) {
        if (t instanceof configuration_1.RightArc) {
            if (this.s1B1RightRelations.has(this.s1.tag + this.b1.tag)) {
                const rel = this.s1B1RightRelations.get(this.s1.tag + this.b1.tag);
                if (rel) {
                    this.c.relations.push(this.rightRelation(rel));
                }
            }
        }
        else if (t instanceof configuration_1.LeftArc) {
            if (this.s1B1LeftRelations.has(this.s1.tag + this.b1.tag)) {
                const rel = this.s1B1LeftRelations.get(this.s1.tag + this.b1.tag);
                if (rel) {
                    this.c.relations.push(this.leftRelation(rel));
                }
            }
        }
    }
    s2S1LeftArgsToPronoun = new Map().set(symbols_1.Tagset.npr + symbols_1.Tagset.vb, [
        symbols_1.DepRelations.nsubj,
        symbols_1.DepRelations.dislocated,
    ]);
    setS2S1Relation(t) {
        if (t instanceof configuration_1.RightArc) {
            if (this.s2S1RightRelations.has(this.s2.tag + this.s1.tag)) {
                const rel = this.s2S1RightRelations.get(this.s2.tag + this.s1.tag);
                if (rel) {
                    this.c.relations.push(this.rightRelation(rel));
                }
            }
            else if (this.isTwoNodesInStack()) {
                this.c.relations.push(this.rightRelation(symbols_1.DepRelations.root));
            }
        }
        else if (t instanceof configuration_1.LeftArc) {
            if (this.s2S1LeftRelations.has(this.s2.tag + this.s1.tag)) {
                const rel = this.s2S1LeftRelations.get(this.s2.tag + this.s1.tag);
                if (rel) {
                    this.c.relations.push(this.leftRelation(rel));
                }
            }
            else if (this.s2S1LeftArgsToPronoun.has(this.s2.tag + this.s1.tag)) {
                // TODO: to be improved
                const labelsPronoun = this.s2S1LeftArgsToPronoun.get(this.s2.tag + this.s1.tag);
                const rels = this.c.relations.filter(it => it.dependent.token === 'gua');
                if (labelsPronoun)
                    if (this.s2.token === 'gua') {
                        this.c.relations.push(this.leftRelation(labelsPronoun[0]));
                    }
                    else if (this.s2.token === 'che' &&
                        rels.length > 0 &&
                        rels[0].dependent.token === 'gua' &&
                        rels[0].dependent.tag === symbols_1.Tagset.npr) {
                        this.c.relations.push(this.leftRelation(labelsPronoun[1]));
                    }
            }
        }
    }
    parse(nodes) {
        for (let t of nodes) {
            this.c.queue.push(t);
        }
        let guide = new guide_1.Guide();
        let rt = new document_1.Node('ROOT');
        this.c.stack.push(rt);
        if (this.c.stack.length == 1 && this.c.queue.length > 0) {
            // initial configuration
            // shift the first lexeme from queue to stack
            guide.transitions.push(new configuration_1.Shift());
        }
        while (!this.c.isTerminalConfiguration()) {
            let t = guide.getNextTransition(this.c);
            if (t == null || t == undefined)
                break;
            this.setS1S2B1();
            if (this.s1.tag != '' && this.b1.tag != '') {
                this.setS1B1Relation(t);
            }
            else if (this.isQueueEmpty()) {
                this.setS2S1Relation(t);
            }
            this.c = this.apply(t, this.c);
        }
        return this.c.relations;
    }
}
exports.DependencyParser = DependencyParser;
//# sourceMappingURL=parser.js.map