conllu-core/bin/builder.d.ts

A core type to handle CoNLL-U format

import { AdvanceDep, Comment, DepsRelation, EmptyToken, Feature, HeadId, Meta, NominalToken, Relation, Sentence, Token, TokenIdMap, TokenType, UPOS, XPOS } from ".";
/**
 * The policy to adjust a `head` field that point to a token being merge.
 * Every other tokens will have their head value adjust accordingly but the `head` that point
 * to token being merge may need different treatment depending on user requirement.
 */
export declare enum HeadPolicy {
    /** Update all the `head` pointed to the token within merge range to a merged `ID` */
    Adjust = 0,
    /** Remove any `head` linked to the token within merge range */
    Remove = 1
}
export declare class SentenceBuilder {
    meta: (Meta | Comment)[];
    tokens: Token[];
    id_map: TokenIdMap;
    /**
     * Make this builder out of existing Sentence.
     * It will make a shallow copy of existing sentence so any modification
     * with this builder will also immediately reflect on original sentence.
     */
    static from(sentence: Sentence): SentenceBuilder;
    /** Append given meta into this sentence */
    push_meta(key: string, value: string): void;
    /** Append comment into this sentence */
    push_comment(text: string): void;
    /** Find meta by given `key` */
    find_meta(key: string): Meta;
    /** Find index of meta from given `key` */
    find_meta_index(key: string): number;
    /** Append given token into this sentence */
    push_token(t: Token): void;
    /** Find a token by an `id` */
    find_token_by_id(id: number | [number, number], type: TokenType): Token;
    /** Get id of token at given index or undefined if token at given index is Compound token */
    get_id_by_index(index: number): number | [number, number];
    /** Set head and deprel field of token at given `token_index` argument to ID of token at `head_index`. */
    upsert_head_by_index(token_index: number, head_index: number, relation: Relation): void;
    /** Add or replace a dep in deps field of given token index. The dep to be add/replace use head index instead of ID */
    upsert_dep_by_index(token_index: number, head_index: number, relation: DepsRelation): void;
    /** Insert a `token` at given `index`. The index must be <= number of existing tokens. */
    insert_token(token: Token, index: number): void;
    /**
     * Remove a token at given index and update all dependencies to it based on given policy.
     * It return the removed token without update any field value of it.
     */
    remove_token(index: number, policy?: HeadPolicy): Token;
    /**
     * Merge tokens using index, not ID.
     * Both `from` and `to` are index of token.
     * It's inclusive at both end so if `from = 1`, and `to = 2`, it will merge
     * token at index 1 and 2 into 1 token.
     * The field value of merged token will depends on `policy`.
     */
    merge<X extends XPOS>(from: number, to: number, policy?: MergePolicy<X>): void;
    /**
     * Split a token at given `index`. It take `at` argument which is a list of index
     * of location of `form` field of a token to be splitted. All other fields depends
     * on `policy` argument.
     * It update all other dependencies by shifting all the ID accordingly.
     *
     * The token at given `index` must either be NominalToken or EmptyToken.
     */
    split(index: number, at: number[], policy?: SplitPolicy<XPOS>): void;
    /** Build Sentence object out of this builder */
    build(): Sentence;
}
declare type NonCompound = (NominalToken | EmptyToken);
/**
 * A merging policy.
 *
 * It has following attributes:
 * - `headPol` field will determined how all dependants shall be handle.
 * The default value is `HeadPolicy.Adjust`
 * - `lemma` field is a callback that takes all tokens being merged as argument and return merged lemma.
 * The default value is every lemma concatenated together or undefined.
 * - `upos` field is a callback that takes all tokens being merged as argument and return merged part-of-speech.
 * The default value is a part-of-speech of first token being merge.
 * - `xpos` an optional field which is a callback that takes all tokens being merged as argument and return merged language specific part-of-speech.
 * - `feats` field is a callback that takes all tokens being merged as argument and return merged feature(s).
 * The default value is a flatten merge of all unique features from every tokens being merged.
 * - `headRels` field is a callback that takes all tokens being merged as argument and return merged `head` and `deprel` fields.
 * The default is first token being merged head/deprel field if there's no root `Relation` in merging, otherwise, it
 * will become new root.
 * - `deps` field is a callback that takes all tokens being merged as argument and return merged `deps` fields.
 * The default is merged of every unique deps field of tokens being merged.
 * - `misc` field is a callback that takes all tokens being merged as argument and return merged `misc` fields.
 * The default value is all unique of flatten map misc fields of every tokens.
 */
export declare class MergePolicy<X extends XPOS> {
    headPol: HeadPolicy;
    lemma: (tokens: NonCompound[]) => string;
    upos: (tokens: NonCompound[]) => UPOS;
    xpos?: (tokens: NonCompound[]) => X;
    feats: (tokens: NonCompound[]) => Feature[];
    headRels: (tokens: NominalToken[]) => [HeadId, Relation];
    deps: (tokens: NonCompound[]) => AdvanceDep[];
    misc: (tokens: NonCompound[]) => string[];
}
/**
 * Define how each property of splitted will be derived.
 * By default, all properties are copy from original token.
 */
export declare class SplitPolicy<X extends XPOS> {
    lemma: (token: NonCompound) => string;
    upos: (token: NonCompound) => UPOS;
    xpos?: (token: NonCompound) => X;
    feats: (tokens: NonCompound) => Feature[];
    headRels: (token: NominalToken) => [HeadId, Relation];
    deps: (token: NonCompound) => AdvanceDep[];
    misc: (token: NonCompound) => string[];
}
export {};