wiktor_ner.py

from typing import Dict, List, Tuple
from lxml import etree
import json
from collections import defaultdict

# from src.annotation_types_old import
from src.input_parsers.interface import InputParser
from src.annotations import Annotation, MorphosyntacticAnnotation, NerAnnotation


class WiktorNERInputParser(InputParser):
    def __init__(self) -> None:
        super().__init__()

    def parse(self, content: str) -> Tuple[str, List[Tuple[int, int, Annotation]]]:
        """Parse wiktorner file into text and annotations.

        Annotations are returned as a dictionary with channel name as a key and list of tuples.

        Args:
            co z  (str): Path to file containing CCL.

        Returns:
            Tuple[str, List[Tuple[int, int, Annotation]]]: Text and annotations.
        """
        content_parsed = json.loads(content)

        if "text" in content_parsed:
            text = content_parsed["text"]
        else:
            text = ""

        annotations = []

        # Morphosyntactic annotations
        if "tokens" in content_parsed:
            for token in content_parsed["tokens"]:
                if "position" in token:
                    token_start, token_end = token["position"]
                    if "lexemes" in token:
                        for lexeme in token["lexemes"]:
                            if "disamb" in lexeme and lexeme["disamb"] == True:
                                if "mstag" in lexeme:
                                    annotations.append(
                                        (
                                            token_start,
                                            token_end,
                                            MorphosyntacticAnnotation(lexeme["mstag"]),
                                        )
                                    )

        # NER annotations
        if "entities" in content_parsed:
            for entity in content_parsed["entities"]:
                if "positions" in entity:
                    entity_start, entity_end = entity["positions"]
                    if "type" in entity:
                        annotations.append(
                            (entity_start, entity_end, NerAnnotation(entity["type"]))
                        )

        return text, annotations