Python object representing a Perseids Treebank text

greek-text.py

import xml.etree.ElementTree as ET

class Text:
    """
    Object representation of a Ancient Greek text formatted in the Perseids XML schema.
    See GAGDT (Guidelines for the Ancient Greek Dependecy Treebank) 2.0 for schema 
    information and additional metadata.
    """
    def __init__(self, path: str) -> None:
        self.path: str = path
        self.tree: ET.ElementTree = ET.parse(self.path)
        self.root: ET.Element     = self.tree.getroot()
        self.body: ET.Element     = self.tree.find("body")

    def count_sentences(self) -> int:
        sentences: list[ET.Element] = self.body.findall("sentence")
        sentence_count: int = len(sentences)
        return sentence_count
    
    def count_words(self) -> int:
        word_count: int = 0
        # Sum up all the words in every sentence in the document
        for sentence in self.body.findall("sentence"):
            words: list[ET.Element] = sentence.findall("word")
            word_count += len(words)
        return word_count