import textwrap from typing import List from faker import Faker from funcy import identity, iterate, take, last from cv_analysis.utils import star from cv_analysis.utils.rectangle import Rectangle from synthesis.random import rnd from synthesis.text.text_block_generator.text_block_generator import TextBlockGenerator from synthesis.text.line_formatter.paragraph import ParagraphLineFormatter class ParagraphGenerator(TextBlockGenerator): def __init__(self): self.line_formatter = ParagraphLineFormatter(blank_line_percentage=rnd.uniform(0, 0.5)) def __call__(self, rectangle, n_sentences): return self.generate_paragraph(rectangle, n_sentences) def generate_paragraph(self, rectangle, n_sentences): lines = generate_random_text_lines(rectangle, self.line_formatter, n_sentences) return lines def generate_random_text_lines(rectangle: Rectangle, line_formatter=identity, n_sentences=3000) -> List[str]: text = Faker().paragraph(nb_sentences=n_sentences, variable_nb_sentences=False, ext_word_list=None) unformatted_lines = textwrap.wrap(text, width=rectangle.width, break_long_words=False) # each iteration of the line formatter function formats one more line and adds it to the back of the list formatted_lines_generator = iterate(star(line_formatter), (unformatted_lines, True)) # hence do as many iterations as there are lines in the rectangle lines_per_iteration = take(len(unformatted_lines), formatted_lines_generator) # and then take the lines from the last iteration of the function formatted_lines, _ = last(lines_per_iteration) return formatted_lines