42 lines
1.8 KiB
Python
42 lines
1.8 KiB
Python
from funcy import identity, compose, first, juxt, rest, rcompose
|
|
|
|
from cv_analysis.utils import star, rconj
|
|
from synthesis.random import rnd
|
|
from synthesis.text.line_formatter.line_formatter import LineFormatter
|
|
|
|
|
|
class ParagraphLineFormatter(LineFormatter):
|
|
def __init__(self, blank_line_percentage=None):
|
|
self.blank_line_percentage = blank_line_percentage or rnd.uniform(0, 0.5)
|
|
|
|
def __call__(self, lines, last_full):
|
|
return self.format_lines(lines, last_full)
|
|
|
|
def format_lines(self, lines, last_full):
|
|
def truncate_current_line():
|
|
return rnd.random() < self.blank_line_percentage and last_full
|
|
|
|
# This is meant to be read from the bottom up.
|
|
current_line_shall_not_be_a_full_line = truncate_current_line()
|
|
line_formatter = self.truncate_line if current_line_shall_not_be_a_full_line else identity
|
|
format_current_line = compose(line_formatter, first)
|
|
move_current_line_to_back = star(rconj)
|
|
split_first_line_from_lines_and_format_the_former = juxt(rest, format_current_line)
|
|
split_off_current_line_then_format_it_then_move_it_to_the_back = rcompose(
|
|
split_first_line_from_lines_and_format_the_former,
|
|
move_current_line_to_back,
|
|
)
|
|
current_line_is_a_full_line = not current_line_shall_not_be_a_full_line
|
|
# Start reading here and move up.
|
|
return split_off_current_line_then_format_it_then_move_it_to_the_back(lines), current_line_is_a_full_line
|
|
|
|
def format_line(self, line, full=True):
|
|
line = self.truncate_line(line) if not full else line
|
|
return line, full
|
|
|
|
def truncate_line(self, line: str):
|
|
n_trailing_words = rnd.randint(0, 4)
|
|
line = " ".join(line.split()[-n_trailing_words - 1 : -1]).replace(".", "")
|
|
line = line + ".\n" if line else line
|
|
return line
|