utils.py

import regex as re
import os
import nltk
nltk.download('punkt')
from nltk.tokenize import sent_tokenize

def extract_text_by_citation(paragraph):
    citation_regex = re.compile(r'(.*?)(\[\d+\]\.)', re.DOTALL)
    parts_with_citation = citation_regex.findall(paragraph)
    citation_dict = {}
    for part, citation in parts_with_citation:
        part = part.strip()
        citation_num = re.search(r'\[(\d+)\]\.', citation).group(1)
        citation_dict.setdefault(str(int(citation_num) - 1), []).append(part)
    return citation_dict


def correct_citation_format(paragraph):
    modified_sentences = []
    sentences = sent_tokenize(paragraph)
    for sentence in sentences:
        modified_sentences.append(sentence)
    citation_regex = re.compile(r'\[\d+\]\.')
    i = 0
    if len(modified_sentences) == 1:
      has_citation = bool(citation_regex.search(modified_sentences[i]))
    while i < len(modified_sentences):
      if len(modified_sentences[i:i+2]) == 2:
        sentence_group = " ".join(modified_sentences[i:i+2])
        has_citation = bool(citation_regex.search(sentence_group))
        if not has_citation:
            return False
        i += 2 if has_citation and i+1 < len(modified_sentences) and citation_regex.search(modified_sentences[i+1]) else 1
      else:
        return True
    return True

def has_citations(paragraph):
    return bool(re.search(r'\[\d+\]\.', paragraph))

def citations_check(paragraph):
    return has_citations(paragraph) and correct_citation_format(paragraph)