import gradio as gr

import re
import string
import random
from bs4 import BeautifulSoup

contents = ''

def make_body(file):
    with open(file.name, 'r') as xml_file:
        text = xml_file.read()
        xml = BeautifulSoup(text, "xml")
        global contents
        contents = str(xml.find_all('body')).strip('[]')
    allReferences = []

    for tag in xml.findAll('ref'):
        allReferences.append(tag.get('id'))

    with open("body.xml", 'w') as f:

        f.write(make_refs(contents, allReferences))
    return "body.xml"

def index_to_ref_id(reference_text, reference_ids):
    reference_text = reference_text.strip('[]')
    dash_separator = "–" # long dash here
    comma_separator = ", " # workaound
    id_max = len(reference_ids)
    ids = []
    refs = []

    if reference_text.find(dash_separator) != -1:
        s, e = [int(x) for x in reference_text.split(dash_separator)]
        for i in range(s,e+1):
            ids.append(i)
    elif reference_text.find(comma_separator) != -1:
        ids = [int(x) for x in reference_text.split(comma_separator)]    
    else:
        ids.append(int(reference_text))

    for id in ids:
        refs.append(reference_ids[id-1])

    return refs

# https://www.geeksforgeeks.org/python-generate-random-string-of-given-length/
def generate_id(type, length=32):
    N = length

    # using random.choices()
    # generating random strings
    res = type +  '-' + ''.join(random.choices(string.ascii_lowercase +
                                string.digits, k=N))

    return(str(res))


def xref_generator(reference_text, reference_ids):
    id_type = "xref"
    references_str = ' '.join(index_to_ref_id(reference_text, reference_ids))
    generated_xref = '<xref id="' + generate_id(id_type) + '" ref-type="bibr" rid="' + references_str + '">' + reference_text + '</xref>'

    return generated_xref

def make_refs(article_body, reference_ids):
    citations_list = []

    citations_raw = re.findall(r"\[\d+(?:–\d+)?(?:,\s?\d+)*\]", contents)

    for cit in citations_raw:
        citations_list.append(cit) # bugfix

    x = contents

    for cit in citations_list:
        x = x.replace(str(cit), str(xref_generator(cit, reference_ids)))

    return x



demo = gr.Interface(
    make_body,
    gr.File(file_count="single", file_types=[".xml"]),
    "file",
    description=
    """
    # Инструкция:
    1. В левое окно перетащить файл jats xml с размеченным списком литературы (файл 1);
    2. Нажать "Submit";
    3. В правом окне появится строчка "body.xml";
    4. Нажать на неё, загрузится body.xml;
    5. Скопировать содержимое файла body.xml;
    6. Заменить в файле (1) раздел body содержимым файла body.xml.
    """,
    allow_flagging="never"
)

demo.launch()