|
import streamlit as st |
|
from datasets import load_dataset |
|
import csv |
|
import datetime as dt |
|
import random |
|
import os |
|
from huggingface_hub import Repository, HfApi |
|
|
|
|
|
HF_API_KEY = os.environ.get("HF_TOKEN", None) |
|
|
|
api = HfApi(token=HF_API_KEY) |
|
|
|
REPO_ID = "imomayiz/darija-english" |
|
DATASET_REPO_URL = f"https://huggingface.co/datasets/{REPO_ID}" |
|
SUBMISSIONS_DATA_FILE = os.path.join("submissions", "submissions.csv") |
|
|
|
submissions_repo = Repository( |
|
local_dir="submissions", clone_from=DATASET_REPO_URL, use_auth_token=HF_API_KEY |
|
) |
|
|
|
def load_data(repo_id): |
|
dataset = load_dataset(f'{repo_id}', name='sentences', split='sentences') |
|
return dataset |
|
|
|
def fetch_sentence(dataset, column_name="darija_ar"): |
|
|
|
|
|
random_sentence_index = random.randint(0, len(dataset) - 1) |
|
random_sentence = dataset[random_sentence_index][column_name] |
|
|
|
return random_sentence |
|
|
|
def store_submission(sentence: str, translation: str, translation_fr: str): |
|
if sentence and (translation or translation_fr): |
|
with open(SUBMISSIONS_DATA_FILE, "a") as csvfile: |
|
writer = csv.DictWriter(csvfile, |
|
fieldnames=["darija", "eng", "darija_ar", "time"]) |
|
writer.writerow( |
|
{"darija_ar": sentence, |
|
"eng": translation, |
|
"darija": translation_fr, |
|
"time": str(dt.datetime.now())} |
|
) |
|
api.upload_file( |
|
path_or_fileobj=SUBMISSIONS_DATA_FILE, |
|
path_in_repo=SUBMISSIONS_DATA_FILE, |
|
repo_id=REPO_ID, |
|
repo_type="dataset", |
|
commit_message="New submission", |
|
) |
|
st.success(f"Submitted at {DATASET_REPO_URL}/{SUBMISSIONS_DATA_FILE}") |
|
|
|
|
|
|
|
dataset = load_data(REPO_ID) |
|
|
|
|
|
def main(): |
|
|
|
if "sentence" not in st.session_state: |
|
st.session_state.sentence = fetch_sentence(dataset) |
|
if 'translation_input' not in st.session_state: |
|
st.session_state.translation_input = "" |
|
if 'translation_input_fr' not in st.session_state: |
|
st.session_state.translation_input_fr = "" |
|
if 'display_new' not in st.session_state: |
|
st.session_state.display_new = False |
|
|
|
st.title("Translate From Arabic to English") |
|
|
|
st.markdown( |
|
"""This mini-app allows you to contribute to the **darija-english** dataset |
|
as part of [DODa](https://darija-open-dataset.github.io/) |
|
project. To contribute, simply translate the given sentence from Arabic to English. |
|
The translated sentence will be submitted to the dataset |
|
[here](https://huggingface.co/datasets/imomayiz/darija-english).""" |
|
) |
|
|
|
st.text("") |
|
|
|
st.write(f""" |
|
<div style=" |
|
padding: 5px; |
|
border: 1px solid #000000; |
|
border-radius: 5px; |
|
"> |
|
<p style="font-size: 20px;">{st.session_state.sentence}.</p> |
|
</div>""", unsafe_allow_html=True) |
|
|
|
|
|
|
|
st.session_state.display_new = st.button("New Sentence", |
|
on_click=fetch_sentence, |
|
args=(dataset,)) |
|
|
|
|
|
|
|
translation_input_placeholder = st.empty() |
|
|
|
with translation_input_placeholder.container(): |
|
translation_input = st.text_input("Enter translation to english: ", |
|
st.session_state.translation_input) |
|
st.session_state.translation_input = translation_input |
|
|
|
|
|
translation_input_placeholder_fr = st.empty() |
|
|
|
with translation_input_placeholder_fr.container(): |
|
translation_input_fr = st.text_input( |
|
"Enter translation to darija in latin characters: ", |
|
st.session_state.translation_input_fr |
|
) |
|
st.session_state.translation_input_fr = translation_input_fr |
|
|
|
|
|
if st.button("Submit Translation"): |
|
if translation_input: |
|
st.success("Translation submitted successfully!") |
|
|
|
elif translation_input_fr: |
|
st.success("Translation submitted successfully!") |
|
|
|
else: |
|
st.warning("Please enter a translation before submitting.") |
|
|
|
|
|
store_submission(st.session_state.sentence, |
|
st.session_state.translation_input, |
|
st.session_state.translation_input_fr) |
|
|
|
if __name__ == "__main__": |
|
main() |
