import gradio as gr
from toiro import tokenizers

num_input_lines = 3
default_text = "ここにテキストを入力し、Enter を押してください。"
title = "Japanese Tokenizer Comparison"
description = """
This is a demo comparing Japanese tokenizers. You can compare the tokenization results of tools that are available with just a `pip install` in Python.
"""

article = """

# How to install each library

[Janome](https://github.com/mocobeta/janome):
```
pip install janome
```

[nagisa](https://github.com/taishi-i/nagisa):
```
pip install nagisa
```

[sudachi.rs](https://github.com/WorksApplications/sudachi.rs):
```
pip install sudachipy sudachidict_core
```

[mecab-python3](https://github.com/SamuraiT/mecab-python3):
```
pip install mecab-python3
```

[fugashi_ipadic](https://github.com/polm/fugashi):
```
pip install fugashi ipadic
```

[fugashi_ipadic](https://github.com/polm/fugashi):
```
pip install fugashi unidic-lite
```

"""


def tokenize(text):
    words_janome = tokenizers.tokenize_janome(text)
    words_nagisa = tokenizers.tokenize_nagisa(text)
    words_sudachirs = tokenizers.tokenize_sudachipy(text)
    words_mecabpython3 = tokenizers.tokenize_mecab(text)
    words_fugashi_ipadic = tokenizers.tokenize_fugashi_ipadic(text)
    words_fugashi_unidic = tokenizers.tokenize_fugashi_unidic(text)

    return (
        words_janome,
        words_nagisa,
        words_sudachirs,
        words_mecabpython3,
        words_fugashi_ipadic,
        words_fugashi_unidic,
    )


iface = gr.Interface(
    fn=tokenize,
    inputs=gr.Textbox(
        label="Input text",
        lines=num_input_lines,
        value=default_text,
    ),
    title=title,
    description=description,
    article=article,
    outputs=[
        gr.Textbox(label="Janome"),
        gr.Textbox(label="nagisa"),
        gr.Textbox(label="sudachi.rs"),
        gr.Textbox(label="mecab-python3"),
        gr.Textbox(label="fugashi_ipadic"),
        gr.Textbox(label="fugashi_unidic"),
    ],
    live=True,
)


iface.launch()