app.py

import base64
import datetime
import gradio as gr
import pandas as pd
import pytz
import plotly.graph_objects as go
import matplotlib.pyplot as plt
from PIL import Image
from io import BytesIO
import plotly.io as pio

pio.renderers.default = "browser"

df = pd.read_csv("herbologist_almanac_checklist_data.csv")
columns_with_tasks = [
    "task1",
    "task2",
    "task3",
    "task4",
    "color1",
    "color2",
    "color3",
    "color4",
    "color5",
    "color6",
]
PLANTS = list(df["plant"].unique())
# 更改字体设置
# rc("font", family="Arial Unicode MS")

# def decimal2base64(decimal_int):
#     # 定义 64 进制字符集
#     base64_chars = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"

#     if decimal_int == 0:
#         return "0"

#     base64_str = ""
#     while decimal_int > 0:
#         remainder = decimal_int % 64
#         base64_str = base64_chars[remainder] + base64_str
#         decimal_int //= 64

#     return base64_str


def bin_ls2base64(ls):
    # 将二进制列表转换为二进制列表
    binary_str = "".join(str(n) for n in ls)

    # decimal_int = int(bin_str, 2)
    byte_str = int(binary_str, 2).to_bytes((len(binary_str) + 7) // 8, byteorder="big")
    base64_str = base64.b64encode(byte_str).decode("utf-8")
    return base64_str


def base64_to_binary(base64_str):
    if isinstance(base64_str, str):
        # 将base64字符串转换为二进制列表
        byte_str = base64.b64decode(base64_str)
        binary_str = bin(int.from_bytes(byte_str, byteorder="big"))[2:].zfill(
            len(columns_with_tasks) * len(PLANTS)
        )
        # ls = [int(n) for n in binary_str]
        return binary_str


def parse_token(token):
    try:
        if not token:  # 处理空字符串的情况
            token = "\x00"

        if len(token) > 0:
            payload = base64_to_binary(token)
            almanac_data: list = [int(n) for n in payload]
            # print(len(almanac_data))
            parsed_dict = {}
            for _, row in df.iterrows():
                parsed_dict[row["plant"]] = [
                    almanac_data.pop() for _ in range(len(columns_with_tasks))
                ]

        return parsed_dict
    except Exception as e:
        print(e)


# 定义一个简单的函数，模拟接收DataFrame数据
def process_data(*args):
    plot_library = args[-1]
    almanac_dict = dict(zip(PLANTS, args[:-1]))

    almanac_df = df.filter(items=["plant", "name"] + columns_with_tasks)
    almanac_bin_ls = []
    for pl in PLANTS:
        # TODO refactor by using len
        almanac_inp = almanac_dict[pl]

        almanac_bin = [0 for _ in range(len(columns_with_tasks))]

        for n in almanac_inp:
            almanac_bin[n] = 1

        almanac_bin_ls += almanac_bin

        for _, i in enumerate(almanac_inp):
            almanac_df.loc[almanac_df["plant"] == pl, columns_with_tasks[i]] = "✔"

    almanac_reverse_64 = bin_ls2base64(reversed(almanac_bin_ls))
    # 这里可以添加处理DataFrame的逻辑
    return (
        (
            generate_img_by_plotly(almanac_df.drop(columns=df.columns[0]))
            if plot_library == "plotly"
            else generate_img_by_matplotlib(almanac_df.drop(columns=df.columns[0]))
        ),
        almanac_reverse_64,
    )


def show_checkbox_groups(token):
    checklist_inputs = []
    parsed_dict = parse_token(token)
    for index, row in df.iterrows():
        tasks = [
            row[col][0].upper() + row[col][1:]
            for col in columns_with_tasks
            if df.notnull().at[index, col]
        ]
        with gr.Row():
            checkbox = gr.CheckboxGroup(
                tasks,
                label=f"{row['name']}",
                value=[
                    tasks[i]
                    for i, v in enumerate(parsed_dict[row["plant"]])
                    if v == 1 and df.notnull().at[index, columns_with_tasks[i]]
                ],
                type="index",
            )
            checklist_inputs.append(checkbox)

    return checklist_inputs


def wrap_text(text, max_width=20):
    """Manually wrap text based on max_width (character count)"""
    wrapped_lines = []
    words = text.split(" ")
    line = ""

    for word in words:
        if len(line) + len(word) + 1 <= max_width:
            line += word + " "
        else:
            wrapped_lines.append(line.strip())
            line = word + " "

    wrapped_lines.append(line.strip())
    return "\n".join(wrapped_lines)


def generate_img_by_matplotlib(df):
    fig, ax = plt.subplots(
        figsize=(12, 16)
    )  # Adjust the figure size for better readability
    ax.axis("off")  # Turn off the axis

    # Create a table with a light grey background for the header
    table = ax.table(
        cellText=df.values,
        colLabels=df.columns,
        loc="center",
        colColours=["#f2f2f2"] * len(df.columns),
        colWidths=[w / len(df.columns) for w in [1] + [2] * 4 + [1] * 6],
    )

    table.auto_set_font_size(False)
    table.set_fontsize(10)

    # Apply text wrapping and center alignment to non-header cells
    for (row, col), cell in table.get_celld().items():

        if cell.get_text().get_text() == "nan":
            cell.set_text_props(text="", ha="center")
        if row == 0:
            cell.set_text_props(
                weight="bold", ha="center"
            )  # Bold and center-align header text
        if col == 0:
            text = cell.get_text().get_text()
            wrapped_text = wrap_text(text, 9)
            cell.set_text_props(text=wrapped_text, ha="center", linespacing=1)
        else:
            text = cell.get_text().get_text()
            wrapped_text = wrap_text(text)

            cell.set_text_props(
                text=wrapped_text,
                ha="center",
                fontsize=10,
                linespacing=1,
            )  # Enable text wrapping and center-align

    # Manually scale table if needed for better readability with wrapped text
    table.scale(1, 4)  # Adjust row height

    # Ensure the layout is adjusted properly
    plt.tight_layout(pad=0.5)  # Increase padding slightly

    # Save the figure to a bytes buffer
    buffer = BytesIO()
    fig.savefig(
        buffer, format="png", pad_inches=0.2
    )  # Adjust padding around the figure
    plt.close(fig)  # Close the figure to free up memory
    buffer.seek(0)
    image = Image.open(buffer)

    return image


def color_mapping(color):
    color_hex = {
        "turquoise": "#28b6aa",
        "chartreuse": "#dcde82",
        "red": "#981c05",
        "yellow": "#f4ca3a",
        "pink": "#fd8d9b",
        "blue": "#4194bd",
        "white": "#ffffff",
        "black": "#b7b7b7",
        "orange": "#d97413",
        "purple": "#8a659a",
        "viridian": "#a1c42a",
    }
    # if color == "" or color == "✔":
    #     return "white"
    return color_hex.get(color, "white")


def styled_header(header):
    return dict(
        values=[[f"<b>{attr.upper()}</b>"] for attr in header],
        line_color="darkslategray",
        fill_color="royalblue",
        align=["center"],
        font=dict(color="white", size=10),
        # height=40
    )


def styled_cells(cells):
    return dict(
        values=cells,
        line_color="darkslategray",
        fill_color=["lavender", "white", "white", "white", "white"]
        + [[color_mapping(str(el)) for el in col] for col in cells[5:]],
        align="center",
        font_size=10,
        height=30,
    )


def handle_element(el):
    emoji_mapping = {
        "light": "🌞",
        "moisture": "💧",
        "mood": "💗",
        "sell": "💲",
        "harvest": "🌱",
        "collect": "🖐️",
        "hygiene": "🧽",
        "pest": "🐛",
        "overgrowth": "🌿",
        "show": "👩🏻‍🌾",
    }
    if isinstance(el, str):
        if el == "✔":
            return "✅"
        for cond in emoji_mapping.keys():
            if cond in el.lower():
                new_text = el + emoji_mapping[cond]
                return new_text

    return el


def generate_img_by_plotly(df):
    df.fillna("", inplace=True)

    cells_values = [df[col].to_list() for col in df.columns]
    header_values = list(df.columns)

    # add emoji
    cells_values = [[handle_element(el) for el in col] for col in cells_values]

    # Create the table figure
    fig = go.Figure(
        data=[
            go.Table(
                header=styled_header(header_values),
                cells=styled_cells(cells_values),
                columnwidth=[80, 160, 160, 160, 160, 80, 80, 80, 80, 80, 80],
            )
        ]
    )
    fig.add_annotation(
        text=f'Herbology Almanac Checklist Generated on(GMT): {datetime.datetime.now(tz=pytz.utc).strftime("%Y-%m-%d %H:%M:%S")}',
        xref="paper",
        yref="paper",
        x=0.5,
        y=0,
        showarrow=False,
    )

    fig.update_layout(
        width=1200,
        height=1200,
        margin=dict(l=20, r=20, t=20, b=20),
        # title_text=f'Herbology Almanac Checklist Generated on(GMT): {datetime.datetime.now(tz=pytz.utc).strftime("%Y-%m-%d %H:%M:%S")}',
        # annotations=[
        #     dict(
        #         text=f"Generated on(GMT): {datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")}",
        #         xref="paper",
        #         yref="paper",
        #         x=0.5,  # 水平居中
        #         y=-0.1,  # 距离图表底部的位置
        #         showarrow=False,
        #         font=dict(size=10),
        #     )
        # ],
    )
    buffer = BytesIO()
    fig.write_image(buffer, format="png")
    buffer.seek(0)
    image = Image.open(buffer)

    return image


with gr.Blocks() as app:
    gr.Markdown(
        """
        <center><font size=8>👩🏻‍🌾Herbology Almanac Checklist Generator📝</font></center>
        This is a simple web app that generates an almanac checklist for your plants.
        """
    )

    gr.Markdown(
        """
        # RECOVERY TOKEN
        """
    )
    recovery_token = gr.Textbox(
        value="",
        label="Recovery Token",
        info="Save this token or paste your saved one here",
        placeholder="Keep this token to restore your previous input".upper(),
        interactive=True,
    )

    gr.Markdown(
        """
        # YOUR RESEARCH TASKS
        """
    )
    checklist_inputs = show_checkbox_groups(recovery_token.value)

    gr.Markdown(
        """
        # IMAGE STYLE
        """
    )
    plot_library = gr.Radio(
        ["matplotlib", "plotly"],
        label="Plot Library",
        value="matplotlib",
        info="Choose your plot library",
    )

    submit_button = gr.Button("Generate Image and Token")

    # df_out = gr.Dataframe(label="Output Dataframe", interactive=False)

    generated_img = gr.Image(label="Generated Image", format="png", type="pil")

    logs = gr.Markdown(
        """
        # CHANGELOG
        - 2024/08/31: Initial release
        - 2024/09/03: Fix a mistake in the tasks of mimbulus
        - 2024/09/04: Correct Radiant count for water hyacinth
        - 2024/09/05: Support image generated by plotly
    """
    )

    submit_button.click(
        process_data,
        inputs=checklist_inputs + [plot_library],
        outputs=[generated_img, recovery_token],
    )

    recovery_token.change(
        show_checkbox_groups,
        inputs=[recovery_token],
        outputs=checklist_inputs,
    )

    # generate_button.click(
    #     generate_img,
    #     inputs=[df_out],
    #     outputs=[generated_img],
    # )


app.queue()
app.launch()