from nest2D import Point, Box, Item, nest, SVGWriter
import plotly
import numpy as np
import plotly.graph_objects as go
import cv2
import gradio as gr

def transform(point:list, x:float, y:float, rotation:float):
    if point is None:
        return None

    point = np.array([point[0], point[1], 1])
    matrix = np.array([[np.cos(rotation), -np.sin(rotation), x],
                        [np.sin(rotation), np.cos(rotation), y],
                        [0,0,1]])

    return (matrix@point)[:2]

class BinPacking:

    def __init__(self, width:int, height:int, image:np.ndarray, imageScale:float=1) -> None:
        self.width = width
        self.height = height
        self.pgrp = None
        self.total = 0
        self.image = cv2.resize(image, (int(image.shape[1]*imageScale), int(image.shape[0]*imageScale)))
        if self.image.shape[2] == 4:
            x,y = np.where(self.image[:,:,3]==0)
            self.image[x,y] = np.array([255,255,255,0])
            self.imgray = cv2.cvtColor(self.image, cv2.COLOR_BGRA2GRAY)
        else:
            self.imgray = cv2.cvtColor(self.image, cv2.COLOR_BGR2GRAY)
        self.imgray = cv2.bitwise_not(self.imgray)

    @property
    def box(self):    
        return Box(self.width, self.height)

    def pack(self):
        # make margin
        imagem = cv2.dilate(self.imgray, np.ones((3,3), np.uint8), iterations=10)
        _, thresh = cv2.threshold(imagem, 127, 255, 0)

        contours, _ = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
        contourList = []
        for contour in contours:
            contourList.extend(contour)
            
        hull = cv2.convexHull(np.array(contourList).reshape(-1,2))
        hull = np.append(hull, [hull[0]], axis=0).reshape(-1,2)
        item = Item(
            [
                Point(point[0], point[1])
                for point in np.flip(hull, 0)
            ]
        )

        max_item = int(self.width*self.height/item.area)
        self.pgrp = nest([item,]*max_item, self.box)
        if self.pgrp is None or len(self.pgrp) == 0:
            self.total = 0
        else:
            self.total = len(self.pgrp[0])
        return self

    def visualize(self):
        fig = go.Figure()
        widthScale = self.width/self.height
        fig.update_layout(
            autosize=False,
            width=500*widthScale,
            height=500,
            margin=dict(
                l=50,
                r=50,
                b=50,
                t=50,
                pad=4
            ),
            paper_bgcolor="LightSteelBlue",
            showlegend=False,
        )
        if self.pgrp == None or len(self.pgrp) == 0:
            return fig

        _, thresh = cv2.threshold(self.imgray, 127, 255, 0)
        contours, _ = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
        transformedPoints = []
        for item in self.pgrp[0]:
            for contour in contours:
                for point in contour:
                    transformedPoints.append(transform(point[0], item.translation.x, item.translation.y, item.rotation))
                transformedPoints.append([None,None])
            transformedPoints.append([None,None])

        fig.add_trace(go.Scatter(
            x=[0, self.width, self.width, 0, 0],
            y=[0, 0, self.height, self.height, 0],
            fill="toself",
            mode="lines",
            textposition="bottom right",
        )) 
        fig.add_trace(go.Scatter(
            x=[point[0] for point in transformedPoints],
            y=[point[1] for point in transformedPoints],
            mode="lines",
            textposition="bottom right",
        )) 
        return fig


def process(width:int, height:int, scale:float, image: np.ndarray):

    packer = BinPacking(width, height, image, scale)

    packer.pack()
    figure = packer.visualize()
    if packer.total == 0:
        text = "Input too big"
    else:
        text = f"Fit {packer.total} instances"

    return [figure, text]

def change_language(languageSelection):

    return [
        gr.Dropdown.update(value="English" if languageSelection == "English" else "日本語"),
        gr.Slider.update(label="Width" if languageSelection == "English" else "横長"),
        gr.Slider.update(label="Height" if languageSelection == "English" else "縦長"),
        gr.Slider.update(label="Input Scale" if languageSelection == "English" else "入力拡大"),
        gr.Markdown.update(
            """
            # Image fitting
            ### Given Image input, fit as many as possible x number of input on canvas
            """
            if languageSelection == "English" else
            """
            # 画像フィッティング
            ### 画像入力すると、キャンバス上の入力の数 x できるだけ多く収まります
            """),
        gr.Image.update(label="Image" if languageSelection == "English" else "画像入力", ),
        gr.Text.update(label="Result" if languageSelection == "English" else "結果"),
        gr.Button.update("Submit" if languageSelection == "English" else "送信"),
        gr.Plot.update(label="Plot" if languageSelection == "English" else "プロット"),
    ]

if __name__ == "__main__":
    options = ["English", "日本語"]
    with gr.Blocks() as demo:
        with gr.Row():
            with gr.Column(scale=8):
                pass
            with gr.Column():
                languageSelection = gr.Dropdown(options, value="English", show_label=False)

        desc = gr.Markdown(
            """
            # Image fitting
            ### Given Image input, fit as many as possible x number of input on canvas
            """
        )

        with gr.Row():
            with gr.Column():
                image = gr.Image(image_mode="RGBA", label="Image")
                with gr.Row():
                    width = gr.Slider(value=1500, minimum=100, maximum=8000, label="Width")
                    height = gr.Slider(value=1500, minimum=100, maximum=8000, label="Height")
                    scale = gr.Slider(value=1.0, minimum=0, maximum=5, step=0.01, label="Input Scale")
                fit = gr.Button("Submit")
                
            with gr.Column():
                plot_output = gr.Plot(label="Plot")
                text = gr.Text(label="Result")

        languageSelection.change(fn=change_language,inputs=[languageSelection] , outputs=[languageSelection, width, height, scale, desc, image, text, fit, plot_output])
        fit.click(fn=process, inputs=[width, height, scale, image], outputs=[plot_output, text])
        

    demo.launch()