#!/usr/bin/env python3
"""
大型图像分析工具
用于分析《三体》项目的复杂图表结构
"""

import os
import sys
from PIL import Image, ImageDraw, ImageFont
import numpy as np
from collections import Counter
import json

def analyze_image_basic_info(image_path):
    """分析图像基本信息"""
    print(f"正在分析图像: {image_path}")
    
    try:
        # 使用更节省内存的方式打开图像
        Image.MAX_IMAGE_PIXELS = None  # 移除像素数量限制
        
        with Image.open(image_path) as img:
            print(f"图像格式: {img.format}")
            print(f"图像模式: {img.mode}")
            print(f"图像尺寸: {img.size} (宽 x 高)")
            print(f"总像素数: {img.size[0] * img.size[1]:,}")
            
            # 计算文件大小
            file_size = os.path.getsize(image_path)
            print(f"文件大小: {file_size / (1024*1024):.1f} MB")
            
            return img
    except Exception as e:
        print(f"打开图像时出错: {e}")
        return None

def analyze_image_colors(img, sample_size=1000):
    """分析图像颜色分布"""
    print("\n=== 颜色分析 ===")
    
    # 将图像转换为RGB模式（如果不是的话）
    if img.mode != 'RGB':
        img = img.convert('RGB')
    
    # 采样分析（对于大图像，采样会更快）
    pixels = list(img.getdata())
    if len(pixels) > sample_size * sample_size:
        # 均匀采样
        step = len(pixels) // (sample_size * sample_size)
        pixels = pixels[::step]
    
    # 统计主要颜色
    color_counter = Counter(pixels)
    print(f"采样像素数: {len(pixels):,}")
    print("主要颜色 (RGB值, 出现次数):")
    
    for color, count in color_counter.most_common(10):
        percentage = (count / len(pixels)) * 100
        print(f"  RGB{color}: {count:,} 次 ({percentage:.1f}%)")

def detect_content_regions(img, threshold=240):
    """检测图像中的内容区域"""
    print("\n=== 内容区域检测 ===")
    
    # 转换为灰度图
    gray = img.convert('L')
    pixels = np.array(gray)
    
    # 找到非白色区域（假设白色背景）
    non_white = pixels < threshold
    
    # 找到边界
    rows, cols = np.where(non_white)
    if len(rows) > 0:
        min_row, max_row = rows.min(), rows.max()
        min_col, max_col = cols.min(), cols.max()
        
        print(f"内容区域边界:")
        print(f"  行范围: {min_row} - {max_row} (高度: {max_row - min_row + 1})")
        print(f"  列范围: {min_col} - {max_col} (宽度: {max_col - min_col + 1})")
        
        return (min_row, max_row, min_col, max_col)
    else:
        print("未检测到明显的内容区域")
        return None

def extract_text_regions(img, region_bounds=None):
    """提取可能的文本区域"""
    print("\n=== 文本区域分析 ===")
    
    # 如果指定了区域边界，只分析该区域
    if region_bounds:
        min_row, max_row, min_col, max_col = region_bounds
        img_cropped = img.crop((min_col, min_row, max_col, max_row))
    else:
        img_cropped = img
    
    # 转换为灰度图
    gray = img_cropped.convert('L')
    pixels = np.array(gray)
    
    # 简单的文本检测：寻找高对比度区域
    # 计算局部方差
    from scipy import ndimage
    
    # 使用sobel算子检测边缘
    sobel_x = ndimage.sobel(pixels, axis=1)
    sobel_y = ndimage.sobel(pixels, axis=0)
    edges = np.sqrt(sobel_x**2 + sobel_y**2)
    
    # 找到高边缘密度的区域
    edge_threshold = np.percentile(edges, 90)
    high_edge_regions = edges > edge_threshold
    
    # 统计高边缘区域
    high_edge_pixels = np.sum(high_edge_regions)
    total_pixels = pixels.size
    
    print(f"高边缘密度像素: {high_edge_pixels:,} / {total_pixels:,} ({high_edge_pixels/total_pixels*100:.1f}%)")
    
    return high_edge_regions

def create_overview_image(img, output_path="overview.png", max_dimension=2000):
    """创建图像概览（缩略图）"""
    print(f"\n=== 创建概览图像 ===")
    
    # 计算缩放比例
    scale = min(max_dimension / img.size[0], max_dimension / img.size[1], 1.0)
    
    if scale < 1.0:
        new_size = (int(img.size[0] * scale), int(img.size[1] * scale))
        print(f"缩放到: {new_size}")
        overview = img.resize(new_size, Image.Resampling.LANCZOS)
    else:
        overview = img.copy()
    
    # 保存概览图
    overview.save(output_path)
    print(f"概览图已保存: {output_path}")
    
    return overview

def main():
    """主函数"""
    image_path = "/home/ben/code/huhan3000/3body/三体结构3.drawio.png"
    
    print("=" * 50)
    print("《三体》项目大型图像分析工具")
    print("=" * 50)
    
    # 分析基本信息
    img = analyze_image_basic_info(image_path)
    if img is None:
        return
    
    # 分析颜色分布
    analyze_image_colors(img)
    
    # 检测内容区域
    regions = detect_content_regions(img)
    
    # 提取文本区域
    extract_text_regions(img, regions)
    
    # 创建概览图
    create_overview_image(img, "/home/ben/code/huhan3000/3body/overview.png")
    
    print("\n" + "=" * 50)
    print("分析完成！")
    print("=" * 50)

if __name__ == "__main__":
    main()