如何在PIL中实现不规则多边形内的最大字号文本自动排版？

需求概述 #

需要将任意长度文本排版到指定非矩形多边形内，实现类似Word文本绕排的效果，且字号尽可能大。现有PIL/textwrap仅支持矩形区域排版，缺乏不规则形状适配能力。

现有矩形适配代码 #

已实现顶部小矩形区域的最大字号适配：

img = Image.open("Frames/defaultframe.png")
draw = ImageDraw.Draw(img)

selected_size = 1
for size in range(1, 150):
    title = "SOME TITLE HERE"
    titleFont = ImageFont.FreeTypeFont('C:/Windows/Fonts/POORICH.TTF', size=size)
    w, h = titleFont.getsize(title)  # older versions
   
    if w > 300 or h > 45:
        break
        
    selected_size = size
         
titleFont = ImageFont.FreeTypeFont('C:/Windows/Fonts/POORICH.TTF', size=selected_size)

draw.text((435, 135), title, fill='black', anchor='mm', font=titleFont)

目标多边形边界 #

polygon = [(130, 405),(642, 405),(625, 537),(580, 660),(285, 660),(200, 550)]

预期效果 #

解决方案实现 #

核心思路

1. 射线法判断点是否在多边形内：无需复杂向量知识，实现简单可靠的区域校验
2. 逐行文本适配：对每个候选字号，逐字符构建文本行，校验行边界是否在多边形内
3. 二分法快速找最大字号：替代遍历法，大幅提升效率

完整代码

from PIL import Image, ImageDraw, ImageFont

def point_in_polygon(point, polygon):
    """用射线法判断点是否在多边形内部"""
    x, y = point
    inside = False
    n = len(polygon)
    for i in range(n):
        p1x, p1y = polygon[i]
        p2x, p2y = polygon[(i+1)%n]
        # 检查点的y坐标是否在当前边的y区间内
        if ((p1y > y) != (p2y > y)):
            # 计算射线与边的交点x坐标
            x_intersect = (y - p1y) * (p2x - p1x) / (p2y - p1y) + p1x
            if x <= x_intersect:
                inside = not inside
    return inside

def fit_text_in_polygon(text, polygon, font_path, max_size=150):
    """寻找能完整放入多边形的最大字号，并返回排版后的文本行"""
    low = 1
    high = max_size
    best_size = 1
    best_lines = []
    
    while low <= high:
        mid = (low + high) // 2
        font = ImageFont.FreeTypeFont(font_path, size=mid)
        lines = []
        current_line = ""
        valid = True
        line_height = font.getsize("A")[1]
        # 获取多边形的边界范围
        min_x = min(p[0] for p in polygon)
        max_x = max(p[0] for p in polygon)
        min_y = min(p[1] for p in polygon)
        max_y = max(p[1] for p in polygon)

        for char in text:
            test_line = current_line + char
            line_width = font.getsize(test_line)[0]
            # 计算当前测试行的边界坐标
            line_left = min_x
            line_top = min_y + len(lines) * line_height
            line_right = line_left + line_width
            line_bottom = line_top + line_height

            # 校验行的四个角是否都在多边形内，且不超出多边形垂直范围
            if line_bottom > max_y or not (
                point_in_polygon((line_left, line_top), polygon) and
                point_in_polygon((line_right, line_top), polygon) and
                point_in_polygon((line_left, line_bottom), polygon) and
                point_in_polygon((line_right, line_bottom), polygon)
            ):
                # 无法添加当前字符，换行
                if not current_line:
                    valid = False
                    break
                lines.append(current_line)
                current_line = char
                # 检查新行的垂直位置是否超出多边形
                if min_y + len(lines) * line_height + line_height > max_y:
                    valid = False
                    break
            else:
                current_line = test_line
        
        # 处理最后一行
        if current_line and valid:
            lines.append(current_line)
            last_line_bottom = min_y + (len(lines)-1)*line_height + line_height
            if not point_in_polygon((min_x, last_line_bottom), polygon):
                valid = False
        
        if valid:
            best_size = mid
            best_lines = lines
            low = mid + 1
        else:
            high = mid - 1
    
    return best_size, best_lines

# 主执行逻辑
if __name__ == "__main__":
    img = Image.open("Frames/defaultframe.png")
    draw = ImageDraw.Draw(img)
    target_polygon = [(130, 405),(642, 405),(625, 537),(580, 660),(285, 660),(200, 550)]
    input_text = "SOME TITLE HERE OR LONGER TEXT YOU NEED TO FIT"
    font_path = 'C:/Windows/Fonts/POORICH.TTF'

    best_size, text_lines = fit_text_in_polygon(input_text, target_polygon, font_path)
    font = ImageFont.FreeTypeFont(font_path, size=best_size)
    line_height = font.getsize("A")[1]
    min_x = min(p[0] for p in target_polygon)
    min_y = min(p[1] for p in target_polygon)

    # 绘制文本（左对齐，可自行修改为居中/右对齐）
    for idx, line in enumerate(text_lines):
        draw.text((min_x, min_y + idx*line_height), line, fill='black', font=font)
    
    # 可选：绘制多边形边界用于调试
    draw.polygon(target_polygon, outline='red')
    
    img.save("text_in_polygon.png")

优化与扩展建议

• 文本对齐：若需居中对齐，可将每行绘制的x坐标改为 (max_x + min_x - font.getsize(line)[0])//2
• 精确字符校验：如果需要更严格的适配，可以遍历每个字符的 bounding box 所有点，确保完全在多边形内
• 字体选择：不同字体的字符宽度差异较大，可根据需求调整字体参数
• 性能优化：对于非常复杂的多边形，可预先缓存边界信息，减少重复计算

内容的提问来源于stack exchange，提问作者connor mcevoy