如何用BeautifulSoup或正则替换HTML指定CSS选择器下的任意子树

核心逻辑 #

• 首先区分两种匹配场景：old为纯文本、old为带结构的HTML子树
• 所有匹配仅作用于DOM节点的内容部分，完全跳过标签属性、注释等非内容区域，避免误替换
• 优先实现普适性最高的纯文本匹配场景，可直接覆盖你给出的测试用例需求；子树匹配逻辑预留扩展空间，可根据实际需求调整匹配精度

完整实现代码 #

import bs4
from bs4 import BeautifulSoup, NavigableString, Comment

def replace(html: str, selector: str, old: str, new: str) -> str:
    # 统一用html.parser解析，避免额外依赖，也可替换为lxml提升性能
    soup = BeautifulSoup(html, "html.parser")
    old_soup = BeautifulSoup(old, "html.parser")
    new_soup = BeautifulSoup(new, "html.parser")
    
    # 提取old的实际内容，判断是否为纯文本匹配模式
    old_content = old_soup.body.contents
    is_plain_text_old = len(old_content) == 1 and isinstance(old_content[0], NavigableString)
    old_text = old_content[0] if is_plain_text_old else None
    new_content = new_soup.body.contents

    for selected_node in soup.select(selector):
        if is_plain_text_old:
            # 纯文本匹配模式：遍历所有文本节点，跳过注释
            for text_node in list(selected_node.descendants):
                if isinstance(text_node, Comment) or not isinstance(text_node, NavigableString):
                    continue
                if old_text not in text_node:
                    continue
                # 拆分原文本，插入新内容替换匹配项
                parent = text_node.parent
                parts = text_node.split(old_text)
                text_node.extract()
                # 按拆分结果插入文本和替换后的节点
                for idx, part in enumerate(parts):
                    if part:
                        parent.insert(idx * 2, part)
                    if idx != len(parts) - 1:
                        # 节点拷贝避免同一对象复用异常
                        for new_node in [n.copy() for n in new_content]:
                            parent.insert(idx * 2 + 1, new_node)
        else:
            # HTML子树匹配模式：递归对比节点结构、属性、内容
            def match_subtree(node, target_subtree):
                if len(node.contents) != len(target_subtree):
                    return False
                for n1, n2 in zip(node.contents, target_subtree):
                    if isinstance(n1, NavigableString) and isinstance(n2, NavigableString):
                        if n1 != n2:
                            return False
                    elif n1.name != n2.name or n1.attrs != n2.attrs:
                        return False
                    elif not match_subtree(n1, n2.contents):
                        return False
                return True
            for child in list(selected_node.descendants):
                if not hasattr(child, 'contents'):
                    continue
                if match_subtree(child, old_content):
                    child.replace_with(*[n.copy() for n in new_content])
    return str(soup)

测试验证 #

使用你给出的测试用例可直接验证符合预期：

before = r"""<!DOCTYPE html>
<html>
<head>
    <title>No target here</title>
</head>
<body>
    <h1>This is the target!</h1>
    <p class="target">
        Yet another <b>target</b>.
    </p>
    <p>
        <!-- Comment -->
        Foo target Bar
    </p>
</body>
</html>
"""
after = replace(
    html = before,
    selector = 'body',
    old = 'target',
    new = '<span class="special">target</span>',
)
print(after)

内容的提问来源于stack exchange，提问作者s-m-e