使用xmllint批量匿名XML文件时多XPath节点匹配失效求助

问题背景 #

使用bash脚本结合xmllint对XML/XHL文件做匿名化时，单节点XPath（如//Prestation/Libelle/@V）替换正常，但匹配多节点的XPath（如//DonneesIndiv/PayeIndivMensuel/Remuneration/Indemnite/Libelle/@V）仅返回第一个节点的值，无法完成所有节点的替换。

原因分析 #

1. string()函数的局限性：xmllint --xpath "string($xpath)" 中，string()函数仅返回XPath匹配到的第一个节点的内容，后续逻辑只会基于这个值判断，忽略其他节点。
2. xmllint shell模式的定位问题：执行cd $xpath时，如果XPath匹配多个节点，只会定位到第一个节点，set操作也仅修改该节点，其余节点不会被处理。

解决方案 #

方案一：改进bash+xmllint脚本，遍历所有匹配节点 #

通过节点计数和位置定位，逐个处理每个匹配节点：

#!/bin/bash

#Dir with xml and xhl files
dir="/home/batch_spl_ia/jeremy"

while true; do
    # ask for xpath
    read -p "Enter the xpath (ex: /root/element/@attribut or //Prestation/Libelle/@V or //IdVer[text()]) : " xpath

    # ask how many character for the anonymous value
    read -p "How many character do you want ? : " nb_caracteres

    # ask what type of character
    read -p "Do you want alpha (a), digital (n) or alphanumeric (an)? (a/n/an) : " choice

    # Variable to check if a value has been replaced
    value_replaced=false

    # read files in directory
    for file in "$dir"/*.xml "$dir"/*.xhl; do
        if [ -e "$file" ]; then
            # 获取匹配节点的数量
            node_count=$(xmllint --xpath "count($xpath)" "$file")
            # 检查是否存在匹配节点
            if (( $(echo "$node_count > 0" | bc -l) )); then
                value_replaced=true
                echo "Found $node_count nodes in $file"
                
                # 生成统一替换值（如果需要每个节点不同值，将此逻辑移到循环内）
                if [ "$choice" == "a" ]; then
                    replacement=$(openssl rand -base64 "$nb_caracteres" | tr -dc 'a-zA-Z' | head -c "$nb_caracteres")
                elif [ "$choice" == "n" ]; then
                    replacement=$(cat /dev/urandom | tr -dc '0-9' | head -c "$nb_caracteres")
                else
                    replacement=$(openssl rand -base64 "$nb_caracteres" | tr -dc 'a-zA-Z0-9' | head -c "$nb_caracteres")
                fi
                
                # 遍历每个节点执行替换
                for ((i=1; i<=node_count; i++)); do
                    xmllint --shell "$file" << EOF > /dev/null 2>&1
cd ($xpath)[position()=$i]
set $replacement
save
EOF
                    echo "Replaced node $i in $file with: $replacement"
                done
            else
                echo "No nodes found for XPath in $file"
            fi
        fi
    done

    # Check if value replaced
    if [ "$value_replaced" = false ]; then
        echo "Error : no value found for the xpath specified : no replacement"
    fi
   
    echo "-----------------------"

    # Ask user if he wants to anonymize another xpath
    read -p "Do you want to anonymize another xpath ?(y/n) : " continuer

    if [ "$continuer" != "y" ]; then
        break
    fi
done

关键改进点：

• 用count($xpath)获取匹配节点总数
• 通过($xpath)[position()=$i]定位到第i个节点，逐个执行替换
• 可选择统一替换值或每个节点生成不同随机值（调整生成逻辑位置即可）

方案二：改用Python+lxml（更稳定的多节点处理） #

如果bash脚本的局限性难以满足需求，推荐使用Python的lxml库，处理多节点XPath更直观可靠：

import os
import random
import string
from lxml import etree

dir_path = "/home/batch_spl_ia/jeremy"

def generate_random_string(length, char_type):
    if char_type == 'a':
        chars = string.ascii_letters
    elif char_type == 'n':
        chars = string.digits
    else:
        chars = string.ascii_letters + string.digits
    return ''.join(random.choice(chars) for _ in range(length))

while True:
    xpath = input("Enter the xpath (ex: /root/element/@attribut or //Prestation/Libelle/@V or //IdVer[text()]) : ")
    nb_caracteres = int(input("How many character do you want ? : "))
    choice = input("Do you want alpha (a), digital (n) or alphanumeric (an)? (a/n/an) : ")
    
    replacement = generate_random_string(nb_caracteres, choice)
    value_replaced = False
    
    for filename in os.listdir(dir_path):
        if filename.endswith(('.xml', '.xhl')):
            file_path = os.path.join(dir_path, filename)
            try:
                tree = etree.parse(file_path)
                nodes = tree.xpath(xpath)
                if nodes:
                    value_replaced = True
                    print(f"Found {len(nodes)} nodes in {file_path}")
                    for node in nodes:
                        # 区分文本节点和属性节点
                        if isinstance(node, etree._ElementStringResult):
                            # 修改文本节点内容
                            node.getparent().text = replacement
                        else:
                            # 修改属性节点值
                            node.value = replacement
                    # 保存修改后的文件
                    tree.write(file_path, encoding='utf-8', xml_declaration=True)
                    print(f"Replaced all nodes in {file_path} with: {replacement}")
                else:
                    print(f"No nodes found for XPath in {file_path}")
            except Exception as e:
                print(f"Error processing {file_path}: {str(e)}")
    
    if not value_replaced:
        print("Error : no value found for the xpath specified : no replacement")
    
    print("-----------------------")
    continuer = input("Do you want to anonymize another xpath ?(y/n) : ")
    if continuer.lower() != 'y':
        break

优势：

• xpath()方法直接返回所有匹配节点的列表，无需手动计数和定位
• 自动区分文本节点和属性节点，处理逻辑更清晰
• 异常处理机制提升脚本稳定性

内容的提问来源于stack exchange，提问作者Jhilumine