使用str_locate时遇正则语法错误及越南语乱码问题求助

核心问题 #

1. 越南语字符因编码问题显示为乱码（如Tên doanh nghi???p:），导致正则模式无效，触发Syntax error in regex pattern错误。
2. 依赖str_locate+str_sub的文本位置提取方式脆弱，易受页面结构变化影响。

解决步骤 #

1. 修复字符编码 #

乱码源于网页读取或脚本文件编码不匹配：

• 读取网页时强制指定UTF-8编码：

  # 修正搜索结果页读取
  search.result = read_html(link, encoding = "UTF-8")
  # 修正企业详情页读取
  com.page = read_html(paste0("https:", com.link), encoding = "UTF-8")
  

• 确保R脚本文件编码为UTF-8（RStudio可通过右下角编码选项设置），避免硬编码的越南语标签出现乱码。

2. 替换为正确的越南语正则模式 #

将代码中乱码的标签替换为准确的越南语短语，示例：

3. 改用更可靠的元素定位提取 #

放弃依赖文本位置的方式，直接通过HTML结构定位字段，稳定性更强：

方法1：XPath兄弟节点定位

# 提取企业名称
com.name = com.page %>% 
  html_nodes(xpath = '//*[contains(text(), "Tên doanh nghiệp:")]/following-sibling::div') %>% 
  html_text() %>% trim()

# 提取税号
com.mst = com.page %>% 
  html_nodes(xpath = '//*[contains(text(), "Mã số thuế:")]/following-sibling::div') %>% 
  html_text() %>% trim()

# 其他字段同理替换对应标签

方法2：批量提取所有信息项

# 获取所有信息项节点
info_items = com.page %>% html_nodes(".company-info-item")

# 批量解析标签和值
info_df = lapply(info_items, function(item) {
  label = item %>% html_node("span") %>% html_text()
  value = item %>% html_node("div") %>% html_text() %>% trim()
  data.frame(label = label, value = value, stringsAsFactors = FALSE)
}) %>% do.call(rbind, .)

# 按需提取字段
com.name = info_df$value[info_df$label == "Tên doanh nghiệp:"]
com.mst = info_df$value[info_df$label == "Mã số thuế:"]

4. 优化数据存储效率 #

避免循环中反复rbind，改用列表存储后统一合并：

# 初始化空列表
info_list = list()
count = 1

for (mst in list.mst) {
  link = paste0(link.source, mst,'/')
  message(link)
  search.result = read_html(link, encoding = "UTF-8") 
  all.com.link = search.result %>% html_nodes(".company-name a") %>% html_attr('href') %>% unique()
  
  for (com.link in all.com.link) {
    com.page = read_html(paste0("https:", com.link), encoding = "UTF-8")
    # ... 提取所有字段 ...
    
    # 将单条企业信息存入列表
    info_list[[count]] = data.frame(
      com.name = com.name,
      com.mst = com.mst,
      com.active = com.active,
      com.add = com.add,
      com.tel = com.tel,
      com.cap.phep = com.cap.phep,
      com.hoat.dong = com.hoat.dong,
      com.nganh = com.nganh,
      stringsAsFactors = FALSE
    )
    count = count + 1
  }
}

# 合并所有数据
info = do.call(rbind, info_list)

内容的提问来源于stack exchange，提问作者Tann